T H E A M A Z O N IA N L A N G U A G E S T h e A m azo n Basin is arguably b o th th e least-know n a n d the m ost com plex linguistic region in th e w orld today. I t is th e h o m e o f som e 300 languages b elonging to a ro u n d 20 language fam ilies, plus m o re th a n a dozen genetic isolates, an d m any o f these languages (often incom ­ pletely d o cu m en ted and m ostly en d an g ered ) show p ro p erties th a t co n ­ stitu te exceptions to received ideas a b o u t linguistic universals. This b o o k is th e first to provide an overview in a single volum e o f this rich an d exciting linguistic area. T h e e d ito rs a n d co n trib u to rs have sought to m ake th e ir descriptions as clear a n d accessible as possible, in o rd er to p rovide a basis for fu rth e r research o n th e stru ctu ral characteristics o f A m az o n ian languages a n d th eir genetic a n d areal relationships, as well as a p o in t o f en try to im p o rta n t cross-linguistic d a ta for the w ider co n stitu en cy o f theoretical linguists. R. M. w . D I X O N and AL E X A N D R A Y. AiKHENVALD are D ire c to ra o d A ssociate D irector o f the Research C entre for Linguistic Typology at the A ustralian N ational U niversity (the R esearch C entre will relocate to La Trobe U niversity in M elbourne from January 2000). Professor D ixon’s b o o k publications include gram m atical studies of the A u stra jp n languages D yirbal a n d Yidiny, o f F ijian and o f English, as well as The Languages o f Australia (1980) and Ergativity (1994). O ne o f his cu rren t projects is a gram m ar o f Jaraw ara (Brazil). P rofessor A ikhenvald h as published 6 b ooks and nearly 100 p apers in Russian, English, Portuguese and Y iddish, covering a range o f subjects including Berber, H ebrew , Indo-E uropean and N ative S outh A m erican languages. H er m o n o g raph Classifiers: A Typology o f Noun Categorisation Devices will be published in 1999. She is currently com pleting a full-length gram m ar o f T atiana. THE AMAZONIAN LANGUAGES e d ite d b y R . M . W. D IX O N and A L E X A N D R A Y. A I K H E N V A L D R esea rch C entre f o r L in g u istic T ypology L a T robe U niversity, M e lb o u rn e M C a m b r id g e UNIVERSITY PRESS CAMBRIDGE UNÍVERSITY PRESS Cambridge, New York, Melbourne, Madrid,, Cape Town, Singapore, Sâo Paulo Cambridge University Press The Edinburgh Building, Cambridge CB2 2RU, UK. Published in the United States o f America by Cambridge University Press, New York www.cambridgc.org Information on this title: www.cambridge.org/9780521570213 © Cambridge University Press 1999 This publication is in copyright. Subject to statutory exception and to the provisions o f relevant collective licensing agreements, no reproduction o f any part may take place without the written perm ission o f Cambridge University Press. First published 1999 Hardback version transferred to digital printing 2006 Digitally printed first paperback version 2006 A catalogue record f o r this publication is available from the British Library Library o f Congress Cataloguing in Publication data Amazonian languages / edited by R. M. W. Dixon and Alexandra Y. Aikhenvald. p. cm. - (Cambridge language surveys) Includes bibliographical references. ISBN 0 521 57021 2 (hardback) I , Indians o f South America - Amazon River Region - Languages. I. Dixon, Robert M. W. II, Aikhenvald, A. Y. (A leksaidra lUrevna) III. Series. PM5099.ii.A48 1999 498 -d c2 1 98-46270 CIP ISBN-13 978-0-521-57021-3 hardback ISBN-10 0-521-57021-2 hardback ISB N -13 978-0-521-57893-6 paperback ISBN-10 0-521-57893-0 paperback CONTENTS List o f maps List o f contributors Acknowledgements List o f abbreviations Conventions followed p o g e xvii 1 Introduction by R. M . W. D ixon and A lexandra Y. Aikhenvald 1 C u ltural background 2 Linguistic diffusion 3 G enetic relationship 4 T h e punctuated equilibrium model 5 O r|àn iz atio n o f this book Bibliography 2 Carib by D esm ond C.Derbyshire 1 In troduction 2 Phonology 2 . 1 Segmental 2.2 Phonotactics and suprasegm entals 2.3 M orphophonology 3 M orphology, particles and pronouns 3.1 Inflectional m orphology 3.1.1 Person-m arking afflxes on verbs, nouns, adverbials and postpositions 3.1.2 Tense, aspect, m ode and tiumber suffixes on verbs 3.1.3 Possession, tense and num ber suffixes on nouns 3.1.4 Inflectional suffixes on locative postpositions 3.2 Derivational m orphology 3.2.1 Verb derivational affixes 3.2.2 N om inalizing affixes attached to verb stems to form nouns xviii xx xxi xxiv I 3 7 11 16 19 20 23 23 26 26 26 28 31 31 32 37 40 42 43 44 45 Contents 3,2.3 Some adverbial derivations from noun and verb stems 3.3 T he particle word class 3.4 Pronouns 4 Syntax 4.1 M ain clause structuring 4.2 Subordinate clause constructions 4.2.1 Nominalizations 4.2.2 Adverbializations 4.2.3 Postpositions 4.3 Ergativity Bibliography 3 The Arawak language family by A lexandra Y. Aikhenvald 1 2 3 4 5 6 7 Introduction; the Arawak languages and their speakers 1.1 Com parative studies, genetic classification and subgrouping Phonology 2.1 Segmental phonology 2.2 Syllable structure, suprasegmentals and phonotactics 2.3 M orphophonology Typological profile 3.1 W ord structure 3.2 M orphological processes 3.3 W ord classes N om inal morphology 4.1 Possession 4.2 Classifiers and genders 4.3 N um ber Closed classes: personal pronouns, demonstratives, interrogatives, numbers, adpositions Verbal m orphology 6.1 Classification o f verbs and predicate structure 6.2 Cross-referencing and split ergativity 6.3 Valency-changing derivations 6.4 O ther verbal categories 6.5 N oun incorporation 6.6 Relative and negative markers Syntax 7.1 G ram m atical relations 7.2 N oun phrase structure 7.3 Complex predicates and serial verbs 50 53 53 54 55 56 56 57 58 60 61 65 65 73 75 75 78 79 80 80 80 81 82 82 83 84 85 85 86 87 90 93 95 95 96 96 96 97 Contents ix 7.4 7.5 7.6 98 98 99 C onstituent order N on-verbal clauses Complex sentences: relative clauses,com plem ent clauses, coordination, pivot and switch-reference 7.7 G ram m atical means used for m arking discourse functions Bibliography 101 102 4 Tupi by Aryon D. Rodrigues 1 Introduction 2 Phonology 2.1 Vowels 2.2 C onsonants 2.3 Tone 3 G ram m atical overview 4 N ouns 5 Pronouns 6 Verbs 7 Demonstratives 8 Questions 9 Subordinate clauses 10 Pivots B iblio|laphy 107 107 110 110 112 114 114 ' 115 117 118 120 120 121 121 122 5 Tupl-Guarani by Cheryl Jensen 125 1 Introduction 125 2 Identification o f Tupi-G uarani languages 128 2.1 D istinguishing characteristics o f T u p í'ö u a ra n í 128 languages 2.2 Subgroups within Tupi-G uarani 129 3 Froto-Tupi-G uarani phonology 133 3.1 Proto-Tupi-G uaranf phonem es 133 3.2 Proto-Tupi-G uarani allom orphs 134 3.3 Possible Tupi-G uarani m orphophonem ic phenom ena 135 3.3.1 Replacement by nasal consonants 135 3.3.2 Strategies for the loss o f consonant clusters 136 formed at m orphem e juncture 4 Phonological changes within T upi-G uarani 137 4.1 Weakening o f * //a n d 137 4.2 Palatalization 138 4.3 Labialization 140 4.4 M erger o f with 141 Contents Final-consonant phenom ena 4.5.1 Devoicing 4.5.2 N asalization 4.5.3 Loss 4.6 Vowel shift 5 Stems 5.1 Categories o f stems 5.2 Stem classes 6 Person markers 7 N ouns 7.1 Case marking 7.2 Possession 7.3 N oun com position 7.4 Indication o f num ber 7.5 N oun phrases 7.6 Possession classes 8 Postpositions 9 Verbs 9.1 Independent verbs 9.2 Oblique-topicalized verbs 9.3 Serial verb constructions 9.4 Tem poral subordinate clause constructions 9.5 Valency-changing devices 9.5.1 Causatives 9.5.2 Detransitivizers 9.5.3 Object incorporation 10 N om inalizations 11 Semantics Bibliography 142 142 143 143 144 146 146 146 146 148 148 150 151 151 152 152 153 154 155 156 157 157 158 158 159 159 159 160 161 M acro-Jê by Aryon D. Rodrigues 1 H istorical survey 2 D istribution 3 Linguistic scholarship 4 Phonology 4.1 Vocalic systems 4.2 C onsonantal systems 4.3 Tone 5 M orphology 5.1 Inflection for contiguity o f a determ iner 5.2 Inflection for possession 5.3 N um ber 165 165 166 169 171 171 174 180 180 180 182 183 4.5 xi Contents 5.4 N o u n classification 5.5 A greem ent m arking on the verb 6 Syntax 6.1 C onstituent order in declarative sentenœ s 6.2 A dpositional phrases 6.3 G enitive phrases 6.4 D em onstrative phrases 6.5 N um eral phrases 6.6 Adjectival phrases 6.7 Ergativity 6.8 Valency-changing processes 6.8.1 Reflexives and reciprocals 6.8.2 Causativizalion 6.9 Switch-reference 7 A bout the consistency o f M acro-Jê as a genetic group 7.1 A b rie f appraisal o f the gram m atical affinity 7.2 Phonological equations Bibliography 7 Tucano by Janet Barnes 184 185 187 187 188 190 192 192 193 193 195 195 195 197 198 198 198 201 207 1 Introduction 207 2 Phonology 2.1 Syllable structure 2.2 Segmental phonology 2.3 N asalization 2.4 A ccent 3 M orphology 3.1 T he verb 3.1.1 Evidentials 3.1.2 A spect 3.1.3 M ood and m odality 209 210 210 211 212 212 213 213 214 216 3.2 P ro n o u n s 3.2.1 Persona] pronouns 3.2.2 Possessive pronouns 3.3 Classifiers 3.4 Specificity m arker 3.5 N ouns 3.6 Adjectives 3.7 N egation 3.8 N om inalized verbs 3.9 Switch-reference 4 Syntax 217 217 218 218 219 221 221 222 222 223 224 Contents 4.1 Tim e and location 4.2 N oun phrase Bibliography 224 225 225 Pano by Eugene E. Loos 1 Introduction 2 Phonology 2.1 Segmental phonology 2.2 Com m on phonological variations 3 G ram m ar 3.1 W ord classes and noun phrase form ation 3.1.1 N oun phrase 3.1.2 Pronoun classes 3.L 3 Relative clauses 3.1.4 N oun phrase pluralization 3.2 Transitivity concord 3.2.1 Switch-reference 3.2.2 Adverbial suffix concord 3.2.3 Locative phrases 3.2.4 Ergative m arking 3.3 N oun incorporation 3.4 Verb formation 3.4.1 Verb roots 3.4.2 Suffixes 3.5 M ood indicators 3.6 Deictics Bibliography 227 227 230 230 231 234 234 235 235 236 236 236 237 239 239 240 243 243 243 244 248 248 249 IVIaká by Silvana and Valteir M artins 1 Introduction 2 G enetic classification 3 Phonology 4 W ord structure 5 W ord classes 6 N om inal categories: possession, classifiers, num ber 7 Verb structure 8 Valency-changing derivations 9 Incorporation 10 G ram m atical relations 11 N egation 12 Syntax 13 Lexicon Bibliography 251 251 253 255 257 257 258 259 260 261 263 264 265 265 266 Contents xiii Nambiquara by Ivan Lowe 269 269 271 272 272 272 272 273 274 274 277 279 280 283 283 284 284 284 284 287 288 289 290 291 1 2 Introduction Phonology 2.1 The variants o f phonem es 2.2 Syllable structure 2.3 Stress 2.4 M orphophonem ics 3 M orphology 3.1 Verbs 3.1.1 M ain verbs 3.1.2 Subordinate verbs 3.2 Adjectives 3.3 N ouns 3.4 Pronouns 3.5 Adverbs 3.6 Interjections 3.7 Ideophones 4 Syntax 4.1 M ain clauses 4.2 Subordinate clauses 4.3 Clause coordination 4.4 N oun phrases 4.5*- N om inalizations Bibliography 1 Introduction 2 Phonology 3 Word classes 4 N ouns 5 N oun phrase structure 6 Verbs 7 Predicate structure 8 Pronouns 9 Demonstratives and interrogatives 10 C onstruction types Bibliography 293 293 295 298 298 299 300 300 302 304 304 306 Small language families and isolates in Peru by M ary Ruth Wise 1 Introduction; the languages and families 2 Phonology 3 M orphology 3.1 N om inal m orphology 307 307 312 318 319 Arawá by R. M. W. Dixon 12 Contents 3.1.1 Classifiers 3.1.2 Case 3.2 Pronouns 3.3 Adjectives 3.3.1 Com parative and superlative grades 3.4 Verbal m orphology 4 Syntax 4.1 C onstituent order 4.2 Relativization 319 320 321 323 323 324 329 329 332 4.3 Subordination and coreference/switch-reference Bibliography 333 335 Other small families and isolates by Alexandra Y. A ikhenvald and R . M. W. Dixon 1 Brazil 1.1 Y anomami 1.1.1 Phonology 1.1.2 Word structure and word classes 1.1.3 N ominal categories 1 .1.4 Classifiers 1.1.5 G ram m atical relations 1.1.6 Verb structure 1.1.7 N oun incorporation and verb com pounding 1.1.8 Syntax 1.2 Truraai by Raquel G uirardello 1.2.1 Phonology 1.2.2 Features o f the gram m ar 1.3 M ura-P iraha 1.3.1 Phonology 1.3.2 Features of the gram m ar 1.4 Jabuti 1.4.1 Phonology 1.4.2 Features o f the gram m ar 1.5 C hapacura family 1.5.1 Phonology 1.5.2 Features o f the gram m ar 1.6 M áku 1.7 A ikaná and K oaia 2 Bolivia 2.1 T he Tacana family 2.1.1 Phonology 2.1.2 Features o f the gram m ar 2.2 Isolates 341 341 341 345 346 346 347 348 349 350 350 351 352 352 353 354 355 357 357 357 358 359 359 361 362 364 364 365 366 367 Contents 14 xv 3 C olom bia 3.1 Phonology 3.2 M orphology 3.2.1 Word structure and typological profile 3.2.2 N om inal m orphology 3.2.3 G ram m atical relations 3.2.4 Verbal m orphology 3.3 Syntax 4 Venezuela 4.1 P honology 4.2 Features o f the gram m ar Bibliography 369 370 372 372 373 375 375 376 377 377 378 379 Areal diffusion andlanguage contact in the[çana-Vaupés basin, north-west Amazonia by A lexandra Y. Aikhenvald 385 1 2 385 386 386 388 390 39 i 391 394 396 396 397 403 404 405 406 406 Areal diffusion in north-w est A m azonia Linguistic situation in the Içana-Vaupés basin 2.1 Languages spoken 2.2 C ultural setting and language attitudes 2.3 H istorical evidence 3 The Vaupés region as a linguistic area 3.1 G eneral observations 3.2-” Phonological characteristics 3.3 G ram m atical structure 3.3.1 Typological profile and w ord structure 3.3.2 N om inal m orphology 3.3.3 G ram m atical relations 3.3.4 Verbal m orphology and predicate structure 3.4 Syntax and discourse techniques 3.5 Semantics 4 Properties shared by languages o f the Içana and Vaupés region 4.1 Pitch accent 4.2 Topic-advancing verbal derivation 4.3 Possessive classifiers and -ya- possessive m arker 4.4 C om plex systems o f classifiers 5 Conclusions Bibliography 406 407 409 410 411 413 15 The Upper Xinguas an incipientlinguistic area by Lucy Seki 417 1 The U pper X ingu and its languages 1.1 Languages spoken 1.2 H istorical background 417 417 419 Contents 1.3 The mobility o f groups, language loss and further contacts 2 The U pper X ingu as a culture area 3 Linguistic situation 4 Incipient areal diffusion in the U pper Xingu 5 Conclusions Bibliography Index o f authors Index o f languages and language fam ilies Subject index 423 -423 424 426 428 428 431 436 445 MAPS M ap 1 C arib languages w ith approximate locations M ap 2 A raw ak languages with approximate locations page 22 M ap 3 Tupi languages w ith approxim ate locations M ap 4 Tupi-G uarani la n g iia p s with approxim ate locations 126 M ap 5 M acro -Jêlanguages w ithapproxim atelocations 164 208 66 108 M ap 6 Tucano languages w ith approximate locations M ap 7 P ano languages w ith approxim ate locations 228 M ap 8 M aku languages and dialects with approxim ate locations 252 N am biquara languages and dialects w ith approxim ate locations 268 M ap 10 M ap 9 Arawá languages and dialects with approxim ate locations 292 M ap 11 Small la i^u a g e families and isolates o f Peru w ith approxim ate 308 locations M ap 12 Small language families and isolates o f Brazil, Bolivia, C olom bia 342 an d Venezuela with approxim ate locations M ap 13 Languages in the Içana-Vaupés Basin with approxim ate locations 384 M ap 14 Languages o f the U pper Xingu with approxim ate locations 418 CONTRIBUTORS A lexandra Y. Aikhenvald Cheryl Jensen Research Centre for Linguistic Summer Institute o f Linguistics Typology Ag. Cabanagem C P 5040 La Trobe University 66601-970 Belem, Pará, Brazil B undoora, Vic, 3083, A ustralia Al-Cheryl_Jensen@ sil.org a.aikhenvald@ latrobe.edu.au Eugene E. Loos Janet Barnes 7407 Cave Drive ILV - A partado Aéreo 120308 Dallas, TX 75249, USA Bogotá, C olom bia eloos@ cyberramp.net [email protected] Ivan Lowe D esm ond C. Derbyshire Sum m er Institute o f Linguistics 56 Furze Hill R oad Horsleys G reen, High Wycombe Headley Down Bucks., HP14 3XL, England Surdon, G U35 8H A, England [email protected] des_derbyshire@ silorg Silvana and Valteir M artins R. M. W. Dixon Rua G uaianazes 357 Research Centre for Linguistic Bairro Paraiso Typology 16001-970 A ragatuba, S.P. Brazil La Trobe University Bundoora, Vic, 3083, Australia A ryon D. Rodrigues S Q N 4 1 0 K 102 Raquel G uirardello 70865-110 Brasilia, D F, Brazil Rua Ferrucio Celani No. 152 Centro - Valinhos - S.R 13270-000 Brazil L ist o f contributors Lucy Seki M ary Ruth Wise U niversidade Estadual de C am pinas Summer Institute o f Linguistics Instituto de Estudos da Linguagem 7500 W. C am p Wisdom Rd, D epartam ento de Lingíiística, D allas, T X 75236-5699, U SA Cx. P. 6045 13081-970 Cam pinas, S.P. Brazil [email protected] M aryRuth_W ise@ sil.org ACKNOW LEDGEMENTS T he editors w ould like to thank the linguistic scholars o f S outh A m erica (in uni­ versities, museums and missionary organizations) who have discussed with them the wonders of the S outh American languages, and freely shared their m aterials and ideas. And the speakers o f A mazonian languages who have extended friendship; explained what to say to who in what circumstances and why; an d helpeä unravel the intricate structures o f their native tongues. This volume could n o t have come into being w ithout the critical assistance of two members o f the Research C entre for Linguistic Typology. Jennifer Elliott, the C entre’s A dm inistrator, solved com putational conundrum s an d restored phonetic symbols that had strayed. Suzanne Kite, o u r Research A ssistant, spent hundreds o f hours checking the chapters for consistency, style and sense; e-m ailing contributors to clarify what they m eant and the most appropriate way to express it; and inte­ grating what began as a num ber o f heterogeneous parts into a reasonably unified whole. ABBREVIATIONS The following abbreviations are used for the states o f Brazil: AC Acre PE Pernam buco AM A m azonas PI Piauf AP A m apá PR Paraná BA B ahia RJ R io de Janeiro R ondônia ES E spirito Santo RO GO G oiás RS R io G rande do Sul MA M aranhào SC S anta C atarina MG M inas G erais SB Sergipe MS SP SÈO Paulo MT M ato G rosso,do Sul M ato Grosso'" TO Tocantins PA P ará T he following abbreviations are used in the bibliographies: UAL International Journal o f American Linguistics SIL Sum m er Institute o f Linguistics A bbreviations used in interlinear glosses are: 1 1+2 first person ADJ ADV adjective first person inclusive 1+3 first person exclusive ADV Z adverbializer IR relating to first person 2 second person AG A G .N M LZR agentive agent nom inalizer adverbial 3 third person A LIEN alienable A transitive subject allative ABS absolutive A LL A N IM AC action A PPLIC applicative ASSO C associated ACC accusative animate xxii L ist o f abbreviations ATTRIB attributive case IM M im mediate AUX auxiliary verb IM PE R im perative BEN benefactive IM PE R FV imperfective C consonant IM PERS im personal inalienable CAUS causative IN A L C IR .N M L Z R circum stantial IN A N inanim ate IN CL, incl inclusive nom inalizer CL classifier IN D E F indefinite CNT contiguous marker IN D IC indicative IN F infinitive/pseudo­ INST instrum ent C O LL collective C O M IT com itative C O M PL completive aspect infinitive COND conditional IN T E R R O G interrogative CONJ conjunction IN TR intransitive CONT continuative aspect IR R irrealis COREF coreferential LIN K linking o r relational CURR current D EC L declarative mood LOC morphem e locative DEF definite MAL malefactive D EIC deictic DEM deinonstrative M ASC, masc M ED IA L medial masculine DENM LZR denom inalizer N noun D ESID desiderative NCNT non-contiguoiis DET determ inate DETRZR detransitivizer NEG negative/negation determ iner D IF.LOC diffuse locative case NEUT neuter D IM n.fem non-fem inine D IST diminutive distant NM LZR nom inalizer D L , dl dual nominative OS different A/S referent NOM NOM .CA SE EM PH em phatic NP noun phrase ERG ergative 0 transitive (direct) EVID evidential EXCL, excl exclusive OBL oblique EXIST existential O BL.TO P oblique-topical ized FEM , fern feminine FO C focus PART. LOC partitive locative case FUT future PASSV passive G EN genitive PAST.l first past tense nom inal case object verb L ist o f abbreviations PAST.2 second past tense REM rem ote P ER F perfect REP reportative PERFV perfective REPET repetitive PL, pi POSSD plural RES resultative possession/possessed REV reversative POSSR POSSV possessor S intransitive subject possessive SEQ sequential PRES present SER.V serial verb PRO pronoun SF stem formative PROG PROX progressive aspect S G .sg singular proxim ate SO A/S referent sam e as O PUNCT punctual aspect PUN CT.LOC punctual locative case SPEC specificity m arker R EA L realis ss same A/S referent REC recent TA M N tense, aspect, m ood and R EC IP reciprocal REFL reflexive TR transitive REL relative clause m arker V verb/vowel RELZR relativizer V BLZR verbalizer referent num ber CONVENTIONS FOLLOWED 1 SPELLING We have tried to avoid spelling conventions that are particular to one language, using instead those which are m ost generally accepted. (a) Spanish ‘h u ’ and ‘j ’. The sounds [w] and [h] are shown by ‘hu’ and-'j’ respec­ tively in Spanish orthography. We have generally preferred plain ‘w’ and ‘h ’, e.g. W itoto rather than H uitoto, Cashinawa rather than C ashinahua, G uahibo rather than G uajibo. (We have allowed just a few exceptions, e.g. Jivaro [Hivaro] is always spelt with ‘J ’ and this has been retained. C ahuapana and Chayahuita are almost always spelt w ith ‘h u ’ and this is retained here.) The sound [k] is variously written as ‘k ’ or ‘c’ o r ‘q u ’. For nam es that include a [k] we have, as a rule, retained the spelling which is best knovra. (b) English ‘-an’. English-speaking linguists often put *-an’ on the end of the name o f a language family, e.g. Arawakan, C ariban, Tupian. (A lthough this is oot done consistently. MereifuUy we have never seen Jêan or M akùan.) Spanish- and Portuguese-speaking linguists avoid nam es ending with an English suffix. We think it m ost useful to have one universal nam e for each family (rather than a certain nam e when w riting in English and another nam e when w riting in Spanish or Portuguese), and so avoid the ‘-an’. This does mean that sometimes a language family and one language within the family may be referred to by the same label, but context is almost always sufficient to avoid confusion. There are additional reasons for o u r following this convention. The term Ä raw ak' is used by Spanish- and Portuguese-speaking linguists for a well-established language family. A num ber o f English-speaking linguists use A raw akan’ for som ething quite different, a chimerical high-level grouping (see §3 o f chapter 1) which is said to include what South American linguists term 'Arawak* but Is in this classification nam ed ‘Maipuran*. We follow the traditional (and current South American) practice o f using ‘A raw ak’ (not ‘M aipuran’) for the established language family, and avoid the term ‘A raw akan’. Conventions follow ed 2 xxv LANGUAGE AND DIALECT, FAMILY AN D SUBGROUP (a) T h e term lan g u ag e’ can be used with two quite different meanings. One is the linguistic sense, when forms o f speech that are m utually unintelligible are desig­ nated to be distinct languages. Two forms o f speech that are m utually intelligible are then dialects o f one language. The other is the political sense, when each separ­ ate n ation o r tribe likes to say that it has its own language. The two senses o f ‘lan­ guage’ often give the same result b u t sometimes they do not. O ne linguistic language may relate to several political languages, as when the Swedish and Norwegian peoples each say that they have a distinct language, whereas these are m utually intel­ ligible an d are thus, on linguistic grounds, dialects o f a single language. O r one political language may relate to several linguistic l a n g u a ^ as when the Chinese say that they all speak the Chinese language, but in a num ber o f dialects. In fact the ‘dialects’ are n o t m utually intelligible and are, on linguistic criteria, separate lan­ guages. In this book we employ the term ‘language’ in the linguistic sense. N ote that earlier work on S outh Am erica has used the term la n g u ag e’ sometimes in the lin­ guistic an d sometimes in the political sense. For, instance, there are seven M akil tribes, each o f which has its own political language, but there are ju st four separate languages in the linguistic sense. Yanomami has sometimes been described as involving four closely related languages, but it seems m ost appropriate to character­ ize Y anom am i as a dialect continuum (effectively, as a single language). W ithin the A raw á family, Jam am adi, Jaraw ara and Banawá are distinct tribes and have been said to involve distinct languages (indeed, there are currently three m issionary teams, each working on its own Bible translation). In fact these speech form s have about 95 per cent vocabulary in com m on and very sim ilar gram m ara Each is easily intelligible to speakers o f the others and they are clearly dialects o f one language, on linguistic criteria. W ithin the Tupi-O uarani subgroup there is some confusion about what is a language and what is a dialect; the com m entary on table 5.1 attem pts to provide some clarification. (b) We have tried to ensure th a t the s tandard criteria for proving genetic relation­ ship between languages are followed in this book. T he established term for a group o f languages th at is genetically related is ‘language fam ily’. Smaller genetic groups w ithin a family are called ‘branches’ or ‘subgroups’ o f the family. For instance, one talks o f the G erm anic branch, or subgroup, within the Indo-European family. Terms such as ‘stock’ and ‘phylum ’ (together with ‘m icro-phylum ’, ‘m acro-phylum ’, ‘m eso-phylum’) go with the quite different m ethodology o f lexicostatistics an d are best avoided. We thus talk o f the T upi-G uaranl branch, or subgroup, o f the Tupi language family, and so on. {The Portuguese term tronco is sometimes translated xxvi Conventions followed into English as stock, but in many instances fa m ily would be a m ore appropriate ren­ dering.) 3 GRAMMATICAL TERM INOLOGY We intend the linguistic surveys contained in this volume to have lasting value. For this reason, am ong others, we have avoided use o f any o f the current (or past) formal linguistic theories. The contributors have written in term s o f what has recently come to be called Basic Linguistic Theory. This is the cum ulated tradition o f linguistic description, that has evolved over the last 2,000 years. M ost gram mars (certainly, all those that have perm anent value) are written in term s o f the param e­ ters o f Basic Linguistic Theory - what is a phoneme; what is a word; the param e­ ters in terms o f which systems of tense, aspect and evidentiality vary; the criteria for recognizing a relative clause; and so on. (a) Contributors have used standard abbreviations for core semantico-syntactic relations: A transitive subject O transitive object S intransitive subject S that is m arked in the same way as A on a transitive verb S that is m arked in the same way as O on a transitive verb The terms ‘active’ a nd ‘stative’ are typically used for verbs that take and argu­ ments respectively. A language will thus be characterized as, for instance, AVO, SV (rather than just as SVO). There are in fact Am azonian languages where A and S do not occur in the same position, e.g. in Kuikùro, o f the C arib family, the basic constituent orders are SV and OVA. (Here a single formula - using ‘S’ to cover both intransitive and tra n ­ sitive subject - could n o t satisfactorily be employed.) (b) Verbs typically fall into a number o f subclasses: Intransitive - can only occur in an intransitive clause, w ith S core argu­ ment. Transitive - can only occur in a transitive clause, with A and O core arguments. Ambltransitive - can occur in either an intransitive o r a transitive clause; there a re two varieties o f ambitransitive: S = A type, e.g. ‘he (S) has eaten’, ‘he (A) has eaten lunch (O )’ S = O type, e.g. ‘the glass (S) broke’, ‘he (A) broke the glass (O)’ Conventions follow ed xxvii M ost languages have a subtype o f transitive: Extended transitive (or ditransitive) - take A and O argum ents and also an oblique argum ent. N ote that, with an extended transitive verb like ‘give’, in some languages the G ift is in O function with the Recipient in oblique function, in some languages the Recipient is O and the G ift is oblique, while other languages have b o th construc­ tion types available (e.g. in English, John gave the book to M ary and John gave M ary the book). A few languages (e.g. Trum ai, in §1.2 o f chapter 13) also have a subtype o f intran­ sitive: Extended intransitive - take an S argum ent and also an oblique argu­ ment. (c) G ram m atical terms tend to be used in different ways by different authors. It may be useful to provide a characterization o f the way some im portant term s are employed in this volume. A t the clausal level, a language may m ark the syntactic functions o f its core argu­ m ents (A, S, O) either by dependent-marking, which involves function m arkers (case, adpositions, etc.) attached to N Ps which realize the core argum ents; o r by headmarking, which invcMves bound pronom inal elements (relating to core argum ents) attached to the predicate. (Some languages com bine the two strategies.) Passive is a valency-reducing derivation. Prototypically it applies to a transitive clause, taking the A argum ent o ut o f the core and placing it on the periphery, with the old O becoming new S. Antipassive is similar to passive but here O is moved into the periphery with old A becoming new S. (We know o f only one A m azonian lan­ guage which appears to have an antipassive derivation - Cavineña, in §2.1.2 o f chapter 13.) Causative is a valency-increasing derivation. Prototypically it applies to an intransitive clause, bringing in a new argum ent (the C auser) as A, w ith the original S becom ing O. In some languages causative applies only to intransitives but in others it may also apply to transitives. Applicative is another kind o f valency-increasing derivation whose prototypical application is to intransitives. The original S becomes A with w hat was a peripheral argum ent being moved into the core as O. There can be a variety o f applicatives depending on the original peripheral function o f the new O; these include instru­ m ental, comitative, benefactive, dative, locative (see, for example, chapter 3). There is a fuller account (with exemplification) o f passive, antipassive, causative and applicative in A typology o f argum ent-determ ined constructions’, by R. M. W. xxviii Conventions followed Dixon and A. Y Aikhenvald, pp. 71-113 o f Essays on language function and lan­ guage type, dedicated to T. Givôn, edited by J. Bybee, J. H aim an and S. A. Thom pson (published in 1997 by John Benjamins, Amsterdam). A serial verb construction involves a single predicate consisting o f several verbs wihich share certain properties - generally, the same subject (S o r A ) and often other argum ents as well. T he verbs usually share a single specification for tense, aspect, modality, m ood an d polarity; they constitute one prosodic phrase. In some languages certain types of clause com binations (e.g. m ain clause plus rel­ ative clause, main clause plus purposive clause) are obligatorily m arked for whether they have the Same Subject (SS) or Diiferent Subjects (DS). This is referred to as switch-reference m arking. M any A m azonian languages have a set o f classifiers which characterize a given noun in terms o f its inherent properties, typically animacy, shape, form and m aterial. The classifiers may be attached to numerals, deictics, locatives or verbs. O ther languages may have a small closed system o f genders (or noun classes) which typically include m asculine and feminine terms. Each noun m ust belong to one gender class, A num ber o f A m azonian languages show b o th classifier sets and gender systems (e.g. Tucano, in chapter 7). Introduction R . M . W. D I X O N A N D A L E X A N D R A Y. A I K H E N V A L D The A m azon basin is tiie least known and least understood linguistic region in the world. M aps of the language families o f S outh Am erica (with one colour for each genetic group) purvey an im pression o f anarchy - there are dabs o f yellow a nd blue and red and orange and brow n mingled together like a painting by Jackson Pollock. And when one does get hold o f a gram m ar o f an A m azonian language it is likely to show strange properties - multiple sets o f classifiers, oddly conditioned ergativity splits, an d so on - that constitute exceptions to received ideas about typological universals. In other instances one finds the richest examples o f categories th a t are weakly attested elsewhere. For instance, Tucano languages (chapter 7) have the most highly articulated systems o f evidentiality in the world; this is an obligatory specification o f th ^, evidence a speaker has for m aking a statem ent - whether observed, o r reported, or inferred, or assumed. However, a major difficulty is that a high p roportion o f available gram m ars are incomplete, affording a glimpse of some exotic gram m atical property but with insufficient inform ation to enable the reader to fully understand it, and to realize its overall typological significance. In fact, the m ajor language families all have markedly discontinuous distribu­ tions (m ore than is found in any other part o f the world). The Tupi and Arawak families each com prise at least ten separate geographical regions and the C arib family at least five. Even Tucano, one o f the m ost linguistically hom ogeneous of families, is spoken over three distinct areas. In addition, there has been a great deal of linguistic diffusion, over large and small regions within the A mazon River Basin. This can m ake it hard - sometimes impossible - to determ ine w hether points o f similarity between two languages in close proximity are evidence o f shared genetic developm ent, or simply o f m utual borrowing. One o f the editors has devoted several decades to searching for substantive lin­ guistic universals. In case after case, ju st as he thought he h ad achieved some sig­ nificant typological statem ent, a counter-exam ple popped up; and this was invariably from a language o f A m azonia. He decided that the m ost sensible course o f action was to learn Spanish and Portuguese and then go to South A m erica - visit 2 K M. W. Dixon and Alexandra Y. Aikhenvatd universities an d museums, read the published m aterials (much o f it in obscure jo u r­ nals), consult unpublished theses, and talk to linguists who have worked on this or that language (many o f whom will never get around to publishing a full gram m ar). In this way he achieved a degree o f insight into the m ost com plex linguistic area in the world today. This book is p u t forward in a sim ilar spirit. It attem pts to describe - in as clear and accessible a m anner as possible - the current linguistic situation in the Amazon Basin, involving at least 300 languages, belonging to 20 o r so language families, plus more than a dozen genetic isolates. T he volume should be regarded as a first and preliminary step towards elucidating the structural characteristics o f A m azonian languages, and their genetic and areal relationships. There are several unfortunate aspects to the linguistic situation in A m azonia today. Only a m inority o f scholars in linguistics departm ents in S outh American nations work on the indigenous languages, with the rem ainder preferring to confine themselves to looking at Spanish or Portuguese in terms o f the latest type o f formal theory to filter dow n from the north. (Seki’s Gramálica da lingua Kamaiurá, cur­ rently in press w ith E ditera da U nicamp, is the first full gram m ar o f a Brazilian lan­ guage, written by a Brazilian, since A nchieta’s Arie de grainmatica o f 1595.) As a result, a high p roportion o f the descriptive materials published on South American languages over the past few decades have been by missionary linguists o f the Summer Institute o f Linguistics and sim ilar organizations. In other parts o f the world (for instance, A ustralia) there is m utual respect and cooperation between missionary linguists and scholars from the local universities. In contrast, in m ost (although not all) S outh A m erican countries there is antipathy - sometimes even open hostility - between the two groups. M alicious stories are spread that the missionaries are agents o f the CIA o r are engaged in gold explora­ tion or drug trafficking. N one o f this is true. But many o f the missionaries are evangelical Protestants (mostly from N o rth A merica and Europe) operating in pre­ dom inantly C atholic countries. They do have better aeroplanes and other support facilities (provided by their backers in the USA and elsewhere) than are available to m ost local linguists. A nd they are - in m any cases - m ore prolific in writing gram ­ m ars and com piling dictionaries, and publishing them. There is good and bad in every group. Some o f the m issionaries do - as is alleged - attem pt to destroy the native culture and religion an d replace it with their own b rand o f fire-and-brim stone Christianity. (These people should be banned; they also tend to be those who do the poorest linguistic work.) But many o f the missionaries do m uch more good than harm . They may help protect the lands o f a native tribe from invasion by gold miners and the like. They often provide medi­ cines. They will help a people adapt to the outside worid that is gradually intruding I Introduction 3 into their lives. A nd they can also (over and above the business o f Bible translation) provide gram m ars, primers, vocabularies, volumes o f traditional texts; a nd assist in literacy work. In m any places the lack o f cooperation between the two groups is marked. One may find both a group o f missionaries and a group o f academic linguists (o f sim ilar quality) working on the sam e set o f languages, b u t with neither referring to the work of the o th er in their publications. They sometimes won’t attend the same confer­ ence; they decline to com m unicate and cooperate in a way that would be beneficial to ail. A separate point is th a t the standard o f scholarship in South Am erican linguis­ tics is n o t high. M uch o f the am ateur data from before about 1950 has only a limited usefulness, with the transcription often being poor. M any o f the m issionaries have had inadequate training an d produce ‘cookbook’ descriptions (in the 1950s and 1960s these were often cast w ithin the im penetrable formalism o f tagmemics) that cannot d o justice to the genius o f a language. Linguists from universities may employ oth er kinds o f form alism s, th a t will soon pass out o f fashion. H aving made these general observations, we m ust add that there are notable exceptions on both sides - a num ber o f descriptive studies that achieve a high standard o f clarity and explanation. T he H andbook o f Am azonian Languages, edited by Desmond C. D erbyshire and Geoffrey K. PulluiBÍand published by M outon de G ruyter, is a m ost worthwhile enterprise th at has so far run to four volumes that include ten gram m ars (ranging in quality from quite good to very good) together with a number o f typological and historical studies. Yet the H andbook would be m ore useful if its contributors were no t forced to follow an idiosyncratic scheme o f organization: syntax, then phonol­ ogy, then m orphology T h a t this is basically unw orkable is dem onstrated by the fact that five o f the eight gram m ars in the first three volumes have, as the whole o f ‘23, M orphology’, a single sentence along the lines ‘This has been treated in earlier sec­ tions.’ It is o f course necessary to know the basic inflectional m orphology in order to understand the syntax, so this inform ation is slipped in early on in the descrip­ tion (b u t at different places in each gram m ar). T he net result may be th a t nowhere is there any integrated m orphological statem ent, e.g. o f the structure o f the verb. 1 CULTURAL BACKGROUND The A m azon is the w orld’s m ajor river. Its drainage area is over 7 million square kilometres (more than twice th a t o f the Congo). T he volume o f w ater carried is a fifth o f all the flowing w ater in the world (more than four times that o f the Congo). Its 7 m ajo r tributaries are each m ore than 1,600 kilometres in length (com pare this 4 R. M. W. D ixon and Alexandra Y. A ikhenm ld with the longest river in Europe, th e Rhine, which is 1,320 kilometres long). There are distinct wet an d dry seasons, at different times o f year in different parts o f the A mazon Basin, so that the water level may vary by up to 20 metres. Several hundred kilometres up a m ajor tributary, such as the Purus, the river may be about 1 kilo­ m etre wide in the dry season but flood to a width o f 15 kilom etres in the w t season, inundating the adjoining forest lands. N o rth o f the Amazon is the Orinoco, flowing into the Caribbean sea, w ith a drainage area o f alm ost a million square kilometres. In this book we attem pt to cover languages spoken in the A m azon and Orinoco Basins - that is, from the north coast o f S outh A merica, east to the m outh o f the A m azon, west to the Andes, and south to the southernm ost headwaters o f the A mazon tributaries. If m ost o f the languages in a family are spoken in the A m azon/O rinoco Basin (e.g. Arawak) then we cover that family. I f m ost o f the lan­ guages in a family are outside the region (e.g. G uaicuru) then we do not deal with that family M ost of the A m azon and O rinoco Basins is lowland rain forest, rich in vegetable and animal foods. Archaeologists (by and large) consider that it was first populated about 12,WX) years ago, by hunter/gatherer populations who crossed from Asia at the Bering S trait (which was then dry land) and travelled dow n the N orth American coast. It would have taken only a couple of thousand years for A m azonia to be fully populated. T he m ajor protein food is fish, supplem ented by the hunting of pecca­ ries, deer, tapir, rodents and monkeys. D om estication of plants - focusing first on bitter m anioc - is dated to about 5,000 years ago. It is thought to have begun in the higher lands around the foothills o f the Andes. People em ploying slash-and-bum agriculture soon spread out along all the m ajor rivers, while populations o f hunters and gatherers contracted to the more rem ote areas. Each language family tends to have a characteristic profile in terms o f the type o f territory it is found in, m ethods o f food procurem ent, and material culture. A lm ost all tribes speaking languages belonging to the A rawak, C arib andT upi fam­ ilies are found in the rain forest, use agriculture, and m anufacture canoes, ham ­ m ocks and pottery. In contrast, Jê-speaking peoples are mostly found on grasslands; they have little agriculture and no canoes, ham m ocks or pottery, however, they do have the m ost complex systems o f social organization. Scattered between the agricultural tribes, in the heart o f the rain forest, are small tribes of hunters and gatherers, belonging to m inor linguistic families such as M aku, M uraP ira h i and G uahibo. Some o f these may be the rem ainder o f earlier populations that occupied larger tracts o f land before the agricultural expansion. The languages o f the .seven M ak u tribes, for instance, have been suggested to com prise a rather divergent language family, although full p ro o f has still to be provided for thia They may instead constitute scattered relics of an olden-days linguistic area, whose Ian- i Introduction 5 guagcs had - through long-term diffusion - converged towards a com m on struc­ tural prototype. (This is a topic that requires further work.) M ultilingualism was (an d is) the norm am ong the Indian tribes o f A m azonia. Some groups - predom inantly, those with agriculture - were rather warlike, given to raiding their neighbours. W ithin som e tribes there was a caste o f slaves, which originated as captives o f war. There were also symbiotic relationships between different groups o f people. For instance, the agricultural Tucano tribes (living along the Vaupés River) and th e hunting/gathering M aku (living in the forest away from the river) are in a ‘m aster-underling’ relationship. T he M aku supply their ‘m asters’ with the m eat o f forest anim als and with fish poison, receiving in exchange m anioc flour from Tucano gardens and pottery artifacts. M akù people are considered socially inferior io the Tucano and do not .enter into the system o f exogamous inter­ m arriage that characterizes the Vaupés region. There are some clans within Tucano tribes (and even one whole tribe) that are said to be o f M akii ancestry. A lthough they now speak a Tucano language and practise agriculture they are considered socially inferior and to be less desirable as m arriage partners. (In §1.1 o f chapter 3 there is m ention o f the creation o f a mixed language as the result o f C arib conquest over an A raw ak people.) T hen, in the sixteenth century, cam e the Europeans. W hite-skinned invaders quickly to o k over the coastal areas and m ajor rivers as far as they were navigable. (C hapter 15 e x p iait^ how rapids on the Xingu River proved a barrier to the invad­ ers, preserving the. U pper X ingu as a refuge area into which Indian tribes congre­ gated.) T he Indian population rapidly reduced until, by 1900, it was probably no more than one-tenth o f w hat it had been in 1500. T h e biggest killer was involuntary. W hite invaders brought w ith them diseases to which Indians had no im m unity - smallpox, influenza and thé like. European dis­ eases spread ahead o f the E uropeans themselves; many tribes and languages are, likely to have disappeared before even their nam e could be recorded. O thers went a ' little later. The Arawá people (see chapter 11) are known only from a fifty-word vocabulary collected by th e English explorer Chandless in 1867. A lm ost everyone in the tribe died a few years later from a measles epidemic; the few survivors took refuge w ith the neighbouring K ulina who are said to have m assacred them (Rivet an d T astev in 1938: 72-3). A typical situation was for a num ber o f tribes, each reduced in numbers, to merge. The ethnologist Franz C aspar (1956: 221) lived for a while am ong the T upari tribe an d was told by them th a t ‘in the days o f their grandfathers and great-grandfathers several sm all tribes had m erged. O f every m an and every woman Topto was able to say w ith o u t hesitation o f w hat extraction they were. There was only one m an left out o f each o f the “V aikorotá”, “A um eh” and “ M ensiato” tribes. Five were real 6 R. M. W. Dixon and Alexandra Y. A ikhenm ld “Tupari”. All the re s t. . . were “V akarau” . . . Even the present language of the tribe, he said, was n o t the old Tupari, for the minorities had adopted the language o f the Vakarau. But people could still recall how the various tribes had spoken and Topto told me a few words o f the real Tupari language.’ Just one o f the original languages had survived, alm ost certainly with considerable substrata from the others. There was also enslavement by whites. H em m ing (1978) tells how the European settlers at the m o uth o f the A m azon would travel upriver, capture a tribe, bring them back to work on their plantations under harsh conditions, then when they died out (often, w ithin about ten years), go upriver again to capture and enslave another tribe. A nd there was a great deal o f simple murder. A gun is more power­ ful than bows and arrows; if white people w anted some patch o f land they simply took it, together - if necessary - w ith the lives o f the people who had been living there. Indian tribes invariably fought back; they m ight win in the short term but never in the long term . Some o f them simply fled. A t the end o f the sixteenth century, eighty-four Tupinam bá-speaking villages on the east coast o f Brazil became exasperated by the treatm ent at the hands o f the Portuguese and decided to migrate inland. They travelled - slowly - up the Sâo Francisco River, across the grasslands o f M ato Grosso, until they reached what is now Bolivia, where they were dismayed to encounter Spanish colonists. They then turned north-east and followed the M adeira River until it joined the A m azon, settling on a large island in the middle o f the river, which is now called T upinam barana. But there was no escape. In 1639 their island was invaded by Portuguese com ing up the A m azon (Hemm ing 1978: 235), Before A m azonia was invaded, relations between tribes in A m azonia were some­ times pacific, other times turbulent. We know, for instance, that speakers o f TupiG u aran i languages spread out over a wide area, with considerable effect on languages o f peoples they came into contact with (see §2.2 o f chapter 5). There would o f course have been instances o f migration and o f language death. But these happenings greatly increased after 1500. Invasion o f the forest hinterlands intensi­ fied with the rubber boom at the end o f the nineteenth century. This fell away from about 1910 (when cheaper rubber became available from M alaysia), and lands which had been occupied by non-Indians were again released to their original owners. There are, however, rather few Indians remaining, and these are fast being acculturated. A typical example concerns the Yuqui, speakers o f a Tupi-G uarani language. This group lost agriculture and became hunters and gatherers (although they still retained a caste o f slaves). For several hundred years they were able to keep out of the way o f Spanish colonists, retreating further into the Bolivian forest. By the 1 Introduction 7 1950s the colonial expansion left the Yuqui now here to hide. They were in danger of being wiped ou t when, in the early 1960s, they were persuaded to settle on a New Tribes M ission station (Spearm an 1989). Having been saved from one fate, they are now pointed towards another. W ith their original life-style and culture stripped away (and with only Spanish nam es), the ethnic identity and language o f the Yuqui are unlikely to survive for m ore than a couple o f generations. Estimates vary, but it is thought that there were, in 1500, somewhere between 2 and 5 million people living in A m azonia. T he present Indian population is no more than 400,000. O f the 170 languages reported to be still spoken in Brazil, 115 have less than 1,000 speakers with only 4 having m ore than 10,000 speakers (none has m ore than 20,000); sim ilar figures apply for other South American nations. The governments have been slow to extend roads into rem ote areas or to supply schools and medical posts. But these are coming, and w ith them assimilation into m ain­ stream society. Every year the indigenous languages are used less and less, and Spanish and Portuguese m ore and more. Every year another few languages pass into oblivion. O f the estim ated 300 languages now spoken in A m azonia, only a small fraction are likely to be still actively used in 100 years’ time. We should round off this cultural overview with a short com m ent on trade lan­ guages. Several creoles evolved during the early years o f colonization, based on Spanish, French and Portuguese. But there was one lingua franca o f m ajor im por­ tance. W hat was called Lingua G eral (‘language [that is] universal’ in Portuguese) or N heen-gatu (‘speech [that is] good’ in N heengatu itselO evolved on the east coast of Brazil in the sixteenth century. Its m orphology was simplified from Tupinam bá - a language o f the T upi-G uarani subgroup that was spoken all along the east coast - but the syntax is similar to Portuguese. Lingua G eral soon spread up the Amazon and had dialectal variants in different regions. Indeed, it began to com pete with Portuguese as the m ajor language o f Brazil. In 1727 K ing John V (back in Lisbon) banned the use o f Lingua G eral and it began gradually to fade. Lingua G eral was the trad e language to which m ost tribes in the Brazilian A mazon were first exposed and it was only replaced by Portuguese in the twentieth century. This creole is not quite extinct, still being spoken as first language by a small num ber of people in the U pper R io Negro region. 2 LIN G U ISTIC DIFFUSION There are a num ber o f cultural traits that recur throughout A m azonia. These include female initiation rites (which are much com m oner than puberty rites for boys). There are generally a num ber o f sham ans who control spirits that can both cause and cure diseases. (See, for example. Steward and Faron 1959: 284-318.) The 8 R. M. W. Dixon and Alexandra Y. Aikhenvald word for ‘dog’ is often either the same as the word for ‘ja g u ar’ or else related to it. Probably the m ost pervasive lexeme is kuku or koko for ‘m other’s brother’ or ‘fatherin-law’ (in a system where one can m arry a cross-cousin). If a man marries a w oman from another tribe he must know how to address his wife’s father; this has undoubt­ edly led to the widespread borrowing o f the term for ‘wife’s father’. Linguistic traits, by and large, diffuse m ore slowly than other cultural traits. However, given sufficient time, languages from several genetic groups that are located in the sam e geographical area will gradually com e to share certain linguis­ tic features and will, as a consequence, make up a ‘linguistic a rea’. This can be defined as: a region including languages from several different genetic groups, with the languages sharing certain symptom atic features which can be inferred to have diffused across the area. The features will not, as a mle, be found in languages from these genetic groups which are located outside the area. It should be noted that very com m on properties (those found in very m any of the languages o f the world, e.g. a tense system) are less significant as diagnostic m arkers o f a linguistic area than are more unusual properties (e.g. an evidential!ty system). Am azonia can be recognized as a linguistic area in term s o f features like the fol­ lowing, which are shared by all (or m ost) languages in the area. (a) The m ajority o f languages are polysynthetic and head m arking; agglu­ (b) There is typically one liquid phoneme, which is frequently a flap. There tinating with little fusion. are usually more affricates than fricatives. The high unrounded central vowel Í is frequent. A typical A mazonian vowel system has five members: /, e, a, i, tilo. There is typically contrastive nasalization of vowels. (c) M any languages have extensive classifier and/or gender systems. G ender assignment is often semantically transparent, and is not overtly marked on the head noun. (d) There are very few oblique cases - often just a locative and an instrumental/comitative. (e) Possession (either alienable or inalienable) is typically m arked on the possessed noun, no t on the possessor; the m ost w idespread word order is ‘possessor possessed’ (e.g. ‘John his-canoe’). (0 Often, just one core argum ent is cross-referenced on the verb. There may be different bound pronom inal paradigm s depending on which core argum ent is being cross-referenced in each particular instance. (g) The rules for which core argum ent is cross-referenced can be complex (relating to the meaning o f the verb, clause type, etc.) often giving rise 1 Introduction 9 to a ‘split-ergative’ system. Fully accusative systems o f m arking for predicate argum ents are rarely encountered. (h) The bound pronom inal forms m arking a possessor within an N P are typically the sam e as one of the bo u n d pronominal paradigm s for m arking core argum ents o f a clause (sometimes the same as the A or A/Sji series, other times the same as the O o r O/S^ series). (i) M ost (although n o t all) languages have prefixes; there are typically fewer prefix than suffix positions. (j) If there are several prefix positions, the bound pronom inal prefix(es) will typically appear further from the ro o t than prefixes that m ark valency-changing derivations (e.g. causative, applicative). (Tucano lan­ guages are entirely suffixing; the bound pronom inal suffixes appear further from the root than suffixes th a t m ark valency-changing deriva­ tions.) (k) M ost verbal categories (e.g. tense, aspect, modality, direction) are expressed through optional suffixes. (I) Subordinate clauses typically involve nom inalized verbs, w ith the type o f subordination being marked on the verb. (m) If there is noun incorporation, typically only those nouns which are obligatorily possessed can be incorporated, and they typically precede the verb" root. (n) In many languages adverbs and adpositions may be incorporated into (o) There is generally only a small class o f lexical numbers. the verb, typically following the verb root. There are o f course a few exceptions to these pan-A m azonian tendencies. For instance, Tupi-G uarani languages are the only languages in A m azonia which allow incorporation o f unpossessed nouns; however, in m odem languages this technique is falling o u t o f use (Seki forthcom ing, K akum asu 1986). In Nadëb, a M akù lan­ guage from the Middle Rio N egro, incorporated adverbs and adpositions precede the verb root instead o f following it. In Palikur, a N orth Arawak language from Brazil and French G uiana, incorporated body parts follow the root (Aikhenvald and Green 1998). It is interesting to com pare typological characteristics o f the A m azonian lin­ guistic area, in lowlands S outh A merica, with those o f the A ndean linguistic area in the adjacent m ountains, w hich comprises the Q uechua and Aymara families. The A ndean area is clearly different in alm ost all o f the characteristics ju st listed, (b) There are two or three liquids; fricatives rather than affricates; and a three vowel system, i, a and u, with no contrastive nasalization, (c-e) There are no classifier or 10 R. M . W. Dixon and Alexandra Y. A ikhenm ld gender systems; there is an extensive set o f case markers; possession is marked both oi\ possessor an d on possessed, (f-h ) Two core argum ents are m arked on the verb, in an entirely accusative system; bound pronom inal markers o f possession show some sim ilarity to, but are n o t identical with, the forms m arking core argum ents on the verb, (i-k ) There are no prefixes; and there is an obligatory suffixal system for tense and aspect. (1) Subordination does not involve nominalization. (m -n) There is n o incorporation o f nouns, adverbs o r prepositions (Cole 1982; 161 m entions incorporation within nominalization, but this is just a type o f com pounding, as in English), (o) There is a full set o f lexical numbers. As regards characteristic (a), A ndean languages are synthetic, and com bine head and dependent marking; they are basically agglutinating with some fusion (subject, object and tense suffixes to the verb may be fused). There Is no sh arp boundary between the A ndean and A m azonian linguistic areas - they tend to flow into each other. For instance, A ndean features such as lack o f prefixes and an accusative technique for m arking syntactic function are found in languages o f the Tucano family, which are in A m azonia but fairly close to the Andes. There are a num ber o f gram matical properties which are not shared by all A m azonian languages but are found in the languages in certain regions and help to define these as linguistic subareas w ithin a wider linguistic area. These include; (1) contrastive lexical tones are found in two regions - one in southern A m azonia (overlapping the states o f R ondônia and M ato G rosso) and (2) o ne in the north-w est (along the Vaupés river, from Brazil into Peru). switch-reference m arking is found in a group o f languages in western A m azonia. (3) gender assignment is not semantically transparent in a region o f south­ ern A m azonia centred on the P urus river basin (where Bolivia, Brazil an d Peru meet) which includes languages from the Arawâ and C hapacura families and th e Peruvian and Pre-andine subgroups o f Arawak. (4) classifiers are used in different m orphosyntactic contexts in different regions - for instance, they are used in possessive constructions in the U pper R io Negro region but not in Peruvian A raw ak languages. The limited regional distribution o f these features can be im portant for hypoth­ esizing w hether certain features were m ore w idespread at some time in the past, and are being progressively lost, o r w hether they are recent innovations th a t are cur­ rently diffusing more and m ore widely. 1 Introduction GENETIC RELATION SH IP There can be a num ber o f explanations for some perceived similarity between two (or more) languages It can be an indication o f genetic relationship. O r it can be the result o f areal diffusion, borrow ing from one language into another (in one o r in both directions) or borrow ing by both languages from a third language. O r it can be some universal feature, e.g. the word for ‘blow’ typically imitates the activity, having a form som ething like f^ u {the actual form will relate to the phonetic and phonological resources o f each language). O r it can be chance, as where the A ustralian language M babaram and the Indo-European language English both have dog as the nam e for an anim al o f the species Canis. To say that a group o f languages is genetically related (as a language family) is to say only one thing. It is to assert that they go back to a com m on ancestor, each having developed from this ‘proto-language’ by its own set o f historical changes. C oncordant with this, there is only one way to prove that a group o f languages is genetically related. This is to propose w hat the proto-language was like (in some detail) and to describe how each o f the m odern languages developed, by systematic changes, from this com m on ancestor. If two (or more) languages show a certain set o f similarities, these need n o t nec­ essarily be evidence for genetic relationship. They may, alternatively, be due to bo r­ rowing, to universal tendencies o r ju s t to chance. It is true that certain kinds o f similarity (a pronoun paradigm , or a suppletive set) are strongly suggestive o f genetic relationship. But suggestion is not p ro o f A nd the only p ro o f is to demon­ strate the genetic relationship by showing the regular changes which each individ­ ual language has undergone from a postulated proto-language (dealing with phonology, verbal and nom inal morphology, pronouns, markers o f negation, etc., as well as all kinds o f lexemes). In only a few instances d o we have w ritten records o f sufficient time depth to iden­ tify the proto-language and details o f the changes m odern languages have under­ gone - Sanskrit and the m odem Indie languages; Latin and Romance languages; Old Irish and Irish Gaelic, Scots Gaelic and M anx. In other instances, large por­ tions o f a putative proto-language have been reconstructed, together with the changes m odern languages have undergone. A num ber of proto-languages have been proved in this way, so that all scholars are agreed on their genetic unity; they include Indo-E uropean, U ralic, Dravidian, A ustronesian, A lgonquian and M ayan. A lthough m ost o f these language families have been recognized for a fair period o f time, it has n o t proved possible to estab­ lish any higher-level genetic linkage between them. 12 K M. W Dixon and Alexandra Y. Aikhenvald However, some people are not prepared to rest upon w hat can be scientifically proved and w ant to go on - to attem pt to relate families together as higher-level families, and so on back {sometimes, even back to a putative proto-W orld lan­ guage). Such suggestions o f higher-level genetic groupings are generally based on a few odd correspondences (often, between one o r two languages in family X and a few languages in family Y, rather th a n between proto-X and proto-Y ) and thus can n o t be accepted as serious scholarship. T he suggestions that have been m ade are mutually incompatible, which is itself n o t a good portent. For instance, Schuller (1919/20) proposed a genetic link o f C arib with Arawak (also throwing in C hibcha and M ayan); Greenberg (1960) linked C arib with Jê, P ano and N am biquara; G reenberg (1987) kept Carib, Jê and P ane together but now placed N am biquara in a group w ith Tucano and M akii, am ong others; and Rodrigues (1985) suggested a genetic link between Carib and Tupi. Thus C arib has been linked with Arawak, M aya, C hibcha, Jê, Pano, N am biquara and Tupi at one time o r another (although not all at once). Kaufman (1990) surveys a num ber o f higher-level genetic suggestions, showing how these dis­ agree; he has lists o f languages that are ‘M acro-Jê for some [investigators], M acroK ariban for others’, and o f languages that are ‘M acro-Tupian for some, M acro-Jê for others’, etc. A perceptive com m ent on suggestions o f this type concerning higher-level genetic links comes from M ason (1950:162); ‘It is a truism o f linguistic research that, given large enough vocabularies to compare, and m aking allowances for all possible changes in the form o f a word or stem, as well as in its m eaning, a num ber of appar­ en t similarities, convincing to the uncritical, can be found between any two lan­ guages.’ The lack o f scientific basis for posited long-distance genetic links can be illustrated with a couple o f examples (these could be m ultiplied a hundred times over). Ehrenreich (1897) first suggested a genetic link between the A raw á and Arawak families. H e gave 17 ‘correspondence sets’ with examples for each drawn from 1 or m ore o f 3 Arawá languages, and I or m oK o f 29 Arawak languages. A typical set is; ‘bow ’ (as in ‘bow and arrow ’) Arawá languages Paumari; kiidaii A raw ak languages Bare; dm idaja Jam am adi: didiSa K ustenau; tuti A rawá: bigauaha Tariana: shidoa (‘arrow ’) N o p roto-form was suggested (nor is it easy to see one that could be suggested), for pro to -Arawá, o r for proto-A raw ak, or for p roto-Arawá-Arawak. In fact Arawá Ian- I Introduction 13 guages all have different (and unrelated) words for ‘bow ’. For proto-A raw ak, Payne (1991: 396) reconstructs *tapo ‘bow ’ (a form th a t is in fact reflected in Tariana yawithepu ‘bow ’). Yet on the basis o f 17 sets o f ‘correspondences’ like this, it was sug­ gested that the Arawá and Arawak families are genetically related. Rivet and Tastevin ( 1940) followed a sim ilar ‘m ethod’, draw ing o dd correspondences between some forms in a few A raw á languages and som e forms in a few A raw ak languages (and they drew on a pool o f 86 Arawak languages). A typical (short) entry from their list o f ‘cognate’ sets is (Rivet and Tastevin 1940; 8): appeler [‘to call’]; wawma-he, il appelle [‘he calls’], K ulina [Arawá] pi'uäna, bi~uána, Tariana {Arawak] T he idea that Arawá (now called M adi) and A raw ak are related in a family called ‘A raw akan’ was repeated by M atteson (1972). She believed that she was dem on­ strating genetic relationships by 'rigorous application o f standard techniques o f the com parative m ethod’ (M atteson et a l 1972; 21). T he first o f her 353 putative cognate sets o f proto-A raw akan is (M atteson 1972:172): 1. ‘abdom en’ *tia-ri [in pröto-A raw akan] H ika-teiko, proto-Piro-A puriná *tso-mon~tí-a, proto-A shaninka *ato-li, proto-M adi [i.e. proto-A raw a] *tee~IV, proto-N ew iki N ote th a t the only segment that the proposed proto-A raw akan reconstruction and the four interm ediate reconstructions have in com m on i s ‘t’. F u rth er on in the paper, M atteson (1972:203) gives the Piro and A puriná forms; 1. ‘abdom en’ *tiká-lelko [«c, accent included here b u t not above] [proto-Piro-A puriná] ckcte, Piro -íikáko, A puriná and then the forms in her A shaninka subgroup (M atteson 1972; 211): 1. ‘abdom en’ *íjo-mö«-/í-a [proto-A shaninka] -niotia, M achiguenga -tsomonte. Cam pa tsomoné, N om atsiguenga In the proto-M adi section there is no word for ‘abdom en’. However, the first cognate set given is (M atteson 1972: 219); 14 R. M. W. Dixon and Alexandra Y. Aikhenvaid 1. ‘b ark ’ *ato-li [proto-M adi] ete-ru, Culina *aiu-li, pre-Jam am adi There are a n um ber o f possible explanations here. One is that M atteson just copied o u t the first item from her proto-M adi list, assuming that (as in the other lists) it was ‘abdom en’, overlooking the fact that the proto-M adi list has no term ‘abdom en’ but its first item is ‘b ark ’. O r she may have imagined that ‘abdom en’ could be cognate w ith ‘b a rk ’, but neglected to m ention this. There is n o ‘abdom en’ given in the proto-N ew iki list, but we do find this further on, under proto-W estem-N ew iki (M atteson 1972: 231): 1, ‘abdom en’ *tée-lV[sic, accent included here but n o t above] [protoWestern-Newiki] dé-, Piapoco tere, Cabiyari This example is typical o f M atteson’s methods. It shows a lack o f understanding o f the established m ethodology of language com parison an d reconstruction (and an extreme sloppiness). Indeed, the volume begins with ‘reconstructions’ of lexemes in ‘proto-A m erindian’, com paring odd forms from odd m odem languages across two continents. The Arawak family is well established (going back to 1783). Recently, the term ‘A raw akan’ has been used for an unproven higher-level unit which includes Arawak (now called M aipuran), Arawá, C hapacura, G uam o, H arakrabet and Uro-Chlpaya (the details vary a little from au th o r to author). This is p art o f the reason why we avoid the term ‘A raw akan’ (along with other terms ending in ‘a n ’) in this volume, preferring ‘A raw ak’, a label consistently used by scholars from South American countries for the well-established genetic grouping (corresponding to the M aipuran o f the A raw akanists). The first satisfactory reconstruction o f a p a rt of proto-A raw ak vocabulary is by Payne (1991); this includes 203 forms. Dixon has reconstructed over 400 lexemes for proto-A raw â. Three possible cognates can be recognized between these lists, none o f them fully convincing. The gram m atical forms o f proto-A raw á and protoArawak are also quite different. It must be concluded that there is no evidence w hat­ soever th at (despite their similar names and geographical proxim ity) the Arawá and Arawak language families are genetically related. In fact, the m ost perceptive com m entators on S outh Am erican languages do not suggest any genetic connection here, e.g. M ason (1950), L oukotka (1968), Tovar and D eT o v ar(l9 8 4 ) and Rodrigues (1986). However, the eccentric suggestion from I Introduction 15 Ehrenreich, Rivet and Tastevin, and M atteson has been repeated by others, e.g. M étraux (1948), Noble (I96S) and G reenberg (I960, 1987). Even so em inent and (generally) reliable a scholar as Derbyshire uncritically accepts this, m ingling data from A raw á with that from Araw^ak languages in his ‘Com parative study o f m o r­ phology and syntax in Brazilian A raw akan’ (D erbyshire 1986). There can be a tendency to accept w hat is put forward as an orthodoxy, w ithout pausing to question it. When a m issionary linguist from Peru, who had spent over thirty years studying K ulina (an A raw á language), received a copy o f D ixon (1995), she w rote to Dixon: ‘I becam e so intrigued by your declaration o f no kinship with the A raw ak family that I read and re-read that p a rt [of the p a p e r]. . . N o wonder we never did seem to “fit the p attern ” am ong A raw aks here.’ That is, this linguist had been told that K ulina was an A raw ak language. But there are no significant cognates and entirely different gram m ars, which puzzled her. We have dwelt at some length upon the A raw ak/A raw á suggestion to illustrate its vacuous nature. A similar refutation could be provided for alm ost all other (perhaps for all other) suggestions o f higher-level genetic relationships between the estab­ lished language families. There have been a num ber o f com m entators who are sound scholars and confine themselves to summarizing the established families and the isolates - as already m entioned, these include M ason (1950), L oukotka (1968), Tovar and D e Tovar (1984) and Rodrigüès (1986). Rivet has considerable reputation but his work on S outh American com parative linguistics is deeply flawed (he also suggested a genetic connection between South A m erican Indian and A ustralian A boriginal lan­ guages). G reenberg’s (1960, 1987) A m erind’ has attracted considerable publicity but is w ithout scholarly foundation (see the critical com m ents in Dixon 1997: 54-5 and references quoted there). It should be borne in m ind that ju st because som ething has been published in a book does not necessarily mean that it has any value. People writing a gram m ar o f language X from family Y sometimes feel im pelled to say som ething like ‘family Y belongs to the M acro-Tucanoan stock within th e Equatorial-Tucanoan stock within G reenberg’s A m erind’. Such a statem ent should be avoided. It simply adds a veneer o f fantasy to w hat may well be a sound and useful gram m ar o f language X. (C ontributors to this volume have confined themselves to just m entioning those genetic groupings that are well accepted an d have been proved.) As a rough rule o f thum b, readers should be on their guard against any nam e beginning with ‘m acro-’ or any unit labelled as a ‘sto ck ’. With the possible excep­ tion o f M acro-Jê (which, if proved to be a genetic group, ought to be relabelled) all m acro-X ’s are intensely speculative. A sim ilar com m ent applies for alm ost all stocks. (N ote that in this book we refer to the Tupi family and the Tupi-G uarani 16 K M. W. Dixon and Alexandra Y. A tkhem ald branch or subgroup within it - just as one talks o f the Indo-European family and the G erm anic branch o r subgroup w ithin it - in preference to Tupi stock and TupfG uarani family.) 4 THE PUNCTUATED EQ U ILIBRIU M MODEL It is generally recognized am ong historical linguists that one can only recognize his­ torical lelationships between languages (and reconstruct shared proto-languages) to a time-depth o f 5,000-8,000 years. G iven any longer time, individual languages are likely to have changed too much for their original genetic connection to be apparent. Thus, G reenberg’s (1987) views a bout a relationship between all the lan­ guages o f South America and m ost o f those o f N orth A merica are not only unproven but unprovable. But if one views the ‘family tree’as the basic model o f language relationship, one m ust presume that there probably was a family tree joining together family trees. T h at is, that many o f the language families (and isolates) o f the Americas presum­ ably are genetically related, only one would need a time m achine to discover exactly how. In this view, G reenberg’s scheme is a perfectly possible one. But there are many alternative scenarios that are equally plausible, and there is no way o f deciding between them. We believe that this is a mistaken view. T he family-tree model is appropriate for certain kinds o f language development but not for all o f them. H um ankind is believed to have had language for at least 100,000 years (many people would o p t for a longer time-span). C onsider the Indo-European family, which has over 100 m odem languages and for which a tim e-depth o f about 6,000 years is posited. There are about 17 periods o f 6,000 years in 100,000 years. If one language spawns 10^ descendants in 6,000 years, then over 100,000 years it should give rise to 102^ 1?= 10^ or 10 million billion billion billion languages. But how many languages do we have in the world today? A bout 5,000-6,000. This suggests that we should re­ think the idea th at a family-tree type o f language split is the universal model of lan­ guage developmentD ixon’s essay The rise and fa ll o f languages (1997) puts forw ard an alternative idea, the Punctuated Equilibrium model o f language development. He suggests that d uring m ost o f the history o f hum ankind there have been long periods of equilib­ rium , interrupted by short periods o f p unctuation. D uring an equilibrium period a given geographical area would have been inhabited by a num ber o f small political groups o f similar size, each with its own traditions, religion, laws and language. No one group or language would have substantially greater prestige than any other. C ultural traits, including linguistic features, w ould steadily diffuse across the geo­ 1 Introduction 17 graphical area. There would always be gradual changes; some languages w ould split and som e would cease to be spoken, but this would be o n a relatively m inor scale. In due course, as a result o f linguistic diffusion, all the languages in the geograph­ ical area would become m ore similar in structural profile - they would converge towards a com m on prototype. A n equilibrium situation may be punctuated by som e cataclysmic happening. This could be a n atural event (flood or drought, rising o r falling sea levels); o r some material innovation (say, a new m ethod o f food production); o r the rise o f some military o r religious leader, intent on building an em pire; or entry into new terri­ tory. In a period o f punctuation there will be expansion and split o f peoples and of languages. It is during punctuation that a family-tree model is appropriate; we get the divergence o f a steadily increasing set o f languages, all em anating from a com mon proto-language. Eventually, the punctuation m ode will gradually lose power, and merge into a new equilibrium period. When the first people cam e into South Am erica - 12,000 o r so years ago - they would quickly have expanded to fill the continent. Population is likely to roughly double each generation if there is unlimited food and land available. This was a period o f punctuation, describable by a family-tree diagram . O nce the land was fully populated (which m ay only have taken 2,000-3,000 years), a period o f equi­ librium would have com menced within each geographical zone - in the tropical forests, on the grasslands, on the m ountains, and so on. There would have been many small groups o f hunters and gatherers living in a state o f relative equilibrium with each other. Linguistic traits would have diffused across the languages within each region. A m ajo r punctuation w ould have been triggered by the adoption o f agriculture, believed to have taken place about 5,000 years ago. The peoples w ith agriculture had a distinct advantage in food production. As a result they expanded and split, and so did their languages. T his serves to explain the readily provable genetic unity of the A raw ak, C arib and T upi families. The agriculturalists took over the best land, along m ajo r rivers. Scattered between them are hunters and gatherers. G enetic rela­ tionships are here less clear. As mentioned in §1, it is possible that m odern-day M akii tribes are relics o f an earlier equilibrium situation, w ith their similarities being the result o f having belonged to an olden-days diffusion area (rather than being indicators o f close genetic relationship). T he point being m ade is th a t there is no sure way to distinguish between similarities th a t are due to areal diffusion and those th at reflect com m on inheritance. Suppose Europe came to be invaded and settled by the Chinese, leaving ju s t small pockets o f people speaking Italian and Basque an d H ungarian. A later-day linguist might well take the similarities between these three relic languages (their ‘S tandard A verage European’ features) as evidence 18 R. M. W. Dixon and Alexandra Y. Aikhenvald o f genetic relationship. It is possible that the history o f the M akii follows a similar pattern. C hapter 6, on M acro-Jê, discusses the Jê family and 11 other language families, basically all spoken in the grasslands region. Rodrigues presents 39 possible cognates and a num ber o f points of gram matical similarity. All o f these are o f a t)fpological nature, relating to similarities in construction type, constituent order and gram mat­ ical categories between the languages, rather than shared grammatical forms. (Pronom inal forms, for instance, appear to be rather different between Macro-Jê families.) Rather than all the Macro-Jê families being related in a higher-level family tree, it seems to us that they could constitute a long-term linguistic area; this would account for their considerable typological similarities. (There jnay, o f course, be genetic links between some o f the established families, within the linguistic area.) It will be seen th a t - in terms o f the Punctuated Equilibrium m odel of language development - the quest for a family o f language families (a tree o f trees) is mis­ conceived. Each m odern language family probably had its origin in 'the end o f a period o f equilibrium. Similarities that are noted between proto-languages may well be areal features which had - during the equilibrium period - spread to all or m ost o f the languages in a given geographical region. In summary, at the end o f the initial hum an expansion across the whole of South A merica a family tree would have appropriately m apped the period of punctuation that was drawing to an end. A long period o f equilibrium then ensued, with roughly stable population and a great deal o f cultural and linguistic diffusion within each ecological zone. T he genetic relationships between languages, which had been clear at the end o f the period o f punctuation, would have gradually become blurred and finally lost as more and more features diffused. Then the equilibrium would have been punctuated and just a few o f the languages (those whose peoples had some distinct advantage in living and winning, e.g. agriculture) would e.Kpand and split, each starting its own family tree (and obliterating other languages, whose speakers did n o t have this point o f advantage). Establishing that a group of languages is genetically related, as a language family, is generally an easier m atter than deciding on the internal constitution o f the family tree, i.e. subgroupings. To assign a set o f languages to a subgroup there must be evi­ dence that they have shared some historical development (and it should be some rath er distinctive change, not anything th a t com monly recurs all over the world). The difficulty here is to distinguish between similarities that are due to shared genetic developm ent and those which are due to diffusion. C onsider the Arawak family, for instance. A num ber o f subgroups have been tentatively established, each o f them being located in a certain region. T he languages in a given subgroup do 1 Introduction 19 share certain developments. But these may well be characteristic features o f that geographical region, and found in both the A raw ak and the non-A raw ak languages that are spoken there. It is first necessary to exam ine the areal linguistics o f the region, facto r o u t the areal features, and then see whether the rem aining similarities between the Arawak languages spoken thene constitute sufficient evidence for sub­ grouping. N one o f this has yet been done. A full investigation o f subgrouping w ithin an areal perspective - is an im portant topic for future research. The European invasion, com mencing in 15(M), acted as an abrupt punctuation o f linguistic areas across all o f South A merica. The prestige languages - Spanish, Portuguese and a few creoles - are continually expanding their domains. It is likely that m ore th an h alf (perhaps much m ore than half) of the languages spoken in 1500 have already passed into oblivion, and the rem ainder are following at a steady rate. Q uechua is currently estim ated to have about 8.5 million speakers; however, in central Peru (and probably in other regions as well), m ost children o f Q uechuaspeaking parents are preferring to speak ju s t Spanish (A delaar 1991; 50). Only G u aran i appears to be safe, in the medium term , because it is one o f the two national languages o f Paraguay (and is in fact spoken in Paraguay by m ore people than is Spanish). In low land A m azonia there is little hope for even m edium -term survival o f any language. T he tentacles o f European-style civilization have been slow to penetrate th ^ a i n forest, and it is this that has helped some cultures and lan­ guages to m aintain their autonom y. But the outside world is now creeping in. Every decade, each indigenous language is spoken a little less (sometimes, a lot less) and Spanish o r Portuguese a little more (or a lo t more). D escribing these languages, before they disappear, is an urgent task. If everyone who calls them self a linguist - from South Am erican countries and from overseas - were to devote a year o r so to fieldwork, and then write and publish a gram m ar, dictionary an d volum e o f texts for some previously undescribed (or scarcely described) language, then m ost o f the rich linguistic and cultural heritage would be preserved, for posterity. (This would also lead to substantial enrichm ent o f Basic Linguistic Theory.) 5 O R G A N I Z A T I O N OF T H I S B O O K Basically, we have devoted a chapter to each o f the m ajor language families - a long ch ap ter for a large family (w ith several dozen languages) and a short chapter for a small family (with just a few members). Tupi-G uarani (a subgroup within the Tupi family) is the best-known group o f languages and we have accorded it a chapter o f its own. 20 K M. W. Dixon and Alexandra Y Aikhenvald Then, in chapter 12, M ary Ruth Wise surveys the small language families o f Peru. In chapter 13, we provide some inform ation on the small families and language iso­ lates o f Brazil, Bolivia, Colom bia and Venezuela. The final two chapters look at lin­ guistic areas - the established area in the Içana-Vaupés River Basin, and the incipient area on the U pper Xingu River. There are a few languages that we say nothing about, for the simple reason that alm ost nothing is known about them. These include: Awaké, H oti, Irantxe, Kanoé, Puinave, Sape a n d the small K atukina family. Editing the present book has not been an easy task. The contributors come from differing backgrounds and have different kinds o f linguistic training. In a couple of instances one o f the editors had personal contact w ith contributors and she was able to work closely with them on their chapters. Chapters 9 and 15 were written in Portuguese and translated into English by the editors. We have n ot in all cases been able to achieve the degree of coverage that we had hoped for. T he reader will be able to see for themself, from this volume, w hat the state o f the art is - at the end o f the twentieth century - with respect to knowledge o f the various language families in A m azonia. BIBLIOG RAPHY Adelaar, W. E H. 199!. ‘The endangered languages problem: South America’, pp. 45-91 of Endangered languages, ed. R. H. Robins and E, H. Uhlenbeck. Oxford; Berg. Aikhenvald, A. Y. and Green, D. 1998. ‘Palikur and the typology o f classifiers’, Anthropological Linguistics, 40.429-80. Anchieta. J. de. 1595. Ârie de grammatica da lingoa mais usada na costa do Brasil. Coimbra: A ntonio de Mariz. Caspar, F. 1956. Tupari. London; Bell. Cole, P. 1982. Imhabura Quechua, Lingua Descriptive Series. Amsterdam: North-H olland. Derbyshire, D. C. 1986. ‘Comparative survey o f m orphology and syntax in Brazilian Arawakan’, pp. 469-566 of Handbook o f Amazonian languages, vol. I, ed. D. C. Derbyshire and G. K. Pullum. Berlin: M outon de Gruyíer. Dixon, R. M. W. 1995. ‘Fusional development o f gender m arking in Jarawara possessed nouns’, UAL, 61.263-94. 1997. The rise andfall o f languages. Cambridge: Cambridge University Press. Ehrenreich, P. 1897. ‘Materialen zur Sprachenkunde Brasiliens: Vokabulare von Purus-Stiitnmen’, Zeitschrift fiir Elhnologie, 24.59-71. Greenberg, J. I960. ‘The general classification o f Central and South American languages’, pp. 791-4 o f Men and cultures: Selected papers o f the 5th International Congress o f Anthropological and Ethnographic Sciences, ed. A. Wallace. Philadelphia: University of Pennsylvania Press. 1987. Language in the Americas. Stanford: Stanford University Press. Hemming, J. 1978. Red gold: The conquest o f the Brazilian Indians. Cambridge, Mass.: H arvard University Press. 1 Introduction Kakumasu, J. 1986, ‘U rubu-K aapor’, pp. 326^0 3 o f Handbook o f Amazonian languages, ¥ o l 1, ed. D. C. Derbyshire and G. K. Pullum. Berlin: M outon de Gruyter. Kaufman, T. 1990. ‘Language history in South America; what we know and how to know m ore’, pp. 13-73 of Amazonian linguistics: studies in Lowland South American languages, ed. D. L. Payne. Austin: University o f Texas Press. Loukotka, C. 1968. Classification o f South American Indian languages. Los Angeles: Latin American Center, UCLA. Mason, I A. 1950. ‘The languages o f South American Indians’, pp. 157-317 o f Handbook o f South American Indians, vol. VI, ed. J. H. Steward. Bureau of American Ethnology, Bulletin 143. W ashington, D.C.: Smithsonian Institution. M atteson, E. 1972. ‘Proto-Arawakan’, pp. 160-242 o f M atteson et ai., 1972, M atteson, E. et al. (eds.) 1972. Comparative studies in Amerindian languages. The Hague: M outon. Métraux, A. Î948. Tribes o f the Juruâ-Purùs Basins’, pp. 657-86 o f Handbook o f Souiit American Indians, vol. Ill, ed. J. H. Steward. Bureau o f American Ethnology, Bulletin 143. W ashington, D.C.: Sm ithsonian Institution. Noble, G. K. 1965. Proto-Arawakan and its descendants. Publication 38 of Indiana University Research Center in Anthropology, Folklore and Linguistics. (UAL vol. 31, no. 3, p art II.) Payne, Doris L, 1991. Ä classification o f M aipuran (Arawakan) languages based on shared lexical retentions’, pp. 355-499 o f Handbook o f Amazonian languages, vol. Ill, ed. D. C. Derbyshire and G. K. Pulluin. Berlin: M outon de Gruyter. Rivet, P. and Tastevin, C. 1938. ‘Les langues Arawak du Purùs et du Juruá (Groupe Arauá)’, Journal de la Société des Américanistes de Paris, 30.71-114. 1940. ‘Les langues Arawak du Purùs et du Juruà (Groupe Arauà) (suite)’. Journal de la Société des Américanistes de Paris, 32.1-55. Rodrigues, A. D. 1985. ‘Evidence for Tupi-C arib relationships’, pp. 37Î-404 o f South American Indian ^ gu ages; Retrospect and prospect, ed. H. E. M. Klein and L. R. Stark, Austin: University of Texas Press. 1986. Llnguas brasileiras: para o conhecimento das linguas indigemts. Sào Paulo; Ediçôes ■Loyola. ‘ Schuller, R. 1919/20. ‘Zur sprachlichen Verwandtschaft der Maya-Qu'itäé mit der CaribAruác’, Anthropos, 14-15.465-91. Seki, L. forthcoming. Granmtica da lingua Kamaiurá. Campinas: Editöra da Unicamp. Spearman, A. M. 1989, Yuqui: forest nomads in a changing world. New York: H olt, R inehart and Winston. Steward, J. H. and Faron, L. C. 1959. Native peoples o f South America. New York: McGrawHill. Tovar, A. and De Tovar, C. L. 1984. Cataiogo de las lengtias de Américn del Sur. Madrid: Editorial G redoa Carib DESM O N D C. DER BYSHIRE 1 in t r o d u c t io n ' The C arib family contains the second largest num ber of languages (following Arawak) in South America. H undreds of C arib language names can be found in the literature, but after eliminating diiTereot nam es that apply to only one language or to dialects o f a language, the number of languages spoken today is probably about 30. K aufm an (1990: 51) gives the num ber as 43, but this seems to include some that are probably extinct o r are d iaiects (see G ildea 1998, for the K aufm an 1989 classification). HoflT (1992) gives approximate numbers o f speakers o f individual languages, and these ad d up to a total o f between 48,000 and 57,000 C arib speakers. Many o f them also speak the national language (D utch, English, French, Portuguese or Spanish). C arib languages 'are spoken in six countries o f South America; Colom bia, Venezuela, G uyana, Surinam e, French G uiana and Brazil. M ap 1 shows approxi­ mate locations for each language. There are differing views about the original C arib hom eland. Som e scholars claim that it was south o f the A m azon, probably between the X ingu and Tapajos rivers. O thers locate it n o rth o f the A m azon, where m ost o f the C arib groups are currently found, with movements o f some groups (including A rara, Bakairi, K alapalo, K uikiiro and T xikâo) to locations south o f the A m azon as far as the Xingu river basin. See the introduction in Edwards (1980) for a sum m ary o f these views. T he earliest known linguistic m aterials on any C arib language are two works with the title Introduction à la langue des Galibis by the missionary priest Pierre Pelleprat (1606-67). They were based on the research and notes o f Fr. D ionisio M esland, a Jesuit intellectual. The first attem pt at a classification came in the next century, * It is obvious that this chapter could not have been written without the source m aterials pro­ vided by the scholars whose works are listed in the bibliography. My thanks to all o f them for enriching my knowledge o f Carib languages and my apologies for any misrepresenta­ tions or errors I may have introduced. I am grateful to the following for helpful comments on an initial draft o f the paper: Bob Dixon, Spike G ildea, Berend Hoff and Tom Payne, I have especially valued the opportunities to interact in recent years with Gildea and Hoff, both o f whom have made such extensive and im portant contributions to C arib studies. S £ 0 “= S 2 a s S2X <i T3ws Tws3 L§u I l l l l tti tti < ç a H O i2 5 : B s l | I . = â - l'5 -S g a I i I ? < û ‘3 s ‘ S "i |- q S £ 2 25 2 Carib made by Fr. Salvatore Filippo Gilij (1721-89), and was restricted to Venezuelan Carib languages. D urbin (1977) supplies inform ation on these eariy sources. Later classifications appeared, b u t all were based on insufficient and unreliable data. The best internal classifications in recent times (G irard 1971, D urbin 1977, Kaufm an 1989) are reviewed by G ildea (1992, 1998). As he points out, they are all flawed to a greater o r lesser degree by lack o f adequate data and the consequent impossibility o f reconstructing grammatical m orphem es and lexical items. Thus Carib com parative and historical studies lag far behind those o f the other two large A mazonian language groups, Arawak and Tupi-G uarani. In the last twenty years, however, good descriptions o f the m orphology and syntax o f several C arib lan­ guages have become available, a nd ongoing research by several scholars in other lan­ guages should soon lead to more reliable reconstruction and internal classification. Gildea ( 1998) is an excellent com parative study o f certain features o f C arib m orphosyntax (constituent order, person-m arking prefixes, tense-aspect-mode marking, m arking o f possession, nom inalizing morphology, and the development o f ergativity in m ain clauses), based on nineteen languages (some not well docum ented as yet). The most conservative classification to date is that o f K aufman (1989, reproduced in Gildea 1998), based on cognate retention, consonant cluster reduction, and stress patterns. It is this classification that is reflected in table 2.1, which shows: (t) K aufm an’s putative four m ain branches o f the C arib family (the Carib language and Panare constitute sejjarate groupings); (2) the individual languages on which this survey is based (K aufm an includes others); (3) the two-letter abbreviation used in this paper for each language name; (4) som e frequently used alternate names, or the larger groups of which the language is a dialect; (5) approximate populations (Wai Wai from Hawkins, p.c.; the others from G rimes 1996); and (6) the main sources I have used. T he sources in table 2.1 include substantive gram m ars for C arib (H off 1968), Wai Wai (Hawkins 1998), H ixkaryana (D erbyshire 1979, 1985), M akushi (Williams 1932, C arson 1982, A bbott 1991), A palai (Koehn and K oehn 1986), and D ekw ana (Hall 1988). There are also useful w ord lists for C arijona (O tterloo and Peckham ms.), K ashuyana (Derbyshire 1958), Wai Wai (Hawkins, forthcoming), W aim iri-Atroari (Hill and Hill 1994), M akushi (W illiams 1932; 138-342), A rekuna and A kaw aio (Edwards 1980), D ekw ana (H all 1988), P anare (M attéi-M uller and Henley 1990). O ther w ord lists and dictionaries to which I have n o t h ad access include: M osonyi (1978) for K alijia, a dialect o f Carib; A rm ellada (1943) and A rm ellada and Salazar (1981) for the Pemong group. O th er C arib languages still spoken, and included in K aufm an’s classification, are: Yukpa: Japreria; in a separate Y ukpa G roup Yukpa G roup 26 D esmond C. Derbyshire A kuriyo: G u ian a Branch H ianákoto: G u ian a Branch Salum á: Pawishiana: G uiana Branch N o rth A m azonian Branch Patam ona (Ingariko): N o rth Am azonian Branch Mapoyo: C entral Branch Yavarana (M apoyo group): C entral Branch K alapalo: S outh A m azonian Branch M atipu (N ahukw a): S outh A mazonian Branch A rara: S outh A mazonian Branch H off (1992) gives one additional language, Sikiana (33 speakers). Grimes (1996) gives more details about all o f the languages nam ed above, including locations, dialects, alternate names, degrees o f bilingualism, and populations. 2 PHONOLOGY 2.1 S eg m e n tal There is a solid core o f segmental phonology that is found in m ost o f the 17 lan­ guages listed in table 2 .1 .1 present these segments in tables 2.2 and 2.3. Pending a m ore reliable reconstruction based on well-established c o p a te sets, the non-parenthetical consonants and vowels shown in 2.2 and 2.3 may represent something close to a substantial p art o f the proto-C arib system. The forms in parentheses occur less frequently in som e languages, and may prove not to be in th e proto-system. They are explained more fully in the notes that follow. 2.2 Phoiiotactics and suprasegmentals In m ost C arib languages, syllable structuring is as follows; word- o r phrase-initial: (CK QV (V )(C) elsewhere (except AP, see note 3 to table 2.3): C(C)V(V)(C) In the languages with vowel sequences and long vowels (see §2.1 and note 3 to table 2.3), the syllables in which they occur do n o t have final C, except in D E, which can have a final C or € C followiog a long (or short) vowel (H all 1988:269). In languages in which a syllable-final C occurs, it is usually restricted to a few consonants: in CA , only nasals a n d h, k , ?; in MA, h and n; in WA, p, t, k, m, n, h; in AP, ? a n d / ; and in PA, ?, h, m , n. W ord- or phrase-final C has a very restricted occurrence: none occur in H I other than when apocope o f the weak vowel i occurs; only n in TR ; m, 27 2 Carib Table 2.2 Consonant s y ste m fo u n d in 1 7 C arib languages plosive nasal voiceless affricate voiced affricate voiceless fricative voiced fricative flap lateral approximant bilabial alveolar velar glottal P (b) m t k (k“) (?) (d) n (g) (4>) s (J) (P) (z) (3) (J) palatal (tD (<13) r (x) h (1) w y Notes: (t) The non-parenthetical segments are found in all 17 languages, with the following exceptions: p is lacking in W W and DE; /I is lacking in CA, WA-AT. M A. WA, AP, BA and TX; w and y are lacking in TX, but my only source is Emmerich (1994), who lists all of the phonemic inventory, but may have followed Rjrtuguese usage of symbolizing w as u, and y as /; r is replaced by I in WA, but Jackson (1972: 48) describes it as ‘a reverse flap with lateral opening’ (see note 2 below on r a n d s is lacking in TR (s is in the orthography o f Leavitt (1971, part 1, p. iii), but he describes it as a ‘sound between sh and ch’), and also in TX and WA; h is lacking in AR, AK, AP (where h appears in the orthography, but representing 7), and TX. (2) The quality o f /, ra n d jso u n d s is difficult to identify precisely in Carib languages, since the rso u n d s tend to have lateral releases and tend to be made with the blade, rather than with the tip, of the tongue. Many of the sources do not give much detail. The alveolar flap is the common r sound, and is sometimes described as having lateral release (WW in Hawkins 1998), and in some cases as having an / variant (KA in Wallace 1970, and DE in Hall 1988). WW and HI have both the alveolar /• and palatal j with lateral release. Only / occurs in WA (see note I above); / occurs also in BA and TX in contrast with an alveolar r. Franchetto (1990) reports KU as having a palatalized r, but in Franchetto (1995) she describes it as a voiced velar fricative (with a voiceless variant), confirmed by spectographic evidence, (3) # is the only voiceless bilabial in WW; in H I, it corresponds to p in a few words, but it is usually replaced by h. In K U there is a dialectal variation between h and # , and conditioned alternation between h m d p (Franchetto 1995). Durbin (1977: 34) based the main division of his classification primarily on the change from proto-Carib *p (found in N orthern C arib languages) to h (in Southern Carib languages). (4) The voiceless palatal affricate i / i s reported for 9 languages, and the palatal fricative / for 11 languages; 6 of those languages have both forms in contrast. (5) The voiced fricatives are rare: P occurs only in C R , but is reported as a variant o f w in 28 D esmond C. Derbyshire Table 2.2 (com.) Notes (cont.) MA and DE; alveolar z occurs in CR, AK , AP and BA; palatal j occurs in AK, A P and BA; and the voiced affricate d j in CR, A R and AK. (6) The labialized velar is reported as a distinct segment only in CR. Descriptions o f other languages show a two-segment sequence o f kw, some o f these may represent a single k" segment. (7) The voiced stops b, d and g are reported as marginal in some languages (e.g. CA (Hoff 1992), and HI, in which g does not occur at ali (Derbyshire 1985)). All three stops occur in CR, AK, BA; only b and d in WA-AT; only palatalized rf'in KU; only b and g in TX; and g in KA, but only as a noncontrastive variant o f k. (8) The glottal stop is reported for 6 languages: CA, CR, KA, AP, DE and PA; in HI it occurs as a variant o f voiceless consonants that precede nasal consonants (m, n, jt} or their voiced stop counterparts (6, d, cP). WA-AT has pre-glottalized nasals 'm and 'n which contrast with m and n, and glottalized (i.e. checked) vowels that contrast with non-glottalized vowels and with each other; there is also non-contrastive glottalization which occurs word finally after short vowels. With all this glottalization going on in Waimiri-Atroari, it is o f interest that Hill and Hill (1994) do not show glottal stop as a separate segment in the phonemic inventory, on the grounds that it does not fill a normal C role in the language. It always occurs in conjunction with a vowel or nasal consonant. (9) In CA, [h] occurs in word-medial syllable-final position in some dialects and [x] in other dialects, as a result o f a historical process o f syllable reduction (Hoff 1968, 1992; Gildea 1995a). n, w, y in WW; m , n, and their glottalized counterparts in WA-AT; h and g (a condi­ tioned variant o f n) in M A. Stress, to my knowledge, is contrastive only in PA. It is predictable in terms o f syllable structure for some languages: in CA, it occurs on any second heavy syllable (one w ith a long vowel, diphthong, or final C) counting from the left; otherwise on the final syllable (H off 1992). Such languages usually also have the prosodic vowel length described in §2. 1 and note 3 to table 2.3. In other languages stress is normally on the final (WW, PA) or penultim ate (WA-AT, A P) syllable. In H I, prim ary stress occurs on final o r penultim ate syllables, varying according to the dialect; secondary sttïss occurs on heavy syllables, i.e., those with final C or (prosodic) long vowels. 2.3 Morphophonology The following processes are found in C arib languages, usually occurring at m or­ pheme boundaries within a gram m atical word; vowel harm ony; other assimilatory processes; deletion; insertion; and metathesis. Vowel harm ony across m orphem e boundaries is reported for the following lan­ guages. In K A , W W and H I i is replaced by u when u occurs in a preceding or fol­ lowing syllable; and in KA and H I, o is [«placed by a in prefixes to harm onize w ith a in a stem-initial syllable. In WA-AT, under certain conditions (listed in Hill and 2 Carib 29 Table 2.3 Vowel system fo u n d in 17 Carib languages high mid low front central back i e j (ë) a u o (Ö) Norn: (1) The non-parenthetical segments are found in all 17 languages, with the following exceptions: / is non-contrastive in HI, occurring only as a variant o f e, preceding or following a palatal consonant; il occurs in some D E examples (Hall 1988), b ut it is n o t clear to me front the description o f the vowels whether it has phonemic status (2) M id-central vowels are restricted, occurring only as: e in T R , WA, DE and PA; and as ö in A R and AK. T he schwa is reported for K U in Franchetto (1990), but is replaced by i in Franchetto (1995). The schwa is also found in BA examples in Souza(1995), but it is not incl uded in the list o f vocalic sounds in footnote 2 o f that paper. (3) Long vowels contrast with short vowels in some languages (TR, CR, WW, WA-AT, MA, AR, AK and DE). In other languages there is prosodic lengthening of vowels in CV syllables, predictable in terms o f the syllable patterning in words and phrases (CA, HI; in one CA dialect a few suffixes have as part o f their form the lengthening o f the preceding vowel, attd this can perturb the regular prosodic pattern). In AP, nasalized vovwls are long in contrast wfth non-nasalized vowels. Languages that have contrastive long vowels also usually have sequences o f two non-identical vowels. AP has sequences o f three vowels, with each V constituting a separate syllable (Koehn and Koehn 1986: 122). Hill 1994), i can be replaced by / o r e before a syllable containing a, e, or ». In K U , vowel harm ony is restricted to the vowels o f two person-m arking prefixes and one possession suffix (F ranchetto 1995). F or BA, in which voiced consonants are many an d occur frequently, Souza ( 1995) Imports a complex system o f consonantal harm ony, with predictable patterns o f alternation, dissimilation and spreading o f the voicing feature, affecting lexical items, prefixes and suffixes. O ther assimilatory processes include the following. In m ost languages for w hich m orphophonological processes have been reported, there is consonantal fusion (e.g. n -l-j;> ji) and palatalization (e.g. r > < /b e fo r e i and/or after /)■ Palatalization and affricativization processes are reported for K U (F ranchetto 1995). In WA-AT both palatalization and labialization o f k an d h occur; syllable-initial k’’ and can occur as the result o f the optional deletion of the vowel i between k or h an d y , and k'" and I f optionally occur when the vowel i immediately follows the initial C; loss o f palatalization also occurs in WA-AT: ji> n when followed by a C. 30 Desmond C. Derbyshire In CA, a morpheme-fina! n before p , t o r k results in the sequences mb, nd or the velar nasal q respectively. In some CA dialects there has been loss o f word-medial syllables which h as resulted iti breathiness of the vowel in the preceding syllable, which sometimes takes the form o f a velar or glottal fricative in syllable-final posi­ tion (H off 1992). In T R , n > jj before k and w ord-inally; n > m befo rep ;p>4> following k; and k > X following h. In WW, k > h before t, k, s, i f at m orphem e boundaries (cf. H I below); # > A in consonant clusters and h then deletes if it is preceded by tw o syllables, a long V, a CC cluster o rm . In H I, morphem e-initial n > p following e; r > j before e; and s and i > / a n d t f respectively when they follow a syllable containing the vowels e o r a; morphemefinal k > h when it precedes a voiceless non-syllabic other than h\ and morphemeinitial h > p following m orpheme-flnal m or h. (The last two rules avoid non-perm itted CC clusters in a word.) In WA, p > h beforep, m and w; p > m before n; t> h before t, I and n; t> n before m; k > li before k; and k > a before m and n. In AP, a sequence o f V plus nasal C in morphem e-initial position, and preceding another C, becomes a nasalized V; any morpheme-final stop > ? before C. D eletion processes include the following. High vowels, especially f and u, are deleted in several C arib languages when they occur morpheme-finally following a non-clustered C and preceding a morphemeinitial V o r unclustered C, provided th a t the resulting CC is a perm itted one. This process is reported for WW, H I, WA-AT, M A and AP. W hole syllables containing these weak vowels can delete in some languages (CA, M A). For fuller discussion, see G ildea (1995a). In WW, morphem e-initial y deletes following s and if. In H I, a sim ilar process occurs; morphem e-initial y deletes when it follows a voiceless alveolar or palatal C, but in this case h is inserted before the C; this results in ty > h t, sy> h s, f y > h f ( t f and y d o not com e together at morphem e boundaries). In M A, certain C i syllables (pi, ti, ki, si) are weak, and when certain alBxes are added the whole syllable may be deleted or be reduced to h. A few insertion processes have been reported. In WW, when i f h preceded by a long V, the V is shortened and h is inserted fol­ lowing the V. In HI and M A, there is insertion of y between a prefix that consists of a single V (o or a in HI; u o r a io MA) and a stem-initial V (only a V o f the same quality as that o f the prefix in HI). A similar /-insertion process occurs in PA (Gildea p.c.), (See 31 2 Carib Gildea 1998, for a diachronic hypothesis which traces this y back to a morphological inverse m arker in proto-Carib.) In M A, ther« is also insertion of t between the thirdperson prefix I- and a stem-initial vowel. In HI, there are also epenthesis rules for the insertion o f a brief vowel sound between syllables when certain CC sequences occur. D E has the following m etathesis rules at m orphem e boundaries: i- + t > ty > tj' i- + k > ky > tj i- + s > sy > J i- + m > my > ji i- + n > ny > ji i- + r > ry y > HI has one invariable m etathesis rule: m orphem e-final j o r /follow ed by h changes to hs an d / i / respectively (cf. the above deletion o f y process, which also entails inser­ tion o f h). There are other m ore idiosyncratic changes, where certain CV sequences become VC when followed by certain suffixes o r particles; hu>uh; h i> ih ; ti> it. M A has one similar process th a t regularly occurs at m orphem e boundaries: hi and /iu > ih an d lA respectively. 3 M ORPHOLOGY, PARTICLES AND PRONO UN S This section describes: inflectional m orphology (§3.1), derivational m orphology (§3.2), the particle word class (§3.3) and pronouns (§3.4). Verbs in C arib languages are especially com plex morphologically, and nom inats and adverbials also abound in prefixes and suffixes, both derivational and inflectioiial. T he structure o f the verb in m ost C arib languages is: Person prefix - detransitivizer prefix - root - derivational suffix tense/aspect/m ood/num ber suffix T he ergative languages have an A person m arker attached to an ergative marker (§3.1.1). F or M A, A bbott (1991) treats the whole sequence as a suffix o f the verb, in final position following the T A M N suffix. For K U , Franchetto (1990) treats the sequence as a separate word, im mediately following the verb. 3.1 Inflectional morphology There are four main types o f inflectional m orphology discussed in this subsection: person-m arking affixes on verbs and nouns, and on som e adverbials and postposi­ tions (§3.1.1); tense, aspect, m ode and num ber suffixes on verbs (§3.1.2); possession, tense and num ber suffixes on nouns (§3.1.3); and inflectional suffixes on locative postpositions (§3.1.4). 32 Desmond C. Derbyshire 3.1.1 P e r s o n -m a rk in g affixes o n verbs, n o u n s , a d v e rb ia ls a n d p o s tp o s itio n s In o rd er to get a clear picture o f the person-m arking systems o f C arib languages, it is necessary to distinguish the three gram m atical functions to which the verb person markers relate: intransitive subject (S), transitive subject (A) and object (O). There is only one prefix per verb. A characteristic o f the m orphosyntax o f these languages is that they show varying degrees o f ergatively organized m orphology and syntax; in some, ergativity is strong, in others, it is weaker. (See §4.3 for further discussion.) The m ajority o f the languages have split systems, partly ergative and partly accusative. The person-m arking systems form four distinct sets that can be classified as follows: Transitive A -oriented prefixes Transitive 0-oriented prefixes Intransitive S^-oriented prefixes Intransitive S^j-oriented prefixes (See Derbyshire 1991 and 1994 for the H I system based on this classification; and G ildea 1998, for an alternative analysis which includes an inverse system and a sum m ary o f o th er possible analyses.) Sjj prefixes occur with one set o f intransitive verbs (in general, ‘active’ verbs), while the prefixes occur with another set (in general, ‘stative’ verbs). However, the sem antic distinction between ‘active’ and ‘stative’ does not always determine to which set a particular verb belongs. Some languages listed in table 2.5 have only a partial set o f S^-oriented forms (restricted to one o r two persons). In PA the distinc­ tion is n o t prim arily between and S„, b ut between two tense/aspect paradigms; ‘p ast perfective’, which takes the A o r S^, prefixes, and ‘non-past-perfective’, which takes the prefixes: a - ‘IS ’, a- ‘2S’, y - ‘3S’ (only the first and second person forms are the same as those in the O set shown in table 2.5; see G ildea 1998). G ildea (1998) has reconstructed the proto-C arib forms o f the person-marking prefixes based on cognate sets from nineteen languages. In tables 2.4 and 2.5,1 use his reconstructed forms, but have modified the cognate sets to conform to the data in my own sources and to include only the languages on which this study is based. I exclude A K , A R , M A and KU, the strongly ergative languages, whose personm arking systems are somewhat different (see table 2.6 for the M A and K U sets), and also two languages for which my sources do not provide the necessary data KA and TX. There is one category o f person not shown in the tables: first person exclusive (1 + 3); except in M A and K U (see table 2.6), the 1 + 3 prefix is identical in form and function w ith third person, and a free pronoun ana (o r cognate) ‘1 + 3’ is always present. In PA, the 1 + 2 prefix is also identical with third person 33 2 Carib Table 2.4 Comparative set o f transitive A -o rie n ted p ra xes which occur when the O is third person Language iA 2A I+2A 3A i-/0s-/kà- m(i)- s(i)- s(i)iww-/i- m(i)m(i)mm(i)- kis(i:)klt(i)kt(i)- n(i)-/kin(i)n-/in(i)n(t)n- t{i)- m{t)- n(i)- w(i)-/tW(i)-/0- m(i)m(i)- W(})-/0- m(i)- proto AP BA CA CR DE HI PA TR WW WA ki-/kize- n(i)- n(i)n-/kinn(l)-/0-/ji- k-/kut-/h-/s- n{i)-/n{i)-/mën-/k(})-/0 - Notes: The BA prefix s- ‘lA /30’ is presented here as A-oriented, i.e., with t h e s - referring to ‘iA’; some linguists present it as O-oriented, i.e., with the j- referring to ‘30* (e.g. Souza 1994: 23-4; ‘the patient [= 0 ] is always the term referred to’ [my translation]). Souza, however, makes exceptions for the 2A/30 and I + 2A/30 prefixes, which refer to the A (‘subject’ is the term she uses). Pending a fuller description of BA, I tentatively place s- ‘1A/30’ in the Aoriented set. (both in the I + 2 A /3 0 prefix (table 2.4) and the 3A/I + 2 0 prefix (table 2.5)), thus collapsing the distinction between 1 + 2 and 1 + 3 in the verb prefix, retaining it only in the free pronouns (table 2.11). T h e intransitive verb prefixes are identical, o r very nearly so, with m ost o f the forms in table 2.4. The only significant differences are with the first person forms in five o f the languages: BA is n o ‘IS ’ (cf. s~ikâ- ‘1A’; according to Souza (1994) there distinction in BA); CA /0- ‘1A’); H I k f ‘lSg'(cf. h >-/í - 0- ‘ISj,’(cf; s f i j - ‘lA ’); W W k f t j - ‘lS^’ (cf. ‘lA ; the i- variant occurs w ith all C-initial stems in the transitive set, but in the intransitive set, Í- occurs with only a small subset o f C-initial stem s); and PA w (i)- ‘IS^’ (cf. i f / / - ‘lA’). In PA there is also a suffix, -ne ‘1 + 2 3 ’ (cognate with the M A suffix -ni ‘1 + 2 S ’ in table 2.6), which occurs In a subset o f m ain clauses th a t are part o f the reanalysis hypothesis o f G ildea (1998); see also §4.3. T h e absence o f third-person prefixes in table 2.5 requires explanation. In m ost o f these languages the O -o rien ted prefixes function as p a rt o f a system governed by a person/agentivity hierarchy; first person and second person o u tra n k third person; w hen the higher-ranked person in a transitive clause is th e subject, the A -oriented prefix is used, m arking the A; when the higher-ranked person is the 34 Desmond C. Derbyshire Table 2.5 Comparative set o f transitive O-oriented prefixes which occur when the A is third person Language 10 20 1+ 20 proto AP BA CA CR DE HI PA TR WW WA *u(y)y(i)i0-/yyi0-/yr(o)0-/yy./yjy. o(y)y-/i- *a(y)00-/Í-/Ía(y)-/oëyi0(d)o(y)-/a(y)a(y)a(t)a(w)ë(w)- *k(i)k(i)kk(i)k(i)k(u)k(i)n(i)k(i)k(i)k(i)- object, the O -oriented prefix is used, m arking the O. T he O -oriented set occurs when the subject o f a transitive verb is ‘3’ and the object is ‘1’ o r ‘2’ or ‘1 + 2 ’. W hen only first and second persons are involved (1 A /2 0 o r 2 A /1 0 ), the person m arkers vary am ong the languages, and usually one o f the participants is expressed w ith a free pronoun. (This is not, however, the case in CA , where both lA /20 and 2A/10 have the sam e préfixai form , and use o f a free pronoun does n o t disam biguate; only situation and context can do so.) W hen b o th subject and object are third person, neither o u tra n k s the other; the prefix then used is from the A -oriented set. Some features o f these O -oriented transitive sets have been explained in term s o f an inverse system (D oris L. P ayne 1993; G ildea 1994, 1998). The intransitive S^, prefixes are identical, or nearly so, w ith the O forms in table 2.5 in four languages: CA, DE, T R and WA. In two languages, A P and HI, pre­ fixes occur only for second person and the forms are identical with the O forms for that person; there is no prefixes occur for all other persons. In two languages, CR and WW, set; all intransitive verbs take the same prefixes, which in m ost cases are from the A/S^, set; in CR, however, the yi- ‘1S’ prefix is from the 0/S^ set. In the remaining languages inform ation is lacking as to w hether there is a distinct intran­ sitive Sjj set or subset of prefixes. The following examples ( la - ld ) from Dekw ana illustrate the person-m arking prefixes on transitive and intransitive verbs (taken from H all 1988): (la ) Transitive A -oriented; w-edant(ö)-a lA/30-m eet-PRES ‘I m eet him /her.’ 2 Carib 35 Table 2.6 Person-marking affixeslclitics in ergative languages M akushi Kuikiiro Person S/O prefix or suffix A suffix S/O prefix/clitic A post-V clitic 1 u-/u(y)a(w)-/a(y)-n(i) ‘S’ u(y)- ‘O ’ i-/aw- ‘S’ i(t)-/0 - ‘O ’ (only free form anna) -u-ya -0 -ya ueku- u-héke e-héke ku-péke i-/0 - i-héke ti- (free form Usure) ti-héke 2 1+2 3 1+3 (lb ) -0 -i-ya Transitive O -oriented: y-edant(ö)-a 3A/10-m eet-PRES ‘H e/she m eets m e.’ (Ic) Intransitive S -oriented: w -amo-a IS -cry-P R E S ‘I am c r y i n g .’ (Id ) Intransitive S -o rien ted ; y-a:wo-a 1 S-swell-PRES ‘I am sw elling.’ T he strongly ergative languages have different person-m arking systems, although the form s are clearly cognate to the forms found in table 2.5, which shows the Ooriented set o f the mixed ergative-accusative languages. The four ergative languages fall into two groups geographically: A R , A K and M A, which are located north of the A m azon in G uyana and Brazil; and K U, located in the Xingu River basin well south o f the A m azon, M A and K U are shown as representative o f the two groups in table 2.6. The m ost striking difference between these two sets and the A- and O -oriented sets in tables 2.4 and 2.5 is th a t the (ergative) A forms are suffixes (M A), or are post­ posed to the verb as p art o f a separate word, com posed o f the pronom inal proclitic and the ergative enclitic (K U ). In general, the same forms are used for m arking both the A (suffix or clitic) and S/O arguments. In M A , b u t not in KU, the 1 + 2S form -n (i) is unusual in two ways: it is a suffix, and its form is totally different from the corresponding 1 + 2 0 prefix, and also it does n o t follow the norm al C arib pattern for 1 + 2S. (C f the PA suffix -në ‘1 + 2 S ’ described earlier.) There are also different forms in MA for 2 S /0 {a(w )-la(y)~ prefix) an d 2A (-e-ya suffix). 36 D esmond C. Derbyshire T he A form s in table 2.6 include the ergative clitic: -ya in M A , and -h ik e in K U . (In o th er C arib languages -ya is cognate w ith form s like the H I postposi­ tion -wya ‘to ’ (indirect object m arker) o r ‘ergative/agentive’; -h ik e is probably cognate with form s like the H I postposition hoko ‘concerning, about, w ith’.) These ergative clitics occur on the free form noun when there is one, o r are attached to the pronom inal suffix (M A ) o r proclitic (K U ), as described above. In M A , the free form an d person-m arker never co-occur. T he form s are clearly not cognates with each other; the M A -ya, however, is cognate w ith clitics th a t som e­ times occur w ith an ergative m eaning in the languages listed in tables 2.4 an d 2.5. In M A the -ya clitic does not occur for ‘1 + 2 ’ person, but -héke does occur in K U , in the form ku-péke. K U is unusual am ong C arib languages in having a ‘1 + 3’ free form pronoun, tisûre, th a t is n o t a cognate o f a m a (see table 2.11 ), and whose corresponding prefix {H-) is not the third person. T he person/agentivity hierarchy is n o t applicable to the person-m arking in the ergatively organized systems o f these languages, but in K U it is relevant to w hat Franchetto (1990) calls the ‘interactive m oods’, in which ergative constructions occur with third person A, an d accusative constructions occur w ith first person and first person inclusive A (w ith first person exclusive a n d second person A there is free alterna­ tion between th e ergative and accusative patterns). See §4.3 for further discussion on ergativity Examples o f person m arkers in M akushi (2a-c) and K uikùro (3a-c) are; (2a) i-kon ek a-hp i-u-ya 30-make-PAST-1A - erg ‘I m ade it.’ (2b) u-koneka-hpi-i-ya I O-make-PAST-3 A - erg ‘H e m ade m e.’ (2c) u -k om am i-h p i 1S-stay-PAST ‘I stayed .’ (3a) u-ta-li-igo 10-hear-PUNcr-FUT ‘H e will hear me.’ (3b) u -k a tju n -ta ri 1S-work-coNT ‘I am w orking.’ léha i-héke aspect 3A - erg 2 Carib (3c) i- 37 k a t|u n -ta fi-k o 3S-work-coNT-PL ‘They are w orking.’ Possessed nouns, certain derived adverbials, and some postpositions (mostly locatives) have prefixes th a t refer to the possessor (which is the underlying S o r O in forms derived from verbs; see §3.2). The forms are nearly identical with the O-oriented prefixes in table 2.5, for the languages listed there, and the S /0 prefixes in table 2.6, for the ergative languages. The m ain difference is the addition o f a third-person prefix, which does not occur in the O -oriented set, for reasons explained earlier in this section; the form o f th e prefix is f-, eh ‘3PO SSR’ in m ost languages (recon­ structed as proto-C arib *y- by G ildea 1998). There is also a third-person reflexive possessor form, reconstructed by Gildea as *t- ‘3 R E F U , and having that same form In nearly all languages. See §3.1.3 for possession, tense and num ber suffixes on pos­ sessed nouns, and §3.2 for derivational affixes which form nouns and adverbs. Examples from H ixkaryana, a mixed ergative-accusative language (4 a-b ) and M akushi, an ergative language (5), are; (4a) (4b) (5) 3.1.2 ay-am ori 0-a m o ri i-h a n a ri t-hanari 2-hand 3-hand 3-ear 3REFL-ear ‘your h and’ ‘his h an d ’ ‘his ear’ ‘his own ear’ ro-hyawo i-hyaw o tà-hyaw o 1-with 3-with 3REFl,-with ‘with m e’ ‘w ith him ’ ‘w ith h im se lf’ u-w ahkari i-w ahkari ti-p o n 1-axe 3-axe 3REFi-clothes 3-behind ‘my axe’ ‘his axe’ ‘her ow n c lo th es’ ‘behind h im ’ i-p ik iri T ense, a s p e c t, m o d e a n d n u m b e r suffixes o n verbs Verb suffixes in the indicative m ood are portm anteau forms expressing tense, aspect, m ode and number. (PA, an d possibly some other C arib languages, do n o t have the num ber com ponent.) There is another set o f suffixes for the imperative (including hortatory) m ood; these have com ponents of num ber and m otion, and differ in form according to the person o f the subject. H ere I will describe only the indicative T A M N suffixes. T he mixed ergativeaccusative languages for which T A M N d a ta are available vary widely as to the number, tense an d aspect distinctions th a t are made. H off (1968) shows nine for CA. G ild ea (1998) reports three p ast tenses w ith no aspectual distinction for PA. 38 D esmond C. Derbyshire Table 2.7 Som e tense-aspect suffixes in m ixed E R G -A C C languages Language Past or immediate past Recent past CO M PL/CON T Distant past C O M PU C O N T AP -no BA -da -ma,-wa -dai,-raki,-taki CA -i (?) -yakon -ya,-sa,-e -yae,-yanë Nonpast -V(nas)ko CR -0 -ne DE -i,-anô -ne.-akene HI -no -yakoZ-yaknano -ye/-yakoni PA -yah -i, -e -ya(ke) TR -0 -ne,-0 -kën -a,-ta,-tai -yaha,-aha,-yano -ya(-e) WW -ye,-y,-e,-yakjie,-ekjie -(ya)si,-(a)si, -(e)si WA -ne -ya,-j Notes: ( ! ) The distinction between recent and distant past is neutralized in DE, WW and WA. I am assuming this is also the case in BA, CA and CR, but the data in my sources may not be complete. (2) In DE, in the two past tenses, the first Ibrms -r and -ne also m ark completive aspect, and the other two forms, -anö and -akene, m ark incompletive. (3) The completive vs continuative aspect distinction for the recent an d distant past tense fontis in AP and HI is indicated by the slash (/). (4) In CA. the form shown lor immediate past, -i, does not have as its primary value ‘past’ but a m odal value ‘realis’ (Hoff 1968: 172^3, 177-9; 1986: 81-6). H aw kins (1998) describes three past an d two nonpast tenses, w ith no aspectual distinctions, for WW. D erbyshire (1985) records five p ast an d two nonpast tense paradigm s for H i; there are three basic past tenses, the recent past and distant p ast having aspectual distinctions; the two nonpast tenses have a m odal distinc­ tion. I present the suffixes found in my sources in table 2.7. These are part o f G iidea’s (1998) Set 1 TA M (N ) suffixes, except for the A P nonpast form , which Gildea con­ siders an innovation. His Set 2 suffixes, which occur in a few C arib languages, are also innovations, resulting in inost cases from the reanalysis o f some o f the nominalizing suffixes (see table 2.10; and discussion in §4.3). I suspend final ju d g e­ m ent on the status o f G iidea’s Set 2 suffixes until C arib languages have been sufficiently docum ented so that his reanalysis hypotheses can be thoroughly tested. The collective num ber com ponent o f the T A M N suffixes, which refers to the S, A an d /o r O referents, constitutes a cognate set in all the languages for which data are available; -to (AP, CA, DE); -tJ (e ),-tfo (w ) (HI); -të,-ti,-nto (TR); - ifaw,-lf,-t, 2 Carib 39 -tu QN'Wy, -tërtëo (WA). Some o f the -yaC {C )Y forms in table 2.7 are m orpholog­ ically complex, as can be seen from the placem ent o f the num ber suffix; cf. the fol­ lowing noncollective (6a, c) and collective (6b, d) H I verb forms: (6a) n-am jek-yako 3-hunt-REC.PAST + COMPL ‘He went h u n tin g.’ (6b) n-aiiuek-ya-t/o-ko 3-hunt-REC.PAST-C0LL-REC.PAST + COMPL ‘They went h u n tin g,’ (6c) k-om ok-yakoni 1-COme-DlST.PAST + CONT ‘1 u s e d t o c o m e .’ (6d) m -om ok-ya-tj’-koni 2-COme-DIST.PAST-COLL-DIST.PAST + CONT ‘You all used to com e.’ Sometimes the collective suffix (6f) replaces the noncollective form (6e) and consti­ tutes the com plete tense-aspect marker, again illustrated from HI: (6e) 0 -wayeh-no 3-die-IMM.PAST ‘H e d ied.’ (6f) 0 -w ayeh-tJow i 3-die-l MM. PAST + COLL ‘They died.’ T he ergative languages vary significantly in respect of both the TA M N system (fewer and different categories) and the forms o f the suffixes (or absence thereof). M A has three tense distinctions, analysed by A bbott (1991) as: ‘past’ (previous day or earlier); ‘proxim ate’ (indicated with an auxiliary, n o t a suffix, and referring to a time before, during, o r after utterance, w ithin a 24-hour period); and ‘universal’ {no overt form , and referring to present o r future). There is also a distinct per­ fective/perfect aspect which can occur alone or co-occur with the past (which occurs on an auxiliary) and proxim ate tenses. T he copula a nd two auxiliary verbs function in the M A TA M N system. (O ther M A aspect markers appear to be outside the system; see A b bott 1991.) K U has at least three tense-aspect distinctions: continuative, punctiliar and future (the latter two can co-occur). T here is an o th er clitic, -kh a ‘aspect’, w ithout 40 Desmond C Derbyshire further specification. There is also a suffix which functions as both nominalizer and perfective aspect (F ranchetto 1990). T h e forms in these two languages are: M akushi Kuikiiro Past -hpi Continuative - ta d Proximate V E R B + pi/j + C O PU LA Punctiliar -lël-jiarë Universal U N IN F L E C T E D VERB F uture P erfective/P erfect-io/i -igo Perfective + Nom inalizer -përë Aspect -leha The collective suffixes are; K U -ko; M A -kon, -nikon. Examples showing full in tran ­ sitive verb forms with only noncollective {7a, 8a) and collective (7b, 8b) suffixes (where (7) is M akushi and (8) is K uikuro) are: (7a) l-pih 3S-at aw -entam ohka-hpi 3S-eat-PAST ‘H e ate him .’ (7b) aw -entam ohka-hpi-kon 2S-eat-PAST-C0LL ‘You all ate.’ (8a) t /u e hard i-k atju n -tari 3-work-coNT 'H e is w orking hard.’ (8b) tjxie hard i-katfun-tari-ko 3-work-coNT-coLL ‘Tliey are working hard.’ 3.1.3 P o ssessio n , te n se a n d n u m b e r suffixes o n n o u n s Possession param eters include inalienable/alienable and present/past time. N oun suffixes indicate possession and time. The person o f the possessor is indicated by a prefix (§3.1.1). Collective num ber is m arked by a postposed particle (H I komo) for simple, non-derived nouns and also for m any derived nouns; other H I derivations take the suffix -yamo ‘C O LL’. (See §3.2.2 for examples o f the derivational m or­ phology.) I will first illustrate the nom inal inflection system from H I, and then note some significant differences that have been reported for o ther C arib languages. a) H ix k a r y a n a n o m in a l inflection system The following suffixes occur to m ark possession in HI. T he basic form s are lexically conditioned, but m ost possessed noun stems occur with -ri or -ni: 41 2 Carib basic possession: -ri, -ni, -ti, -tfe, -0 past/form er possession: -th k U tP iiri, -tlioltflio ‘simple past/form er possession’; an d -nhiril-nho ‘rem ote past/form er possession’ (the alternate form s in the first two pairs are phonologically conditioned; the -tho and -nho forms occur w ith the preflx.es for first person, first person exclusive, and third person with a preceding N P; the -thiri and -n h k i form s occur with the other person-m arking prefixes). Inalienably possessed nouns (mainly body parts and kinship terms, but also a few others) norm ally have the possession suffix. (There is a suffix -nano ‘depossession’ which replaces the possession suffix on inalienably possessed nouns, and gives the item m ore general reference, w ithout specifying any possessor, but it is rarely used in HI, except for one o r two items. Cognates o f this depossession suffix occur in other C arib languages.) T he possession suffix is obligatory (and always the -r i form) in three H I derivational processes (see §3.2.2). Alienably possessed nouns usually occur w ith the -n i form when a possessed form is required, b u t -ri occurs w ith some nouns, including kanawa ‘canoe’ (9), which is alienably possessed; the others in (9) are inalienably possessed. H I non-derived possessed nouns: (9) i-kanaw a-ri ro-yo-ni k-oso-ti o -h e-tje 3-canoè-POSSD 1-mother-POSSD 1 + 2-name-POSSD 2-wife-POSSD ‘his c a n o e ’ ‘my m other’ ‘o u r (tw o) nam es’ ‘your w ife’ . C om pare the use o f the collective particle: ik m a w a ri kom o ‘their canoes’; kosot homo ‘o u r (three o r more) nam es’. T he suffixes occur when there is a free form possessor: romuru kanawari ‘my son’s canoe’; co w tiyo so ti ‘my bro th e r’s nam e’. The only prefix that co-occurs with a free form possessor is y- ‘G E N ’ when the possessed noun is vowel-initial, as in y-osoti (G E N -nam e) in the preceding example. H i past/form er possession: rokam w atho ‘my old (or form er) canoe’; ikcm aw m hiri ‘his old/form er canoe’; ohetjenhiri ‘your form er wife’; ro m u m betfenho ‘my son’s form er w ife’. b) Variations in o th e r C arib languages Some languages have a very similar set o f suffixes to H I for marking basic posses­ sion (AP, CA, W W ). The past/form er possession suffixes in two of these languages are: AP -hpiri or - Vnpiri; W W -thiri, -tho, -nhiri, -nho (the same forms as H I, except for the H I initial í / variant). In CA, the suffix sequence -ri~mbo (-POSSD-PAST) corresponds most closely to the HI, A P and WW forms, but -mho is not restricted to possessive constructions, and its semantic reach is wider than ‘past’ (H off p.c.). 42 Desmond C Derbyshire D E suffixes are -dii, -i and a few less com m on variants; the -dii is probably a cognate o f -r i which occurs in other languages. There are three collective suffixes and these replace the noncollective suffix: -tomo m arks plurality o f the entity referred to by the noun; -komo m arks plurality o f the person o f the possessor; and the sequence o f the two takes the form o f -tonkomo. T here is also a past suffix, -hii, which co-occurs with, and precedes, the possession suffix -dii. T R possessed nouns usually have no overt suffix, although - ri optionally occurs under certain conditions; the person prefixes are the primarj- indicators o f posses­ sion; the past/form er possession suffix is -hpë; collective suffixes are -konl-komo and -tonl-tomo. WA possession suffixes are -n, -er, the collective suffix is -tern; past/form er posses­ sion suffixes are: -( V)tpë, -( V)npë, -(V )tp i, •( V )npi The ergative language MA has the possession suffixes -ri and -ri, with less com mon variants -si, -ni, -ka, -no. T he collective suffix is -kon. The past/former possession suffix is -cihpi on non-derived nouns and -hpi (the verb past-tense suffix) on derived nouns. There are m any obligatorily non-possessed nouns in C arib languages, including all nam es o f plants, animals, persons and natural phenom ena. For some o f these a generic noun th a t has possession m arking is often used as a classifier, with or w ithout the addition o f the specific noun, e.g., HI rokni (kaykusu) ‘my pet (a dog)’ (r-ok-ni 1-dornestic animal-POSSD, kaykusu ‘dog’). See Koehn (1994) on A P gen­ itive constructions; and Carlson and Payne (1989), who first described this system as ‘genitive classifier’, established a typology o f such systems, and provided exam ­ ples from AP, M A and PA. Carib languages tend to have between ten and twenty o f these classifiers, and it is the only type o f noun classifier system found in the family. 3 .1.4 In fle c tio n a l suffixes o n lo c ativ e p o s tp o s itio n s A t the end o f §3.1.11 noted that locative postpositions could take person-m arking prefixes. In C arib languages there is a set o f locative suffixes that occur with'simple, one-syllable postposition stems, and also on m ore complex locative stems. These carry basic locative meanings such as ‘in/on’, ‘to/into/onto’, ‘from ’, ‘(passing) by/through’, and the forms vary according to the meanings o f the nouns which they govern. The prim ary semantic distinctions are usually: liquid (e.g. water), flat surface, open area, enclosed place. M ore specific locative m eanings result when the basic suffixes are attached to complex stems. Table 2.8 shows part o f a basic set from each o f four languages (AP, H I, M A , WW). The distinction between ‘open area’ and ‘enclosed place’ is somewhat fuzzy. It is only specific for M A (A bbott 1991). T he general pattern, however, is clear. The final 43 2 Carib Table 2.8 Basic sets o f locative postpositions and inflectional suffixes in fo u r languages Liquid in into from Flat surface on/at to/onto from Open area in/on/at to/into from Enclosed place in to from Apalai Hixkaryana Makushi Wai Wai kua-o kya-ka kua-e kwa-wo kwa-ka kwa-ye ka ka-ta ka-pai kwa-w kwa-ka kwa-y po po-na po-e ho ho-na ho-ye po po-na po-i <|>o ^ -n a <i»-y ta-o ta-ka ta-e ta-wo/ya-wo ta-ka/ya-ka ta-ye/ya-ye ya ya-plh ya-pai ya-w ya-ka ya-y a-o a-ka a-e ya-wo ya-ka ya-ye ta ta-pih ta-pai ya-w ya-ka ya-y suffixal form s {-na, -e, -o, -ka, etc.) constitute p a rt o f the basic set of locative suffixes (there are a few other basic forms); and they are cognates in three o f the languages (M A, the ergative language, being the exception in some o f its forms). The pre­ ceding forms (the locative stem s) reflect the four prim ary semantic distinctions, and these seem to be cognates in all four languages. M ore complex locative stems occur with the same final suffixes, and am plify the prim ary set o f semantic distinctions. These facts can be seen in the following HI exam ples (Derbyshire 1985; 205-19 lists tw enty-one semantically conditioned sets in HI): (10a) asa ra a tra il y -a h e ta -w o O E N -e d g e -a t ‘a t t h e e d g e o f t h e t r a i l ’ (10b) w ew e tr e e m ah y a -y e re a r-fro m ‘f r o m b e h i n d t h e t r e e ’ 3.2 Derivational morphology C arib languages are rich in derivational m orphology: verb roots derived from nouns and from verb roots; com pounds o f verb and noun roots; nouns derived from 44 D esmond C. Derbyshire nouns, verb stems, adverbials and postpositions; and adverbials derived from nouns, verb stem s and adverbs. In this section I will first describe some verb deriva­ tional affixes (§3.2.1), followed by nominalizing affixes added to verb stems to form nouns (§3.2.2), and then some adverbial derivations from noun and verb stems (§3.2.3). 3.2.1 V erb d e riv a tio n a l affixes In m ost C arib languages, only one verb derivation category occurs as a prefix; the detraositivizer, which is added to a transitive stem and carries the meanings o f ‘reflexive’ o r ‘reciprocal’, or simply Intransitive’ which is often best translated as a passive in languages like English. There are many categories which occur as suffixes. Table 2.9 shows the m ore frequently occurring ones grouped into three main types: valency changers (causative o f an intransitive verb and causative o f a transitive verb), aspect m arkers (ingressive, term inative and cessative), and word class chang­ ers, which derive verbs from nouns; one set of suffixes forms intransitive verbs, and the other two sets form transitive verbs w ith meanings like: benefactive or pro­ ductive for one set, and malefactive o r reversative for the other set. M any o f the forms in the eight languages are clearly cognates. Ergative M A (A bbott 1991) has one construction that is quite different from the o ther languages: the form ation o f causatives o f both transitive and intransitive stems is prim arily by means o f a separate verb, em apuhti ‘cause’ - the subject o f which is the causing agent - and a preceding subordinate clause in which the causee is the subject: (11) im a k u ih p i pe am tri e s-e n u m in k a e m a p u h ti - i - y a badness d e n m lz r 2: pro D E T R Z R -think cause-3-ERG ‘She w ill c a u s e y o u t o t h i n k in a b a d w a y .’ M A has a few intransitive verb stems which take a GAUS suffix to form a transi­ tive stem; the suffixes are cognates of some o f the forms in table 2.9: -nipi, -pa. There is another sutfix which occurs with transitive verbs, -mehpo, with the specific m eaning o f ‘o rd er (to do something)’. For other derivations, M A has affixes which are cognates o f forms in table 2.9: prefix -eh-, es-, at- ‘D etrzr’; and suffixes -piahti ‘Ingressive’, -aretihka ‘Terminative’, -ma ‘Intr.vblzr’, -pa, -ma, -h li ‘Tr.vblzr’, -ka ‘V blznm al’. In all the languages with CAUS suffixes, where the suffix is added to an IN T R to form a T R , the S o f the IN TR becomes the O o f the T R and a new A is introduced; where the suffix is added to TR , to form TR^, in T R j a new A is introduced, the O rem ains the same, and the A o f T R , (the causee) becomes an oblique agentive phrase, m arked w ith the agentive/ergative postposition (see §3.1.1, re table 2.6; an d 45 2 Carib §4.2). Com pare the following W W clauses; ( 12a) is a simple T R and ( 12b) is T R + CAUS: (I2a) weewe n-aam a a<|>a^a tree 3A-fell + iMM.PAST daddy ‘D addy felled the tree.’ (12b) weewe n-ama-<|>o tree 3-fell-CAUS + iMM,PAST grandfather (j’ootj'o a4>a4)a ya daddy to/by ‘G randfather had daddy fell the tree.’ Som e o f the derivations in table 2.9 show th e form ation o f verbs from noun stems by the addition o f a derivational affix. Som e noun stems can also be com ­ pounded w ith verb stems to form a verb, i.e., noun incorporation. This is com m on in C arib languages for noun stems that refer to body parts which are obligatorily possessed. T he possessor becomes the O o f the verb, as shown below from Wai Wai: (13) oy-ew-kraji 10-eye-stick .in to + 1mm .p a s t ‘It stuck into my eye.’ T. E. Payne (1995) reports a few examples o f a type o f ‘classificatory’ incorpora­ tion in PA for describing institutional activities. H e argues also th a t the tightly knit OV (object-before-verb) construction in PA represents an interm ediate historical stage between syntactic and m orphological structure, and could be characterized as ‘loose incorporation’ (following M iner 1986). This would also apply to o th er Carib languages. T he OV structure does n o t constitute canonical object incorporation, since th e V retains its transitivity and the O is n o t p a rt o f the verb inflection, but as Payne shows, the construction has some o f th e formal and functional properties associated with incorporation. 3.2.2 N o m in a liz in g affixes a tta c h e d to v e rb ste m s to fo rm n o u n s M ost nom inalizers are suffixes (one significant exception, the prefix n-, is discussed below), and cognate forms regularly appear in the languages for which data are available. G ildea (1998) presents tentatively reconstructed proto-C arib forms of several nominalizers. Table 2.10 is n o t an exhaustive set o f nominalizers, b u t shows G ildea’s reconstructed proto-form s and the cognate forms which are described in my sources. The functions o f the suffixes in table 2.10 are: *-ne ‘N om inalizer o f A’; *-/-i‘N om inalizer o f A ction/State’; *-iipi or *-/«/7«,‘N om inalizer o f A ction/State; Fast Tetrse’; *-sapo ‘N om inalizer o f S /0 , Past Tense’; *~topo ‘thing, time o r p la œ ■ m Il ai. ü- :i3 C/5 ^ Cl- Z g s S i âs , I , ,< «w à sO *1.(ü O O c c c c 48 D esmond C. Derbyshire Table 2.10 NominaUzing suffixes added to verb stems Language *-ne *-ri •-tu p u -ri AP CA DE HI KA MA PA TR WW WA -ne -nen -ri -ri -dii -(ni)-ri -ri -n(i)/-a -h p iri -xpo -ho/-hë -th iri -tpi-ri -hpi -hpë/hë -sah -sa? -ri -ri -hpê -thir» -/a<f>u -0 -tpê/-tpi -Jie -ne -nen -iie{n) -ne -jie -ne -0 *-sapo -xpo -saho *-topo *-pini -topo -topo -toho -t(o)ho -toho -toh -to7 -topo -to<j>o -pin* -bin -hünü -hini -top(o) -pin -pin -hni -pin associated with the state or action’; I add one other, not included by Gildea: *-pm i ‘N ominalized N egation’. H I examples are: (14a) -jie ‘N om inalizer o f A’ ro-hananih-jie 1-instruct-A.NMLZR ‘one who instructs m e’ (14b) - f / i i j r i ‘N om m alizerof A ction’ IN T R : a-w anota-niri 2-Sing'AC.NMtZR ‘your sin ging’ TR: i-h an an ih -ri 3-instrUCt-AC.NMLZR ‘the instructing o f h im ’ (14c) -th ici ‘Nom inalizer o f A ction, Past Tense’ i-w an ota-th iri komo 3-sing-AC.NMLZR COLL ‘their singing (in the p ast)’ ( 14d) -saho ‘N om inalizer o f the S/O, Past Tense’ S; i-m an h o-sah o u ro /o m o r o /m o k i iMPERS-dance-S.NMLZR 1/2/3 pro ‘I (am ) / you (are) / h e (is) the one w h o da n ced .’ O: t-ono-saho iMPERs-eat-O.NMLZR koso decf ‘The deer (was) what was eaten.’ 49 2 Carib (14e) -t(o )h o ‘N om inalizer o f thing/time/place associated with State/A ction’ IN T R : r-om oh-toho 1-COme-ASSOC.NMLZR ‘lime o f my com ing’ / ‘place to which I ’ll com e’ / ‘circum stances o f my com ing’ TR: i-hkoto-tho 3-CUt-ASSOC.NMLZR ‘saw, thing for cutting it’ (141) - h in il-fm i ‘N om inalizer o f N egation’ IN T R ; i-to -h n i IMPERS-gO-NEG.NMLZR uro IPRO ‘I (am ) one who does not / will n o t go.’ TR; t-ono-hni tM PERS-eat-NEG.NM LZR okoye Snake ‘Snake (is) a thing that is n o t eaten / we do n o t eat.’ T he above examples dem onstrate the following characteristics that generally apply in C arib languagea T he (a) ~pe nom inalization applies only to transitive verbs; the corresponding intransitive S nom inalizer (‘doer o f the action’) is a different construction (see §3.2.3 for the nom inalization o f the t-...-so adverbial stem); thus A a n d S are treated differently. The same is seen in the (d) -saho a nd (f) -hint nom inaüzations - th e subject o f the verbless clauses refers to the underlying , S and O o f the verb that h a s been nominalized, never to A. The pivot o f the other three nom inalizations (b, c, e) is the same: the referent of the person-m arking prefix (which functions as a possessor in these nom inal constructions) is the underlying S or O o f the nom inalized verb, never the A. This highlights the ergatively organized characteristic o f nom inalizations in C arib languages. There is one construction th a t occurs in many Carib languages as a nominalization in which b o th the A and O are in focus. Gildea (1998) refers to this as ‘the idiosyn­ cratic O N ominalizing Prefix n-’. As that label indicates, the pivot o f the construction is a prefix n- which refers to the underlying O. There is, however, also a personmarking prefix (or possessor noun phrase) that precedes it which always refers to the A (but which in the surface nominalized form functions as possessor). It usually co­ occurs w ith one o f the Action Nominalizing suflixes, including Past, and in PA it can occur with other suffixes (G ildea 1998). I illustrate again from H I in (15a, b); (15a) n(i)- ‘O N o m in alization ’ r o -n -a r i-h o -n ifi 1-O.NMLZR-take-CAUS-AC.NMLZR ‘the th in g I send (lit, cause (so m eo n e) to take)’ 50 D esmond C. Derbyshire (15b) r o -n -ari-h o-th iri 1-O.NMLZR-take-CAUS-NMLZR.PAST ‘the thing I sen t’ In CA, the n- prefix does not function as a nominalizer, but is dependent on the co­ occurrence o f a nominalizing suffix. Also in CA, the n- can refer to the action as well as to the underlying O (H off p.c.) 3.2.3 S o m e a d v e rb ial d e riv a tio n s fro m n o u n a n d v e rb stem s ‘Adverb’ is the term I use for what I describe for HI in D erbyshire (1985: 10) as: ‘a heterogeneous class com prising what in other languages are often dis­ tinguished as adjectives and adverbs, and including also locative and time words, and num erals and quantifiers. They are bound together into one class on the grounds o f their com m on syntactic function, either (i) as com plem ent o f the copula in co p u lar clauses, or (ii) as adjunct [i.e., oblique constituent] in any type o f clause.’ There is no recognizable class o f ‘A djective’ in H I; what seem like adjectives, on a semantic basis, function as adverbs or nouns (nominalized forms of adverbs). The above definition covers m ost o f the C arib languages for which I have seen descrip­ tions, although some linguists use the label ‘A djective’ for some forms and may have good reasons for m aking finer distinctions for the languages they have described. Postpositional phrases, at least in H I, have the same syntactic functions as, and certain similar m orphological properties to, adverbs. There is a relatively small set o f basic, non-derived adverbs, perhaps as many as 100, with m eanings such as ‘good’, ‘tali’, ‘high’, ‘how ’, ‘one’, ‘here’, ‘over there’, ‘far away’, ‘with difficulty’, ‘soon’, ‘m any’, ‘all’. The antonym s o f many o f these are formed by adding the suffix -hira ‘N eg’. There is the potential for thousands of derived adverbs, since there are at least 15 derivational processes (in HI; and most o f them are reported for other C arib languages); some of these processes are highly productive. I will describe a few o f them, with WW data (16a-d,I7,18) taken from Hawkins (1998). There are several adverbializing processes which take the following form: if- + N or V stem + Suffix. The prefix ti- is invariable (except for phonological condi­ tioning); the stem can be that o f any semantically appropriate noun or verb; the form of the suffix differs for each process. (16a) ti-...-sol-fi ‘state or process, derived from V ’ IN T R : t-wayih-so ADVZ-die-ADVZ n- 0 -a-si 3S-be-SF-N0NPAST ‘H e is dying.’ (refers to S o f the action) 51 2 Carib TR ; t-am a-Ji ADVZ-slash-ADVz n- 0-a -si 3S-be-SF-NONPAST on weew e this tree ‘T h is tree is slashed.’ (refers to O o f the action) This adverbial construction can be nom inalized by adding the suffix -m i ‘N m izr’. The resulting nom inal derived from the intransitive verb also refers to the S o f the underlying verb (e.g., W W twayihsomi ‘one who dies / is dying’) and this is the ‘nominalization o f S’ construction which corresponds in meaning to -j\e ‘nom inaiization o f A’, the construction described in §3.2.2. (16b) ti-...-(pore ‘obligation/fitness/goodness, derived from V or N ’ IN T R : ti-to-i|)ore ADVZ-gO-ADVZ n-0-a-si 3S-be-SF-N0NPAST ‘It would be good to go.’ TR: tJ-enta-(j)ore n-a-a-si ADVz-hear-ADVZ 3S-be-SF-N0NPAST ‘It o u gh t to be heard.’ N: tJi-(j)i-(i)ore m- 0 -a-si ADVZ-wife-ADVZ 2S-be-SF-N0NPAST ‘You have a good/beautiful w ife.’ (This derived adverb is followed by the copula, and it is always in the third person singular form , for both intransitive and transitive verbs; this gives it som ething o f an im personal meaning, as indicated in the English translations. In the case o f the N stem, however, the copula is inflected for the appropriate person.) (16c) ti-...-fel-ye/-jie ‘state o f being, derived from possessed N ’ tu -w u jiu -re ADVZ-Iaziness-ADVZ ‘la zy ’ (16d) ti-...-ke ‘p ossessin g an entity, derived from possessed N ’ tu-hm e-ke ADVZ-egg-ADVZ ‘having e ggs’ In the forms derived from verbs in (16a) and (16b), the adverbializing suffix has as its pivot the S or O constituent o f the underlying verb, as in the case o f the nom inalizing suffixes (see §3.2.2). The nuclear constituents o f a negative clause are an inflected form o f the copula preceded by a phrase containing an adverb that has been derived from a verb stem by the addition o f the suffix -hra ‘N eg’ (the WW form; all Carib languages know n 52 Desmond C. Derbyshire to me have cognates o f this form). Com pare the N om inaliœ d N egation suffix in table 2.10 (W W -hni). Possessed noun stems and some postpositions and adverbs are negated with the same suffix. W ith forms derived from intransitive verbs and possessed noun stems, the im personal prefix y U e - co-occurs w ith the suffix; with forms derived from transitive stems a personal prefix co-occurs, and it refers to the O o f the verb; in both cases the inflected copula (or other main verb which occa­ sionally occurs) has the prefix which refers to the subject o f the action or state. Examples from W W include: (17) WW; -/ira/-/-a‘negation adverbializer, derived from V or possessed N ’ yi-w inki-ra ÍNTR-. IMPERS-Sleep-NEG w-0 -a-si 1-be-SF-NONPAST ‘I will n ot go to sleep.’ TR: k-eni-hra n-0 -a-si 1+2-see-N EG 3-be-SF-NONPAST ‘H e does n ot see us.’ POSSD.N; yi-(j)i-hra ki-w tje-si IMPERS-wife-NEG IS-gO-NONPAST ‘I ’m goin g w ithout my w ife.’ POSSD.N: 0 -ew u-hra w- 0 -a-sí IMPERS-eye-NEG 1-be-SF-NONPAST ‘I ’m blind / have poor sight.’ A nother adverbial derivation conraion in Carib languages relates to temporal and conditional expressions. There are two adverbializing suffixes in WW which mark the nuclear constituent o f such expressions: -taw ‘when, while, if’; and -tfh e ‘after, if ’. The resulting adverbs have person-marking prefixes o r preceding free form nouns that refer to the S o f an intransitive, and to the O o f a transitive, verb. WW examples are: (18a) -taw IN T R : ‘when, while, if’ tuuna mok-ya-taw to-hra t- 0 -a-si rain c o m e -S F -if 1 + 2S-be-SF-N0N?AST go-N E G ‘If it rains ( i f rain c o m e s ) we (18b) -tjlie IN T R : will not g o .’ ‘after. If’ a-<j>aka-tjhe ti-htf-e-si 2S-wake, up-after 1 + 2S-g0-SF-N0NPAST ‘A fter you wake u p we will go.’ TR : o -n om -tjhe m ariya Je w-0 -a-si I-leave-after knife w anting 1S-be-SF-NONPAST ‘A fter (you) leave m e, I will want a knife.’ 2 Carib 3.3 53 The particle word class Carib languages abound in particles. M ost o f these are always postposed to other constituents, never occurring alone o r as the initial constituent in a phrase. They are never inflected, n o r can they have derivational suffixes added (both o f which are possible with other types o f postpositions, e.g., the locative postpositions described in §3.1.4). I agree with Hofl’s (1990: 495) statem ent, in defending his use o f ‘parti­ cle’ rath er th a n ‘clitic’: ‘Even though the C arib elements show a few ditic-like fea­ tures, these are n o t sufficient to deny them word status.’ Some are like clitics in that they are phonologically attached to the preceding constituent and are subject to the same m orphophonological processes that apply at m orphem e boundaries within words. O thers are not affected by those processes. All, however, are more mobile than b ound affixes and their general syntactic behaviour justifies their being defined as a d istinct word class. They fall into distinct sub-classes on the basis o f their meanings and their syn­ tactic and discourse roles. Linguists have labelled and described them in difierent ways. In D erbyshire ( 1985) I distinguished three sub-classes for H I:(1) modifier p ar­ ticles, which function alm ost like adjectives to modify the immediately preceding constituent; (2) discourse particles, which usually refer to something in the dis­ course-pragm atic context; and (3) verification (i.e. evidential) particles, which express the speaker’s attitude to w hat they are saying, including degree o f certainty and authority. In any particle sequence in H I, members o f each class usually occur in th at sam e order; any verification particle is always in the final position in the phrase. H ofT(1986,1990) distinguishes two classes for CA: (1) non-m odal particles, including b o th my modifier and my discourse particles, and (2) m odal particles, which include my verification particles and a few others. The following examples are from H I (m ost other languages have sim ilar cate­ gories and sometimes what appear to be cognate forms): (1) modifier: komo ‘Collective’; ym o ‘A ugm entative’; heno, which has two meanings: ‘Eteceased’ (refer­ ring to a person), and ‘Large group’ (referring mainly to anim als and plants; e.g. torono heno ‘flock o f birds’); (2) discourse; haxa ‘C o n tra st’; rma ‘Same referent’ or ‘C ontinuity’,; harha ‘R eturn to form er state or location’; an d (3) verification; (h a )ti ‘H earsay’; ( ka )n a ‘Uncertainty*; and ( /w j» » ‘D eduction’ (which the speaker makes from know n facts). 3.4 Pronouns Table 2.11 displays the pronoun sets from nine languages. T he first set o f third person forms are referential, in the sense that they usually refer to entities that have already been introduced into the discourse. T he other third person pronouns are 54 D esmond C. Derbyshire Table 2.11 Pronouns, singular ( noncolkctivej form s Person AP CA DE HI 1 iw i kiraoro ina om oro a:wu kihko a?na am oro ÜWÜ küwû jijtaa öm ödö Ufö kiwro amna otnoro 1+2 1+3 2 PA TR WW WA uurj uurihkon anna am irt yu yuto ana amën wt këmë ainya êmê owî ktiwi amna amoro iu/yu kunmê emna ëmë m tík iri ken nërë noro inëlë mên irë ero inëlg mëe nioso tnëi/sin mëhe sen ont helë MA 3R E FEREN TIA L AN IM in o ro inoro mohko tüwü noro IN A N iro sero i:ro iiyöö ko m öídö m o san » m iseri mëh ont seni sirtri sih 3DEICTIC: PROXIMATE A N IM mose rao:se m ö jiö INA N s en t e:ni ee7dö e;ro serë 3DEICTIC: M EDIA L AN IM m okiro mo;ki mö?kü m okro m iikiri mah muku m ëërë m ikt mio INA N m oro mo;ro m önö m oro s ii n i mu m ërë m oi’o m ëlë jiö ? d ö moki m iik iri kën m ëki m ik i mëklë 3DEICTIC: DISTAL ANIM m oki m o :k i m îkro IN A N m oni tno:tii iiy ö ö m ont s ii n i m ën mën m in i m if ir t dem onstratives an d usually have the deictic function o f relating the person or entity directly to the extraiinguistic context o f the utterance (as also do the first and second person pronouns). The third person deictic pronouns have two param eters for anim acy (anim ate and inanimate) and three for spatial scaling (proximate, medial and distal). CA has one form for medial and distal animate. MA has the same forms for medial and distal, and lacks ‘3 R E F E R E N T IA L .IN A N ’ (the sources are n o t specific, but it seems that 3D EIC forms can be used with referential, as well as deictic, meaning). In this section I describe some of the m ore distinctive characteristics o f C arib syntax under three main headings: main clause structuring (§4.1); subordinate 2 Carib 55 clause constructions (§4.2); and the ergativity which, to a greater or lesser degree, governs the case m arking, person m arking, derivational processes and constituent order patterning in C arib languages (§4.3). M uch o f the syntax is found in the m or­ phology, which is described in §3. 4.1 M a in clause structuring There are six basic clause types in C arib languages: transitive (with A and O argu­ ments); intransitive (S argum ent); copular with an inflected copula; copular with an auxiliary verb (occurs mostly in the languages with main clause ergativity); verbless copular; and quotative (having a main verb ‘to say’ and an em bedded clause of direct speech th a t norm ally precedes the ‘say’ verb). There is one dom inant feature in the ordering o f constituents in main clauses: the fairly rigid OV sequence. This correlates w ith basic phrase constituent orders: G E N -N and N -Postposition. (F or the N oun-M odifier type o f phrase, see below on paratactic sequences.) W hat appears to be an exception to the rigidity o f the OV order is found in PA, in which VO occurs a t least as often as OV, but as D oris L. Payne ( 1994) points out, this is due m ainly to the fact th a t the final O is often paratactically adjoined to the clause, rather than being an integral part o f the clause syntax (again, see below on paratactic sequences). It has proved more difficult to determ ine w hether the basic position for A and S N Ps is pre- o r post-verbal, since both orders occur, but noun phrases and free pronouns do not often occur once they have been introduced as topics in a discourse. T he verb person markers are usually ■the only indicators o f the A and S (and O) referents. Where N Ps do occur, the evi­ dence points to transitive AOV and intransitive SV as being basic in some languages (e.g. C A , D E), a n d VS and OVA as basic in others (e.g. AP, H I, PA). In th e ergative languages (e.g. KU, M A) SV and OVA are the dom inant orders, although in the transitive clause AOV also occurs. T he evidence for these basic patterns is not pri­ marily frequency o f occurrence, but the fact that alternant orders can be explained as being pragmatically m arked (Derbyshire 1986). T he above applies prim arily to the basic configuration in clauses which have only third person referents (A , S, O). H off (1995, for the C arib language) and G ildea (1995b, for C arib languages in general) provide insightful studies that show a nonconfigurational pattern for clauses with first and second person referents, and for certain constructions which, according to G ildea’s diachronic hypothesis, are his­ torically more innovative. T here is considerable flexibility in the ordering o f oblique constituents (adver­ bial an d postpositional phrases). They usually occur after the verb o r any post­ verbal A/S argum ent, but can be fronted to clause-initial position for emphasis, co n trast, o r continuity w hen referring back to the previous clause(s). T here is a 56 Desmond C. Derbyshire strong tendency n o t to have more than one constituent preceding the main predica­ tion (V o r OV). Paratactic constructions are a feature o f C arib sentences and clauses. A sentence m ay consist o f two or more juxtaposed clauses usually in a coordinating function, but w ithout any conjunctions or particles (such as ‘and’, ‘b u t’, ‘o r’, etc.). Only the final clause in such a sequence has the term inal intonation pattern. Within a clause, constituents o ther than the verb (i.e. A, S and O noun phrases, and adverbial and postpositional phrases) can consist o f two o r m ore juxtaposed phrases. The first can usually be considered the head phrase, and the phrases that follow may have one o f several functions: coordinating, modifying, adding a greater degree o f specification, o r as an afterth ought clarification. Each phrase in the sequence is usually separated by a brief pause. These sequences can be ieft-dislocated for em phasis or as a topicalization device. M ore often they are right-dislocated, following the main predica­ tion. A single n oun or phrase can also be left- or right-dislocated, as is often the case with the PA post-verbal O constituent m entioned above. There can also be dis­ continuous sequences in which the head phrase precedes, and the other phrases follow, the verb. F or fuller discussion and exemplification, see Derbyshire (1985: 129-35) on H I, and Doris L. Payne (1994) on PA. 4.2 Subordinate clause constructions In some Carib languages (e.g. CA, H I, WW ) there are, strictly speaking, no sub­ ordinate clauses. T he subordinate constructions take the form o f phrases: nominal phrases (§4.2.1), adverbial phrases (§4.2.2) and postpositional phrases (§4.2.3). These phrases have as their nuclear elements nominalized or adverbialized forms derived from verbs. T he more strongly ergative languages have a few constructions o f that kind, but there are also regular subordinate clauses, including some with finite verb forms. A s I pointed out in §3.2.2 and §3.2.3, m ost of the nominalized forms and some o f the adverbialized forms take person-m arking prefixes which refer to the S and O o f the underlying verb, thus indicating the ergative character o f the constructions, and in the derived structures these prefixes express a possessor rela­ tionship rather than S or O (although this is sometimes difficult to translate into English). Free form (pro)noun possessors can substitute for third person prefixes. The A o f the underlying verb, when it occurs as either a noun phrase or personmarking prefix, is followed by a postposition which has cognates in m ost Carib lan­ guages with the meanings ‘to, by, ER G ’ (H I o~wya in (23c) and ro m u ru wya in (24b)). 4,2.1 N o m in a liz a tio n s T he various types o f nom inalization and the derivational sufikes (and, in one case, prefix) are described in §3.2.2. Here I give a few examples from W W and M A, of 2 Carib 57 the way these function as A , S and O, and aiso as modifiers of other noun phrases (corresponding to relative clauses in languages like English) in m ain clauses. The other functions o f nom inalized constructions occur in postpositional phrases (§4.2.3). Examples (19a-c) are W W and (20a-b) are M A. (19a) S o f an intransitive clause m iya 0-tO 'tJow m a ra ri y -am a-jie away 3-go-cOLL + im m .p a s t field o e n - cuI-A . n m lzr kom o c o ll ‘The field cutters w ent away.* (19b) O o f a transitive clau se a-m ok-ri w -enta 2-come-AC.NMLZR l-h ear + im m .past ‘I heard you/your com ing.’ (19c) modifier o f an o th er noun, the w h o l e noun phrase being the A o f a transitive clause ji-esk-e-si tak yawaka i-yo-hto-Ja^u 3S30-bite-SF-NONPAST now axe iM PERS-edge-VBLZRS/O.NMLZR.PAST ‘The axe that has been sharpened now bites (cuts) thing&’ (20a) O o f a transitive clause aw -ennahp o-hp i-kon epuhtt-hp i-i-ya 2-return-PAST-coLL know-PAST-3-ERG ‘He knew y ou all returned.’ The derived nom inal has a finite verb form but is syntactically nonfinite (see Gildea 1998, ch ap ter 2, for discussion). (20b) m odifier (relative clause with finite tense suffix); post-verbal, rightdislocated , m od ifyin g its pre-verbal head (m irifi): m iriri erah m a-h p i-i-ya ti-rui n i-k u p i-h p i that see-PAST-3-ERG 3REFL-brother O.NMLZR-do-PAST ‘He saw that, what his brother 4.2 .2 had d o n e.’ A d v e rb ia liz a tio n s As noted in §3.2.3, some o f the adverbs derived from verb stems have person-marking prefixes (or, like derived nominals, a preceding possessor noun phrase) and othera do not. T he latter usually have the impersonal prefix, which in some languages has the same form as the third person prefix but does not function as such. WW examples are givenin §3.2.3 o f the more com mon adverbial derivations. Examples (21a-b) are also WW, showing another frequently occurring derivation, formed by the suffix -sol-/i. 58 Desmond C. Derbyshire with the meaning ‘purpose or goal o f a verb o f motion’. This derivation is not to be confused with the -sol-fi derivation that co-occurs with the adverbializing if- prefix (§3.2.3); the two sets o f suffixes have the same allomorphy in all the C arib languages for which they have been described (G ildea p.c.), and both result in adverbial func­ tions, but the ‘purpose o f m otion’ derivation has person-marking prefixes on transi­ tive verbs and the impersonal prefix on intransitive verbs, while the other derivation has the invariable prefix ti- ‘A dvz’. Examples from W W are: WW: -sol-fi ‘purpose/goal o f m otion’ (21a) IN T R : yi-win-so ki-wtj'-e-si IMPERS-Sleep-ADVZ:PURP0SE.0F.M0TI0N 1-gO-SF-NONPAST ‘I am goin g (away) to sleep.’ (21b) TR: aw -akronom a-Ji k-mok-ya-si am jie 2-help- A0VZ:PURPOSE.OF. MOTION i-com e-SF-N O N PA ST later ‘Later I will com e to help you.’ MA has the sam e construction, w ith the suffixes -U se, and it is nonfinite. Examples include: (22a) IN T R : suhm inan-se toh play-A D V ZrPU R TO SE.O F.M O TIO N 3 :c o l l epahka-piti-hpi po ro pona go OUt-ITERATIVE-PAST OUtside tO ‘They went outside to play.’ (22b) TR: tiaron-kon w iti-h p t m oroh other-coLL go-PAST fish yapih-se Catch-ADVZ:PURPOSE.OF.MOTION ‘O thers went t o catch fish.’ 4.2.3 P o stp o sitio n s In postpositional phrases the object o f the postposition can be any type o f noun or noun phrase, including nom inalized constructions. These em bedded subordinate constructions are com m on in C arib languages. Two o f the more frequently occur­ ring constructions o f this kind in HI will now be described. In HI, the prim ary means o f conveying the concepts o f ‘desire’, ‘w ant’, ‘like’ and ‘love’ is with the postposition Je, which has the basic m eaning ‘desiderative’. It co­ occurs with the copula, which carries the inflections for tense and person of the subject. The following examples show that /e : ( 1) can be inflected with the O person marker (23a); (2) can be preceded by a simple noun and suffixed with the negation 2 Carib 59 adverbializer -/lira (23b); and (3) can be preceded by a m ore complex construction such as a derived nom inal (23c). HI occurrences o f f e ‘Desid’ are: (23a) o- Je 2-DESÍD w -ehj-aha IS-be-NON PAST ‘I love you.’ (23b) a-kanaw a-thiri fe-h r a w -ehJ-aha 2-canoe-PCBSD.PAST d esid - n eg IS-be-NONPAST ‘I d o n ’t want your o ld can oe.’ (23c) ro-m in yaka ro -k a t|h o y-ok-niri Je 1-house to I-thing G E N -bring-N M L Z R DEsrD w -ehJ-aha o-wya IS-be-N O N PA ST 2-by i w a n t you t o bring my things to my house.’ (lit. ‘I am w anting the b rin g in g o f m y things to ray house by you.’) In (23c) the noun phrase which is em bedded as the object o f the postposition fe has as its nuclear constituent a possessed noun that is derived from a transitive verb, -oki- ‘bring’. T he A o f that verb is indicated in the sentence-final postpositional phrase owya by the prefix o- ‘2’. The root -wya has several usages, including ‘to ’ when referring to recipients or addressees. Here it refers to the A o f the nom inalized verb, and thus functions as an agentive or ergative marker. In H I, as in many other C arib languages, ergativity operates only in subordinate constructions. The second em bedded construction of this kind is one which occurs with the post­ position ke, which has the meanings 'because o f’, and ‘by means o f . H I 'ke’ is not inflected for person, nor does it occur with the suffix -hira ‘N eg’, but it has the same em bedding potential as fe , and an em bedded derived nom inal functions ergatively. (24a) i-to -h ra w-ahko, thejiehra tuna iMPERS-go-NKi 1-be + iMM.PAST m uch water y -o m o k -n iri ke GEN -Come-AC.NM LZR bccause ‘I didn’t g o because it was raining heavily.’ (lit. ‘I d i d n ’t g o b ecause o f the c om in g o f m uch rain.’) (24b) k-eryehotehe hojiko 0 -wo-niri- ke 1- b e .h a p p y + n o n p a s t p ec ca ry O EN -shoot-A C .N M L Z R because ro -ra u ru w ya 1 -s o n by ‘1 am happy because my son killed a peccary.’ (lit. ‘I am happy because o f the shooting o f a peccary by my son.’) 60 Desmond C. Derbyshire In (24a) the noininalized verb yom okniri is intransitive and the preceding pos­ sessor noun tuna refers to the S of the verb. In (24b) the nom inalized verb w onirí is transitive, the preceding possessor noun hopko refers to the O o f the verb, and the A o f the sam e verb is expressed by romuru with the ergative m arker wya. 4.3 Ergativity The ergativity th at is rife in Carib languages has already been discussed in several sections. H ere 1 first review the categories and constructions which are affected by it and the varying degrees o f ergativity that are found in different languages, and then present two opposing views which have been proposed on the direction of diachronic change, one from earlier ergative to accusative systems, and the other from earlier accusative to ergative systems. The languages in which ergativity is dom inant are: A K , A R , M A (all in K aufm an’s N o rth A mazonian Branch) and K U (South A m azonian Branch - also K alapalo in this Branch). In these languages there is case m arking in main and sub­ ordinate clauses, and the person-m arking systems and constituent order patterns are also ergatively organized. BA, another language in the S outh A m azonian Branch, has been described as not having nom inal ergative case m arking or constituent order, but as having a split ergative-accusative person-m arking system, which is intrinsically tied to a syntactic S /0 pivot in complex sentences, both coor­ dinate and subordinate constructions (Souza 1994). All the other languages described in this survey have mixed ergative-accusative systems, b u t differ in the degree o f ergativity a nd accusativity. In m ost o f them the ergativity does not occur in main clauses b ut is restricted to subordinate constructions with verbs that have been nominalized o r adverbialized (§3.2.2 and §3.2.3). G ildea (1998) agrees with the above set o f five dom inantly ergative languages. He divides the rest into different groupings depending on the degrees o f ergativity and accusativity they display in what he considers to be m ain (finite) clauses. These groupings are based prim arily on hypotheses concerning developments in the main clause person-m arking and tense-aspect systems, constituent order changes, reanalysis o f derivational affixes as tense-aspect markers, and reanalysis o f dem on­ strative p ronouns as auxiliary verbs. His groupings can be sum m arized: (1) CA, CR, D E, H I and W W are exclusively accusative in m ain clauses; (2) AP, K A, TR and WA (also Yukpa and K alijia, a Venezuelan dialect o f CA) have at least one main clause construction that is ergative, but otherwise are dom inantly accusative; and (3) PA is a complex system o f construction types, some o f which are ergative and others accusative. H e concludes that the dom inantly accusative languages represent the earlier proto-C arib system. G ildea’s research has been extensive and his diachronic approach is sound and persuasive. 2 Carib I 61 have taken the view that the languages with dom inant ergativity represent the earlier stage o f C arib m orphosyntax and that the other languages, w ith varying degrees o f mixed ergative-absolutive patterning, have developed from that earlier more ‘pure’ ergative stage (D erbyshire 1991,1994). This was consistent with a view o f constituent order change in the C arib family th a t I had proposed earlier (Derbyshire 1981), based on a study o f constituent order in three languages (CA , H I and M A). In the 1991 paper I com pared the ergativity-related characteristics o f M A, PA an d H I, arguing that the facts of PA, as described in T. E. Payne (1990) and by G ildea (in eariier work culm inating in his 1992 dissertation), did n o t support their reanalysis hypotheses for that language and that the relevant PA constructions could be m ore plausibly construed as nonfinite subordinate constructions (i.e., not reanalysed as finite clauses). A reference gram m ar o f PA is in preparation (T. E. Payne and D. L. Payne, ms.), and this may help to determ ine which o f the com pet­ ing views is m ore convincing for th a t language. My view of the direction of change in the Carib family has been reinforced by a more general factor: the ram pant ergativity that is found in so many A m azonian lan­ guage families (Arawá, Carib, Jê, Pano, Tacana, Tupi-Guarani, Yanomami). This suggests a long history of ergativity in the area (Derbyshire 1987). Preliminary comparisons o f some o f the linguistic features of Carib and Tupi-Guarani suggest the possibility of a distant genetic relationship (see Goeje 1909 - cited in Rodrigues 1985: 373, 3 9 8 - 9 - for an early conjecture about such a relationship; Rodrigues 1985 for phonological correspondences and a list o f likely lexical cognates; Rodrigues 1996, and D erbyshire 1994 for morphosyntactic similarities). Historical and com parátive studies are well advanced for Tupi-Guarani, and Jensen (1998) has m ade a strong case th at the present mixed ergative-accusative systems in the languages o f that family can be traced back to more dom inant ergativity in proto-Tupi-Guarani, A m ore definitive assessment o f the C arib-T upi relationship (whether genetic or areal), and o f diachronic change in all areas o f C arib m orphosyntax, is dependent on extensive com parative studies aimed at producing reliable phonological and lexical reconstructions and a m ore reliable internal classification o f the C arib family. Such studies can only be undertaken when fuller descriptions become avail­ able o f m ore C arib languages. BIBLIO GRA PHY Abbott, M. 1991. ‘M acushi’, pp. 23-160 of Derbyshire and Pullum 1991. Armellada, C. de. 1943. D kciom rio de la lenguapemon. Caracas: Artes Gráflcas. Armellada, C. de and Salazar, M . G. 1981. Diccionariopemon. Caracas: Ediciones Corpoven. Carlson, R. and Payne, Doris L. 1989. “Genitive classifiers’. Proceedings o f the Fourth Annual Pacific Linguistics Conference. Eugene: University o f Oregon. 62 Desmond C. Derbyshire Carson, N. 1982. ‘Phonology and m orphosyntaxof MacuKÍ (Carib)’, Ph.D. thesis, University o f Kansas. Derbyshire, D. C, 1958. ‘Ka'chuyana (Questionário Padrâo para a Pesquisa nas Línguas Indígenas Brasileiras)’, Brazil: SIL. 1961. 'N otas comparativas sobre très diaietos karíbe’, Boletim do Musmi faraense Emilio Goeldi, Nom Série, Antropologia, 14.1-10. 1979. Hixkaryaim, Lingua Descriptive Studies 1. Amsterdam: North-Holland. i 981. Ä diachronic explanation for the origin o f OVS in some C arib languages’. Journal o f Lmguistics, 17.209-20. 1985. H ixkaryam and linguistic typology, Dallas: SIL and the University of Texas at Arlington. 1986. ‘Topic continuity and OVS order in Hixkaryana’, pp. 237-306 o f Native South American discourse, ed. J. Sherzer and G. Urban. Berlin: M outon de Gruyter. 1987. ‘M orphosyntacfic areal characteristics o f Amazonian languages’, ÍJJL, 53..11I-26. 1991. ‘A re Cariban languages moving away from or towards ergative systems?’, H'or/c Papers of the Summer Imtitute of Linguistics, University o f North Dakota Session, 35.1-39. 1994. ‘Clause subordination and notninalization in Tupi-G uaranian and Cariban lan­ guages, pp. 179-98 o f Wise 1994. Derbyshire, D. C. and Pullum, 0 . K. (eds.) 1986. Handbook o f Amazonian languages, vol. L Berlin: M outon de Gruyter. 1991. Handbook o f Amazonian languages, vol. III. Berlin: M outon de Gruyter. 1998. Handbook o f Amazonian languages, vol. IV. Berlin: M outon de Gruyter. Durbin, M. 1977. ‘The Carib language family’, pp. 23-38 o f Carib-speaktng Indians: culture, society and iangiiage, ed. E. B. Basso. Tucson: The University o f A rizona Press. Edwards, W. F. 1977. An introduction to the Akawaio and Arekuna peoples o f Guyana. Georgetown, Guyana: University of Guyana, 1980. A short grammar and dictionary o f the Akawaio and Arekuna languages o f Guyana, Georgetown, Guyana: University o f Guyana. Emmerich, C. 1994. ‘The Txikào language: fricatives or no fricatives?’, pp. 65-72 of Wise 1994. Franchetto. B. 1990. ‘Ergativity and nominativity in K.uikuro and other Carib languages’, pp. 407-27 o f Doris L. Payne 1990. 1994. ‘A ergatividade Kuiktiro: quadra gerai, hipdteses explicativas e uma visao comparativa’, pp. 7-16 o f Wise 1994. 1995. ‘Processos fonologicos eni Kuikuro: uma visao auto-segmental’, pp. 53-83 o f Esltidos fonologicos das lingiias indlgenas bmsileiros, ed. L. Wetzels. Rio de Janeiro; Editera Oniversidade Federal do Rio de Janeiro. Gildea. S. L.O. 1992. 'Comparative Cariban morphosyntax: on the genesis of ergativity in independent clauses’, Ph.D. thesis. University o f Oregon. 1994. ‘Semantic and pragmatic inverse: “inverse alignment” and “inverse voice” in Carib of Surinam ’, pp. 187-230 o f Voice and inversion, ed. T. Givon. Amsterdam: John Benjamins. 1995a. ‘A comparative description of syllable reduction in the C ariban language family’, Í // I Í , 61.62-102. 1995b. ‘Word order change in the Cariban family’. Ms. 1998. On reconstructing gramnum comparative Cariban morphosyntax, Anthropological Linguistics Series. Oxford: Oxford University Press. G irard, V. 1971. ‘Proto Carib phonology’, Ph.D. thesis, University o f California, Berkeley. Goeje, C. H. de. 1909. Etudes llngihstiqms Caraïbes, Verhandelingen der Koninklijke Akademie van Wetenschappen, afdeeling ietterkunde, n. r. 10(3). Amsterdam: Johannes Miiller. 1946. Etudes linguistiqim Caribes, Verhandelingen der Koninklijke Akademie van Wetenschappen, afdeeling Ietterkunde, n. r. 49(2). Amsterdam: North-Holland. 2 Carih 63 Grimes. B. F. (ed.) 1996. Ethnologue: languages o f the world. 13th edition. Dallas: SIL. Hall, K.. L. 1988. ‘The m orphosyntax o f discourse in D e’kwana C arib’, Ph.D. thesis, Washington University, St Louis. Hawkins. R. E. 1998. ‘Wai Wai’, in Derbyshire and Pullum 1998. forthcoming. ‘Wai Wai’, international dictionary series, vol. 1, ed. M . R, Key. Hill, J. H. and Hill, T. E. 1994. ‘Phonological sketch of Waimiri-Atroari (Cariban)’, Ms. Hoff. B. J. 1968. The Carib language. The Hague: M artinus Nijhoff. 1986. ‘Evidentiaiity in Carib: particles, affixes, and a variant o f Wackernagel’s law’, Lingua, 69.49-103. 1990. ‘The non-modal particles o f the C arib language o f Surinam and their influence on constituent order’, pp. 495-541 o f Doris L. Payne 1990. 1992. 'C ariban languages’, pp. 213-17 of International encyclopedia o f linguistics, vol. I, ed. W. Bright. New York: Oxford University Press, 1995, ‘Configurationality and nonconfigurationality in the Carib language o f Surinam ’, / / J i . , 61.347-77. Jackson, W. S. 1972. ‘A Wayana gram m ar’, pp. 47-77 o f Languages o f the Guyanas, ed. J. E. Grimes. N orm an: SIL and University o f O klahoma Press. Jensen, C. 1998. ‘Comparative Tupi-G uarani m orphosyntax’, pp. 487-618 o f Derbyshire and Puilum 1998. Kaufman, T. 1989. ‘Classification o f K ariban languages’. Ms. 1990. ‘Language history in South America: what we know and how to know m ore’, pp. 13-73 o f Doris L. Payne 1990. Koehn. E. and Koehn, S. 1986. ‘A palaí’, pp. 33-127 o f Derbyshire and Pullum 1986. Koehn, S. 1994. ‘The use o f generic terms in Apalai genitive constructions’, pp. 39-48 o f Wise 1994. Leavitt, C. W. 1971. ‘Trio gram m ar’, parts 1 and 2. Ms. Mattéi-Muller, M. and Henley, P. 1990. Los Tamanakti: su lengua, su vida. San Cristobal: Universidad Catolica del Tachira. Miner, K. L. 1986. ‘Empty categories, case and configurationality’, Natural Language and Linguistic Theory, 2.39-76. M osonyi, I C. 1978. ‘Diccionario básico del idiotna cariña’, thesis for the title o f Profesor Agregado, Universidad Central de Venezuela, Caracas. Otterloo, R. V. and Peckham, J. 1975. ‘Carijona: Swadesh-Rowe word list’. M a Payne, Doris L. (ed.) 1990. Amazonian linguistics: studies in Lowland South American lan­ guages. Austin: University o f Texas Press. Payne, Doris L. 1993. ‘Nonconfigurationality and discontinuous expressions in Panare’, Proceedings o f the Nineteenth Annual Meeting o f the Berkeley Linguistic Society. Special Sesmm on Syntactic fow et in Motive American Languages, 19.121-38. 1994. ‘OVSu versus VSuO in Panare (Cariban): do syntax and discourse match?’ Text, 14.581-610. Payne, T. E. 1990. ‘Transitivity and ergativity in Panare’, pp. 429-53 o f Doris L. Payne 1990. 1995. ‘Object incorporation in Panare’, /X4Í,, 61.295-311. Payne, Thom as E. and Payne, Doris L. ms. ‘Panare reference gram m ar’. Rodrigues, A. D, 1985. ‘Evidence for Tupi-Carih relationships’, pp. 3 7 1 ^ 0 4 o f South American Indian languages: Retrospect and prospect, ed. H. E. M. Klein and L. R. Stark. Austin: University o f Texas Press. 1996. *A case o f grammatical affinity among Tupi, Karib, and M acro-Jê’. Ms. Souza, T. C. de. 1994. ‘A syntaxe de um a lingua ergativa: o Dakairí’, pp. 17-38 o f Wise 1994. 1995. ‘O traço sonoto em Bakairi (Carib)’, pp. 29-51 of EstudosfonoUgicos das linguas indigenas brasileiros, ed. L. Wetzels. Rio de Janeiro: E ditera Universidade Federal d o Rio de Janeiro. 64 Desmond C. Derbyshire Wallace, R. 1970. ‘N otas fonologicas da Lingua kaxuyána’, Soletim do Museu Paraense Emilio Goeldi, Norn Série, Antropologia (Belém), no. 43. Wheatley, J. 1969. ‘Bakairi verb structure’. Linguistics, 47.80-100. 1973. ‘Pronouns and nominal elements in Bacairi discourse’, Linguistics, 104.105-15. Williams, J. 1932. Grammar notes and vocabulary o f the language o f the Makuchi Indians o f Guiana. St G abriel-M odling near Vienna, Austria: Verlag der Internationalen Zeitschrift ‘Anthropos’. Wise, M. R. (ed.) 1994. Lingiiistica Tupi-GuaranllCaribe, vol. 8 oSRevista Latinoamericana de Estudios EtnolingtVslicos, ed. I. P. Pastor. Lima, Peru. The Arawak language family A L E X A N D R A Y. A I K H E N V A L D 1 in t r o d u c t io n : the araw ak la nguages an d t h e ir SPEA K ER S The A raw ak language family' contains the largest num ber of languages in S outh America. G eographically, it spans 4 countries o f C entral America - Belize, H onduras, G uatem ala, N icaragua - and 8 o f South A merica - Bolivia, G uyana, French G u ian a, Suriname, Venezuela, Colom bia, Peru, Brazil (and formerly Argentina an d Paraguay). There are ab o u t 40 living A raw ak languages. N um bers given in the literature vary from 154 (L oukotka 1968) to 89 (N oble 1965). As is often the case in S outh America, a single nam e may be used for what are in fact two or more distinct lan­ guages (a single language can also be referred to with several different nam es). A striking exam ple is the nam e ‘Baniwa’ which etymologically comes from a TupiG uarani term for bitter manioc.^ It is applied to two languages, Baniwa o f Içana (also know n as K urripako) and Baniwa o f G uainia, which are about as different as English an d Russian. (These peoples do not call themselves Baniwa.) For extinct languages there is often not enough data to make an inform ed decision. Table 3 .1 presents a cautious assessment o f what we believe to be distinct languages. Their approxim ate geographical locations are shown on m ap 2. Arawak languages played an interesting role in conquest. The first native American peoples encountered by Colum bus in the Bahamas, H ispaniola and Puerto Rico were A raw ak-speaking Taino - their language became extinct within the first h u ndred years o f the white invasion (Rouse 1992). Spanish - and thus many other languages - contains a num ber o f loans from A raw ak languages, including ' I am grateful to all my teachers o f Arawak languages - Cändido, Graciliano, Olivia, Jovino and José Brito (Tariana), Hum berto Baltazar and Pedro Angelo Tomas (Warekena), the late Candelário da Silva (Bare), Afonso, Albino and Joao Fontes, Celestino da Silva and Cecilia and Laureano da Silva, Januário Paiva and the late Marciiia Rodrigues (Baniwa). Also to David Payne, José Alvarez and Peter van Baarle. ^ The name ‘W arekena’ is also misleadingly used with reference to several distinct Arawak peoples; see discussion in Aikhenvald (1998). 66 Alexandra Y. A ikhem akl I > o |Q s ÿ' t d > I ■C o i Q 2Ô öS ‘C eu — O u3 1 o T3 2 ffl .2 îj m 1 ■a 0 jj 'i s s ■c 8 g 12 '5J °o I I I 5 ^<*-. O S) T« o c c S -I Ü 8 g â 2 ffl o 1C! 1ffl a X i2 g ii g .s lî X •w *w '2 3 £ s t .Ë O ? '> ■| ..g 3 ôa 5 U X Ë o y 1o 1^ 3 2 ^ 3 I I g ÿ- •2 I "H •S :> :s 2 äá; ^ i ® 1 T3 Ä f II •H S '5 § a á2SiÄ £H</í Il s C i3 ÿ '' o á Q" Q -S -o 1 1 1 ° O < I ■ si I(3 '5 i S II ä I' fC It- Ïíö I öI I ■c o fc cu U < S < < 'S' ^ <u cj ^ -S â i ö ^ ^ "•f ■«■ -^ s J §6 IS il I«»I i y ■3 á Ig ■S O I S‘ ■á-5 3 g I« ■11 I I o z I O' »JS 1O 'C^ Ë u c: It 2 < B S J-£ î| 1 ô | ■^g g-â hF oc3 i H o S “ 'C si ^0^ O «vÎ S s _ ■s < 1 N Sá^ m ^ o ^ & sio a 4? - 3 S 2Ü (i JD .2 ^ S .Í ^ » tti' ^ a ° d ii â o ■H I §I I I I > > § -I ., t < È 3- s > < i£ á â s Î ÿ o 1> ’C “ 8 2 :sc j« 3 13 . & Ô 2 ^ -R e g O s3 I M ■” II o Ô t JS S 1° 4 S E & íl^ , “o §■ ^ a.o M « à -a - s s .i o s « o o 72 Alexandra Y. Aikhenvald hammock, tobacco, potato, guava and other nam es for flora and fauna (Oliver 1989, Valenti 1986). Languages in areas settled by the European invaders soon became extinct. Those on the north coast o f South America perished early, before 1700. When the search for gold and rubber spread up the A m azon a nd its tributary, the Rio Negro, further languages succum bed. Sometimes the Indians retaliated, attacking settlements and missions; but the invaders always returned. Indian rebellions often provoked forced m igrations which sometimes ended up creating a new dialect or even a new lan­ guage. For instance, in 1797 the British authorities removed the rebellious inhabi­ tants o f St Vincent (an island in the Lesser Antilles) to Belize on the mainland. Racially, these were a m ixture o f Indians and black slaves who spoke an Arawak language called Island Carib. This resulted in the creation o f a new dialect o f Island C arib - known as Central American Island Carib, Cariff, Black C arib or G arifuna - which by the twentieth century had developed into a separate language, now one o f the Arawak languages with the largest num ber o f speakers (Taylor 1977b). The overwhelming m ajority o f A raw ak languages are now endangered. Even in the few com munities with over 1,000 speakers, a national language (Portuguese or Spanish) or a local lingua franca (Lingua G eral A mazônica, Q uechua or Tucano) is gradually gaining ground am ong younger people. A massive switch to Lingua Geral A m azônica took place around 1900 in the region o f the R io Negro, and resulted in the rapid loss o f a num ber o f languages. N um erous dialects o f Baniwa o f Içana spoken on the Lower Içana in Brazil are almost extinct. Tucano is rapidly replacing Tariana in the Vaupés river basin; Yawalapiti is yielding to K am aiurá as a lingua franca o f the Xingu Park (M ujica 1992). The few healthy A raw ak languages are G uajiro in Venezuela and Colom bia (esti­ mates vary from 60,000 to 300,000 speakers) and G arifuna in C entral America (from 30,000 to 100,000). The C am pa languages (total estim ate 40,000-50,000) are one of the largest indigenous groups in Peru. The majority o f materials collected between 1600 and 1900 consisted of word lists, phrases and a few paradigms. For languages which became extinct eariy (e.g. Taino, Caquetio, Shebayo) just a few words survived in the early Spanish chronicles. There are materials on a few Arawak languages spoken in the northern regions of South America (Achagua, M aipure, Island Carib mixed pidgin, L okono Arawak) which go back to the mid seventeenth century. Valuable materials - including word lists, and sometimes gram m atical notes, short texts and dialogues (though not all equally reli­ able) - were collected by travellers (for instance, Wallace 1853, M artius 1867, Von den Steinen 1886, Chaffanjon 1889, K och-G riinberg 1911,1928,Nimuendajii 1932). M ost o f the m aterials on Arawak languages collected during the second half of the twentieth century are by SIL linguists. Their quality and quantity varies. A t present, no A raw ak language - with the possible exceptions o f Lokono and 3 Arawak 73 Resigaro - has been provided with a comprehensive gram m ar (phonology, m or­ phology, syntax), dictionary and text collection. (See table 3.1.) 1.1 Comparative studies, genetic classiflcation and subgrouping Com parative and historical studies o f the Arawak family have a long history. The genetic unity o f Arawak languages was first recognized by F ather Gilij in 1783, three years before Sir William Jones’s fam ous statem ent about Indo-E uropean. The recognition o f the family was based on a com parison o f M aipure, from the O rinoco Valley, an d M oxo from Bolivia. He nam ed the family M aipure. Later, it was ‘renam ed’ A raw ak by Brinton (1891) and Von den Steinen (1886), after one o f the m ost im p o rtan t languages o f the family, Arawak (or Lokono), spoken in the G uianas. This nam e gained wide acceptance during the following decades. C om parative studies o f A raw ak languages initiated by Gilij were continued by Von den Steinen (1886) who proposed the first subdivision o f the A raw ak lan­ guages. H e distinguished N u-A raw ak and Ta-Arawak divisions (based on the form o f the Isg pronom inal prefix - see table 3.5). F u rth er studies were done by A dam ( 1890), B rinton ( 1891, 1892) and others. The limits o f the family were established by the early twentieth century. The Arawak affiliation o f a few languages earlier considered ‘problem atic’ has been proved within recent years. These are A muesha (Taylor 1954b), Resigaro (D avid L. Payne 1985), Iñapari (Valenzuela 1991), and Bahw ana (Ram irez 1992). T hough there are no doubts concerning the genetic affiliation o f the Araw ak lan­ guages listed in table 3.1, problems still exist concerning internal genetic relation­ ships w ithin the family and possible genetic relationships with other groups. Even the nam e o f the family has been a subject o f controversy. The m ajority o f native South Am erican scholars use the nam e ‘A raw ak’ (‘A ru ák ’) to refer to the group of unquestionably related languages easily recognizable by pronom inal prefixes such as mi- or /a- ‘ 1sg’, pi- ‘2 sg’, relative prefix ka- and negative ma-. A num ber o f schol­ ars, m ainly N o rth A mericans, prefer to use the term ‘A raw ak(-an)’ to refer to much more doubtful genetic unities o f a higher taxonom ic order, and reserve the term ‘M aipuran’, o r ‘M aipurean’ for the group o f undoubtedly related languages (see David L. Payne 1991, K aufm an 1990; and the introduction to this volume). H ere I follow the South American practice and use the nam e ‘A raw ak’ for the family o f def­ initely related languages, following Rodrigues (1986). R econstruction, internal classification and subgrouping o f A raw ak languages is still a m atter o f debate; further detailed work is needed on both the descriptive and the com parative fronts. As m entioned in the introduction to -th is volume, the putative studies o f A raw akan’ by M atteson (1972), N oble (1965) and others are deeply ñawed. 74 Alexandra Y. Aikhenvald Unfortunately, these have been adopted as the standard reference for the classifica­ tion o f A raw ak languages, especially am ong anthropologists, archaeologists and geneticists, influencing ideas on a putative proto-hom e and migration routes for proto-A raw akan - see the criticism in Tovar and De Tovar (1984). The earliest classifications - D e G oeje ( 1928), M ason ( 1950) and even Loukotka (1968) - were predom inantly based on the geographical distribution o f languages. A com parative vocabulary o f 67 Arawak languages with 151 cognate sets (all of mixed quality) was compiled by D e Goeje (1928) as a chapter o f his gram m ar of Lokono Arawak. Later com parative studies were m ade by Shafer (1959) (see the criticism in Taylor 1961, David L. Payne 1991)^ and Valenti (1986). The first truly scientific reconstruction o f proto-A raw ak phonology - over 200 lexical items a nd a few gram m atical morphemes - was published by David L. Payne (1991). However, his subgrouping o f Arawak languages, which is based on lexical retentions, rather than on innovations, remains open to discussion. Taylor (1977a, b) was the first to put forward scientific argum ents in favour of N orth Arawak as a separate subgroup. His contribution to Arawak com parative phonology and to com parative studies o f Caribbean Arawak (Island Carib, G arifuna, G uajiro, Lokono) cannot be overestimated. A few low-level reconstruc­ tions have been suggested recently (proto-L okono-G uajiro, see C aptain 1991; and proto-X ingu, see Seki and Aikhenvald forthcoming). The main problem of Arawak com parative linguistics and subgrouping is the lack o f adequate data for many languages. G eographical expansion and consider­ able linguistic diversity within the family pose the problem o f distinguishing areal from genetic phenom ena. This is crucial for m orphological reconstruction. In some cases the source o f areal diffusion can be established. Q uechuan influences are dis­ cernible in A m uesha (Wise 1976). Resigaro was strongly influenced by Bora and Ocaina (W itoto family). The gram m ar of Tariana was drastically restructured under the influence o f East Tucano languages during the past 300 or 400 years (Aikhenvald 1996a). In other cases the origin o f a substratum is n o t known. The Arawak languages o f Peru show structural similarity to neighbouring non-A raw ak languages (H arakm bet, W itoto). P alikur is said to have arisen from the merger of 8 dialect or language groups (D iana Green p.c., 1996); these unknow n substratum languages may account for the fact that it is one o f the tnost divergent N orth Arawak languages, and has lost a num ber o f com m on A raw ak features. The history o f the Arawak language family offers interesting examples of language contacts. The creation of a ‘mixed’ language o f A raw ak-Carib origin in the Lesser ’ He established 18 phonetic correspondence sets between 36 Arawak languages. This was a start; unfortunately most of these data were of mixed quality and strongly biased towards N orth Arawak. O f the 36 languages, 13 are, in fact, dialects o f Baniwa o f Içana/Kurripako, and only 3 languages are not N orth Arawak. 3 Arawak 75 A ntilles is one of th e m ost surprising pieces of evidence on language h isto ry in pre­ conquest times. Speakers o f Iñeri, a dialect o f the Arawak language now (mislead­ ingly!) known as Island Carib, were conquered by C arib speakers. Their ‘mixed’ A ra w a k -C a rib pidgin survived until the seventeenth century (Taylor 1977b, Hoff 1994). In this pidgin, the ‘speech o f men’ and the ‘speech of women’ were dis­ tinguished in the following way. Women used morphemes and lexemes o f Arawak origin, while men used lexical items o f Carib origin and grammatical morphemes mostly o f Arawak origin. This pidgin coexisted with the Carib used by men and the Iñeri used by women and children. This diglossia gradually died out with the spread of competence in Iñeri/Island C arib am ong both men and women. However, as a result. Island Carib underwent a strong lexical and gram matical influence from Carib. A num ber o f gram m atical phenom ena distinguish N o rth A raw ak from nonN o rth A raw ak languages (see also Tovar and D e Tovar 1984: 120-45). T hough there are a few com parative studies o f various subgroups (Wise 1986, 1990a, 1991a and b; D erbyshire 1986; C aptain 1991), further investigation is needed to decide w hether this division is genetic, or is due to different patterns of areal diffusion, and exactly w hat the subgrouping is. For the time being, we can only be certain o f sub­ groupings o f Arawak languages on a very low taxonom ic level (e.g. X inguan lan­ guages, S outh Arawak o f Brazil, Pi-Ap). The prelim inary subgrouping in table 3.1 is based on this areal-geographical principle."* The linguistic argument in favour of an Arawak proto-hom e located between the Rio Negro and the Orinoco river, or on the Upper Amazon (advocated by Lathrap 1970, Oliver 1989') is the higher concentration o f structurally divergent languages in this region than in other Arawak-speaking areas. This is highly suggestive and cor­ roborated by a few mythical traditions concerning a northern origin am ong Arawakspeaking peoples in southern regions (e.g. in Xingu). The origin myths of the Tariana and Baniwa in the north also suggest that they came from the Caribbean coast. The expansion of Lokono-lñeri from the N orthern Amazon to the islands in the Caribbean is estimated to be quite recent (ju st a few hundred years before the conquest). 2 PHONOLOGY 2.1 Segmental phonology A core of segmental phonology found in most of the languages is presented in tables 3.2 and 3.3. The likely phonem e system of proto-A raw ak comprises these consonants ■' In all subsequent lists o f Arawak languages I follow the order given in table 3.1. ■' Note, however, that Oliver’s scheme is based on N oble’s (1965) and M atteson’s (1972) classifications and his own lexico-statistic comparisons o f highly unreliable data from 30 languages. 76 Alexandra Y. Ä ikhem ald Table 3.2 Composite statement o f consonants in Arawak languages labial voiced stop voiceless stop aspirated voiceless stop voiceless fricatives voiceless affricates laterals vibrants nasals glides (b) p (p'') ((jj) dental alveolar lamino-(alveo)palatal velar glottal k (?) (g) (k") (t*>) rtj w Table 3.3 Composite statem ent o f vowels in Arawak languages front high middle low central i, K e, e: a, a; The following comments expand on tables 3.2 and 33. (1) N orth Arawak languages have some variation o f places o f articulation; only bilabial, apical, velar and glottal are found in LAr. Aft has bilabial, dental, palatal, velar and glottal. Blç has the maximum of seven; bilabial, dental, alveolar, alveopalatal, palatal, velar and glottal. There is an additional set o f palatal stops in Re (palatalized alveolar stops), Bhw (palatalized dental stops) and Am (palatalized bilabial stop). All stops have a palatalized counterpart in Perene Ca (J. Payne 1989). Blç distinguishes dental and alveolar stops. Wp may have glottalized stops. (2) Phonemic voice distinctions in bilabial and dental stops are found in most North Arawak languages and, possibly, in No (Wise 1990); they are absent from most o f the other languages. Diachronically, stops are unstable; thus, p> h in Ya, Pa, Acha, Mild, Mn, a n d ? i n Bhw; / > 0 before a front vowel in Yu, Acha; d>li or 0 tn Am and Ba; rf> 0 before a back vowel in Re, Pia, Ca, Blç, Ta, Yav-WX; before a front vowel In Ya, Re, Yu, Acha, Ca, Blç, Ta; ^ > 0 in W X/BGua. and also in Te, Bau, Ig, Pi. (3) The phoneme b is often restricted to onom atopoeia and a few loans (e.g. Pa, Blç, Ta, Ba); or results from a phonological process (WX); b and p are in a complementary distribution in Bhw and Acha (b corresponds to p in other languages before *a\ p corresponds to h before /, «). Mw and Yav have no bilabial oral stops. (4) A voiced velar stop is found only in Re and in Ga. In Re it developed from the vibrant and from *d, possibly due to the areal influence o f Bora and W itoto languages. (5) A glottal stop is only found in Te, Bau, Ig, Cha, Re; Wa, Mw, Guaj; Yu, Acha, Pia; Ka and Bhw. G lottalization is a word prosody in Wa, Ya atid Ta (§2.2). ( 6 ) A full set o f aspirated voiceless stops is found only in some N orth Arawak languages (1C, Re, Yu. Blç and Ta). Ba has pit and kh', and lit is extremely rate. These languages J Arawak 77 Tables 3.2 an d 3.3 com m ents (cont.) (7) (8 ) (9) (10) (11) (12) (13) developed aspirated labial and velar stops as the result o f a metathesis with h (which, in turn, comes from *k before a high vowel). An example o f the sequence stop- V-h in proto-North-A rawak is *ku:paki ‘fish’> *ku:pahi>Blt;, Ta ku:phe. LAr also has aspirated apical and velar stops; both LAr and G a have a voiceless fricative/ developed from ph (David L. Payne 1991). Aspiration in stops may have been innovated independently by different Arawak subgroups (pace David L. Payne 1991). Aspirated voiced labial and dental stops are found only in Ta; they result from vowel fusion across a m orphological boundary, e.g. di-hipa (3sg.nfem -take)>tfti>a *he takes’. Bilabial fricatives ate rare.‘ Voiced labiodental i>is found in Pi, Ap. Ca and G a, and Ka; Bhw has (p. Am has a voiced p. Only Re distinguishes/ and v. M ost languages have two sibilant fricatives. Some N orth Arawak languages have just one: it is s in G a. LAr, Yu, Pia, and / i n Pa; Aft; WX; BIç, Ta. A voice contrast in sibilant fricatives is found in Re (alveolar z and s) and in Am (alveopalatal 3 and J). BIç and Acha have alveopalatal 5 (derived trom * r before a high vowel). Alveopalatal fricatives are retroflexed in BIç (voiced) and in Am (voiceless). A distinctive innovation o f Ya is an alveodental fricative r (from proto-Xingu *kH_u, o r from *tf: Seki and Aikhenvald forthcoming). Am is unusual in having a voiced velar fricative r. A glottal fricative / 1, or a velar fricative x, has restricted occurrence in Ap, Wp, Afl, BJç, Ta, Ba, Wx (see §2.3) and may alternate with zero in BIç, Ba and Bhw. M ost South and South-W estern Arawak languages (Ig, Wa, Ya, Pa, Pi, Ap; Ca; Am, Cha, Re) have two affricates, while N orth Arawak languages tend to have either alveolar ts o r alveopalatal i/w h ic h are often allophones. There are no aflfricates in Te, Pr, LAr and IC. Am also has a palatal retrofiex tf. Voiced alveolar d z and alveopalatal d j are found in Re. Alveolar d z in BIç corresponds to y in many Kurripako dialects, e.g. Hohôdene dm w i, Kum andene yawi ‘jí^ u a r’; in W X d z and d j are variants o f y before high vowels. Re, BIç, Ba, have aspirated affricates. A typical Arawak system is similar to other languages o f Amazonia in having a single liquid phoneme with a flap and/or, more rarely, lateral articulation, as in Bau, Ig, Ya, Ap, Ca, Mw, Wp, An, Yu, Acha, Pia, WX. Ba. Bau and Ig have an alveolar vibrant, and A n and G a have a trill. Palatalized laterals are found in some-Ca dialects and in Ya. There are four alternative systems: (i) one rhotic, one lateral, as in Te, Wa, Pa, Pi, Am, Cha, Pr, G a, K a, Yav; (ii) two rhotics, no lateral: a lateral flap and a trill in Guaj; dental and velar vibrants in Bhw; and-sinaple and preaspirated flaps in Biç; (iii) three liquids: flap, tap/trill and lateral in LAr and trill, flap and lateral in Ta; (iv) neither a rhotic nor a liquid; Re (Allin 1975:67), where g, d, n are reflexes o f proto-Arawak *r, possibly as the result o f areal diffusion from Witoto. Aschmann (1993) reconstructs r as an allophone o f rfin proto-W itoto. Ba, Blç and Ta have preaspirated nasals, while Re has voiceless nasals. These are historically derived from a metathesized sequence, nasal + A, a process similar to the one described in ( 6 ) above, e.g. Blç hnua<nu-ha ’lsg-EM PH ’. n is often unstable in syllable-initial and syllable-flnal positions. In Te it is replaced by ^ A few languages (Bau, Ig, Pia) have jJ (Baptista and Wallin 1968, David L. Payne 1991) in free alternation with w before - 1, -e and -0 and in some forms before suffixes. G uaj also has jS in free alternation with w in some cases (José Alvarez p.c.). 78 (14) (15) (16) ( 17) (18) (19) (20) Alexandra Y. Aikhenvald Tables 3.2 and 3.3 c o m m e n ts (cont.) nasalization on the following vowel word-initially, and in W a by nasalization on the preceding vowel word-flnally. The phoneme ñ is found everywhere except Bau, Ig, Pr, LAr, Acha, Pia, WX. Re, BIç. Ta, Ba also have a preaspirated bilabial glide. A lternation o f p and tv is characteristic o f South Arawak Bau and Ig. Nasals and glides form a natural class, being the only phonemes found in syllable-final position in almost every language (§2.2); consonantal length in Pi is phonemic only for nasals and for glides. The minimal system of three vowels found in Am. and sotne dialects o f Ca in Peru, consists o f high front unrounded (/ o r e), high back rounded (« or o) and low (a). Languages can add to this system; (a) a contrast between two front vowels, / and e, as in South Arawak languages Te, Bau, Ig and Pa and C a, in the Southwest, and in Pia, BIç, Ba, WX from the N orth Arawak group; (b) a middle central vowel j, as in Ya; (c) a combination o f (a) and (b), as in Wa, Ap, LA r and Bhw; (d) a combination of (a) and a contrast between the high and middle back unrounded vowels » and u, as in Pi, Re, and Cha, Pr and Yu; (e) a combination o f (d) and (b), as in N orth Arawak languages Wp, Mw, G a, Guaj, Añ; and in Ta. Glottalized vowels are phonemic in Re and Yu. In Am they arise from the loss o f a velar stop (David L. Payne 1991 ). Nasalized vowels usually occur contiguous to nasal consonants, o r to It (Ya, Yu, WX, BIç, Ba). They are phonemic in Wa, Ap, Wp, Mw, Ta. Extensive vowel nasali?;ation in Ta is due to areal diffusion from East Tucano, All Arawak languages - with the exception of Te, Bau and !g, in the south, and Pr, Bhw and Ba in the north - have phonologically long vowels. A contrast between two back unrounded vowels may be a secondary development in some N orth Arawak languages: in Yu o is derived from the vowel sequence mi; and in Ta o is the result of areal diffusion. and vowels, excluding those in parenthesis. These will need to be confirmed by systematic reconstruction. (They are similar to the set proposed by David L. Payne 1991.) 2.2 S y lla b le s tru c tu re , su p raseg m en tals a n d p h o n o ta ctics In most languages syllable structure is (C,)V(V)(C,), where C^ can be any consonant, an d C , can only be a glide (w o r /) , a nasal, or, rarely, a liquid or h. Only Am and Cha, in the south, and P r in the north allow other consonants in the coda position. In Bau and Wp closed syllables arise from vowel elision in rapid speech registers, e.g. Bau firofcon-flcte»Tuir(Baptista and Wallin 1968:9); Wp, norm al speechdap, slow speech dap“ ‘house’ (M anoel Santos p.c., 1993). C onsonant clusters are found only in Pi. Wa and Pi at^ unusual in yet another way: ? or h is inserted before a word-initial vowel. 3 Arawak 79 T here are a few phonotactic restrictions. The phonem e h often occurs only wordinitially o r word-finally as a w ord boundary m arker (A ikhenvald 1996b). In G uaj the glottal stop cannot occur at word boundaries. Word-initial liquids are found only in cross-referencing prefixes (table 3.5). Stress is contrastive in som e languages (Bau, Wa, Ya, Ba, W X, Acha), but n o t in others (C a, Bhw). Nasalized o r long vowels are often stressed. Stress assignment may depend on syllable weight (Ashe: J. Payne 1990; G uaj: Alvarez 1994). There are a t least two tone languages - Te in the south, and Re spoken in the far northeast o f Peru (formerly spoken on the C aquetá river in Colom bia); N om atsiguenga also distinguishes two tones (D avid L. Payne 1997), Te has two tones - falling an d level.’ Re distinguishes high an d low tones, similar to the neigh­ bouring W itoto languages. Some N o rth A raw ak languages o f the U pper Rio N egro and C olom bia (A cha, Pia, BIç, Ta) have pitch accent. G lottalization, aspiration and nasalization are often word prosodies in N o rth A rawak an d in the X inguan languages, while /j-insertion and translaryngeal vowel harm ony are used to m ark a phonological word followed by a pause (A ikhenvald 1996b). 2.3 Morphophonology The following m orphophonological processes are found in Arawak languages, usually at m orphem e boundaries within a gram m atical word: vowel harm ony; m etathesis o f stop and h; vowel fusion. M etathesis o f stop and h yields preaspirated consonants in Re and in a num ber o f N orth A raw ak languages (Allin 1975: 184; Aikhenvald 1995a, 1996b). N asalization and vowel harm ony ate word prosodies in Te an d in the closely related K inikinao, G uaná an d C hané (Bendor-Samuel 1966), and are used to m ark 1st and 2nd person respectively; in these languages prenasal­ ized stops an d fricatives arise as the result o f nasalization. M any N orth A raw ak la n ­ guages have progressive translaryngeal vowel harm ony, e.g. Ba wa- -I- -ihisa> wahasa (Ipl-look for) ‘we look for’. ^ Tone in Te is associated with a unit word rather lhan a syllable; it is assigned to one o f the first three syllables in a word (the placement depends on the m orphological structure - Ekdahl and Butler 1979). Falling tone entails vowel lengthening (Ekdahl and Butler 1979); Te lacks phonologically long vowels. A preliminary com parison o f cognates shows correspondences between falling tone in Te and long vowels in some N orth Arawak lan­ guages; e.g. falling tone: Te h’êe; Blç, Ta ku:phe ‘fish’ (proto-Arawak *kopaki: David L. Payne 1991); level tone: Te kámo, BIç, Ta, WX (h)em a ‘tapir’ (proto-A raw ak *kema: David L. Payne 1991). T he nature and origin o f tones in Te requires further investiga­ tion. 80 Alexandra Y. Aikhenvald There is the following interdependence between syllables and morphemes: prefixes; alm ost all monosyllabic suffixes: mostly monosyilabic, a few disyllabic roots; mostly disyllabic, a few monosyllabic.* A m orphological word usually contains at least one root morpheme; m ono­ syllabic verbal roots have an obligatory prefix (or a suffix, to make them bimoraic, as in Asheninca: D avid L. Payne p.c., 1997). Thus, every gram m atical word contains at least two syllables. 3 TY PO LO G ICA L PROFILE 3.1 Word structure All Arawak languages are polysynthetic and predom inantly agglutinating w ith a few elements o f fusion. They are mostly head-m arking. This accounts for the lack o f core cases used for m arking gram m atical relations in the m ajority of A raw ak lan­ guages. Elements o f dependent marking found in Ta, in the n orth, and in Ap, from the Piro-A purina subgroup, may be due to areal diffusion. Arawak languages usually have a rich noun class and/or classifier system and a separate system o f two or three genders restricted to pronouns and cross-referenc­ ing markers. N om inal m orphology is fairly sim ilar across the family. Verb m o r­ phology varies m ore and can be very com plicated, especially in South Arawak and South-W estern A raw ak languages, and also in some N orth Arawak languages (G uaj, Ta). 3 .2 Morphological processes Arawak languages are mostly suffixing, with just a few prefixes. These are: A/S, pro­ nom inal prefix to verbs and possessor prefix to nouns; relative-attributive ka- and negative ma- to nouns and verbs. Causative and applicative prefixes on verbs are found in S outh A raw ak, Pared-X inguan and Peruvian languages. All types o f suffixes are historically less stable than prefixes; there is much more diversity even am ong closely related languages. M any A raw ak languages have reduplication o f stem -initial CV- (Te, Pr), or • All nominal stems which contain more than two syllables in the proto-Arawak reconstruc­ tion proposed by David L. Payne (1991) can be decomposed into a disyllabic root and a suffix, or are bans. 3 Arawak 81 CV CV - (Te, Wa, Bhw); or o f stem-ftna! -V (Te) or -CV (Ig, Ba, WX). Reduplication is used m ostly with verbs, to m ark intensive, repetitive o r continued action, or plurality o f participants. In W X , reduplication o f an intransitive verb generally indicates intensive action; alternatively, it may refer to (a) plural A /S w ith A = S anibitransitives; or (b) plural O with non-am bitransitives. Total reduplication of the ro ot is used in C a to express ‘continuing excessive action’. Inherently redupli­ cated roots are found in som e languages (Bau, Pi, G uaj, A cha). Som e languages have two (G uaj) or three (Te) kinds o f reduplication w ith different m eanings, e.g. Te CV-: há-harara káye ‘it is very red ’; CVCV-: tuvo'tuvo káye ‘(it) is m oving many times’. Te (together with G uaná, Chané, K inikinao) an d Wa (Xinguan) are unusual in employing apophony (internal vowel change) and vowel harm ony to m ark mood. In Te, realis is m arked with suffix -o, while irrealis is m arked w ith a suffix -a on active verbs and a prefix a- (or o- if the first vowel o f the root is o) on stative verbs; in irrealis, all the o vowels o f the root becom ea. Thus, (w rite-TH EM A TIC- R EA L-3sg + 0/Sj,) ‘he w rote it’ (realis), and yutá-f-á-a (w rite-T H E M A TIC -IR R 3sg + 0 /S g ) ‘let him write it’ (irrealis) (Ekdahl an d Grim es 1964: 263). Realis is unmarked in Wa, while irrealis is marked by a change a > e in the last syllable o f the verb (R ichards 1988; D erbyshire 1986). Suppletion is rare and is restricted to possessed forms o f some nouns, e.g. Te cavâne ‘garden’, issáne ‘his garden’; Me am aka ‘ham m ock’, a-iunumala ‘his ham m ock’ (M edeiros 1990).^ Some N orth A raw ak languages have irregular plural forms o f hum an nouns, e.g. W X neyawa ‘w om an’, pi. neye-pe; BIç, Ta ( a) tfä-Ji ' 'm an ’, pi. a :tfa. 3.3 Word claæes All Arawak languages have nouns and verbs; m ost also have an open class o f adjec­ tives. The size o f the adjective class varies from language to language. Adjectives share numerous properties w ith verbs; in G uaj they can be considered a subclass of stative verbs. In some N orth A raw ak languages (Ta, BIç) adjectives share som e properties with nouns and som e w ith verbs. ’ In one case partial suppletion o f a possessed form may go back to proto-Arawak. The common Arawak term for ‘hotise, home’ typically displays an alternationpelpan: of. Tercna péti, possessed pêno ‘house’; Ignaciano peti, possessed pena; Baniwa pan-(i, possessed -pana; Tarianapani-si, possessed -pana; Palikurpayl, possessed -pi-; Bahwana (pani-ci, pos­ sessed -ana; P ared hati, possessed -hana ‘house’ (David L. Payne 1991 reconstructs *pe and *panali as two different items). 84 (¥) Alexandra Y. Aikhenvald Only a few Arawak languages - IC (Taylor 1956; 31 -2 ), A cha (M eléndez 1989), Bhw (Ram irez 1992:46) and Pr - have a small set o f generic classifiers employed in possessive constructions with alienably possessed items. Bhw has one classifier for game, and one for domestic animals, e.g. m i-iR a ÄaátfÄíi (Isg-POSSV.CLrDOMEST IC .A N IM A L parrot) ‘my parrot’. Classifiers show great diversity from one language to an o th er in semantics and form. A classifier in one language often corresponds to a lexeme, or to a derivational suffix, in another. Classifiers appear to have developed on the level o f individual subgroups. Typical pronom inal genders are m asculine and feminine; Ig and P r also have a neuter gender.*^ N o genders are distinguished in the p l u r a l . T h e markers are uniform across languages, and they go back to proto-A raw ak third person singular cross-referencing markers (table 3.5). Feminine is the functionally unmarked gender in the Caribbean Arawak languages G uaj, Aft, LAr and lO G a ; in the other languages masculine is unmarked. G ender assignment is rather opaque in these languages, as well as in Ca and Pr; elsewhere it is straightforward; feminine gender is used for females, and masculine in all other cases. Pronom inal genders have been lost in Te, Am, Cha, Pa, Wa and Bhw; all these languages do have classifiers.” 4.3 Number All Arawak languages distinguish singular and plural on nouns. Typical markers are reflexes o f proto-A raw ak *-m /-ni ‘anim ate/hum an p lural’, *-pe ‘inanimate/ anim ate non-hum an plural’. Paucal is optionally distinguished from plural in WX and BIç. Re is the only Arawak language with a category o f dual in independent personal pronouns and in nouns; this m ust have evolved under the influence o f Bora and O caina (Allin 1975:164). There is a residue of dual in G uaj in verbal cross-ref­ erencing on about forty verbs (José A lvarez p.c.). in a few N o rth Arawak languages - Ta, BIç and Re - nouns w ith an inanimate referent have a collective meaning; a n oun classifier in a derivational function has to be added for singular reference, e.g. Ta de:ri ‘banana (collective)’, de:rs-pi (banana-C L :L O N G ) *a banana’. Only a noun with a classifier can be pluralized. Pi has two genders, and the relative-interrogative pronoun kla also has a special form for inanimate gender. There are two exceptions to this: Ap also distinguishes two genders in 3rd person plural; Ta has an optional distinction o f two genders in all persons in the plural in independent pro­ nouns. Wa distinguishes two genders in demonstratives only; Pa distinguishes two genders in nominalizations, and Te has an opposition anim ate/inanimate in adjectives and descriptive verbs. 3 Ärawak 85 e.g. de:ri-pi-pe (banana-C L :L O N G -pl) ‘bananas’. This feature is also found in neighbouring languages, G u ah ib o and Tucano (see chapter 14 below). 5 CLOSED c l a s s e s : p e r s o n a l p r o n o u n s , d e m o n s t r a t i v e s , INTERROGATIVES, NUMBERS, A D POSITIO NS Personal pronouns are used to em phasize the subject in topicalization, and in copuia clauses. They consist o f a cross-referencing prefix plus an em phatic one-syllable particle - see table 3.5, e.g. Pa na-tyo, Ba nu-ni, W X nu-ya. Ta m -h a ‘F .’* D em onstrative stems are always monosyllabic. In alm ost every language third person pronouns can be used as proxim ate dem onstratives, and as definite articles with- nouns. M ost languages have a two-term system, near/far {Wp, Re, Yu, Acha, Pia, BIç, W X) o r a three-term one, near/m id-distant/far {Pi, Ca, Ta, LAr, Ba). Am has ju st one dem onstrative pronoun {an ‘this/that’: Wise 1986: 572), and G uaj and Pr have four: near speaker and hearer, far from speaker and near hearer, far from both, very far from both. Interrogatives vary even am ong closely related languages; they can often be used as relative pronouns; in C a, A m, Ta, BIç they also function as indefinites. M ost languages have ju s t the numbers ‘o ne’ (proto-A raw ak *pa-) and ‘two’ (proto-A raw ak * (a)pi an d *yama: David L. Payne 1991). It is com m on to use numeral *pa- ‘one’ in the function o f an indefinite pronoun ‘one, som eone’, ‘another’. South Arawak, Ca, Am, Cha and P r have reflexes o f *mapa ‘three’ (Payne 1991: 421); o th er languages show different forms (Pa hanama; Ta, BIç, A cha mata, macta; G uaj apimûin, LA r kabyn', W p idikinhay 'day, Mw itukm a; Ba kuikunam a (cf A polista erikomj). The proto-language may have had only ‘one’ and ‘tw o’. Only a few languages (Pr, LAr, G uaj) have underived numbers up to ten. A dpositions (postpositions and prepositions; see §7.2) are often derived from body parts; there are some underived adpositions. A n adposition in one language can correspond to a verbal valency-changing suffix in other languages, e.g. Te -yaa ‘with, in, to ’. Pi ->w/-i‘dative/locative applicative’; Pa -kakoa, Pr k a k ‘w ith’. Pi, Ap, Te, Ig, Ta -kaka ‘reciprocal; associative’; S outh Arawak and Pi-Ap mUim- ‘com itative’, C a -imo ‘circum stantial applicative (in the presence of)’. N orth A raw ak lan­ guages -ima ‘w ith’ (cf. §6.3). 6 V E R B A L M ORPHOLOGY The verb is the m ost com plicated p a rt o f the gram m ar, and the only obligatory constituent o f a clause. '• In Ca, Am and Pi different particles are used to form pronouns o f different persons; e.g. Ashe 1sg n-aakaln-aro; 3sg masc ir-iro(ri); Am 1sg na\ îp l n-eht. In Pi 1»g and 3sg pronouns do not involve a pronominal prefix. 88 Alexandra K Aikhenvald Table 3.5 Composite statement o f AlSJpossessor prefixes, OIS^ suffixes/enclitics^' prefixes person sg 1 2 nu- o r ta(p)iri-, ithu-, upai-, a- 3ntem 3fem ‘impersonal’ non-focused A/S^ dummy S / 0 suffixes wa(h)inana- -na, -te -pi -ri, -i -thu, -u -wa -hi -na -na Notes: (1) First person singular forms «- and t- may have coexisted as variants in proto-Arawak. Their distribution is as follows; t- is found only in the extreme north (LAr, Gyaj and A i), while all the rest have n-.“ Exclusive/inclusive oppositions are atypical for the Arawak languages, in Ca (Reed and Payne 1983) and Te, the original Isg now means both ‘I* and *we (excl)’, while the original Ipl form now has a ‘we (inol)’ sense. In Ba and Ta impersonal forms are used in the sense o f 1st inclusive. (2) Second and third person plural forms underwent modification across the family. Com binations o f a prefix and a suffix are used for 2pl in Cha, and for 3pl in Pi and Ap. 2pl and 3pl fall together in Ba (me-) and WX (ni-). Pia has lost the reflex of *(h)i and uses the combination of pi- and -cué instead (Klumpp 1995), while Te has lost pi-, (3) Te, Am, Wa, Pa use a zero prefix for 3sg. Both forms are used in Bfç and a few other northern languages; their choice depends on the diæourse prominence o f the A/S, argument (Aikhenvald 1995b). (4) The evidence for the proto-Arawak origin o f *pa- ‘impecsonal’ (also used as reflexive and as ‘same subject’ marker) is as follows; Wa pV-, Pa ha- ‘3rd possessor coreferential with the subject’; Am p ( V )- ‘3rd possessor’ (Wise 1986: 570); W ppa- (Tracy 1974; 123-4), Mw a- Bhw tjm- ‘same third person subject’. Guaj pa- ‘reciprocal’; Yu p a -.. .(-c/rnca ) ‘reciprocal* (Wise 1990); BIç, Ta, WX pa-, Ba, Guaj ba- ‘impersonal’. (5) A prefix for non-focused A/S^, is found only in the North Arawak languages. The distribution is as follows: /- is found in BIç and Ta (Aikhenvald 1995a), Pr /‘indeterminate possessor’, and possibly Pia while a- is attested in Guaj, Añ, Ba; Pr; and in LAr and IC (Taylor 1977b; Peter van Baarle p.c.). (6) Suffix -ni is found in Pa -fsj/ie ‘object m arker’, Wa -ni ‘anaphoric 3rd person object pronoun’ (Richards 1991; 163); Am -a n ‘surface object m arker' (Wise 1986; 572), BIç and G uar have merged *-«/ and •-»; cf. BIç masc. j/ -, fern. Jii -; masc. -ni, fern, -m; G uar masc. ni-, fern, ice-; masc. -n i, fem. -we. Pr has a neuter ni- ‘definite’ as well as a(see Aikhenvald and Green 1998, for discussion). The majority o f prefixes and suffixes in table 3.5 are listed in M atteson (1972) and David L. Payne ( W l ) , Impersonal *pa-!-pa was added by Wise (1990). IC/Ga, closely related to LAr, have Isg «. Cha innovated ii-lw- for Isg (an influence from Ipl?). In Wp Isgisun-. In Terêna and closely related languages (G uana, Chané, Kinikinao), the A/S Isg prefix nu- has become a nasalization prosody, which continues until the first stop o r fricative; it yields nasal vowels and prenasalized stops. In Pia, /'- ‘3sg.masc’ is hom ophonous with a verbal cross-referencing prefix used with a 89 3 Arawak and as the pronom inal possessor to nouns; while O (direct object) is m arked with the sam e suflBxes as (subject o f intransitive stative verbs). A m ong these split erga­ tive languages are Te, Ig, Bau, Pi, Ap, Bhw, K a, LAr, BIç and WX.^'* A typical Arawak split ergative pattern is illustrated with examples (4)-(7), from BIç. Example (4) shows a cross-referencing prefix which refers to A, and a suffix which refers to O. (4) ri-kapa-ni 3sg.nfem + A-see-3sg.nfem + O ‘H e sees him /it.’ In (5) the same prefix is used for on a verb o f m otion; and in (6) it is used to indicate possessor. (5) ri-emhani 3sg.nfem + S^-walk ‘He walks.’ (6) ri-tfinu-ni 3sg.nfem + possR-dog-POSSV ‘his dog’ In (7), the suffix refers to the (7) argum ent o f a stative verb ‘be cold’. hape-ka-ni be.cold-DECL-3sg.nfem + ‘H e/it is cold.’ However, this pattern is either partially lost or reinterpreted in m any individual languages. For instance, in P r the split ergative pattern survived only with interrog­ ative predicates (Aikheiwald and G reen 1998; app. 1). Pr is also extremely unusual in that b o th suffixes and prefi.xes can be used for all o f A, S and O depending on verbal aspect (Green and Green 1972). In C am pa languages, the choice o f cross-referencing prefixes o r suffixes depends on a num ber o f factors: gram matical function, aspect, and the discourse prominence o f the argum ent. Prefixes are used for A and if the verb is in progressive aspect or fronted extracted (topicalized) A. The difference between /- ‘topicalized A’ and /- *3sg masc’ is that the former refers to any person and number. In Lokono and IC, the vocalic prefix is used on the main verb in a verb-auxiliary construction. Parauhano also has a vocalic suffix used to mark an indefinite patient (Patte 1987; 187); it may be an innovation to m ake the system symmetrical. Possibly also Mn (De Goeje 1948: 158) and G u ar (Oonzález Ñáñez 1991). In Am suffises are used for O and for o f equative and locational predicates; one ‘irregular’ m otion verb, aw ‘go’, takes S suffixes. BIç has elements o f fluid S marking; for a discussion o f ergativity in BIç and WX. see Aikhenvald (1995a). 90 Alexandra Y. Aikhenvald if this argum ent is topically discontinuous and less foregrounded, while suffixes are used for and O, and also for if it is topically continuous and foregrounded, or is in stative aspect (David L. Payne and J. Payne 1991; also see Dixon 1994; 211). Languages which have preserved suffixes ju st for 3rd person continue using pre­ fixes for A/S,. A m ong these, Wa, P a and W p use a suffix to m ark O, while Acha, Yu and Pia use 3rd person suffixes to cross-reference a new topic in A/S^ function. G uaj and An have two kinds o f verbal conjugation: suffixal (or ‘analytic’) and préfixai (or ‘synthetic’). Cross-referencing suffixes in an analytic verbal conjugation refer to the subject (A/S) o f a verb which is low on the transitivity hierarchy. Crossreferencing prefixes used in a préfixai conjugation with transitive verbs with a ref­ erential definite object (Alvarez 1994: 91) refer to the A o f a verb high on the transitivity hierarchy, and cross-referencing suffixes refer to the O. In subordinate clauses, prefixes m ark A/S^ while suffixes markO/S^. Ya, Re, Ta, Ba have lost their cross-referencing s u ffix e s .R e , Ya and Ta have no split ergativity; Ba and WX have elements o f a S/O pivot (also see §7.6). Ta has inno­ vated a case system o f a nominative-accusative type (see §7.1). Example (8) illus­ trates object case m arking on a pronoun in Ta. (8) diha he:suB JE C T di-ka-m akha 3sg.nfem-see-REC.PAST.N0N. v i s u a l nu-na Isg -O B jH c r ‘He has seen m e.’ Some N o rth A raw ak languages (WX, LAr, Acha, Blç, Ta, and Ba) have a small class o f intransitive verbs, the S o f which is marked with a dative adposition or with an oblique case (Ta). They refer to physical and mental states. O a (M unroe ms.: 6ff.) has several subtypes o f ‘oblique’ subjects (m arked with prepositions): dative, instru­ mental and benefactive, used with verbs denoting physical or mental states. Transitive verbs o f knowing and liking take dative subjects. 6.3 Valency-changing derivations M ost Arawak languages have both valency-increasing an d valency-decreasing derivations. Valency-increasing derivations are m ore com plicated in South Arawak, Pareci-Xingu, Piro-A puriná, C am pa and Arouesha than they are in N orth Arawak. All Arawak languages have m orphological causatives. The valency-increasing (causative)m arkers found throughout the family me. (-ji(~ j\a -; (-)ta (-) , (-)m i(-), ( - ) k a (-) (Wise 1990). N orth Arawak languages tend to have ju s t one derivation,** while S outh A raw ak and Pre-andine usually have more. T he possibilities are: “ Mp may also have done so (Zamponi forthcoming). Onty Wp distinguishes direct and indirect causation (-ki and -kizi(i)). See Aikhenvald forthcoming-a, on different causativizing mechanisms in Ta. 3 Arawak (i) 91 causatives o f intransitive or am bitransitive verbs are m arked differently from tliose o f transitive verbs (e.g. the distinction o f i- and ko-!ka- in Te). (ii) in Am and a few C a languages a causative prefix and a suffix are used when the causer is directly involved, otherwise only a suffix occurs. (iii) forceful causation may be distinguished from non-forcefui causation, as in M a; Pi, and Ap (Wise 1986, 1990). Some C a languages have a special causative ‘m ake do by proxy’ (Shaler 1971 ; 33). W ithin the N orth A raw ak branch, G uaj {-ira), A ñ (-jera), IC (-god(a)), LA r (•kyty), an d Re (-i«) form causatives on both transitive and intransitive v e r b a I n other languages, m orphological causatives are form ed on intransitive verbs^* only (Acha -da, Yu, Bhw, Ba, W X -la, Pia -klalda; Ta, BIç -i-(ta)}. Periphrastic caus­ atives are then employed to causativize transitive verbs. South A raw ak, Piro-A puriná, C am pa and A m uesha have very com plicated applicative derivations which p u t an oblique constituent into the O slot. These valency-increasing derivations apply to comitative, dative, benefactive o r malefactive, reason, purpose, instrum ent, allative, elative, presential (i.e. in the presence o f somebody) an d to any ‘thetnatic’ argum ent (i.e. the one with reference to which the action is done).-’ T he oblique constituent which becomes O is cross-referenced on the verb, as shown in (9), from A sha (K indberg 1961: 537): (9) y-ow et|ik-i-ni-ro 3masc-make- n o n .f u t - ben -3 fern ‘He m ade it for her.’ In m ost Peruvian languages applicatives are typologically unusual. O ne verb can contain several applicative markers. The applicative suffixes show that a peripheral In ü u a j, causatives cannot be tbnned on stative verbs. 1n W X, Ba, Ta and Blç causatives are also derived from transitive ingestive verbs (e.g. ‘eat, drink’). ® The num ber o f morphemes, and their behaviour, diffei-s from one language to another. Thus, ‘purpose’ applicative requires a reflexive in Ashe; in No and PC purpose + reflexive indicfates purposeless action. In Asha, presential applicative -mo is also used as a benefac­ tive and comitative m arker unlike other Ca languages (Wise 1990:111 ). In Am, -amp’' ‘the­ matic argum ent advancem ent’ is often used idiosyncratically, to refer to a vehicle in which the action is performed, o r to a meteorological condition. The same morpheme is often used for causative and comitative applicative derivations (e.g. Tr im-lem. Wise 1990; 98; Ap mi, Wise 1990; 108; N o and other C a -agl-akal-ag. Wise 1986; 593-4). Comitative applicatives and causative meanings can be ambiguous, e.g. Ma: no-panki-t-ag-ak-e-ri l-plant-EPENTHETIC-CAUS/COM IT-PERFV-NOK.FUT-3sg.maSC ‘I ordered h im to p l a n t / i planted w ith him.’ (Wise 1990:95) 92 Alexandra Y. Á ikhenm ld constituent has become obligatory. Example (10), from PC (Shaler 1971: 45), con­ tains an applicative derivation, -ako ‘benefactive; with reference to ’; there is another suffix -imo ‘in the presence o f’ which indicates that a peripheral constituent (some­ body in whose presence the action was done, in this case, Irene) has to be overtly expressed. (10) no-p-ako-ts-im o-tsi-ro-ri Isg-give-REFERENCE.TO-EPENTHETIC-IN.PRESENCE.OF-ASPECT- 3sg.fem-3sg.masc Irena Irocarto paño Irene R ichard scarf ‘I gave R ichard the head s carf in Irene’s presence.’ Example (11), from Pi, shows that one verb can contain two causative markers, two applicative derivations - com itative and dative - and an anticipatory passive (M atteson 1965; 81). (11) r-u -m u n -k a k -y eh itx -i-k o 3masC-C0MIT-C0me-CAUS-MANDAT0RY.CAUS-DATIVEANTICIPATORY.PASSV ‘H e is to be com m anded to be caused to com e concerning som ething.’ Pa, Wa, Pi and also Guaj have a different strategy for deriving applicatives. They incorporate adverbials and postpositions; the argum ent o f a postposition then becomes an O or a derived S (cf the discussion on how an adposition in one lan­ guage can correspond to a verbal valency-changing suffix in another language, in §5). Example (12) illustrates the incorporation o f the postposition naku ‘in’ in V/a. (1 2 ) n u -h u k u -n a k u -t-e n e Isg-wash-IN-CAUS-PROLONGED m a k u la bow l ‘I am washing (the inside oQ a bow l.’ ' In contrast, m ost N orth Arawak languages have fewer applicatives. Guaj has a suffix -hiraa which m arks comitatives (Olza Zubiri and Jusayu 1978: 227-9; Wise 1990: 100-1) and also reflexives and reciprocals.^® Lo (Pet 1987: 46) has non-productive ‘them atic’ suffixes which m ark derivations similar to purposive applicatives, e.g. áiba ‘go away’, áiba-ha ‘pursue’, áciiga ‘to fell a tree’, ácuaha ‘to go for w ood’. M ost A raw ak languages have one agentless passive. N o rth Arawak languages tend to employ the same m arker for reflexives and reciprocals (A ikhenvald forthAccording to the explanation in Olza Zubiri and Jusayu (1978; 228), the main meaning of this valency-changing derivation is that ‘many subjects take part in the action’, 93 3 Arawak coming-b). Pi, Pr, Ta and G uaj have agentive passives. Passive in P r is illustrated in (13) (A ikhenvald and Green 1998). (13) g u -ap itiw 3fem -h ea d w a n a k -p ita -k a a -k a k tie-vCL;iRREGULAR.SHAPE-PASSV 3 n e u t-w ith m aw ru c o tto n ‘Its h e a d [o f th e rattle ] is tied w ith c o tto n .’ Passive markers typically contain -k, -n and/or -vv; the marking is often quite dilferent even between closely related languages. G uaj is unusual in that it has two passives; -na and -uu; the latter implies a special effort from the agent (Olza Zubiri and Jusayù 1978; 233-9). Reflexives and reciprocals are typically marked by verbal suffixes. Reflexive pro­ nouns are found in LA r and G a, and in Mw.^' 6.4 Other verbal categories All A raw ak languages have rather complex systems o f tense-aspect, m ood, m odal­ ity, directionals and aktionsarts; only a few have evidentials. These are alm ost always expressed with optional suffixes or e n c litic s .L a n g u a g e s show great vari­ ability in the categories and forms used.^^ The m ajority o f South A raw ak, Pareci-Xingu and Peruvian Arawak languages have a three-fold aspect distinction: completive (com pleted, perfective or telic action); progressive (action/state in progress; also a durative meaning); and habit­ ual. A few languages have a future marker. O ther aspect-like distinctions include ' m om entaneous, prolonged, im m inent action, custom ary, etc. Peruvian Arawak lan­ guages (C a, Am and Cha) as well as Te and Wa distinguish future/irrealis and nonfuture/realis (Wise 1986; 586-7).’'* M any N o rth A raw ak languages distinguish ju s t past (or completive) and non­ past. Yu, A cha, P ia-ra; ‘past, com pletive’ and R e - m i‘recent past’ (Allin 1975; 342) Pet ( 1987) reports that reflexive pronouns are ousting reflexive derivations in young people’s LAr; on IC, see M unroe (ms.), Taylor (1977b). This may reflect the influence o f neigh­ bouring languages - English, Spanish or Creoles. Aspect m arking is obligatory in Te (progressive -ti vs 0 ); it is neutralized in negation and under topicalization. Payne (1991; 380-1) suggests a proto-Arawak origin for a number of directional {*-ape ‘arriving’, *-ane ‘leaving’), aspectual i'*-pe ‘perfective’, *-ena ‘progressive, gerund’, *-kha ‘perfective, habitual’) and m odal (*-j;a ‘future, irrealis’, *-mi ‘conditional’, *-ka ‘interrog­ ative’) suflixes. There are a few more recurring suffix shapes, e.g. W p -nina ‘non-present’, BIç, Ta -nina ‘remote past’; Acha -mina ‘dubitative’ and Bhw -miña ‘desiderative’; Pr -ki ‘veriflcational’ and Ba -Vki ‘verificational, emphatic’. However, more comparative and descriptive work remains to be done to establish the nature o f these correspondences. ” In Caquinte (Swift 1988: 209), non-future reflexive is a portm anteau morpheme (-n), but future reflexive is not {-e-Npa ‘future-reflexive’). 94 Alexandra Y. A ikhenm ld are cognate with Ta, Blç, WX, Ba -m i ‘past, used, pejorative’. Ta is unusual in having a com plicated tense-aspect-evidentiality system possibly due to areal diffusion from Tucano (see chapter 14 below). There is typically ju st one imperative which involves a zero-m arked verbal stem. A few languages have politeness distinctions in imperatives (Pi -i/i and A cha ~ju ‘polite im perative’, P r -m u ‘intim ate request’, -k a ‘perm issive’; Wp -na ‘strong im perative’). T he rich system o f im perative clitics in Ta m ay again be the result o f areal diffusion. M arking o f interrogative m ood is restricted to Pi, Ca, G uaj and Ta. A declarative m ood m arker -ka is a characteristic feature o f N orth Arawak lan­ guages (and is also present in Pi). Ig, Bau, Pa, Wa and the Peruvian languages have rich systems o f m odality (e.g. desiderative, reason, frustrative, m istaken, optative, pretence, purposive, assertive, dubitative - see O tt and O tt 1983; Wise 1986; 598). N o rth Araw ak languages, with the exception o f G uaj, Pr and Ta, have fewer distinctions; but they often have o p ta­ tive, necessitative, potential, dubitative, conditional and frustrative. E v id en tial are rare, with just a few languages having a m arker for ‘reported’, o r ‘hearsay’ (e.g. Te, Ig, Wa, Pa, Pi; A cha, Re, Pia, Blç, LA r). Bhw (Ramirez 1992; 64-5) distinguishes -bi ‘reported, q u o ta tio n a f and -hi ‘inferred’. Ta has a rich system o f evidentials which emerged as the result o f areal diffusion from Tucano. South Arawak and Peruvian languages have a num ber o f verbal affixes which indi­ cate the direction o f movement o f the subject (sec J. Payne 1982 on their tem poral and discourse functions), e.g. Te -opi-pol-p. Pi, Ca -ap, A m -ahp, Ig -pa ‘point o f time reference’, Bau -piko ‘tow ards’, Wa -pe ‘tow ards’; A she -an, Asha -ahant ‘from ’; Am -an, n ‘leaving’, Bau -na ‘action to one place then to another’, Wa -ani ‘going o u t’. These languages also have a great num ber o f aspect-type miscellaneous suffixes which refer either to the manner in which the action is perform ed (e.g. Te - p fo ) ‘another tim e’. N o - í í ‘intensive’, Pi -m ‘distributive*, -lew a ‘characteristic action’, Asha -im a ‘in terrupted’, -na- ‘frequently back and forth’, Pajonal - li ‘reversative’) or to time (e.g. C a -aman ‘early in the m orning’, -nink ‘late afternoon’), in Ca and A m, verbal suffixes m ark number o f participants; there is one m arker for A/S and ano th er one for S /0 . Directionals are rarer in N orth A raw ak languages (e.g. ’W p m a -...-k a n ‘do back and forth’, Pr -elipi ‘do while moving’, -rike ‘in’, -wa ‘far away’, -pagi ‘near by’. Re -hne ‘tow ards’). They also have fewer aktionsart distinctions. A m ong these are W p - k ‘suddenly’, -aan ‘interrupted’, -dan ‘aimlessly’, G uaj -p u ’u ‘frequentative’, -iwii ‘distributive’, -pim aa ‘do in passing’, A ñ -chaa ‘diminutive’, -rK K ‘augm entative’, Pia -ta ta ‘m ore o r less’, Ba -nia ‘inchoative’. Ta recently developed a rich system o f directional and other verbal markers from com pounded verbs. i Arawak 6.5 95 Noun incorporation Noun incorporation o f direct objects and locative constituents is found in Pi and Ca. In G u aj, any possessed noun can be incorporated; the possessor becomes direct object. Exam ple (14) shows incorporation o f an O constituent in Pi. (14) 0-k o stJe -k a /re -ta 3sg.masc-pick .u p -a rro w -E P E N T H E T ic ‘He picked up his arrow.’ (M atteson 1965:40) Pr, frorn the N orth A raw ak branch, has a very unusual, albeit limited, pattern o f b ody-part incorporation. W hile in the overwhelming m ajority o f languages in the A m azon incorporated body parts are placed preverbally (e.g. M unduruku (Tupi), Y anomami, T upi-G uarani, Pano, N adëb (M akù)), they occur post-verbally in Pr. These incorporated body parts have the function o f the O o f a transitive verb, o r the S o f a stative verb. I f a verb contains an incorporated body part in the O slot (note th a t body parts in Palikur are obligatorily possessed), the possessor is raised to be direct object. This is a well-known strategy in incorporating languages (type 1Í in M ithun 1984). T he possessor may be cross-referenced on the verb with an object suffix or it may be expressed with a full NP, as in (15) (see A ikhenvald and Green 1998). (15) a-daha-ni hak is-ota bakim ni-ayh 3neut-for-POSsv ru b -E Y E c h ild -P L ‘in order to ru b the eyes o f the children’ (lit. eye-rub the children) 6.6 Relative and negative mariiers Besides the cross-referencing prefixes, the m ost stable prefixes in A raw ak lan­ guages are Aa- ‘relative, attributive’ and its negative counterpart m a- (M atteson 1972; 165; Taylor 1977b). Ka- derives possessive predicates and denom inal verbs, e.g. W p o~ka~dako-in-(a)-n (3sg.fem -A TTR lB -m outh-R EFL-(EPEN TH ETIC )R E A L .M O O D ) ‘she shouted’ (lit. ‘she m outhed’: Tracy 1974: 125), o r m arks the predicate o f relative clauses (BIç, Ta). Prefix ma- is usually the negative counterpart o f ka-, e.g. Pa ka-yanityo-ki (A TTR l B -w ife-TR A N SITIV lZER) ‘find a wife for som eone’, m a-ym U yo-ki (N E G -w ife-T R A N S IT IV IZ E R ) ‘take away a wife from som eone’ (Rowan and Burgess 1979: 103); ka-tai ‘have fruit’ (o f a tree), m a-tai ‘be w ithout fruit’ (R ichards 1988: 200); Pi ka-yhi (A T T R IB -tooth) ‘having teeth’, ma-yhl (N EG .A T TR IB tooth) ‘toothless’ (M atteson 1965: 119). It is w idespread throughout the family as a general verbal and nom inal negator. 96 Alexandra Y. Aikhenvald O ther m eans o f m arking negation differ even am ong closely related languages.-*^ The majority o f languages use suffixes or clitics;^® there is typically a separate m arker for prohibition (negative imperative). Person/gender/num ber oppositions are frequently neutralized in negative and/or prohibitive constructions (Pr, Ta, BIç); tense/aspect oppositions are neutralized in negative constructions in Te and Ba. Different im perative meanings are neutralized in prohibitive constructions in Acha. 7.1 Grammatical relations As we saw in §6.2, m ost Arawak languages employ cross-referencing on the verb to m ark gram m atical relations. Only Ta and Ap have core cases linked to the discourse properties o f nouns. Suffix -ne in T a (hom ophonous with instrum ental -m ) goes on a focused subject. A clitic -nuku m arks topical non-subject. There is also an oblig­ atory m arker for pronom inal non-A/S. A p (Facundes m s-b) has an absolutive (S /0) case morphem e -«/, used to m ark a fully affected constituent. There is typically just one core case m orphem e in an NP, usually on the last word. All A raw ak languages have peripheral case suffixes (locative, allative, ablative, comitative, instrum ental, benefactive). The same m orphem e often means both ‘to ’ and ‘from ’. Languages often have locative/directional -kil-ku, locative/ablative -tV , benefactive or instrum ental - ( m ) in V (cf David L. Payne 1991: 380; Wise 1990). Two peripheral case morphemes often com bine in one word, as in Ba -wa ‘perlative’ + -ukii ‘locative’> - ira te ‘along the location’, e.g. nisa-ni-waku (Isg + c an o ePOSSV-PERLATIVE + LOC) ‘along the inside o f my canoe’. 7.2 Noun phrase structure In attributive NPs, dem onstratives and numerals usually precede the head noun and adjectives follow it; modifiers agree w ith the head noun in tioun class o r gender, and in num ber (if referring to hum ans or animates). Third person pronouns are often used as definite articles. In possessive NPs, the possessed noun is always the head. The order Possessor-Possessed is found m ore often than Possessed-Possessor; * One o f the m ain points o f differentiation between the diaîects o f Baniwa o f Içana and Kurripako, the majority o f which are reported to be mutually intelligible, is the form of the n eptive particle: ñam(e) is used in Hohôdene, Siuci, etc; t o j i is used in Kumandene and Ayanene (hence the name o f Kurripako (NEG -IND EF.PERSON -speak) ‘the ones who say kiiji for no'). Some dialects use both: Dzawi-nai/Yawarete-tapuya uses na(me) in main clauses, and kusi in interrogative and subordinate clauses Only !g has a prohibitive prefix -ku- which follows the cross-referencing clitic and precedes the root. 97 3 Arawak however, som e N o rth Arawak languages have b o th orders. Inalienably possessed nouns often function as adpositions, the adpositional argum ent being crossreferenced w ith A/Sj/Possessor prefixes. In m any N o rth Arawak languages {Pr, BIç, W X , Ta) the use o f adpositions as prepositions o r as postpositions depends on the discourse status o f the head noun. Postpositions are used when the head noun is n o t individualized; otherwise preposi­ tions are employed. Example (16), from P r (A ikhenvald and Green 1998), shows that a postposition is used when the head noun is not individualized, and the adpositional phrase refers to a habitual activity. In (16), the ‘field’ is not individual­ ized. T he sentence describes the habitual activity o f a w om an. N ote that -madka is here a postpositional enclitic which forms one phonological word with was. (16) eg ka-annipwi-yo w a s -n ia d k a 3fem A TTR lB -w ork-D LiR .fem fleld-CL:iN .FLA T ‘She worked in the fie ld .’ Prepositions are used if the head noun is individualized. Person, num ber and gender o f the head noun are obligatorily cross-referenced on a preposition. This is illustrated with (17). In this exam ple was ‘field’ has nothing to do with any habitual activities associated with a field; the example com es from a story o f a P alikur m an who was p a rt o f a pacification team carefully crossing the A rara Indians’ field while trying to m ake contact with them. (17) wis-uh pes amew-e ay-ta-re Ipl-EX CL com aout s n e a k .u p -C O M P i. there-DiRECTiON-AN APHORiC a-m adka in g i-w a s-ra -k is 3neut-CL:iN.FLAT th i s .n e u t 3masc-fieId-P0SSV»PL ‘We w e n t out s te a l t h i l y t h e r e a c r o s s t h e i r [A raras’] fie ld .’ In Pr, cross-referencing is always obligatory w ith prepositions w hether th e head noun is present or absent; but when the same item is used as a postposition, crossreferencing-is omitted. In contrast, in Ta and BIç, postpositions take the cross-ref­ erencing prefix Í- ‘non-focused A I S,,’ (A ikhenvald 1995b). 7.3 C o m p lex p red icate s a n d seria l verbs Auxiliary verbs are a characteristic feature of IC /G a and LAr. In LA r the auxiliary verb -a takes the cross-referencing and other markers, and the main verb takes suffix -/I ‘su b o rd in ato r’ (Pet 1 9 8 7 :7 6 ,1 20ff.). This construction is used (a) with a preposed m anner adverbial when the adverbial conveys new inform ation - then the auxiliary precedes the m ain verb, as in (18); (b) with a negated verb, and then the auxiliary follows th e m ain verb, as in (19). 98 Alexandra Y. Aikhem ald (18) ab a re i-a s im a k y - n suddenly 3sg.masc-AUX yell-suB OR DiNATO R ‘Suddenly he yelled.’ (19) (Pet 1987: 121) ma-siki-n th-a NEG-give-SUBORDINATOR 3sg.fem-AUX no it ‘She did not give it.’ Auxiliary verbs in Te have a very different function. An auxiliary verb construc­ tion with kS e is used as a valency-reducing mechanism with an additional m eaning o f ‘provisional state’ (e.g. ‘be w rapped’), while its causativized counterpart, k{xo, is used to m ark the causative o f a resultative. Serial verb constructions are an areal property of the languages o f the U pper Rio Negro and C olom bia. Ba, WX, Blç, Ta, A cha and Pia have productive serialization with m otion, posture and modal verbs (see K lum pp 1985: 150; Reinoso 1994, for Pia; Melendez 1994; Wilson 1992: 32-3; 156-7, for A cha). They express aspectual meanings. M ost languages allow no m ore than two verbs in a serial verb construc­ tion. Ta has unusually complicated verb serialization (Aikhenvald forthcoming-c). Outside this group, W p has incipient verb serialization (restricted to m odal verbs). M otion verbs can be serialized in Pa (Rowan and Burgess 1979: 66-7). A nd there is limited verb com pounding o r serialization in Pi (examples in M atteson 1965: 83) an d C a (David L. Payne p.c.). 7.4 Constituent order Wa, Ba an d W X have a split ergative pattern in constituent order used to distinguish core argum ents: AVO, S^^V, (see §6.2). In most other languages constituent order has discourse functions. Fronting o f focused constituents is found throughout the family. The S outh A raw ak languages Te, Bau, Ig, Ca and A m , and also Re, LAr, IC /G a, G uaj and Añ o f the N orth Arawak branch tend to have a verb-initial constituent order. M ost languages of the U pper Rio Negro are verb-medial o r verb-final; Ta tends to be verb-final as the result o f areal diffusion. Pa, Ya and Ap tend to be verb-medial, while Pi is m ostly verb-flnal. 7.5 Non-verbal clauses In m ost languages any non-verb can be used as a predicate o f existential, locative, equational o r attributive clauses; the subject receives cross-referencing. Verbalizers and nominalizers are obligatory in just a few languages (e.g. Wp, Te). 99 3 Arawak Exam ple (20) is from Ig (O tt and O tt 1983:19) and (21 ) from Bhw (Ram irez 1992: 52). ( 20) achichu-havi-ri-pa tomorrow-1 pl.O/Sg-COM PL-MOM ENTANEOUS ‘Tomorrow we will be (here).’ (21) w aituranaw i-na man-lsg.O/Sjj ‘I am a m an.’ Only very few N orth A raw ak languages show a verb meaning ‘have’ (e.g. A cha -wàari (Wilson 1992: 120); WX -deka. Ta -de, e.g. nu-de-ka tfinu (Isg-have-D ECL dog) ‘I have a dog’). There are several strategies for possessive clauses. (I) Attributive-possessive construction: ka- ‘attributive’ followed by noun or an adjective is found throughout the family, as in (22), from Te (Ekdabl and Butler 1979: 158). ( 22) co-xé’exa-ne ATfR IB-child-ALREADY ‘He/she has children already.’ (lit. ‘he/she is childed’) (cf Colloquial English ‘H e is m oneyed’) As we have noted in §6 .6, the negative coim terpart o f ka- is ma-, (ii) A construction ‘to/for + Possessor-Possessed’ is used in a num ber o f N orth A raw ak languages. (23) is from BIç: (23) nu-J*iu 1 tjin u sg-to dog ‘I have a dog.’ (lit. ‘to me a dog’) (iii) O ther N orth A raw ak languages em ploy the possessed form o f a noun in the predicate slot. E.xample (24) is from Bhw (Ram irez 1992: 52). (24) panera nu-panera-ni pan Isg-pan-possv ‘The pan is m ine.’ 7.6 Complex sentences: relative clauses, complement clauses, coordination, pivot and switch-reference Subordinating verb suffixes are the m ost com m on strategy for m arking non-m ain clauses. All the N orth A raw ak languages share a subordinating suffix -kal-ko. Relative clauses are m ost frequently posthead, with very few restrictions on the 100 Alexandra Y. A ikhenm ld function o f the com m on argum ent in either clause. The strategies for m arking rel­ ative clauses are: (a) relative verb forms; (b) nominalized verbs; (c) relative pro­ nouns; (d) com bination of several strategies. (a) Relative verb forms are found in Te, Ap, Ashe and the m ajority of N orth A raw ak languages. T he predicate o f the relative clause is m arked with a gender- and number- sensitive suffix (often the same as the 3sg O/S^^ cross-referencing suffix, in table 3.5). Te uses progressive-ti in this function. Relative verb forms are often used if the com m on argum ent is the subject o f the relative clause; they agree with the com m on argum ent in gender and number. In (25), from Pia (K lum pp and Burquest 1983: 394), the relative clause is enclosed in square brackets. (25) áiba astjeli [yà-anè-eri i-páchia-ca] other m an 3sg.masc-come-REL:MASC 3sg.masc-visit-ASPECT yà-anàa -0 i-w énda -0 amàca 3sg.masc-come-ÄSPECT 3sg.masc-sell-ASPECT ham m ock ‘The other man who arrived to visit cam e to sell a ham m ock.’ In Ashe the cross-referencing prefix on the relativized predicate is deleted if the subject o f the relative clause is coreferential w ith that of the main clause (Wise 1986: 618). Ta (Aikhenvald in prep.), Yu (Schauer and Schauer 1978: 24-5) and LA r (Pet 1987) use different relativizing suffixes on the verb depending on the function o f the com m on argum ent. LA r has two sets: one (-tho ‘feminine’ and -thi ‘masculine’) is used if the com m on argum ent is subject o f the relative clause, and the other (sid ) if it is a direct o r an indirect object. (b) In South Arawak languages and in P a and Pi, nominalized verbs are used to m ark the predicate o f a relative clause if the com mon argum ent is an oblique constituent, otherwise relative forms are used, e.g. Pi mkyawaka (underlying form: 0 -n/Ara-j>a-t«te 3sg.nfem -eat-LO CA TIV E.A DV A NCEM ENT-SU FF:PLACE)‘the place where he eats’ (M atteson 1965: 83). (c) Some languages outside the N orth A raw ak subgroup have relative pronouns (Wa: D erbyshire 1986: 545; A m and Ca: Wise 1986); Pi and N o use demonstratives as relative pronouns. Some N orth A raw ak languages (Ta, Ba) use interrogative pro­ nouns as relativizers. (d) A com bination o f these strategies is found in some N orth Arawak languages. For instance, in LAr, posthead relative clauses may have a relative pronoun, while prehead relative clauses cannot have one (Pet 1987: 168), G uaj (Alvarez 1994: 147) m arks relative clauses with a dem onstrative pronoun (masc. chi, fem. tii, pi. na) before the head which is followed by the relative form o f the verb (masc. -kai, fem. -kalii, pi. -kana). A nom inalized verb is widely used throughout the family to m ark the predicate o f a subordinate clause. 101 3 A rm m k S ubordinating particles, adverbials and enclitics are found in a few Peruvian lan­ guages, in South Arawak, in X inguan and in Pr. In N o rth A raw ak languages, sub­ ordinating particles may com e from interrogatives, as in W X , or are loans, as in Ba. An interesting com piem entation strategy consists in juxtaposing two predicates; the cross-referencing on the predicate o f the main clause indicates the presence of a subordinate clause. In (26), from Pi, the argum ent o f the predicate o f the com plem ent clause (‘be very thin’) is cross-referenced on the S^-type predicate of the m ain verb (‘see’) with an 0 / (26) suffix (M atteson 1965: 45). pom enota wet-lu be.very.thin 1pi + see-3sg.nfem.O/Sj, ‘We see that he is terribly thin.’ Simple juxtaposition is the m ain mechanism for the coordination o f clauses and o f N Ps in the majority o f A raw ak languages. Com itative adpositions can be used to coordinate N Ps; only a few languages have connectives, e.g. A m aly-. Pi hiyahni, BIç hnele, kamena, Ba idi, W X wa. Ta diwesewhya ‘then’. C oordinating conjunc­ tions are quite rare (e.g. L A r ken ‘an d ’ for coordinating clauses and NPs, mathi ‘and/w ith’ for coordinating NPs). There is little data on syntactic pivots for the m ajority o f languages. A mixed S/A and S/O pivot is attested in W X and Ba; while BIç and Ta have no pivot restrictions (Aikhenvald 1995b). Switch-reference is very rare. Wp, Bhw, Wa and Pa use the im personal cross-referencing prefix as ‘same subject’ m arker; and Te uses a valency-increasing deriva'tion (them atic argum ent advancem ent) to indicate ‘different subject’ (E kdahl and G rim es 1964). Ta is the only A raw ak language w ith a com plicated system o f switch-reference; this developed as the result o f areal diffusion from Tucano lan­ guages. 7.7 Grammatical means used for marking discourse functions M any A raw ak languages use cross-referencing on the verb to m ark discourse func­ tions o f core arguments. N uraetous N orth A raw ak languages have a special set of cross-referencing affixes used for focusing a fronted A /S,, o r the argum ent o f an adposition. ffh e s e are prefixes in Bîç, Ba and Pr, and suffixes in A cha and Yu.) In other languages cross-referencing prefixes on the verb are om itted if the A/S^ is fronted (W X , Ba, LAr, IC, Re, and also Ap, Am and A sh e -W ise 1986:619-20; and possibly Ya - see A ikhenvald 1995b). In addition, many languages have verbal affixes to focalize the action - e.g. Acha 4 is used to m ark a significant event (Wilson 1992: 176-7), and A m -ña for ‘foregrounded action’ (Wise 1986: 626). Relative verb forms are used in cleft 102 Alexandra Y. A ikhenm ld constructions to m ark AIS^ as a contrastive topic in Ashe, WX, BIç, Ta, Pia (K lum pp and Burquest 1983: 397). Several techniques can be used simultaneously. Exam ple (27), from BIç, illus­ trates the use o f cross-referencing prefix i- ‘indefinite person’ for the fronted A I S^, a n d the relative verb form to make it contrastive: (27) hjia-pia i-de-ji-hnua h e-PE R F tN D £ F .A /S g -ta k e -R E L -lsg ‘It was h i m who took m e away.’ BlBLiO G RA PH Y Adam, L. 1890. ‘Trois fiittiilles linguistiques des bassins de l’Amazone et de l’Orénoque’, Annales du Congrès International des Am érkm istes, 7.489-96. Aikhenvald. A. Y. 1994a, ‘Grammatical relations in Tariana’, Nordic Journal o f Linguistics, I7 .2 0 H 7 . 1994b, ‘Classifiers in Tariana’, Anthropological Linguistics, 36.407-65. 1994c. ‘Classe nominal e gênero em linguas A ruák’, Boletim do Museu Paraense Emilio Gaeldi. 10.137-259. 1995a, Bare. Languages o f the World / Materials 100. Munich: Lincom Europa. 1995b. ‘Person m arking and discourse in N orth Arawak languages', Stadia Linguistica, 49,153-95. 1996a. ‘Areal diffusion in Northwest Amazonia: the case o f Tariana’, Anthropological Linguistics, 38.73-i 16. 1996b. ‘Words, phrases, pauses and boundaries: evidence from South American Indian lan­ guages’, Studies in Language, 20.487-517. 1998. ‘Warekena’, pp. 225-439 of Handbook o f Amazonian languages, vol. IV, ed. D. C. Derbyshire and G. K. Pullum. Berlin: M outon de Gruyter. forthcoming-a. ‘Transitivity in Tatiana*, to appear in Changing valency: case studies in transitirity, ed. R. M. W. Dixon and A. Y, Aikhenvald, Cambridge: Cambridge University Press. forthconiing-b. ‘Reciprocals in Tariana: their genetic and areal properties’, to appear in Typology of reciprocals, ed. V. P. Nedialkov. Munich: Lincom Europa. forthcoming-c. ‘The typology o f serial verb constructions and verb compounding: the case of Tariana (N orth Arawak)'. To appear in Studies In Language. in prep. A grammar o f Tariana. Aikhenvald, A. Y. and Green, D. 1998. ‘Palikur and the typology o f classifiers’. Anthropological Linguistics, 40.429-80. Allin, T. R. 1975. *A gram m ar o f Resigaro’, Ph.D. thesis. University o f St Andrews. Alvarez, J, 1994. Estudios de lingiiistica Guajira. M aracaibo: Gobernacion del Estado Zulia, Secretaria de Cultura. Aschmann, R. P. 1993. Fro(o (•K/w/ofln, S lL a n d University o f Texas at Arlington Publications In Linguistics, 114. Dallas: SIL and University o f Texas at Arlington. Baptista, P. and Wallin, R. !967, ‘Bauré’, pp. 27-84 o f Bolivian Indian grammars, vol. I, ed. E, M atteson. N orm an: SIL and University of Oklahoma, . 1968. ‘Baure vowel elision’, Linguistics, 38.5-11. 3 Arawak 103 Bendor-Samuel, J. T. 1966. ‘Some prosodic features in Terêna’, pp. 30-9 o f In memory o f J. S. Firth, ed. C. Bazell et al. London: Longmans. Brinton, D. O. 1891. The American race. New York: Hodges Publisher. 1892. 'Studies in South American native languages’. Proceedings o f the American Philosophical Society, 30.45-105. Butler, R E. 1977. ‘Derivaçào verbal na lingua Terèna’, Serie Linguistica, 7.73-100. Captain, D. 1991. ‘Proto-Lokono-Guajiro’, paper presented at 47th International Congress o f Americanists, 7-9 July, New Orleans. Chaffanjon. J. 1889. L'Orénoque et le Caurá: relation de voyages exécutés en 1S86 et 1887. Paris. Derbyshire, D. C. 1986. ‘Com parative survey o f m orphology and syntax in Brazilian Arawakan’, pp. 469-566 o f Handbook o f Amazonian languages, vol. I, ed. D. C. Derbyshire and G. K. Pullum. Berlin: M outon de Gruyter. Dixon, R. M. W. 1994. Ergativity. Catnbridge: Cambridge University Press. Ekdahl, E. M. and Butler, N. E. 1979. Aprenda Terêna, vol. L Brasilia: SIL. Ekdahl, E. M. and Grimes, J. E. 1964, T erena verb inflection’, UAL, 30.261-8. Facundes, S. 1994. ‘N oun classification in Apurina’, M.A. thesis, University o f Oregon. ms.-a ‘Possession and unpossession in A purina’. ms.-b ‘Split case m arking in A purina’. Goeje, C. H. de. 1928. The Arawak language o f Guiana. Amsterdam: Koninklijke Akademie van W etenschappen. 1948. ‘La langue M anao (famille Arawak-M aipure)’, Proceedings o f the International Congress o f the Americanists, session 28, 157-71. González Ñáñez, O. 1991. ‘Transitividad e intransitividad en Guarekena (Arawak)’. Ms. Green, M. and Green, D. 1972. ‘Surface structure o f Palikur gram m ar’. Brasilia: SIL. Ms. Hanke, W. 1960. ‘Die Sprache der Kaiäana am Lago M apari’, Zeilschrift filr Ethnologie, 85.54-81. HofT, B. 1994. ‘Island Carib, an Arawakan language which incorporated a lexical register of Cariban origin, used to address men’, pp. 161-8 o f M ixed languages: 13 case studies in language intertwining, ed. P. Bakker and M. Mous. Amsterdam: Institute for Functional ■Research into Language and Language Use. Kaufm an, T. 1990. ‘Language history in South America; what we know and how to know m ore', pp. 13-73 o f Amazonian linguistics. Studies in Lowland South American languages, ed. by Doris L. Payne. Austin; University o f Texas Press. Kindberg, W. 1961. ‘Cam pa (Arawak) m orphology’, pp. 519-33 o f A William Cameron Townsend en el vigésiino-guinlo aniversario del Instituto IMgiíutico de Verano. Mexico: Instituto Lingüistico de Verano. Klumpp. D. 1985. ‘La oracion simple en Piapoco', Articulas en Lingüistica y Campos Afines, 13.11.1-68. 1995. Vocahulario Piapoco-Español. Santafé de Bogotá: Asociaciôn Instituto Lingüistico de Verano. Klumpp, J. and Burquest, D. A. 1983. ‘Relative clauses in Piapoco’, UA L, 49.388-99. Koch-Grünberg. T. 1911. Aruak-Sprachen Nordwestbrasiliens m d der angrenzenden Gebiete. Vienna; Mitteilungen der Anthropologischen Geselischaft in Wien. 1928. Vom Roroima zum Orinoco: Ergebnisse einer Reise in Nordbrasilien und Venezuela in den Jahren 1911119Í3, vol. IV: Sprachen. Stuttgart. Lathrap, D. W. 1970. The Upper Amazon. New York: Praeger Publishers. Loukotka, C. 1968. Classification o f South American Indian languages. Los Angeles: Latin Am erican Center, UCLA. M artius, K. F. P. von. 1867. Beiträge zur Ethnographie und Sprachenkunde Amerikas, zmnal BraslHens. Leipzig: Friedrich Fleischer. 104 Alexandra Y. Aikhenvald M ason, J. A. 1950. ‘The languages o f South American Indians’, pp. 157-317 o f Handbook o f South American Indians, vol. VI, ed. I H. Steward. Bureau o f American Ethnology, Bulletin 143. Washington, D.C.: Smithsonian Institution. Matteson, E. 1965. The Piro (Arawakan) language. Berkeley atsd Los Angeles: University o f California Press. 1972. ‘Proto-Arawakan’, pp. 160-242 of Comparative studies in Amerindian languages, ed, Ë. M atteson el al. The Hague: M outon. Medeiros, M. C. 1990, ‘Posse nominal na lingua Mehit)aku’, paper presented at the Annual meeting of Asociacion de Linguistica y Filologia da America Latina, Campinas. Melendez, M. A. 1989. ‘El nominal en Achagua’, Lengms Aborigines de Colombia, 4.3-66. 1994. ‘Esquemas sintácticos de la predicacién e interpretacion semântica de la lengua achagua’. Bulletin de l'Institut Français d'études andines, 23.465-79. M ithun, M. 1984. ‘The evolution of nuun incorporation’, Language, 60.847-94. Mujica, M. I. O. 1992. Äspectos fonolôgicos e gramaticais da lingua Yawalapiti (Aruak)’, M. A. thesis, Universidade Estadual de Campinas. Munroe, P. ms. ‘Oblique subjects in G arifuna’. Nimuendajü, C, 1932. ‘Idiom as indigenas del Btasii’, Revisla del Instituto de Etnologla de la Universidad Nacional de Tuaimán, 2.543-618. Noble, G. K. 1965. Proto-Arawakan and its descendants. Publication 38 o f Indian University Research Center in Anthropology, Folklore and Linguistics. {U AL vol. 31, no. 3, part II.) Oliver, J. R. 1989. ‘The archaeological, linguistic and ethnohistorica! evidence for the expan­ sion o f Arawakan into Northwestern Venezuela and Northeastern Colombia’, Ph.D. thesis. University o f Illinois. Olza Zubiri. J, and Jusayii, M. A. 1978. Gramática de la lengua GuaJIra. Caracas: Universidad Catélica Andrés Bello. Ott, W. and Oit, R. 1983. Diccionario ignacimo y castellano can apuntes gramaticales. Cochabamba: SIL. Parker, S. G. 1987. Vocabulario y textas Chamicwo. Comunidades y Culturas Peruanas no. 2L Yarinacocha: SIL. 1988. ‘Some universal aspects o f coalescence processes confirmed by Chamicuro phonol­ ogy and m orphology’, M. A. thesis. University o f Texas at Arlington. 1995. Datos de la lengiia liktpari. Yarinacocha: Instituto Lingiiistico dc Verano. Patte, M. 1987. ‘Variations d'actance en A ñun’, Actances, 3.178-94. 1989. Estudio descriptivo de la lengua Añun ( o 'Paraujano'). San Cristobal: Universidad Catolica del Tachira. Payne, David L. 1981. Phonology and morphology o f Axininca Campa. Dallas: SIL and University o f Texas at Arlington. 1985, ‘The genetic classification of Resigaro’, UAL, 51.222-31. 1991. ‘A classification o f M aipuran (Arawakan) languages based on shared lexical reten­ tions’, pp. 355-499 of Handbook o f Amazonian languages, vol. H I,ed, D. C. Derbyshire and G. K. Pullum. Berlin: M outon de Gruyter. 1997. ‘Nomatsiguenga noun pitch patterns’. Ms. Payne, David L, and Payne, J. 199L ‘Quantifying split intransitive agreement in Asheninca’, paper at the Symposium on Arawakan Linguistics, 47th International Congress of Americanists, New Orleans. Payne, David L., Payne, X and Sánchez, J. 1982. Morfologia.fonologlay fonética delAshéninca del Apurucayali. Série Lingiiistica P eruana no. 18. Yarinacocha; SIL, Payne, J. 1982. ‘Directionals as time referentials in Asheninca’, Anthropological Linguistics, 24.325-37. 1989. Lecciones para el aprendiiaje del idioma Ashéninca, Série Lingiiistica Peruana 28. Yarinacocha: Instituto Lingiiistico de Verano. 3 Arawak 105 1990. ‘Asheninca stress patterns’, pp. 185-212of Amazonian Linguistics. Studies In Lowland South American languages, ed. Doris L. Payne. Austm: University o f Texas Press. Pet, W. 1 A. 1987, ‘Lokono Dian, the Arawak language o f Suriname: a sketch o f its gram tnatical structure and lexicon’, Ph.D. thesis, Cornell University. Ramirez, H. 1992. Le Bahuana, une nouvelle tangue de la famille Arawak. Chantiers Ainerindia. Amerindia (Paris), 17, supplement l. Reed, J. C. and Payne, David L. 1983. ‘Los pronominales en el Ashéninca’, pp. 9 H 0 0 of Estudios Lingülsiicos de textos Ashéninca, Série Lingüistica Peruana 21. Yarinacocha: StL. Reinoso, A. G. 1994. ‘Estructuras sintâcticas de la oraciôn en el piapoco’. Bulletin de l'Institut Français d'études andines, 23.481-508. Richards, X 1973. ‘Difiouldades na anâlise de possessâo nominal na lingua W aurá', Série LingiVslica, 1.11-29. 1977. ‘Oraçôesem Waurá’, Série LingOlstica, 7,141-84. 1988. ‘A estrutura verbal W aurá’, Série Lingüistica, 9.197-218. 1991. ‘Presentaciôn y gram âtica de los pén-afos en la narrativa W aurá’, Revista Latinoamericana de Estudios Etnolingüistkos, 6.157-82. Rodrigues, A. D, 1986. Linguas brasileiras: Para o conhecimento das linguas indlgenas. Sâo Paiilo: Ediçôes Loyola. Rouse, I. 1992. The Tainos: rise and decline o f the people who greeted Columbus. New Haven: Yale University Press. Rowan, O. and Burgess, E. 1979. ‘Parecis gram m ar’, Arquivo Lingüistico 149. Brasilia; SIL. Schauer, S. and Sohaucr, X 197S. ‘Una gram âtica del Yucuna’, Artlcuhs en Lingüistka y Campos Afnes, 5.1-52. Seki, L. and Aikhenvald, A. Y. forthcoming. ‘Proto-Arawak o f Xingii’. Shafer, R. 1959. ‘Algumas equaçôes fonéticas em A rawakan’, Anthropos, 54.542-62. Shaler, D. 1971. ‘Identification o f clause types and participant roles in Pajonal C am pa’. Ms. Shephard G., Jr. 1997. ‘N oun classification and ethnozoological classification in M achiguenga, an Arawakan language o f the Peruvian Amazon’, Journal o f Amazonian Languages, 1.29-57. .Sleinen, K . von den. 1886. Durch Zentral-Brasilien. Expedition zur Erforschungdes Schingu ini Jahre ISS4. Leipzig: F. A. Brockhaus. Swift, K. E. 1988. Morfologia del Caquinle (Arawak Preandino), Série-Lingüistica Peruana 25. Pucallpa: Institute Lingüistico de Verano. Taylor, D. M. 1954a. ‘A note on the Arawakan affiliation o f Taino’, UAL, 20.152-4. 1954b. Ä note on the status o f Amuesha’, UAL, 20,240-1. 1956. 'Island Carib II; word-classes, affixes, nouns, and verbs’, UAL, 22.1-44. 1961. ‘Review o f R, Shafer “Algumas equaçôes fonéticas em Arawakan”, Anthropos 54, 1 9 5 9 ', 2 7 . 2 7 3 - 8 . 1977a. ‘A note on Palikur and N orthern Arawak’, U A L, 43.58-60. 1977b. Languages o f the fVest Indies. Baltimore: Johns H opkins University Press. Tovar, A. 1986. Las lenguas arahaucas. Hacia una delimitaciôn más précisa de la familia Arahuaca. Bogota: Institute C aro y Cuervo. Tovar, A. and Tovar, C. L. de. 1984, Catálogo de tas lenguas de America del Sur, Madrid: Editorial Gredos. Traoy, F. V. 1972. ‘W apishana phonology’, pp. 78-84 of Languages o f the Guianas, ed. X E. Grimes. N orm an: SIL and University o f Oklahoma. 1974, ‘An introduction to W apishana verb m orphology’, UAL, 40.120-6. Valenti, D, 1986. ‘A reconstruction o f the Proto-Arawakan consonantal system’, Ph.D. thesis. New York University. Valenzuela, P. 1991. ‘Com probaciön del lugar de la lengua Iñapari dentro de la ram a pre- 106 Alexandra Y. Aikhenvald andina de la familia arahuaca’. Revista Latinomnericana de Estudios Etnolingulsticos, 6.209-40, Wallace, A. R. 1853, A narrative o f travels on the Amazon and Rio Negro. London, Wilson, P. J. 1992. Una descripcionpreliminar de la gramálicadel Achagua (Arawak). Bogotá: SIL. Wise, M. R. 1976. Äpuntes sobre la influencia inca entre los amuesha, factor que oscurecc !a clasificadon de su idioma’, R em la del Museo Nacionat, 42.355-66, Í986, ‘Grammatical characteristics o f Preandine Arawakan languages O f Peru’, pp, 567-642 o f Handbook o f Amazonian languages, vol. I, ed. D. C. Derbyshire and G. K. Pullum, Berlin; M outon de Gruyter. 1990. ‘Valence-changing affixes in M atpuran Arawakan languages’, pp. 89-116 o f Amazonian linguistics. Studies in Lowland South American languages, ed. Doris L. Payne. Austin: University o f Texas Press, i991a. ‘Un estudio comparativo de las formas pronominales y sus funcionesen las lenguas arawakas norteñas'. Revista Latinoamericana de Esimlios Etnolingiilsticos, 6.83-199. 1991b, ‘Morfosintaxis y subagrupaciones de las lenguas arawakas maipurán', Revista Latinoamericana de Estudios Etnolingiilsticos, 6.259-82, Zamponi, R. forthcoming. Maipure. Munich: Lincom Europa Language Materials. Tupi A R Y O N D . R O D R IG U E S 1 IN TR O D U CTIO N The Tupi family is one o f the largest linguistic groups in South America. N ine of its branches are entirely in the A m azon basin while the tenth, Tupi-G uarani, has most o f its languages in A m azonia. This chapter presents an overview of the family, while chapter 5 provides greater detail on the best-know n and best-described branch, Tupi-G uarani. Early com parative work treated the T upi-G uarani lan­ guages as ‘pure’ members o f the family, and others as ‘im pure’ (N lm uendajii 1948: 214; Loiikotka 1950: 26). Later work treated Tupi-G uarani as ju st one branch of a larger Tupi family (see Rodrigues 1986 and further references therein). W ork is pro­ ceeding on a full reconstruction o f proto-T upi (dealing with phonology, m orphol­ ogy an d lexicon) and the systematic changes through which m odem languages evolved (Rodrigues 1995, Rodrigues and D ietrich 1997); this is required in o rd er to .provide a proper proof o f genetic relationship. Table 4.1 lists the languages o f each branch (except for Tupi-G uarani, which is reserved for the next chapter), together w ith locations and approxim ate num bers o f speakers. II, Aweti, and IV, Mawé, have som e similarities with X, Tupi-G uarani, and fu rth er work may be able to show them to constitute an interm ediate sub­ grouping. M ap 3 shows the approxim ate locations o f the Tupi languages, excluding T upi-G uarani and those languages which are now extinct (marked by f in table 4.1). It will be seen that several o f the branches now have a single living member. Only a little inform ation has been gathered on P uruborá, which is only rem embered as a second language by a few people. X ipáya is rem em bered by just two old women. T he M undurukù are one o f the best-known tribes in the recorded history o f Brazil. D uring the second h alf o f the eighteenth century they waged fierce w arfare against European settlers in the Tapajos region, taking as trophies the heads o f their victims. In 1795 (a decade after that o th er warlike group, the M ura, had surrendered - see §1.3 o f chapter 13) the M undurukù m ade peace w ith the invaders (H em m ing 1987: 19-24). 108 Aryon D. Rodrigues Five o f the Tupi branches - Arikém , M ondé, P uruborá, R am arám a and Tuparí - are spoken d o se together in or near the state o f R ondônia. (They have been badly affected by the clearing o f forests during recent decades.) This concentration sug­ gests Rondônia as a likely location for proto-Tupi, the putative ancestor language (Rodrigues 1958, 1964). Languages o f the Tupi-G uarani branch were docum ented from the sixteenth century. However, languages from other branches only began to be recorded in the nineteenth century. Word lists on M undurukù were published by Von M artius (1867) and Tocantins (1877); on Juruna, M anitsawá and Aweti by Von den Steinen (1886, 1894); and on Mawé and Juruna by C oudreau (I897a,b). In the first part of the twentieth century N im uendajii( 1924,1925,1929a,b, 1930,1932a,b) contributed good data on Xipáya, Juruna, K uruáya, Mawé and Arikém. From the 1950s more systematic work has been attem pted, firstly by missionaries with some linguistic training and then by students from Brazilian universities. Some 4 Tupi 109 Table 4.1 The languages o f the Tupi fam ily I Arikém branch a Karitiána; Candeias river, tributary of the upper M adeira, Area Indigena Karitiana, RO; c. 200 speakers, b tA rikém : Candeias and Jam ari rivers, tributaries o f the Upper M adeira, RO, II Aweti branch Aweti (Aueti, Aueto): U pper Xingu river, MT; c. 90. III Juruna branch a Juruna (Yuruna, Yudya): Middle Xingu river, M T (formerly Lower Xingu, PA); c. 2 0 0 . b Xipáya (Sipaja); Lower Xingu river, PA; 2. c tM anitsaw á: Middle Xingu river, MT. IV Mawé branch Mawé (M aué, Sateré-Mawé): Andirá and Maués rivers, between the Lower Tapajos and the Lower M adeira, A M ; c. 6,000. V M ondé branch a Mondé: around Pimenta Bueno, RO; a few speakers. b Aruá: Área Indigena Rio Branco and Área Indigena Rio Guaporé, RO; a few speakers. c Gaviao (Tkoro, Digiit): Área Indigena Igarapé Lourdes, RO; c. 350. d Surui (Paitér): Ârea Indigena Sete de Setembro, RO and MT; c. 450. e Cinta-larga: Área Indigena A ripuanä, Parque A ripuanà, Area Indigena Roosevelt, and Área Indigena Serra M orena, M T and RO; c. 600. f Zoro: Area Indigena Z orô, MT; c. 300. VI M undurukii branch a M undurukù: Middle and U pper Tapajos and M iddle M adeira rivers, PA and AM; c. 7,000. b Kuruáya: Curuá and Iriri tributaries of the Lower Xingu river, PA; c. 5. VII Puruborá branch Purubora (Borubora): Sao Miguel river, RO; a few speakers. VIII Ram arám a branch a K aro (Arara, Urukii, Itogapuk, Ntogapid, Ram arám a); Area Indigena Igarapé Lourdes, RO and MT; c. 150. b fU rum i: Upper M arm elos river, tributary of the Middle Madeira, RO. IX Tupari branch a Tupari: Branco river, tributary of the Guaporé, RO; c. 300. b Wayorô (Ajurû): RO; c. 80. c Mekéns: Area Indigena Rio Guaporé, RO; c. 150. d M akurap: Area Indigena Rio Branco, Area Indigena Rio Guaporé, and Area Indigena Rio Mequéns, RO; probably c. 700. e Sakirabiát: Area Indigena Rio Mequéns, RO; probably c. 70. f fKepkiriwát: Pimenta Bueno river, RO. X Tupi-G uarani branch - see chapter 5. 110 Aryan D. Rodrigues o f the main publications, from which inform ation has Ijeen taken for this chapter are: I Arikém branch. K aritiána - Landin (1984), S torto (1994, 1997a,b). S to rto is currently com pleting a Ph.D. at M IT on this language. II III Aweti - Emmerich and M onserrat (1972); M onserrat (1976). Juruna branch. Ju ru n a - Fargetti (1992). X ip á y a -C . L. R. Rodrigues (1990,1995), Fargetti is currently working on a Ph.D. on Juruna at the State University o f Cam pinas, Brazil. IV Mawé - Brandon, G raham and G raham (1983), G raham , G raham and H arrison (1984), G rah am (1995), Suzuki (1997). V M ondé branch. Gaviao ^ S tute (1985), M oore (1984, 1989, 1994, 1997). Surui - Van der M eer (1982, 1985), C. Bontkes (1985), W. Bontkes and D ooley (1985). VI M u ndurukù - Burum (1978), Braun and C rofts (1965), Crofts (1971, 1973, 1985) V lli R am arám a branch. K aro - G abas (1989,1990, 1994,1996). G abas is com pleting a Ph.D. on this language at the University of California, Santa Barbara. IX Tupari branch. Tupari - C aspar and Rodrigues (1957), Alves (1991), Rodrigues and Alves (1992). Wayorô - M oore and Galiicio (1993), G alucio (1996). Mekens - H anke, Swadesh and Rodrigues (1958). M a k u ra p - Braga (1992). G aliicio is working on a Ph.D. on Mekéns at the University o f Chicago, and Braga is w orking on a Ph.D. on M akurap at the University o f Toulouse. 2 PHONOLOGY 2.1 Vowels The most extensive system o f oral vowels is in Káro o f the R am arám a branch, with seven members; there are also four nasal vowels - see table 4.2. The Awéti, Mawé and Tupi-G uarani branches have six oral vowels (om itting a from the K áro system) an d sÍK nasal vowels. The rem aining branches have just five oral vowels (i, e, i, a, o) and a corresponding set o f five nasal vowels. C ontrastive length (for both oral and nasal vowels) has been reported for the A rikém and M ondé branches, and may also apply in some other branches. Rodrigues and Dietrich (1997: 268) suggested that proto-T upi had a six-vowel system: i, e, i, a, u, o. However, correspondence sets Involving u and o show an 4 Tupi Table 4.2 Vowels in Káro nasal front œ ntral back front central back high mid low Source: Gabas (1989) alm ost com plem entary distribution (suggesting an original five-vowel system, wiith [u] an d [o] as allophones o f one phoneme). Tlius o occurs either following a labial consonant (as in *po> T upinam bá (Tb) po ‘h an d ’), or as the result o f a change o f proto-Tupi *e before a labialized consonartt (*eA:''>Tb ok ‘house’, *ep"'> Tb ofi ‘leaf’), whereas « appears in other phonological environments (*fc^o>Tb ?u ‘to eat’, * k o p > T b kiiP ‘to be m any’). There are, however, som e cases in which u and o con­ trast (* tfo k'‘o > T h su?u ‘to chew’, * ifo > T h s o ‘to go’), providing evidence in favour o f a six-term proto-system (reconstructing *tfo for ‘to go’, but *tfuk^u for ‘to chew’). F u rth er com parative research is needed. In the languages o f the Arikém branch there has been a cyclic shift o f vowels (Rodrigues 1986, Storto and Baldi 1994 and S torto p.c.). C om paring cognates in languages from other branches we find many examples o f each o f the shifts e > a , a > 0, an d o > i; and some examples o f the shifts i> e , « > / and i> i. It appears that i rem ained unchanged. Thus, a six-vowel system developed into one with five vowels, but w ith five o f the vowels shifting their values: Examples include: ‘bark/sheir In other branches K aritiána G aviâo sabee opa proto-T upari *ape? proto-T iipi-G uarani *pe ‘feather/wing’ proto-T upari *upi + ?a sipi proto-T upari *pep -I- ?o papi proto-T upi-G uarani *pepo ! 12 Aryan D. Rodrigues ‘heavy’ proto-T upari *potsi ‘ja g u a r’ K áro, Tupari ameko piti om baki ‘lo u se ’ proto-T upi-G uarani *kib ‘m oon’ G aviâo kati gep oti K áro ?ip ?ep ‘tree’ proto-T upari *kip proto-T upi-G uarani *?ib (Note, however, th a t there are many unexplained forms. F or example, ‘tree’ is ?iip in G aviâo and ip in M undurukii, w ith vowel i and not i as in the other three branches ju st quoted.) Systematic com parative work across all branches o f the family will be needed to fully validate and contextiialize the proposed Arilcém vowel shifts. There is evidence o f vowel shifts o > a and i > i in the Juruna, M ondé and R am arám a branches but further work is required to investigate how systematic these are. 2.2 Consonants The num ber o f consonants reported for individual languages ranges from ten to nineteen. O ne o f the largest systems is that in Gaviâo, from the M ondé branch, shown in table 4.3. Table 4.4 summarizes the consonant systems o f one language from each branch (also repeating G aviâo for com parative convenience), excepting P uruborá for which there is n o certain d ata and Tupi-G uarani, which is covered in chapter 5. (N ote that some o f these inventories should be regarded as tentative. We have found inconsis­ tencies between different accounts o f the phonem es in a given language.) The first row gives those phonem es found in all languages included in the table. Later rows show the additional phonemes found in each specific language. Sunti also has a voiceless lateral interdental fricative, and a voiceless palatal semi-voweL Every language in the table has a labia! phonem e that may be a semi-vowel o r a fricative (or have both realizations). T he phonem e represented as fyi often has a range of realizations. In both K aritiána and M akurap it can be a palatal semi-vowel [y], nasal [ji] or affricate [dj]. (Storto 1997b gives argum ents that in K aritiàna what we have shown here as y is most appropriately treated as a nasal phoneme.) For Gaviâo, M oore (1984:225) states that the phonem ic status o f y is uncertain since he has recorded no minimal pairs distinguishing it from [tj], [ds] o r [ji). N ote that in Aweti /z/ is reported to be a retroflex fricative (Emmerich and M onserrat 1972; 9). N asals often have a range of realizations, especially in languages with no phone- 4 Tupi 113 Table 4.3 Consonants in Gaviäo voiceless stop voiced stop voiceless affricate voiced affricate voiced fricative nasal flap lateral semi-vowel bilabial dental P b t d ts dz P m palatal velar glottal k ? $ t; d3 n r B 1 y Source: M oore ( i 984: 223-6) Table 4.4 Consonant system s across languages o f the Tupi fam ily jti all languages I 11 in IV V V .VI V ni IX Karitiána Aweti Juruna Mawé Gaviào Suruí Mundurukii Káro Makurap p labial dental/alveolar m t w/p n I b d h b d 1 d 1 d b b 1 r s ts z s z s Is dz s velar palatal y tj J glottal k tj 0 7 0 ? ? d3 tJ d3 tJ d3 J tJ- d3 t| tJ P g 0 8 tl 0 g J} ? 0 0 h h 7 ? ? h 7 h g t) mic voiced stopa Storto (1994) reports that in K aritiána tm l has a num ber o f allophonic realizations: (I) as [bm bj between two oral vowels; (ii) as [b] at the beginning o f a word, followed by an oral vowel; (Hi) as [bm] when following an oral vowel but no t followed by another oral vowel (i.e. a t the end o f a word o r followed by a nasal vowel); (iv) as [mb] following a nasal vowel and followed by an oral vowel; an d (v) as [m] when between two nasal vowels, o r at the beginning or end o f a w ord next to a nasal vowel. T he other nasals. Ini and /ij/ behave in essentially the sam e way. Similar kinds o f allophony are reported for M undurukii and Tupari and for the Tupi-G uarani branch (see c hapter 5). For K áro, G abas (1989) reports that It/ patterns w ith Ibl and /g/ (i.e. effectively filling the em pty Idl slot). For instance, there are parallel phonological rules; k -> g , p b and t —>r. 114 Aryan D. Rodrigues 2.3 Tone A lthough there is no distinctive tone in languages o f the Tupi-G uarani branch, nor in Aweti or M awe (and probably n o t in Puruborá) there are two tones in the M ondé branch, and in the Juruna language, and there is contrastive pitch accent in Káro. In K aritiána an d in languages o f theT upari branch there is salient pitch accent, but it is predictable from stress which is itself predictable from other phonological and m orphological factors. M unduruku has been reported to have four tones (Crofts 1985: x v )b u t low tone is said to be associated with glottal consonants and high tone is said to carry a message ‘m any/m ost’, suggesting that there may in fact be just two phonologically contrastive tones. 3 G R A M M A T I C A L O V E R V IE W * Tupi languages are head-m arking and mildly agglutinative. They typically have one pronom inal prefix to the verb indicating a core argum ent and may also have a prefix th at marks change o f valency. N ouns bear prefixes that m ark inalienable and some­ times also alienable possession. Some languages are reported to have a distinct class o f adjectives while in others adjectival concepts are coded through a subset o f verbs. In Surui adjectives are similar to verbs but differ from them in that only verbs (not adjectives) may take a stative suffix (Van der M eer p.c.). M oore ( 1984:164) reports that in G aviào there are about sixteen verbs that would be translated into English through prepositions, with meanings such as ‘be for’ (in a benefactive sense), 'be below’ and ‘be in’. In contrast, M undurukii has a set o f postpositions which may follow an NP, e.g. (Crofts 1973: 50; note that raised numbers indicate tones): (1 ) i-*Ji^^ piy^bit^Q o ? ^-r)îm ^ m other food d3e-’ -? it^ pe^ 3sgA-3sgO + give 3sgC0REF.iNAL.P0SSD-child to ‘A m other gave som e food to her ch ild .’ Alternatively, the postposition can take a pronom inal prefix, e.g. o ^w é-b é ‘(give) to m e’ (Crofts 1985: 180). In most languages the predom inant order o f core clausal constituents is AOV, SV. ' There are only two Tupi languages for which gram m ars have been provided - M undurukii (Crofts 1973, 1983) and Gaviào (M oore 1984); neither o f these is complete and each is written in a style that could be described as not user friendly. The grammatical notes that follow are based on these, and on various papers and conference presentations, and on per­ sonal communications from a number o f linguists. They should be regarded as program­ matic and tentative. 115 4 Tupi It is reported that O must be im mediately before the verb, with nothing intervening; however, inform ation is lacking as to whether (as one m ight expect) the same applies to S an d V. K aritiána is an exception with a fairly free constituent order within texts; the unm arked orders are said to be AVO and VS. W ithin an NP, adjectives norm ally follow the head noun while a possessive phrase will precede it. In (2), from K áro (G abas 1994: 141), there is an adjective modifier w ithin the A NP: (2) (ma?w4t pât mâygâraQ w ï-n m an han d so m e sn a k e kill-ASPECr ‘A handsom e m a n k illed th e sn a k e.’ Generally, an adjective ca n n o t m ake up an N P by itself, but it can have as its ‘head’ ju st a pronom inal prefix. C om pare (2) w ith (3): (3) a?-pât^ m àygâra wï-n 3sg-handsome snake kill-ASPECT ‘A handsom e one killed the snake.’ ft appears th a t a noun in core function within a clause is not marked for case. In some languages peripheral syntactic functions are m arked by postpositions, as in (1) from M undurukù, and in others by case suffixes. Tupari has inessive, ablative and allative/instrum ental suffixes, e.g.: ■(4) (C aspar and Rodrigues 1957) te-karo-na wa(p)-m 3sgS-fall-ASPEcr hammock-ALL/iNSTR 'H e has fallen into the ham m ock.’ Tupari also has a suffix th a t m arks a constituent which is the focus o f that p a rt o f the discourse, as in: (5) (C aspar and Rodrigues 1957) te-sito-t 3 C 0 R E F -f0 0 t-F 0 C ‘H e 0-k-a 3 + O -eat-A SP EC T eats his own foot.’ T he structure of the noun in M undurukù involves an optional suffix and a pos­ sessive prefix. The suffixes are ‘dim inutive’, ‘augm entative’, ‘plural’, ‘entire’, ‘speci­ fied’ and ‘d ead ’ (obligatorily used when referring to dead people, rather like the late in English). Thus: 116 Aryon D. Rodrigues (6) (Crofts 1973: 85) ( a ) a ’y a ^ t / a t ^ - î i t ^ îi t ^ (b ) F jF m a ^ - b t f w o m an -D iM iN U T iv E nig h t-E N T iR E ‘a g i r l ’ ‘t h e w h o le n i g h t ’ As in other Tupi languages, inalienably possessed nouns (body parts and kin) take one set o f pronom inal prefixes (e.g. o^-ba* ‘m y-arm ’) while other nouns may optionally take one o f a slightly different set o f prefixes m arking alienable posses­ sion (e.g. w é-k o ^b é ‘m y-canoe’) (Crofts 1985; 87). M ost Tupi languages have alternative 3rd person prefixes to a noun according to whether or not the referent is co-referential with some other argum ent in the clause (as in (5) from Tupari). C om pare (7), where the pronom inal prefix indicates co-referentiality, with ( 8), where it doesn’t, both sentences being from K áro (G abas 1994: 137): (7) na?to_^ to-wirapQ tapir 3sgcoREF-food ?o-t eat-ASPEcr T h e ta p ir ate its ow n fo o d .’ (8) n a ? to ^ ta p ir a ? -w ira p o ? o -t 3sgNON.coREF-food eat-ASPECT ‘T he ta p ir is e a tin g its (so m e th in g else’s] fo o d .’ Tupi languages appear to lack gram m atical gender but in at least K âro and M unduruku there are classifiers, which take part in extensive agreement systems. Gonçalves (1987) lists over 120 classifiers for M unduruku. These function as suffixes to the n o u n they describe o r to a modifying numeral or dem onstrative and as prefixes to a verb. F or example ba* ‘a rm ’ is also a classifier for long, rigid, cylin­ drical objects such as a banana. Tills classifier occurs on each o f the words in: (9) (G onçalves 1987: 23,45) LTep^Jep-’-pa"* a^ko^-ba'*]Q two-CL banana-C L o?-’-sii^-ba‘'-?o-’ 3sgA-3sgO-CL-eat ‘He ate two b a n a n a s .’ A set o f about 10 classifiers is reported for K áro (G abas 1996). For instance, kap is used for clusters o f things, such as a bunch o f bananas or the teeth in the mouth; p i? is used for long things, such as a tail o r a hum m ing bird (because it has a long tail); and ?a? is used for anything round, such as the moon, an eye, a turtle o r the liver. In K áro, classifiers are used in possessive constructions, and following both a noun and its modifier in an NP. In (10) the classifier pap ‘long, big and round’ is used w ith the noun ‘alligator’ and also following ‘big’: 117 4 Tupi (10) w ayo nàk pap tjù pap a llig a to r m o u th cl b ig CL ‘a llig a to r’s big m o u th ’ All Tupi languages have pronom inal prefixes to verbs. In m ost languages there can be ju st one pronom inal preñx p er verb (M unduruku is an exception, allowing two). In the T upi-G uarani branch there are several prefix sets (one for A o r S^, one for O or etc.) - see §6 o f chapter 5. It is probably the case that in all Tupi languages p a i x e s to nouns marking aliendjle and inalienable possessors are sim ilar in form to the verbal prefix sets. C rofts (1973: 83-94) lists four pronom inal prefix paradigm s for M unduruku. A lthough the details are not totally clear it appears that one prefix set is used for the S argum ent o f an intransitive verb o f m otion; a second set for the argum ent o f an active intransitive verb and the A o f a transitive; a third set for the argu­ m ent o f an adjective-like stative verb (this is sim ilar to the inalienable possessive prefixes on nouns) and the final set for subject o f a limited set o f verbs including ‘know’ (this is similar to the alienable possessive prefixes on nouns). Languages from the M ondé branch (and at least som e from the Tupari branch) include two verbal elements in each clause - a lexical verb, and an auxiliary element that m arks tense, aspect and m ood. T he basic structure o f a transitive clause is: (a) either a n N P o r a p ro n o m in a l prefix (to th e fo llo w in g a u x iliary ) (b) an auxiliary (c) either an N P or a pronom inal prefix (to the following verb) (d) a verb. realizing the A argum ent realizing the O argum ent In an intransitive clause, (a) m arks the S argum ent and a pronom inal prefix at (o) repeats this inform ation (there atB m inor differences between the forms o f prefixes in slots (a) and (c)). We can ill ustrate this with three example sentences from Gaviao (M oore 1984; 74, 80,90). In (1 1) the A argum ent is realized as a p ronoun and the O argum ent as an NP: ( 11) m li dza-Pipip pogô-à IsgA + AUX ho u se -w a ll cover-BOUNDARY. m a r k e r ‘I covered th e w alls.’ 118 A iyon D. Rodrigues In (12) the S argum ent is realized as a pronoun in slot (a) and repeated in slot (c): ( 12) d saá paa-gá-á Ipi.incIS + AUX Ipl.inclS-gO-BOUNDARY.MARKER ‘Let’s go!’ A nd in (13) the S argum ent is realized as an N P in slot (a) and as a pronoun in slot (c). (13) [a-tsap k otj] dzano m ága aa-kaà 3sgcoREF-house to ls g + b ro th er 3sgS + A ux 3sgS-go ‘My brother goes to his ow n h ou se.’ (N ote th at in (13) the prefix a- to tsap ‘house’ cross-references dzäno ‘my brother’.) Thus, in G aviao (and also in Surui and M ekéns) the pre-auxiliary element is A or S and the pre-verbal one is O or S (a m ixture o f accusative and ergative character­ istics). Table 4.5 shows the forms o f pronom inal prefixes in a selection o f Tupi lan­ guages. It will be seen that 2sg is or begins with e- in each language except K aritiána where (as described in §2. 1) *e has shifted to a. Isg exhibits more variety, being uor o- or we- or wi- or a- (the fin K aritiána is a development from *o-). An original form u'e- could naturally have developed into u- or o- or we- (but not so plausibly to wi- or a-). 2pl begins with e- in m ost languages, suggesting that it was originally based on 2sg; the increments to e- vary considerably. Ipl.excl begins with o or ti in five rows, followed by / o r r or z or (/’, followed by o or u or e; it may be possible to relate these, once systematic phonological correspondences between the languages are worked o u t (paying attention to all parts o f the gram m ar and lexicon), Ipl.incl shows the m ost variation, suggesting that the inclusive/exclusive contrast is a recent innovation, an d has developed separately in each language (X ipáya is reported not to have this distinction). There is often a single 3rd person prefix, irrespective o f number; the forms vary widely between languages. The languages vary a good deal in how they m ark categories o f aspect, tense and m ood. As already m entioned, in the M ondé and Tupari branches these are suffixed to the auxiliary. T he limited data available on K aritiána suggests that there are tense suffixes to the verb in a positive but not in a negative clause (Landin 1984). In M undurukii aspect and tense particles can either precede or follow the nucleus of the clause (predicate plus its core argum ents), with aspect being nearest to the nucleus and tense further out (with a further peripheral constituent, marking loca­ tion, etc., being further out than tense) (Crofts 1973; 54). M undurukü also uses 119 4 Tupi Table 4.5 Pronominal prefixes to the verb in some Tupi languages Isg I K aritiána II Aweti III Juruna IV Mawé V Surui VI MundurukCi VIII K áro X proto-Tupi-G uarani (see chapter 5) ja(y)-, i(t)uU-, aoo-'-, we’oa-, wi-, tjé- 2 sg Ipl.incl Ipl.excl 2 pl ae(y)eeee-’ee(re)-, né- iykay-, tise(w)apawiy’(e’)-/a-'i?ya-(né/re)- itaozouluurutoyo ’tje^téîoro-, oré- aye?iesee(we)méyey’-, e’pe’karope(ye)- reduplication to indicate present continuous, e.g. a^d^ok^djoif ‘to be taking a b ath ’ (Crofts 1985: 68). M any languages appear to use the bare verb stem in imperatives. Evidentiality is a pervasive feature o f parts o f A m azonia. On the inform ation available, it is n o t a m ajor characteristic o f the Tupi family, being found in only a few Tupi languages. Surui (W. Bontkes & D ooley 1985: 167-9) has a ‘hearsay’ par­ ticle, e.g.: (14) é M érésôr ïy à aka be káne DEM N am e h ea r sa y kill c o m p l e m e n t iz e r w ant ‘He wanted to kill M eresor, I h eard.’ Surui also has particles iná which ‘indicates factuality for a present event or state, o r for a past event or state with present relevance’; and éná, which ‘indicates factual­ ity for a simple past event (whose present relevance is not being alluded to)’. (Evidentiality is also reported for K aritiána, G aviâo and Káro.) In G aviâo there is m arking on the verb for the num ber (singular or plural) o f the S or O argum ent. A few verbs have suppletive forms, e.g. kaà ‘go (singular S)’ and mala ‘go (plural S)’; other verbs use a suffix to m ark plural S or O (M oore 1984: 158-9). It appears that in Tupi languages there is a categorization of verbs as either strictly transitive or strictly intransitive. A nd there are, as would be expected, deriva­ tional processes for changing transitivity. Reflexive and reciprocal involve a verbal prefix (following the pronom inal prefix) that derives an intransitive stem. M awé has separate prefixes: -rewe- for reflexive and -to?o- for reciprocal, e.g.: (15) (G raham , G raham and H arrison 1984: 184) u ru -to ’o-kuap Ipl.exclS-REC iP-know ‘We know each other.’ 120 Aryan D. Rodrigues ( ! 6) e-rewe-kuap 2sgS-REFL-know ‘You know yourself.’ In other languages a single prefix covers both reflexive and reciprocal, e.g. -dse^we-^in M unduriiku (Crofts 1985:187). A num ber o f languages have a general intransitivizing prefix, which covers reflexive, reciprocal and passive, e.g. -we- or -iw - in Surui (V ander M eer 1985:228 and p.c.),-pe-in K á r o a n d - ^ i n G av ilo (M oore 1994:159). In most (perhaps all) Tupi languages an intransitive verb can be made transitive by the addition oF a causative prefix. This has the form -m- in K aritiána, -ma- in Gaviào and K áro, -mo- in Aweti, -mu- in Tupi-G uarani, -ma-, -mo- or -mu- in Mawé, atid -m ¥ - in M undurukii, It can be exemplified for K áro; (17) (M oore 1994: 159) iyit sq u eeze w-e-t lsg-AUX-PAST{?) a-ma-wiy-a 3sg-CAUS-g0.0Ut'PARTlCIPLE ‘I squeezed it out (referring to a foot worm).’ (lit; ‘I m ade it go o u t’) 7 DEMONSTRATIVES A lthough there is only inform ation on a few languages, it seems that the Tupi family is cliaracterized by rich systems of demonstratives. For Surui, Van der Meer (1985: 225) gives: ää- ‘this’ (within speaker’s reach, or in their hand) yá - ‘th a t’ (close to hearer, or ju s t out o f easy reach o f both speaker an d hearer) anô-, ajto- ‘th a t’ (away from b o th speaker and hearer) T here are a num ber o f other dem onstratives in Surui, including two that are sizerelated: ââflùn ‘this sm all’ and ajionán ‘that big’. (M oore 1984: 142 gives a similar system for Gaviào.) Suzuki (1997) devotes a whole M A thesis to discussing the set o f fifteen o r so deictics in M awé, involving such param eters as ‘visibility’, ‘gestural’, ‘distance’, ‘in direction of/in trajectory o f’ and ‘sitting/standing/scattered about’. 8 QUESTIONS Polar questions are generally marked by a special particle. In G aviào the question particle té comes a t the beginning o f a clause (M oore 1984: 105) but in Mawé the particle apo com es a t the end o f the clause. Thus; 121 4 Tupi ( 18) (B randon and Seki 1984: 84) ere-ket kahato apo 2sg-sleep very q u e s t io n ,p a r tic le ‘D id you sleep well?’ B randon an d Seki (1984) have a useful discussion o f questions, w ith exam ples from a num ber o f Tupi languages. In m ost (or perhaps all) languages, content question w ords (‘w ho’, ‘w hat’, ‘w hen’ and the like) com e at the beginning o f their clause. 9 SUBORDINATE CLAUSES As in m any A mazonian languages, subordinate clauses are often achieved through nom inalization - see, for example, M oore (1989) on Gaviao. Adverbial a nd com ple­ ment clauses may be m arked by special particles, such as be in (14) from Surui. B randon an d Seki (1981) provide a useful discussion o f complementizers, inc ing the following example from Mawé: (19) pote [k a t WHAT FOR ‘1 w a n t to k iu-wiwo e re -to i NEG 2sg-gO NEG Isg-WITH k n o w w h y y o u w o n ’t go w ith hap] a ti-k u a p NMLZR I s g - k n O W te ra n w ant me.’ C. Bontkes (1985) describes the techniques for form ing subordinate clauses in Surui. T he com plem entizer ewe, for instance, typically m arks a com plem ent clause .in Ö function to ‘see’ o r ‘hea r’, as in: (20) été [awiri tjá d é THEN dog a-peyare ewe iMPERFV 3sg-bark ] ikin o -or COMPLEMENTIZER sce Isg-come é BOUNDARY. MARKER ‘Tlien I came and saw the dog barking.’ T he sentence is, literally, ‘T hen the dog was barking, I came [and] saw [that]’. 10 As we have seen, Tupi languages have some ergative characteristics - for instance, in som e languages bound pronouns cover ju st S and O functions. There is a single m ention in the literature o f the syntactic orientation o f a Tupi language. Van der M eer(1985: 210) suggests th a t Surui has an S/A pivot (that is, it is accusative at the interclausal syntactic level). C om pare: 122 Aryan D. Rodrigues ( 21) 6*5 man déékaa, m ekôp aka é PERFv + g o ja g u ar kill decl ‘T he m an went and [the m an] killed the jag u ar.’ (22) o ij déékaa, m an PERFV + go m eko^ ôi^ aka é jaguar man kill decl T h e m an went and the ja g u ar killed the m an.’ In (21 ) ‘the m an’ is in pivot function in each clause (S in the first and A in the second) and its second occurrence is omitted. In (22) ‘the m an’ is in a non-pivot function, O, in the second clause and thus cannot be omitted. B IB L IO G R A P H Y Alves, P. M. 1991. ‘Análise fonolôgica preliminar da lingua Tuparí’, M A thesis, Universidade de Brasilia. Bontkes, C. 1985. ‘Subordinate clauses in Suruí’, pp. 189-207 o f Fortune 1985. Bontkes. W. and Dooley, R. A. 1985. ‘Verification particles in Surui” , pp. 166-88 o f Fortune 1985. Braga, A. de 0 . 1992. ‘A Fonologia segmental e aspectos morfofonotogicos da lingua M akurap (Tupi)’. MA thesis, Universidade Estadual de Campinas, Brandon, F. R., G raham , A. and G raham , S. 1983. Dkíonário bilmgiie Saleré-portugtiês, Portiigiiês-Sûteré, [Versâo preliminar 2. Unpublished. A previous version dated 1982 was included in the xeroxed series Arquivo Lingiitstico 224, Brasilia: SIL,] Brandon, F. R. and Seki, L. 1981. ‘Interrogativosecom plem entizadores era linguas Tupi’, pp. 107-14 of Aimis de Seminários do GEL. Sào Paulo. 1984. 'M oving interrogatives without an initial + W H node in Tupi’, pp. 78-103 o f The syntax o f native American languages, ed. E.-D. Cook and D. R. Gerdts, Syntax and Semantics, 16. New York: Academic Press Braun, I. and Crofts, M. 1965. ‘M undurukii phonology’, Anthropological Lmguistks 7,7,23-39. Burum, M. (cd.) 1978. AypapayiT i'm elcawin: histárim dos aniigos, Mimduriikü-Pormguês, vol. II. Brasilia: SIL. C aspar F, and Rodrigues, A, D. 1957. ‘Versuch einer G raminatik der Tupari-Sprache'. Ms. Coudreau, H. 1897a. Voyage an Xingii. Paris: A. Lahure. 1897b. Voyage ati Tapajoz. Paris: A. Lahure. Crofts, M, 197!. ‘Repeated m orphs in M undurukû’, pp, 60-80 of Estudos sôbre Linguas e Cuituras Indigenas, ed. S. I, Gudschinsky. Brasilia: SIL. 1973. Gramàtica Mimdunàù, Série Lingüistica 2. Brasilia: SIL. !985. Aspectos da lingua Mundurukû, Brasilia: SIL, Emnnerich, C, and M onserrat, R. M, F. 1972. Sobre a fonologia da lingua Aweti (Tupi). Boletim do Museu Nacional, A ntropologia, No 25, Rio de Janeiro: Museu Nacional. Fargetti, C. M, 1992. ‘Análise fotiolégica da lingua lu rü n a’, MA thesis, Universidade Estadual de Campinas. Fortune, D. L, (ed.) 1985. Porto Velho workpapers. Brasilia: SIL. Gabas, N„ Jr. 1989. ‘Estudo fonologico da lingua Karo (Arara de Rondônia)’, M.A. thesis, Universidade Estadual de Campinas. 4 Tupi 123 1990. ‘Os segmentos fonolôgicos complexos da lingua K áro’, Cadernos de Estudos Lingaisticos, 18.143-51. 1994. ‘O sistema pronominal de m arcaçâo de pessoa na lingua Karo (Arara de R ondônia)’, Revista latinoamericano de Estudios Einolingillsticos, 8.135-50. 1996. ‘Classifiers in K aro (Brazil)’, handout for talk at Society for the Study o f the Indigenous Languages o f the Americas meeting in San Francisco. Galiicio, A, V. 1996. ‘Mekens syntax: a preliminary survey’, M A thesis. University o f Chicago, Gonçalves, C. H. R. C. 1987. Cmcordäncia em Mmdurukii. Campinas: Editora da Universidade Estadual de Campinas, G raham , A., G raham , S. and H arrison, C. 19S4. ‘Preflxos pessoais e numerals da lingua Sateré-M awé’, pp. 175-206 of Estudos sobre Linguas Tupi do Brasil, Série Linguistica 11, ed. R. Dooley. Brasilia: SIL. G raham , S. 1995. ‘Sateré Mawé pedagogical gram m ar’, Arqiiivo LingOlslko 210. Brasilia: SIL. Hanke, W,, Swadesh, M, and Rodrigues, A. D. 1958. ‘N otas de fonologia M ekens’, pp. 187-217 of Miscellanea Paul Riwet octogenario dicata, vol. II, ed. J. Comas. Mexico. Hemming, J. 1987. Amazon frontier: the defeat o f the Brazilian Indians. London: Macmillan. Landin, D. J. 1984. ‘A n outline of the syntactic structures o f K aritiâna sentences’, pp. 219-54 of Estudos sobre Linguas Tupi do Brasil, Série Lingüística 11, ed. R. A. Dooley. Brasilia: SIL, Loukotka, C. 1950, ‘Les langues de la famille Tupi-guarani’, Etnografia e Lingua Tupi-Guarani 16. Boletim 104, Faculdade de Filosofla, Ciências e Letras, Universidade de Sâo Paulo. M artius, K, P, F. von. 1867. fVârtersammlung Brasilianischer Sprachen. Glossaria linguanm Brasiliensium. Glossarios de diversas lingoas e dialectos. quefallao os Indios do imperio do Brazil. (Beitrige zur Ethnographie und Sprachenkunde Amcrikas zumal Brasiliens, 11.) Leipzig: Friedrich Fleischer. Meer, T. H. van der. 1982. ‘Fonologia da Hngua Surtii’, M .A. thesis, Universidade Estadual de Campinas. 1985. ‘Case marking in Surui’, pp. 208-30 o f Fortune 1985. M onserrat, R, M. F. 1976. Préfixas pessoais em Aweti. Rio de Janeiro; Museu Nacional, ■Universidade Federal do Rio de Janeiro. Moore. D. 1984. ‘Syntax o f the language of the Gaviao Indians o f Rondônia, Brazil’, Ph.D, dissertation. City University of New York, 1989. ‘Gaviao nominalizations as relative clause and sentential complement equivalents’, UA L, 55.309-25. 1994. ‘A few aspects o f comparative Tupi syntax’, Revista Latlnoamericana de Estudios Etnolingiiísticos, 8.151 -62. 1997. ‘E strutura de cláusulas em G aviêo de Rondônia’, ABRALIN, Boletim da Associaçâo Brasileirn de Lingiiislica, 20.91-105. M oore, D. and Galùcio, A.V. 1993. ‘Reconstruction o f Proto-Tupari consonants and vowels’, pp. 119-37 of Surrey o f California and other Indian languages. Report 8 , ed. M. Langdon and L. Hinton, Berkeley; D epartm ent o f Linguistics, University o f California. Nim uendaju, C. 1924. ‘Z ur Sprache der Sipáia-Indianer’, Anthropos, 18-19.836-57. 1925, ‘A s tribus do A lto M adeira', Journal de la Société des Américanistes de Paris, n. s., 17.137-72. 1929a, ‘Zur Sprache der M aué-lndianer’, Journal de la Société des Américanistes de Paris, n .s„ 21,131-40, 1929b. ‘Wortliste der Sipáia-Sprache’, Anthropos, 24.821-50,863-96. 1930. ‘Z ur Sprache der K.m\xáyñAnA\tLWiT', Journalde la Société des Américanistes de Paris, n. s., 22.317-45. 1932a. ‘Wortlisten aus Amazonien’, Journal de la Société des Américanistes de Paris, n. s., 24.93-119. 124 Aryan D. Rodrigues 1932b. ‘Idiomas indigenas del Brasil’, Revista del Inslititm de Elnologia de la Vniversidad Nacional de Tiicmián, 2.543-618. 1948, 'Tribes o f the lower and middle Xingù river’, pp 213-43 of Handbook o f South American Indians, vol. Ill, ed. I. H. Steward. Bureau o f A merican Ethnology, Bulletin 143. W ashington, D.C.; Smithsonian Institution. Rodrigues. A. D. 1958. ‘Die Klassifikation des Tupi-Sprachstammes’, pp. 679-84 of Proceedings o f the 32nd International Congress o f Americanists, ed. I Yde. Copenhagen. 1964. ‘A classificaçâo do tronco lingüistico Tupi’, Remsta de Anlropologia, 12.99-104. 1986. Língiias hrasileiras: para o conhecimento das llnguas indigenas. Sáo Paulo: Ediçôes Loyola. 1995. ‘Olottalized stops in proto-Tupi’, paper read at the Summet Meeting o f the Society for the Study o f the Indigenous Languages o f the Americas, University o f New Mexico, Albuquerque. Rodrigues, A. D, and Alves, P. M . 1992. ‘Sobre laringalidade e nasalidade em Tupari', paper read at the 4th Encontro Nacional de Fonética e Fonologia, Universidade Federal Fluminense, Niterôi, Brazil. Rodrigues, A. D. and Dietrich, W. 1997. ‘On the linguistic relationship between Mawé and Tupí-G uarani', Diachronica, 14.265-304. Rodrigues, C. L. R. 1990. ‘Langue Xipaya, étude phonologique’, dissertation de D. E. A., Université Paris VII. 1995. ‘Etude m orphosyntaxique de la langue Xipaya (Brésil)’, Ph.D. dissertation. Université Paris VII. Steinen, K. von den. 1886. Durcit Central-Brasilien. Leipzig: F. A. Brockhaus-Verlag. 1894. Unter den naturvâllcern Zentraî-Brasiliens. Berlin: Geographische Verlagsbuchhandlung von Dietrich Reimer. Storto, L. 1994. ‘Closure and release: pre and post oralization o f nasal stops in Karitiána (Arikem family, Tupi stock)’. Ms. 1997a. ‘Verb raising and word order variation in K aritiana’, ABRALIN, Boletim da Assoa'nçâo Brasileira de Lingiiistica, 20.107-32. 1997b. ‘Vowe! epenthesis in Karitiana’, Ph.D. generals essay, MIT. Storto. L. and Baldi, P. 1994. ‘The proto-Arikem vowel shift’, paper read at the Linguistic Society o f America Annual Meeting, Boston. Stute, H. 1985. ‘Os auxiliares dinàmicos da lingua Gaviào’, pp. 1-41 o f Fortune 1985. Suzuki, M. D. S. 1997. ‘Isto ou quilo: um estudo sobre a sistema deictico da lingua SateréMawé', M.A. thesis, Universidade Federal de Rondônia. Tocantins, A. M. G. 1877. ‘Estudos sobre o tribu “M undurucü’” , Sevista do Iratituio Historico e Geographico Brasileiro, 40.73-161. Tupí-Guaraní CHERYL JENSEN 1 IN TR O D U CTIO N T he T upi-G uarani branch (o r subgroup) o f the Tupi family is perhaps the bestknow n genetic grouping in A m azonia.' It is nam ed for the language groups which were m ost prom inent during the period o f the colonization o f eastern S outh A merica: T upinam bá and G uarani. T he T upinam bá were coastal Indians, living m ainly around the area where R io de Janeiro is now located and northw ard, and were the m ain group with w hom the Portuguese colonists had contact. They covered such an extensive area that their language was referred to in the seventeenth and eighteenth centuries as th e ‘Brasílica’ language and as ‘Brasiliano’, with the nam e Tupinam bá appearing in the eighteenth century (Rodrigues 1986a). It was extensively docum ented by P adre Joseph de A nchieta (1595), who referred to it as ‘the m ost-used language on the Brazilian C oast’. This language, now extinct,^ has .had a m ajo r influence on Brazilian Portuguese. A ccording to Rodrigues, o u t o f a list o f 550 fish native to Brazil, 46 per cent have com m on names borrow ed from Tupinam bá. O ut o f 1,000 bird names, around 350 have Tupinám bá origin. M any place nam es in Brazil also have Tupinam bá origin, such as Ipiranga ‘red w ater’ (from y’- pirátj-a ‘water-red-nominal.suffix’). A few T upinam bá words have even ' Extensive linguistic documentation has been done on various languages o f the TupiGuarani subgroup, the result o f years o f work by many linguists. The author thanks alt o f the linguists whose material is cited in this chapter for their contributions which have made so much reconstruction possible. Special thanks are given to those who contributed specific details to this chapter through personal communication: José Barrientos, A na Suelly Cabral, Wolf Dietrich, Bob Dooley, Car! Harrison, Perry Priest, Lucy Seki and especially Aryon Rodrigues. Rodrigues has spent over forty years studying the T upi-G uarani family, revising his phonological reconstructions, refining the genetic classification, and inspiring and orienting the present generation o f Tupi-G uarani comparativists, o f whom I am one. He has also made accessible a wealth o f historical information. Thanks also go out to Risoleta Juliao, Kelly Smith, Helen Pease and La Vera Betts for supplying me with the spe­ cific locations o f many o f the T upi-G uarani languages. ^ The Tupinam bá communities in Parâ disappeared in the early eighteenth century. The Tupinikin (of Espirito Santo) and Potiguára (of Paraiba), the only remaining indigenous communities descended from the Tupinambá, speak only Portuguese (Rodrigues p.c.). 126 Cheryl Jensen been borrow ed in to the English language: tapioca (from typy?6k-a), jagm T (from jaw ár-a), m anioc (from m ani?ék-a), toucan (from tukán-a) and tapir (from iapi7/r-a), the first word being borrow ed with its nom inal suffix and the latter 4 without. The G u aran i Indians lived to the south o f what is now Sâo Paulo. Speakers o f what is now referred to as Old G uarani, docum ented and described by A ntonio Ruiz de M ontoya (1 6 3 9 ) ,were the main language group with whom the Spanish explorers had contact. Unlike the Tupinam bá language. G uarani is not extinct. ■' See Dietrich (1995b). 5 Tupi-Guarani 127 There are actualiy several languages which belong to th e G uarani subgroup (i.e. the languages o f subgroup 1, table 5.1) that are still spoken today by groups in south­ ern Brazil, Bolivia, northern A rgentina, and Paraguay. Some o f these are referred to in th eir respective localities as G uarani (for example, M byá in Brazil and C hiriguano in Bolivia). It cannot be said with certainty that any o f the varieties of M odern G u aran i are a direct continuation o f O ld G uarani. However, the m ost likely candidate, other than Paraguayan G uarani, spoken extensively by a nonindigenous population, is N handéva. A n o th er group, the Tupi, whose language is closely related to T upinam bá yet lacks final consonants like G uarani, lived in w hat is now Sao Paulo, in the area o f Sâo V incente and the upper Tieté River.'* AH o f these languages cam e to be used extensively am ong colonists in their respective regions. A t the time o f the colonization, the colonists were far out­ num bered by the indigenous population. Since m any o f the colonists cam e to Brazil w ithout wives, they ended up taking indigenous women as wives, with the result that the indigenous language becam e the m other tongue o f their mestizo children.-'’ In this way the indigenous language began to be spoken in a completely new social context, becom ing the predom inant m eans o f com m unication am ong the colonists and between them and indigenous groups. In the process o f its expansion am ong the colonial population, the language underw ent creolization, involving the pro­ gressive simplification o f gram m atical forms. In this way Tupi A ustral (also called Lingua G era! o f Sâo Paulo) grew out o f Tupi in the Sâo Paulo region. It was the predom inant language o f that area in the seventeenth century, but was replaced by Portuguese in the eighteenth. N heengatù (Lingua G eral o f A m azonia), which m eans ‘good talk’, is still spoken today as the first language o f a non-indigenous population in the upper A m azon region. It grew o u t o f theT upinam bá language in the area o f M aranhao and P ará and was the predom inant language o f com m unica­ tion by th e Portuguese occupying the A m azon region during the seventeenth and eighteenth centuries.* Paraguayan G uarani developed from Old G uarani - in a way th at was heavily affected by Spanish - and is today one o f the two national lan­ guages o f Paraguay (Spanish an d Paraguayan G uarani exert a considerable mutual influence on each other - see D ietrich 1993, 1995a). In older writings Tupinambá has also been referred to as Tupi. ’ Padre Antonio Vieira wrote in 1694: i t is true that today the families of the Portuguese and the Indians in Sao Paulo are so interconnected t h a t . . . the language spoken in these fam­ ilies is that of the Indians, and Portuguese is learned by the children at school’ (Rodrigues 1986a). * In the eighteenth century the Portuguese government sent a royal decree prohibiting the use o f the indigenous language by the Luso-Brazi!ian population (Rodrigues 1986a). 128 ClierylJensen The use o f the term T upi-G uarani preceded any attem pts at linguistic classifica­ tion. The term was used, for example, by Alfred M étraux, the first anthropologist to systematically study and make use o f the data from chronicles o f the 1500s and 1600s about Tupinam bá and G uarani. A lthough there were attem pts at classifica­ tion o f the Tupi-G uarani languages in the 1940s and 1950s, the classification which Rodrigues (1958) presented to the 32nd International Congress o f Americanists in Copenhagen in 1956 is the first to distinguish actual languages by the detailed use o f linguistic d ata rather than to attem pt classification based on a list o f names and geographical factors. This classification also distinguished Tupi-G uarani languages from other Tupi languages with a high degree o f accuracy, although data o f many languages was insufficient a t that time to perm it much internal classification. 2 IDBNTlFlCAXrON OF TU PI-G U A R A N Î LANGUAGES T upi-G uarani languages are found throughout Brazil and beyond its borders, to northern A rgentina (subgroup 1, see table 5.1), Paraguay (subgroup 1), Bolivia (subgroups 1 an d 2), and French G uiana (subgroup 8). A few Tupi-G uarani lan­ guages are now extinct, while others have been recently discovered. Initial contacts were m ade in the Brazilian state o f Pará with speakers o f two previously unknown language groups as recently as the late 1980s: Z o ’é (originally called Poturu) and another language simply referred to as ‘the language o f A urá e Aurê’, the two know n speakers of that language. T he question o f what is a language in the linguistic sense (as opposed to the polit­ ical sense - see ‘C onventions followed’ above) is a vexed one for Tupi-Guarani. In subgroup 1, Avâ and Izocefio are m utually intelligible and could be regarded as dialects o f one language. The Paraguayan G uarani. Guayaki same applies to Kaiwá, M byá, N handéva and and X etá appear to be sufficiently different to b considered distinct languages. In subgroup 2, Sirionô/Yuqui and Jora may comprise one language. 2.1 Distinguishing characteristics of Tupi-Guarani languages As one o f the seven branches o f the Tupi family, Tupi-G uarani is noted for a high degree o f lexical and m orphological sim ilarity am ong its m em ber languages in spite o f their extensive geographical separation. Over forty languages or dialects have been identified as members o f the TupiG u aran i subgroup, in terms o f the structural characteristics outlined in this chapter, an d through having recurring lexemes such as forms phonologically derivable from: 5 Tupi-Guarani 129 * ja tfÿ 'm o o n ’, * yp á k ‘sky’, *jepe?áp ‘firewood’, *yfiyrá ‘w ood’, *ap6 ‘ro o t’, * íra ? á ‘ju n g le’, * e f / i ‘eye’, * /i‘nose, beak’, * Ju ru ‘m o u th ’, *nam {'outer ear’, *jyPá ‘lower a rm ’, *poti?á ‘chest’, * etym a ‘lower leg’. A shared phonological change from ■"IV in proto-Tupi to * j is another criterion by which Tupi-G uarani languages are distinguished from other Tupi languages, as is the shared use o f the w ord p e t f m ‘to b acco ’ (derived from p é -tÿ m 'tobacco-planted’) rather than p é in o ther Tupi lan­ guages. T here are several languages, used by Indians, whose structure is sufficiently altered, phonologically an d /o r morphologically, to justify a hypothesis th a t the ancestors o f these people m ust have originally spoken a non-Tupi language. These include C hané, Tapieté, Izoceño, and Guayakí (Aché) from subgroup 1; Sirionô and its d o s e relatives Yuqui a n d J o rá from subgroup 2; and K okám a/O m áw a from sub­ group 3 (Rodrigues 1984/5). Cabral (1995) presents convincing evidence th a t the K okám a”' language is the result o f language shift after coming in contact with T upinam bá (or a closely related group). She shows that there was a total disregard o f the T upinam bá morphology, since affix plus stem were interpreted as a simple morpheme. Furtherm ore, the gram m ar o f K okám a is n ot Tupi, n o r can it be shown to com e from any other know n language. The geographical location o f K okám a is also p ro o f o f extensive travelling by the Tupinam bá far up the Am azon river to what is now the B razil-Peru border, and not ju s t along the coast. 2.2 Subgroups within Tupi-Guarani As a b ranch o f the Tupi family, T upi-G uarani is likely to have had its origin in w hat is now the Brazilian state o f Rondônia. Rodrigues (p.c.) suggests four basic waves o f m igration to account for the present w idespread dispersion o f languages: a wave o f m igration to the south where the G uaranian languages developed, a wave to the east in to Bolivia where Sirionô and G uarayu developed, a wave east to the A tlantic C oast where Tiipinam bâ developed, and a wave (or several waves) to the n o rth and east in to the greater A m azon region. For the Tupi-G uarani subgroup Rodrigues (1984/5) proposes eight tentative further subgroups; table 5.1 is based on his work. The first colum n in this table gives the code by which each language is referred to on m ap 4. T he second colum n gives the language name, while the third colum n gives the abbreviation by which a language is referred to in this chapter (e.g. M byá is M). ’ According to Cabral (p.c.), Kokám a, Omáwa an d Kokam iya are variations o f a single language. The Om áwa variety is now extinct. In ihe group referred to as Kokam iya, in the area o f Iquitos, the only rem aining speakers o f the language are over forty years of age. r ■R Ö 28 ? ^ L I I- I- §3 ° ■c biî e ^ g j Û- a. ^ « a: CL 0^ c 3 ^ Ç0 — m < '~ ^ Û- •2 “S =1 o 5. ñ '§ ^ gÏ cH il ■ ñ i 5. |f 2 ■è Í A CQ 03' a < E I ‘5. c f2 a Z 1 f 8 3 § I g fe' 8. 3 a Ï “ g-S i á l l i o ? .2 SS “ ■ |â - s | . .S m 'C "o S E c « « i i l ° 3 's M S ä l s § ’■5 " J O I o g 15 g iH" IiS 1CQ 1en 1m I I - I I - •f i S g ^ I « Is s -I O ‘C -22 « 1 ÎIÎ á s s t c s 1.5*o I3g M5 5 O C Ä •— & -3 :s lït'il a . : VI '• ÍHII •-0 Ä í íH ! lill Ç 1- c o bo CÖ ■o « g .2 § I •= á 3 tS 'ffl s 'S’ o g g o 0 8 ,2 I I l l l l | | | p r i I t.Ë. •s I § ^ -S 8 l |.l o b 'i l l 2 o s ex t l Ö Î3 O 4fi o c g i §'5 !•< S I I . s Xv g ^ ê B 1 - 1 ^ I §■ 8 Ä § !5 ^ ' < O c U O _ i w>’^ 4Í "§ -S 2 Í illll S s: ^ o g ^ z s gI S3 I CQ 03 Q o < Ëi g a g I cù m m ÛS p i § I 0 -Ô 1 g *i 6 •= g s § g1 £ S' s < I „ ^ -f --S E < lî I ^ ■§ »3 * i i E P < 5 Tupi-Guarani 133 Table 5.2 Proto-TG vowels high low front central back i e y [i] a u o [o] D etails o f the locations where the language is spoken and the approxim ate num ber o f speakers'^ are provided in the last colum n o f the table. R odrigues based his subclassification prim arily on phonological data, but rec­ ognized th e need for the developm ent o f a m ore com plete set o f criteria, includ­ ing gram m atical and lexical features. T he obvious difficulty is in acquiring sufficiently com plete inform ation for each language to employ such criteria systematically. D ietrich (1990b) has w orked on developing a set o f m orphological features in ad d ition to a set o f phonological features for establishing the linguis­ tic proxim ity o f languages. M ello (1992) is currently developing a lexical base for com parison. 3 PR O TO -TU PÍ-G U A R A N Í PH ONO LO GY 3.1 Proto-Tupi-Guarani phonemes Proto-Tupi-G uarani (proto-T O ) had a predom inantly CV syllable pattern in non• finai syllables. T he final syllable could be CV or CVC. Stress in the protolanguage, as in m ost o f the descendent languages, occurred on the final syllable o f the stem. Stress occurs on the penultim ate syllable o f stems in Chiriguano and the languages o f subgroup 2 and o f unsuffixed o r uncliticized stems in Wayampi. In A surini of Tocantins the stress is also on th e penultim ate syllable, except for nouns, which have stress on the third-to-last syllable, due to the perm anent attachm ent o f the (former) suffix -a (see §7.1) to nom inal stems. The proto-T G vocalic system had six vowels as shown in table 5.2. F o r two vowels a convenient orthographic symbol is used, with the phonetic value indicated in square brackets. ” The principal sources of information used in the preparation o f the population statistics in table 5.1 include Rodrigues (1984/5; 1986a) and Grimes (1996). Whenever possible, an atierapt was made to use the m ost recent information, including persottal communication and, for those Indians living in Pará, recent statistics in the Liberal newspaper. Where there is extensive use o f the national language among an indigenous group, it is sometimes difficult to obtain accurate data regarding the number o f speakers o f the indigenous lan­ guage. 134 Cheryl Jensen Table 5.3 Proto-TG consonants alveo-palatal voiceless stop labialized stop palatalized stop voiceless affricate fricative nasal labialized nasal palatalized nasal liquid semivowel p p* pi t glottal k k” k' ts (5 m m" m' (?) velar tJ n g 0” g' r w i Since only two vowel heights occur, the precise degree o f height varies from language to language. Similarly, the high central vowel is m ore fronted in some lan­ guages than in others. Stressed vowels (stem-final) may be oral or nasal, but nasalization is a property m ore o f the morphem e than o f the vowel and acts regressively. T he proto-T G system of consonants (Rodrigues and D ietrich 1997) is given in table 5.3.“' T he following consonants could occur word-finally: *P, *r, *k, *m, *n and *g. The semivowel * j and possibly * ir also occurred in this position. Medially only * j and perhaps *?occiirred at the end o f a syllable. M ost languages do not perm it word-initially, and this was a possible restric­ tion In proto-T G as well. T he language which shows the least change from proto-TG , both phonologically and morphologically, is Tupinambá. 3 .2 P ro to -T u p í-G iia ra n í allo m o rp h s The most salient cases of allophonic variation were those o f the nasal consonants, which were pure nasals ([m], [n], [oD a nasal environment (preceding a nasal con­ sonant or a nasalized vowel) and post-occluded nasals ([mb], [nd], [og]) in an oral environment. This reconstruction differs from Lemle (1971) in the addition of the alveopalatal affricate (/, and in the palatalized and labialized series of consonants, which were originally consid­ ered to be sequences. The reconstruction of the palatalized and labialized series is recent enough for it not to have been incorporated in Jensen (1998b). 5 Tupi-Guarani 135 T h e alveolar semivowel *J also h ad a nasal allophone [ñj in a nasal environ­ m ent. R egardless o f the actual pronunciation o f the * j phonem e (v^hich is in various languages: d j, 5, j , z o r f /) the syllable-final allom orpli was a true semi­ vowel. In some languages the phonetic realization o f the labial semivowel *»yis [gw]. There was probably an optional devoicing o f non-nasal final consonants (*P and *r) as unreleased stops ([p"'] and [t^]). 3.3 Possible Tupl-Guarani murphophonetnic phenomena Several likely m orphophonem ic rules for T upi-G uarani languages are discussed in detail in Jensen (1989). In this chapter I will only m ention a couple o f types. 3.3.1 R e p la c e m e n t b y n a s a l c o n s o n a n ts ( i) P rogressive a pplication A nasalized vowel (1) o r a nasal (2) causes the initial voiceless consonant o f the fol­ lowing m orphem e in the same phonological word to be replaced by a correspond­ ing nasal. For example, from Tupinam bá: (1)'s n u p a ‘beat’ - k a t ù ‘good’ > n u p âo atù ‘b eat hard’ (2) m o ‘CAUS’ + so ‘g o '-> m ono ‘send (cause to go)’ ( ii) R e g re ssiv e application A nasal consonant has the etfect o f optionally replacing by a nasal, a voiced conso­ nant that precedes it in the same word, as in the following example from Tupinam bá: (3) ero ‘cOMfT.CAUS’ + sém ‘go o u t (exit)’->enosém ‘to go out, causing O to go out as w eir Evidence in several languages supports the existence o f both o f these rules in the protolanguage. Even in languages where they are no longer productive, numerous fixed forms still occur, for example, m ono and en o ë in Wayampi, even though the only productive nasalization is the type in ( 1). Some m orphophonem ic rules depend on the type o f morphem e juncture that occurs. A juncture o f two stems is indicated by = . A juncture o f stem and affix or o f two affixes is indicated by +. 136 Cheryl Jensen 3.3.2 S tra te g ie s fo r th e loss o f c o n s o n a n t c lu ste rs fo rm e d a t m o r p h e m e ju n c tu r e ( i ) E p e n th esis When a consonant-final m orphem e combines with another morpheme, various m orphophonem ic strategies may apply, depending on the nature o f the second m or­ pheme. If the second morphem e is a consonant-initial suffix, an epenthetic vowel, y or i, is inserted. F or example, in the A m apari dialect o f Wayampi: * w fr ‘under­ side’ + p e ‘P U N C T.L O C ’> w yrype ‘under’. In Kayabi: ‘m iddle’+ p e > m yterip e ‘in the middle’. The environm ent in which epenthesis takes place is limited since several suffixes have vowel-initia! allom orphs for consonant-final stems, alter­ nating with consonant-initial allom orphs for vowel-final stems. For example: ‘large’. ( ii) L o s s W hen both m orphem es are stems and the second m orphem e is consonant-initial, the final co n so n ant o f the first m orphem e is lost: * ta r e ? fr ‘traira fish’ = *fi'jj ‘w hite’> ♦ fareP /fi'o ‘white traira (species)’. If the final consonant o f the first stem is a nasal, the feature o f nasalization rem ains on the preceding vowel, as in Tupinam bá: (4) akáo ‘head’ = p é p ‘fiat’-> akáo = mép (through rule in §3.3.1)-^ akäm ép ‘flatheam d’ W hen the initial consonant o f the second morpheme, w hether stem or suffix, begins with a glottal stop, in m ost languages the glottal stop is lost and the final consonant o f the preceding stem is retained, as in Wayampi: (5) *akái) ‘head’ = *?ök ‘remove’—>akágo ‘decapitate’ *tak*ár ‘b am boo species’ + *?i ‘DtM’—»tak'*ári ‘ba m b o o species’ This inversion of loss strategy can be explained by a rule of m etathesis prior to loss. In Parintintin and Kayabi this metathesis occurs w ithout loss o f the other consonant. ( 6) (7) (?) til) ‘w hite’ + 7i ‘dim*—>ti?oi ‘very white* (K b) ipit ‘skin’ = ?ok ‘remove’ —»ipi?rok ‘(to) skin’'* The occurrence o f syllable-final glottal stop as the result o f m orphem e com bina­ tion in these two languages suggests that the same may have occurred in the proto­ language. See §4.5.1 (on devoicing) regarding alternation o f t and c. 137 5 Tupi-Guarani Table 5.4 Syllable-initial consonants a n d sem ivowels Subgrp *pi •ts * tj *t/-i *j *W *p *pw *k* 1 ts , J h, tJ J k tj d3 gw p k* k* 2 ts, s ts, s pj k ,k i tx .i j? w p k** k» t 3. j (g)w kw w p w pw z, tJ ,s 0 3 ts , s ts, s pj k 4 h, h, k ,s 5 0 0 t; s s s d3 w w # kw 6 h pj k t j gw k" h, 0 ts k tJ j w p W' k* 7 h h, 0 hw kw 8 h, h, 0 s k, L j w w ,( P ) k« k"- 0 0 0 ,ts ,tJ s J .S k* ( n i ) Voicing o f b ilabial consonant Several languages have rules by which *p (or its reflex), when preceded at m orphem e boun d ary by a voiced non-nasal continuant (i.e. *P o r *r), becomes *j8 (or its reflex w), w ith the preceding consonant being lost. ( 8) (Tb) *?áp = p u k ü > ? á p u k ii‘long hair’ ( li t.‘h a ir= lo n g ’) (9) (Tp) * ts-öp = p u k u > h á w o k ö ‘its long le a f * o k á r= p y té r> o k á P y té r ‘centre o f the plaza (plaza-centre)’ 4 PH ONO LO GICAL CHANGES W IT H IN TU PÍ-O U A R A N Í The phonological diversification o f Tupi-G uarani languages took place prim arily through w eakening o f *pi, * t f and *ts; through various mergers, com plete or partial; through other changes in m anner and point o f articulation which did not result in merger; and through the partial or com plete loss o f final consonants. Vocalic shift, loss o f nasalization, and stress .change alse co;itributed to this diversification. T he consequences o f these changes can be lexical, as in the creation o f homonym s, or morphological. Some o f these changes can be seen in a general way in table 5.4.” 4.1 Weakening o f * í / and * ts The weakening o f * t f and o f * ts follows a path com m only attested in languages: * i / > ts o r f > s > h > 0. The only subgroup in which the distinction between these ” For a m ore detailed account o f changes, language by language, see Dietrich (1990b). 138 CherylJensen two phonem es is systematically retained is subgroup 1, as in the M byá reflexes o f * t f y ‘m other’ and *is<5 ‘(to) go’: f /y 'a n d 6. In other subgroups the two phonemes have merged. For example, the T upinam bá forms for these two morphemes are s ÿ and sá, in G uajajára they are b y and h á, and in Wayampi / and 6. Nevertheless, even in languages where merger has taken place, there are indications that * f /i n certain environm ents (or certain morphem es) was more resistant to weakening than was *ts. For example, in Wayampi, where norm ally * t f and * ts became 0, the reflex of * t/ u ? u ‘(to) bite, chew’ is sii?u, and in A surini o f Tocantins, where normally * t / and * ts became h o r 0, the reflex o f the first person singular pronoun * t f é is sé. C om parative d ata show that h is particularly resistant in the final, (originally-) stressed syllable. For example, the h is lost in initial (prestressed) position (10) but retained in stressed position (11) b o th in Parintintin and in Asurini of Tocantins. (10) * tsepo?i ‘w orm *>ePo?í (P), ew<5?ia (A) (11) * t s é ‘( t o ) g o ’> h ö ( P ) , h á ( A ) Medially in (originaily-)prestressed syllables, the h is retained in Parintintin and lost in A surini, as in example (12).'* (12) * tsy k y jé ‘(to) fear’> k y ts y jé > k y h y ji (P), kyyse (A) The ft is also resistant to loss in the m onophonem ic person marker for third person: *ts- ‘3 (allom orph for Class II stem s)’. Presumably there is resistance to the form ation o f a 0 morpheme. Hence, h is retained in the words /i-oryw éte (A) and h -ù ryw eté (G j) ‘{to be) happy' {< * o rfP -e té ‘happy-genuine’), even though the same phonem e was lost in Asurini in the word-initial position in the reflex of *tsejio?i ‘worm*: ew6?ia. In Wayampi, the reflexes o f both * t j and *ts were lost, leaving a 0 m orphem e as the reflex o f *is-, as in o-éa ‘his/her eye’ (< * ts-etfá ), except for when it com bines with a few monosyllabic stems in the Jari dialect, such as ‘3-nam e’> /i-e e . 4.2 Palatalization Various phonological changes, either w idespread or localized, in some way involved palatalization.'^ In many languages the protophonem e *p>, which occurs in the reconstruction *ep^dk '(to) see’, underwent an unconditioned change of point of articulation to thealveo-palata! position, followed by weakening, as in example (13). Metathesis occurred in this word in languages from subgroups I and 4-8. I am using the term broadly to refer to changes which involved palatalization in some way, even if the final result was not-a palatalized phoneme, such as s. J Tupi-Guarani (13) 139 * p '> tJ > tJ > X or ts > s ® In Tupinam bá this consonant underw ent a structural reanalysis as a consonant plus a semivowel, as in e p já k ‘(to) see’. T he phonem e *k^ occurs in the reconstructed m orphem es * iy é ‘(to) en ter’, *k>ér ‘(to) sleep’ and *k>é ‘here, near the speaker’, which were previously reconstructed as *iké, and *ké, respectively. This change in analysis accounts for the palatal­ ized form k^e tn Moeller’s data (1932) o f G uarayu.*' In Kayabi this phonem e resulted in the alveolar fricative s: set ‘(to) sleep’ and se ‘(to) en ter’. In G uajajára it resulted in the alveo-palatal affricate in isé [itfê\ ‘(to) enter’; that this was caused by the palatalized consonant rather than the preceding high vowel can be seen in ik á (< * iká ‘to be in m otion’).^^ In P arintintin *1^ provoked the raising o f the sub­ sequent vowel, then merged with *k, resulting in to 'r‘sleep’ and M 'h e re'. In most oth er languages, generally merged with *ic(and * ÿ w ith * g )P O th er palatalized forms developed in various languages as allophonic variation. T he palatalization o f t before i, as in example (14), occurred in m ost languages, but not in T upinam bá or Parintintin. (14) *t (preceding i ) > t ' > t J > J o r t s > s The palatalization o f k (including the originally labialized and palatalized phonem es) in U ru bû-K aapor contiguous to i resulted in / In Tupinam bá the frica­ tive s i< * ts , * tf) was palatalized a s / contiguous to i. In G uajajára the sam e sort o f p alatalization takes place, though the sources o f s in this language are dilîerent from those in Tupinam bá. Som e o f these changes are limited to allophonic variation, such as the case o f Tupinanibá. However, w hat begins as allophonic variation has the possibility of leading to a phonem ic split, as the palatalized allophone merges with another phonem e or with the palatalized allophone o f another phoneme. Various com bina­ tions o f mergers took place in the Tupi-G uarani subgroup as the reiexes o f several palatalized consonants o r o f allophones o f consonants converged in the alveolar and alveo-palatal positions. In G u ajajára the phonem e s, with allophone t f contiguous to i, has several sources: *p’ (*cp 'á k> esári), *l {*a^ati> aw asi [aw atfil) and *í:-'(*il-'é > isé). In U ru b û -K aap or (K akum asu 1986) there are cases o f / which are clearly * A nam bé has h as the end result o f this change, one step beyond s in the weakening process *epUik ‘(to) see’ > ahä (Juliâo 1993). M ore recent dala from G uarayu (Newton 1978) lists ke rather than lée. The morphem e *kfêr ‘(to) sleep’ is kér in G uajajára, even though presumably at one time the reflex o f *i-tíér would have been phonetically [i-t JerJ. In Kayabi the nasalized equivalent o f s (< *k) is g\ m ope ‘cause to enter’. 140 Cheryl Jensen recognizable as having originated from k because o f an alternation in the initial con­ sonant o f the stem when it is preceded by the third person prefix For example: k w á r (< * k“'ár) ‘hole’, /- /w 'ir lts hole’; k ^ ( < * k ÿ fl‘louse, lice’), / - / / ‘his louse, lice’. Such alternation is not observable medially, although the conditioning environment is still intact: í / é ‘(to) enter’ (< * iV é), i j ö ‘continuative aspect, in m otion’ (<*ikd). O ther sources o f / com e from the palatalization o f f before /: aw ajt ‘com ’ (< *apati). A t times the conditioning environment has been absorbed into the frica­ tive, perm itting alternations such as fa g w é r and ifa g w é r (< *i-kág'^ér) ‘his bone’. O ther m orphemes have the appropriate conditioning environm ent but are clearly n o t of Tupi-G uarani origin, such as p a fiá (from Portuguesepassear ‘to take a walk or a trip’). Thus what began in large p art as palatalized allophones o f two separate phonemes developed into a new phonem e in U rubu-K aapor. Similarly, the phonem e s in this language comes from several sources: * t f {*tfu?u ‘(to) bite, chew’> s u ?ii) and *ts {*tsefio?i ‘w orm ’> sow 'o?/), perhaps through Nheengatu; * k h (se ‘here, near the speaker’); as well as borrowings from Portuguese: sapé (from citapéu ‘h at’) and sapât (from sapato ‘shoe’). In Wayampi the phonem e s developed through the m erging o f the reflex o f *p-> {* ep 'á k ‘(to) see’> ésä), the palatalized allophone o f *f ( i i ‘nose, beak’> s r ) , and some vestiges o f * t f which escaped further weakening {* tfu ? á ‘(to) bite, chew’> stiPu), with additional sources com ing into the language through borrowing {kam isa ‘clo th ’< P ortuguese camisa ‘shirt’; sem árí ‘m anioc grater’< A p alai /iraan ). In M byá, the palatalized allophone o f *t ( * ify ti‘(to) cu t’> í:y ís r) merged with the reflexes o f * p ' (*ep’á k > etsá ‘(to) see’) and * t f (* e tfá > e ts ä ‘eye’). In Asurini o f Tocantins, the reflex o f the palatalized allophone o f ‘white’ > sig ) and the reflex o f *pj {*ep'ák ‘(to) see’> é s a g ) merged w ith the reflex o f *j (*Juká ‘(to) kiU’>s(5/ca).^‘‘ To these were added the two forms of the first person singular pronoun s e ( < * tfé) and ise ( < which resisted further weakening. The variety o f mergers dem onstrate the care that m ust b e taken In determining correspondences in these languages. 4.3 Labialization In a number o f languages there occurred a merger o f the reflex o f *p“‘ with *if" (and o f * 01“’ with This happened in subgroups 1 ,2 ,4 , 6 a nd 8, as in example (15). In m ost o f these languages the m erger was complete, but in Wayampi it was limited to what were originally stressed syllables. The data in Lemle (1971 ) does not reflect the merger of *J in Asurini of Tocantins with these other sources. 5 Tupi-Guarani 141 (15) ‘tie’ ‘resuscitate’ ‘hole’ *p*ár ’ p ^eráp * k -ár proto-T G M byâ (subgrp 1) k*á k*erá k -á G uarayu (2) k^á k^éra k^ar G uajajàra (4) k*ár k*eráw k*ár Parintîntin ( 6) k*ár k'-erâp k^ár ök*a poéra k»a In T upinam bá, the reflex o f * p ^\s unchanged exœ pt for a structural reanalysis as a sequence instead o f a labialized consonant, as in example (16). In this language a whole series o f sequences o f consonant plus semivowel are created at m orphem e boundaries, so there is no reason to distinguish a set o f labialized consonants from these sequences. ( 16) proto-T G ♦p^ár ♦p^eráp Tupinam bá (3) pw ár pw eráp kw ár K am ayurá (7) hw át hw eráp -kwat^ä ♦k"-ár The distinction between the reflexes of *p"" and *ir“' is retained in Kayabi and K am ayurá as well. However the reflexes themselves have changed to f [(j)] in Kayabi and h w in Kamayurá.^® A parallel change also took place in these two languages with the sequence *pu, which becam e h u in K am ayurá, as in *puká ‘laugh’ > huká, and /'u in K ayabi, as in *puká ia u g h ’> f u t á . I f ♦p'"were reinterpreted as a sequence in these languages, as in Tupinam bá, a single rule could account for the change before vk and u. This parallel change did not take place in the languages in which *p”' merged with * t \ 4 .4 M e rg e r o f * 0 w ith * w In subgroups 4 , 5 , 7 and 8, with the exception o f the A m apari dialect o f Wayampi (subgroup 8), there was a m erger o f oth er languages o f subgroup with *w. This was a complete m erger in the 8 (although no reflex o f *j8 occurs word-flnally) as well as in the Tenetehára cluster o f subgroup 4. Examples ( 17) and (18) show this merger in syllable-initial and syllable-final positions, respectively. (17) * yp y tiJ ‘w ind’> y w y tü (Gj, U), yw ytu (W J), but yPytu (WA) According to Seki (p.c.) this morphem e does not occur in isolation in K amayurá, but occurs in the com pound word ywy-kwat ‘hole in the ground’. “ The nasal equivalent o f hw (see §3.3.1) is m as in *emi-p"'aj ‘person who receives orders’ >emimaj, whereas *p“'aj>hwaj(Seki p.c.). In Kayabi the nasal counterpart o f /[<])] is m and o f Ar'^is ÍJ (Dobson 1988:136). 142 Cheryl Jensen (18) *o-páp ‘it (was) finished’> u p á w (Gj), upá (U ), é p a (W) In other languages the m erger was incomplete, being limited to syllable-initial posi­ tion, as in example (19). In syllable-final position (which in fact only occurs wordflnalty), the reflex o f final *^is m in A surini o f Tocantins (20). There was a prim ary split in languages o f subgroups 5 and 7 (Kayabi and K am ayurá), with the reflex o f final *P [p’’] norm ally becoming reassociated with the *p phonem e. (19) *yPytii ‘w ind’> y w y to a (A), yw ytd (K m ) (20) * -p a P ‘com pletely’> -p am (A ),-p ap (Kb) 4 .5 F in al-co n so n an t phenom ena Various phonological phenom ena deal specifically with word-final consonants. 4.5.1 D e v o ic in g F or Tupinam bá, Rodrigues describes an optional rule by which j3 and rin final posi­ tion become voiceless unreleased stops, For example; (21 ) and r respectively (Jensen 1989: 53-4). s jé flip ‘my father’ [Jérup^] o r P'criiP] s jé ra?ÿ r ‘my son’ [féraîit^ ] o r [Jéra?írj This devoicing apparently became obligatory in the developm ent o f such languages as Kayabi, K am ayurá andT apirapé. W hereas in Tupinam bá the unreieased bilabial stop is easily recognized as an allophone o f the bilabial fricative, such an analysis is m ore com plicated in the other three languages due to the merger o f *fS with *w. Linguists working in these languages have apparently opted for an analysis o f a split, the [P] allophone merging with w an d the [p"" ] allophone w ith p, even though the alternation o f p with w still shows up upon m orphem e com bination, as following example (22) of the morpheme in the * u ? fp ‘arrow ’ from Kayabi,without and with the addition o f the nominal suffix -a. ( 22) u?yp ‘arrow ’ ta p y ? y ja ru?ÿw -a ‘white m an’s arrow ’ (i.e. gun) By analogy [t’’J was treated as an allophone of f and not o f r, even though the same type of alternation occurs with r on m orphem e com bination, as can be seen in the following example o f the m orphem e *a?yr ‘son, child’. (23) ta ? ÿ t ‘son’ ka?i ra?yr-a ‘monkey child’ 5 Tupí-Guaraní 143 There are o th er languages in which devoicing does not occur at all, such as G uajajára. In this language there was a com plete merger o f *P with *w, and the final consonants which com e from */J and * ra re phonemically and phonetically w and r, respectively, as in the agentive and circum stantial nom inalizers, -har (< *-tsár) an d -h a w ( < *-tsáP), respectively. 4 .5 .2 N a s a liz a tio n In A surini o f Tocantins all final consonants became corresponding nasals, as in the following p airs o f unsuffixed verbs, which term inate in bilabials (24), alveolars (25) and velars (26). (24) * p á p ‘finish’> p a m *tsém 1eave’> h e m (25) *k.jér‘sleep’> k en * e tu n ‘smell’> é to n (26) * ? ö k ‘rem ove’> ?ar) * p irá ij‘(be) red’> piroi) W hen the stems which term inate with formerly non-nasal voiced consonants (*ft *r) com bine with another m orphem e - for exam pie, the oblique-topicalized suffix -i (see §9.2) - the n on-nasal consonant is recuperated as w (due to the merger o f with *w) o r r. i-paw-i and l-kcr-i, respectively. T he final *k, which was the only voiceless co n so n ant which occurred in final position in proto-TG , is n o t recuper­ ated in this environment: N ouns originally terannating in * k also show a perm anent change to g: *m ani?6k-a> m ani?aoa ‘m anioc’. N asalization at a word boundary is in fact n o t uncom m on in the languages o f low land S outh America and can be accounted for phonetically by the fact that the norm al resting (and breathing) position o f the velic is lowered, as it is for nasals (Rodrigues 1986b). The nasalization of *k, i.e. the replacem ent by its correspond­ ing nasal, occurred in a few words in the A m apari dialect o f Wayampi: párj (* < p á k) ‘p aca’, / o p ( < * y ífá k ) ‘larva’. In Tapirapé there is optional nasalization o f final *r. 4.5 .3 L o ss In a n u m b er o f languages, final consonants have been com pletely or partially lost, as in the following exam ples (27)-(28) from M byá, o f subgroup 1. W hen the final co n so n an t was a nasal, the feature o f nasalization is retained in the final vowel (28). (27) *aku(J ‘(be) h o t’> a k d *k "ár ‘hole’> k * á ♦ pytsÿk ‘(to) grasp’> p y ÿ 144 Cheryl Jensen (28) * k á m ‘breast’> k à *p y tûn ‘night’> p y tû "•kái) ‘bone’> k i Loss o f final consonants is also a characteristic o f subgroup 2, as in the follow­ ing example (29) from G uarayu. (29) *eym áp ‘domestic anim al’> e ^ m a ’*pytsÿk ‘(to) grasp’> p ÿ ts y * t/ám ‘string, c ord’> ts à ♦ y é r ‘(to) steep’> k e However, in this language the syllable-final r is retained in nouns (30). A ccording to Newton (p.c.) the nom inal form o f ‘sleep’ retains the final r, which the verbal form does not. (30) * k * á r‘hole’> k * a r In subgroup 8 there is some variation as to the loss or retention o f final conso­ nants. In U ru b ù -K aapor all final consonants are retained w ith the exception o f *j8. In Wayampi all final consonants have been lost in the Jari dialect. In the A m apari dialect a few nouns retain final ij or g and several, but not all, retain r. Comparative data o f Wayampi from a century ago by the French explorer Coudreau show the loss o f final consonants to be a relatively recent phenom enon, since he registers the presence o f all a x final consonants (Jensen 1989). The loss o f final consonants has contributed tow ards the formation o f homonyms, including the nom inalizations o f circum stance *-tsár and of agent *-tsá§: both have become - h i in Kaiwá and -a in the Jari dialect o f Wayampi. 4 .6 Vowel sh ift Languages o f subgroup 4 are characterized by vowel shift in central and back vowels (in the general direction of u > o > a > y ) and by loss o f nasalization in some languages. Soares and Leite (1991) describe the two types o f changes as interacting to produce the present vowel systems in these languages through a redistribution o f allophones, which could have happened simultaneously for the various vowels. This is an alternative to the already-existing hypothesis o f a push-chain of gradual changes. G uajajára has a seven-vowel system and no nasalization. T he nasal allophone o f *a, being higher in its point o f articulation than its oral counterpart, retained this 5 Tupi-Guarani 145 position with the loss o f nasalization and developed into a distinct phonem e a In some environm ents {prestressed) the reflex o f *o was raised to u. A surini o f Tocantins has a five-vowel system and no nasalization. T he reflex o f *o in the originally stressed (i.e. final) position changed to a in an (originally) oral environm ent. (W here *o occurred in both the final and the penultim ate syllables, bo th occurrences underw ent the sam e change.) The reflex o f *u changed in all environm ents to o. The reflex o f originally stressed *a in an (originally) nasalized environm ent usually became o, but occasionally became y. C om pare the data from these languages: (31) *proto-T G G uajajára Asurini *kutiik k u tû k kétor) ‘make hole in’ *m onö m ené uré m âna dre ‘we (excl)’ pirwt k uzâ pirog ‘(be) red’ *kujâ késo ‘w om an’ *akár| skérj àkyo[-a] ‘head’ *tî si si ‘nose, beak’ *oré *piráo ‘send (away)’ In Tapirapé the changes *o > a and * u > o were both general. N asalization was not lost. The nasalized phonem e *S became ÿ and the (originally) oral *a became nasal­ ized ä (phonetically a nasalized [a]). Vowel shift is also described for A nam bé o f subgroup 8, although only for the final (stressed) syllable. U nlike the languages o f subgroup 4, its vowel shift involves (formerly) low front and centra! vowels rather than (formerly) low central and back ones. T he changes in central vowels are sim ilar to Tapirapé, with the nasalized allophone o f *a raising to ÿ and merging with *ÿ, and the oral allophone becom ing nasalized: * a > â . However, in this language the front vowel rather than the back one shifted to a C om pare cognates in A m m b é with those o f its relative W ayampi (32). (32) ♦proto-TG A nam bé Wayampi (W J) *pirár) pirÿ *pir-ér pirá pirâ pire ‘(to be) red’ ‘skin (rem oved)’ *pirá pirâ pira ‘fish’ ♦pi?û pi?û nupÿ pi?ô ‘gnat’ nûpâ ‘(to) beat’ *nupä 146 5.1 Cheryl Jensen Categories of stems There are three basic categories o f stems which can be inflected: nouns (§7), post­ positions (§8) and verbs (§9). Verbs are subclassified as transitive, active intransitive and stative intransitive. O ther categories o f noninflected stems include quantifiers, dem onstratives and time words. Adjectives do n o t exist as a separate category. Inform ation which in many languages is com m unicated through adjectives is com ­ municated in Tupi-G uarani languages on a syntactic level through nouns in appositional phrases o r through stative intransitive verbs. Such inform ation is also com m unicated on the morphological level through suffixation o r through com posi­ tion (§7.2). 5.2 Stem classes Inflected stems, regardless o f their gram m atical categorization, are divided into tw o arb itrary classes based on whether o r n o t they com bine w ith a linking o r rela­ tional prefix *r-. This prefix, which occurs only w ith class II m orphem es, indicates a tight gram m atical relationship between the noun o r person m arker from set 2 (§6) which precedes it and the m orphem e to which it is prefixed. T he same rela­ tionship is unexpressed in class I m orphem es. T he following examples in (33) show the com bination o f class I w ithout, and class II with, the r- m orphem e in nouns, transitive verbs (O), stative intransitive verbs (S^,), and postpositions, respectively. In (33) the preceding m orphem e is a person m arker; in (34) it is a noun. (33) C lass 1 Class II *t Jé akáo ‘my head’ * t/é r-etsá ‘my eye’ *tj'é p o tá r * tjé k a tá ‘like m e’ ‘see m e' ‘I am good’ ♦ tjé r-ep^ák * tJ é r-o rÿ p ‘1 am * tjé kotÿ ‘tow ard m e’ ■•tjé r-etsé ‘about me’ (34) *pirá akáo ‘fish’s head’ *pirá r-etsá ‘fish’s eye’ 6 PERSON MARKERS happy’ Four sets o f person markers have been reconstructed for proto-Tupi-G uaranf, as shown in table 5.5. Their function is discussed in more detail in §7-§9. Set 1 m arkers are used with transitive and active intransitive verbs and refer to A 5 Tupi-Guarani 147 Table 5.5 P roto-T G person markers Isg 1 excl 1 inel 2 sg 2 pl 3 Set t Set 2 Set 3 aoroja-^’ erepeo- t jé [ r - l oré [r-] jan é [r 4 né {r-] pé [n-] i-, ts-, t- wiorojere-^* ep ejeo- Set 4 oroopo- Free pronouns itjé oré jan é ené p e . .. ë and Sjj, respectively. In m ost languages their use is restricted to independent clauses.^’ Set 2 m arkers are used to indicate the genitive in nouns and the object o f post­ positions, except when their referents are coreferential with the subject (A o r S) o f the clause. They also occur with verbs in the following constructions: transitive verbs, referring to O, in subordinate, serial and oblique-topicalized constructions, and also in independent constructions when O is superior to A in a 1 > 2 > 3 hierarchy;* active intransitive verbs, referring to S^, in subordinate and obliquetopicalized constructions; stative intransitive verbs, referring to in all constructions. As w ith nouns and postpositions, the verbs also carry coreferentiality restrictions. ‘W hen first and second person m arkers occur with class II stems, they require the presence o f the r- prefix on the stem. This prefix has an ailom orph n- for second person plural. N ote that the five forms for first and second person are the same as, o r derived from, the free pronouns.^' The third person m arkers are prefixes: /- occur­ ring w ith class I stems, ts- occurring with m ost class II stems, and t- occurring with a subset o f class II stems. Set 3 markers are coreferential markers, occurring with nouns, postpositions and verbs (particularly intransitive serial verbs), substituting the markers in set 2, when Some languages have a prefix based on *ti- which occurs with transitive verbs. See Jensen (1987, 1998b) for a proposal of its derivation. Rodrigues and Dietrich (1997) gives a different reconstructed form *jare-. ^ In some languages their use has been extended to other (non-independent) clause types. See Jensen (1990). Details o f this split-ergative m arking system are discussed more fully in §9. This system has been replaced in U rubu-K aapor by strictly nominative marking. In several languages, including Asurini o f Tocantins, G uajajára and Wayampi, first and second person markers o f set 2 are presently analysed as prefixes, though there are strong reasons to consider that they were originally independent words. 148 CherylJensen their referents are the same as the subject (A or S) o f the independent verb (or, in the case of nouns and postpositions, the dom inating verb o f the clause they are in).*^ Set 4 m arkers occur with transitive verbs and are portm anteau forms indicating th at A is first person and O is second person. The two forms * o r o and *opo- dis­ tinguish the num ber o f the second person O, singular and plural, respectively, but do not distinguish the num ber of the first person. The basic divergencies from the protosystem which may be observed in the various languages are the following: — an extension o f the use o f set — a reduction o f the num ber o f coreferential forms to only third person, 1 forms to subordinate clause types. and their replacem ent by other forms (set 1 or set 2) in certain constructions. — various types o f changes in set 4 morphemes, including their loss, their merging to a single form, oro-, and the occurrence o f a form for second-person plural n o t directly derivable (phonologically) from *opo-. In Parintintin and Kayabi the use o f the reflexes o f */- and *ts- for third person (set 2) has been restricted to non-hum an referents. For hum an referents these two languages have three forms, specific for num ber and gender (in the singular), which are not found in the other subgroups o f Tupi-G uarani: (35) Parintintin Kayabi (men’s speech) 3sg.masc 3sg.fem ga hê ? 0a 3pl i)a ?gä 7 NOUNS 7.1 Case marking ëë In proto-TG, as dem onstrated in many languages o f the branch, any noun term i­ nating in a consonant received a nom inal case suffix -a, whenever it occurred in a subject, object o r genitive function (36)-(38). N ouns term inating in a vowel did not com bine with this suffix. There is quite a bit o f variation from one language to another in the extent o f usage o f the coreferentiality markers. U nibii-Kaapor has lost the system altogether. For a complete exposition, see Jensen (1998a). 149 5 Tupi-Guarani There a re four oblique cases: attributive case *-am o (C__)~-ram o (V__) and three locative cases: diffuse (spread out in a certain area) *-jSo, punctual (at some specific p oint) *-pe, and partitive (at som e specific p a rt o f the whole, such as a body p art) *-i. These are illustrated in T upinam bá in examples (36)-(39), respec­ tively. (36) e -i-m o m e ? é T u p â -0 r-a?ÿr-am o IMPER-3-teIl G od-N O M .C A SE LINK-Son-ATTRIB 2sg né r -e k ö -0 LINK-be-NMLZR ‘Tell about your being C o d ’s Son.’ (37) ka?á-Po ja w á r-a r -e k ö -w jungle-DIF.LOC jagUar-NOM.CASE LINK-be-OBL.TOP ‘Jaguars live throughout the jungle.’ (38) jawár-a s jé k é -p e s-e k ö -w jaguar-NOM.CASE Isg garden-PUNCT.ixx; 3-be-OBL.TOP ‘The jaguar is in m y garden.’ (39) s jé jiSr-i a-rek 6 Isg neck-PART.LOC Isg-have ‘I have it at my n eck .’ In some present-day languages, such as A surini o f Tocantins, the use o f the .nominal suffix has been extended to all nouns, whereas in other languages, such as G uajajára, its occurrence has been completely lost. Some o f these locative cases have com bined w ith nouns indicating spatial relationships to create words similar to postpositions. F o r example, in the A m apari dialect o f Wayampi: (40) * w ÿ r‘underside’ + * - p e ‘puNCT.LOC’ > w ÿ r y p e ‘under’ *?ár ‘to p ’ + *-p o ‘DiF.LOc’> ? a r ÿ p o ‘above’ *pÿr ‘p a rt next to ’ + *-1 ‘p a r t . l o c ’ > pÿri ‘next to’ *pÿr ‘part next t o ’ + *-Po ‘DiF.LOC’> p y r y P o ‘near, a lo n g ’ Kayabi distinguishes punctual a nd diffuse forms o f ‘under’: ? w frip e and ? w ÿrim ü. N ouns can also function as the predicate w ithout any derivational m arking other than the absence o f *-a. In this case they arc inflected in the same way as stative intransitive verbs. F or example: (41) ’^i-memyr-a ‘her child’ *i-m em ÿr ‘she gave birth’ 150 Cheryl Jensen 7.2 Possession Pronominal possession is normally expressed by the person markers of set 2. The forms for first and second person were derived from free forms and are reconstructed for proto-TG as free forms. However, in many languages they are now analysed as pre­ fixes. These forms co-occur with the prefix r- when com bining with a stem from class II (43). The third person marker, undisputably a prefix, occurs without the r- prefix. (42) Class I ‘hand’ Isg lexcl lincl 2 sg 2 pl 3 (43) proto-TG Tupinambá Guajajára Wayampi * t;é p ö *oré ptS *janépö ’ né pd sjé pö oré p(5 jané pö né pö pép6 i-p 6 he-pö ure-pö zane-p 6 ne-pö pe-p 6 i-pö é-po oré-po jané-po né-po pé-po í-po *péptS *i-p 6 Class I I 'eye' Isg lexcl lincl 2 sg 2 pl 3 proto-TG Tupinambá G uajajára Wayampi • tjé r-etjá *oré r-etjá *jané r-e tjá *né r-etjá *pé n-etjá *ts-etjá sjé r-esá oré r-esá jané r-esá né r-esá pé r-esá s-esá he-r-ehá ure-r-ehá zane-r-ehá ne-r-ehá pe-n-ehá h-ehá e-r-éa ore-r-éa jane-r-éa ne-r-éa pe-n-éa 0 -éa When the possessor o f the noun is identical with the subject, a coreferential prefix (set 3) is used, as can be seen by com paring Tapirapé (44), where the argum ents are not coreferential, and (45), where the argum ents are coreferential. (44) ere-m a-pén J"e-pá 2sg-CAUS-break Isg-hand ‘You broke my han d .’ (45) â-m a-pén Isg-C A us-break we-pá Isg-hand ‘I broke my (own) hand.’ In some languages, like Asurini of Tocantins and Tapirapé, these prefixes occur for all persons. In others, like G uajajára, the G uarani languages and Wayampi, they 151 5 Tupi-Guarani occur w ith ju st third person, the only person for which there is any possible am bigu­ ity. In W ayampi, there is only one form for ‘my h an d ’: é -p o , whereas there are two forms, non-coreferential and coreferential for the third person reference ‘his/her h an d ’: i-p o and 6 -p o , respectively. U rubu-K aapor has lost the coreferential prefixes altogether. 7.3 Noun composition There are a num ber o f Tupi-G uarani suffixes which occur with nouns to give m od­ ifying inform ation which in other languages tends to be com m unicated through a separate class o f adjectives. These suffixes indicate: size quality status (and its allom orph * -ut/u) ‘large’, *-?i ‘sm all’ *-eté ‘genuine’, *-rán ‘im itation’ *-p'^er (and its allom orphs * -w éran d *-ér) ‘form er’, *-râm (alternating w ith *-wám and *-árn) ‘future’ For example, from Tupinam bá: y -w a s ii ‘great w ater’, p a jé -r ä m -a ‘future sham an’. N ouns may also occur in com position with verbs, particularly stative intransitive verbs, giving inform ation such as colour and sh ap e. For example, from Tupinam bá: itá -p irári-a ‘red rocic (rock-red-NOMiNAL.SUFFix)’. N ouns may also occur with other nouns, the second indicating som e distinguishing feature of the former. For example, G uajajára: k a ? i-a ? ^ r ‘y o im g monkey (monkey-child)’; Tupinam bá arára k d ç ‘macaw distinguished by its head (macaw -head)’; Wayampi; p é k ö -á k ä - m ír ä ‘re d -h e a d e d w o o d p e c k e r (w o o d p e c k e r-h e a d -re d )’; Parintintin: -a ta -tir) ‘smoke (fire-white)’. 7.4 Indication o f number T upi-G uarani languages do n o t have a com mon plural morpheme. T here are some suffixes which are used to indicate a group. For example: *-p'^ér^^ (occurring with hum ans), * - ty r and perhaps *-iyj8. The word * y f iy tÿ r ‘hill’ is a derivation o f * y jiÿ ‘d irt’-i- *iyr. Some languages have their own plural morphemes: Wayampi k ô and Em erillon k o m were borrow ed from a C arib language. In G uajajara a clause-final clitic W3 is used when third person subject, object or both are plural and anim ate (H arrison 1986: 431); this m orphem e has a cognate wa in Kayabi. A nother strategy for indicating num ber is the use o f a quantifier, as in (46), although the numerical system in proto-T G was rudim entary. A nother strategy is This is homonymous with the ‘former’ morpheme and seems to have the same allomorphs. 152 Cheryl Jensen the use o f the diffuse suffix, as in Tupinam bá ka?á~po ‘throughout the jungle’ in (37), when the subject is countable. 7.5 Noun phrases N oun phrases tend to be short and consist o f three basic types: specifier-noun, genitive-noun and noun-appositive,^'' as in examples (46) and (47), (48), and (49), respectively, from G uajajára (Bendor-Samuel 1972; 110-11). The specifier may be a quantifier or a dem onstrative (note that example (47) has two specifiers). In the genitive construction, the r- prefix occurs when the possessed stem is class II. These phrase types may also combine, as in example (49). (46) m oköz two ka?Va?ÿr monkey-child ‘two young monkeys* (47) kw éz that om ö certain arár macaw ‘that particular macaw (parrot)’ (48) m am áz m o th e r r-e im á w LiNK-pet ‘m o th e r ’s p e t’ (49) [m a m á z r-e im á w ] z a p u k á z m o th e r LiNK-pet chicken ‘m o th e r ’s p e t ch ick en ’ 7.6 Possession classes Some stems are gram matically unpossessible, reflecting the indigenous world view, such as th e sun, the moon, the jungle. O thers are grammatically unpossessible, though they may in fact be possessed in real life, such as ‘chicken’ in example (49). In this case possession is handled through apposition. Still other stems are obligatorily possessed, regardless of whether the possessor is in focus. These include body parts and personal possessions. When the possessor is not in focus, an unspecified possessor morpheme is still required. Stems are ^ A relative clause (see §10) may occur in apposition to a noun (or as a headless construc­ tion), as in Wayampi: tnoi [poro'Su?u male] snake [people-bite relzr] ‘a poisonous snake’ (lit. ‘a snake that bites people’) J Tupi-Guarani 153 Table 5.6 Unspecified possessor morpheme Stem class Third person Unspecified possessor la lb (p-inltial) Ila Ub lie lid (V-initial) *nasalization of initial C ♦t•t* 0- *subtraction of initial V further subdassified on the basis o f the different form s that this morphem e takes, as in table 5.6. Examples (50)-(53) are from Tupinam bá: (50) (la) i-akáo ‘his/her h ea d ’ (51) (Ib) i-po?ÿr ‘his/her beads’ (52) (Ila ) s-esá ‘his/her eye’ (53) (Ild ) s-apé ‘his/her p a th ’ akáo m o?ÿr t-csá pé ‘(possessed) head’ ‘(possessed) b eads’ ‘(possessed) eye’ ‘(possessed) path’ In some languages, like M byá, th e strategy in exam ple (53) has been replaced by the one used in (52), resulting in forms like t-apé ‘p a th ’. 8 P o stp o sitio n s P ostpositions are similar in form to genitive constructions. Their objects are indi­ cated by th e same person m arkers th a t are used to indicate the genitive. This is n o r­ mally by a person marker from set 2 (54 and 56) o r a noun (55 and 57). (54) Class I ‘in’ lag lexcl lincl 2sg 2pl 3 proto-TG Tupinambà Guajajâra Asurini (T) *t/é pypé” *oré pypé *jané pypé •né pypé •pé pypé •i-pypé sjé pupé oré pupé jané pupé né pupé pépupé i-pupé he-pupé ure-pupé zane-pupé ne-pupé pe-pupé i-pupé se-pÿpe ore-pÿpe sene-pÿpe ne-pÿpe pe-pÿpe i-pÿpe Some languages havepupé and otherspypé. The latter allows for the possible interpretation of this morpheme as a derivation of py?ä-pe (liver-at). 154 Cheryl Jensen (55) (56) ‘in the water’ *y pypé ÿ pupé y pupé ÿa pype Class I I ‘about, with respect to, at’ Isg lexcl lincl 2sg 2pl 3 (57) proto-TG Tupinambá Guajajàra Wayampi *tjér-etsé *oré r-etsé *jané r-etsé ♦né r-etsé *pé n-etsé ’‘ts-etsé sjé r-esé oré r-esé jané r-esé né r-esé pé r-ése s-«sé he-r-ehé ure-r-ehé zane-r-ehé ne-r-ehé pe-n-ehé h-ehé é-r-e oré-r-e jané-r-e né-r-e pé-n-e 0-ée ‘about, at the bird’ *wyrá r-etsé wyrà r-esé wyrà r-ehé wyrà r-e Like nouns, the postpositions take coreferential person markers when their referent is the same person as the subject. U nlike nouns, these markers do not combine directly with the postpositions. A nother prefix, reflexive *je- or reciprocal * Jo , occurs between the coreferential prefix and the noun. For example, norm al (58), reflexive (59), and reciprocal forms in A surini * etsé ‘a t’: héh e (< ^ts-etsé) ‘at him /her’, eséh e ( < *e-je-etsê) ‘at yourself*, and orosow éhe (< *oro-jo-etsé) ‘we excl, at each o th e r’. (58) em á?e héhe ‘Look at him /her.’ (59) em á?e eséhe ‘Look at yourself.’ In languages w hich have a coreferential prefix only for third person, the forms from set 2 are used for first and second person, together with the reflexive or reciprocal prefix, as in G uarayu: ja n e -je -u p e % e incl, for ourselves’. There are three basic subcategories o f verbs; transitive, active intransitive and stative intransitive. These are distinguished m orphologically by details o f the person-m arking system they employ. D itransitive verbs also occur and have the same inflection patterns as ordinary transitives. The stative intransitive verbs, to a large degree, are those stems which in many languages would be classified as adjec­ tives. However, in Tupi-G uarani languages they share many features with other verbs; they occur w ithout a copula, they are inflected for person, they occur as the ■predicate o f a clause, and they take tense and aspect m arking, including the dis­ continuous negation morphem e specific to predicates. 5 Tupí-Gmrartí 155 T here are two basic types o f person-m arking system s in T upi-G uarani verbs. In independent verbs there is an active-stative m a rking system for intransitive verbs an d a mixed system governed by a person hierarchy for transitive verbs.’* In o th er verbal constructions there is an ergative-absolutive system, which also includes the use o f coreferential prefixes when appropriate. In some languages the use o f the form er system has been extended to include oth e r constructions (Jensen 1990). 9.1 Independent verbs Active intransitive verbs receive A/S^, prefixes from set 1 (60). Inactive verbs receive O/S^, person markers from set 2, as in exam ples (61) (class I) a nd (62) (class II), (60) *ere-ts 6 ‘you (sg) go’ ere-sö (Tb), ere-h 6 (Gj), eré-o (W) *o-ts 6 ‘he/she/they go(es)’ o-s 6 (Tb), o-h<5 (Oj), 6-o (W) (61) *né k a tu ‘you (sg) are good’ *i-katu ‘he/she/they is/are good’ né katii (Tb), nekatii (Gj) i-katu (Tb, Gj), i-kátu (W) (62) *tjé r-orÿp ‘1 am happy’ sjé r-oryp (Tb), he-r-urÿw eté *tJ-orÿ|3 ‘he/she/they is/are happy’ (Gj) s-orÿjJ (Tb), h-urÿw eté (Gj), 0- 6 ry (W) In transitive verbs there is a person hierarchy ( 1 > 2 > 3 ) which indicates •whether the verb should be cross-referenced w ith A , O, or a com bination o f the two. H ierarchically superior O is m arked w ith person markers from set 2, as in exam ples (63) (class I) and (64) (class II). W hen O is third person, the verb receives an A prefix from set 1, followed by a third person prefix from set 2, as in exam ­ ples (65) (class I) and ( 66) (class 11).^'^ W hen A is first person and O is second person, special prefixes from set 4 are used which distinguish the num ber o f O (67). (63) * tjé p o tá r ‘(X) like(s) me’ s jé p o tá r (Tb), h e-p u tár (Gj), e -p ö ta (W) (64) * tjé r-ekár ‘(X) sought m e’ s jé r-ekár (Tb), he-r-ekár (G j), e-r-éka (W) Harrison (1986) contains a thorough discussion of the types of splits that affect the TupiGuarani person-marking systetn, particularly in reference to independent and oblique-topicalized verb constructions. Special O allomorphs *jo- (class 1) and *jots- (class II) occur with monosyllabic stems. Several languages no longer include the O prefix in this construction. For example; o-pota ‘he wants’ (W) rather than o-i-pota. 156 CherylJensen (65) *a-i-potár ‘I like it’ a-i-potár (Tb), a-p u tár (Gj), a-p ö ta (W) (66) *a-ts-ekár ‘I sought it’ a-s-ekár (Tb), a-ekár (Gj), a-éka (W) (67) *oro-ekár ‘I/we sought you (sg)’ oro-ekár (Tb), uru-ekár (Gj), oro-éka (W) These forms, particularly the equivalent o f * o p o , have been m odified in some languages. In K ayabi and U ru b ù -K aap o r they have been replaced by forms from set 1, indicating the first person A , w ith O being indicated through a free pronoun.^* 9.2 Oblique-topicalized verbs The oblique-topicalized (O BL.TO P) verb construction is used when some sort of setting (time, location, o r other circum stance) occurs in the initial position o f the sentence as an adverb (68), postpositional phrase (69) or subordinate clause (70). In this case the O B L.TO P verb occurs at the end o f the (independent) clause, receives absolutive m arkers from set 2, and receives a special O B L.TO P suffix (-t with consonant-final stem s and -n, -w o r -J, depending on the language, with vowel-final stems). In some languages, such as Tupinam bá and Kayabi, this construction occurs only when A or is third person. In others, such as G uajajára an d A surini o f Tocantins, it occurs when A o r is first o r third person, but never when it is second person. In some languages only a few rem nants o f this system are retained. (68) (Tb) [kwesé] yesterday p ajé s jé sham an I s g supán-i suck-O B L.TO P ‘Yesterday the sham an sucked (to cure) me.’ (69) (A) [Tokorohi(-a) -pe] i-ha-i Tucurui to 3-go-OBL.TOP ‘He/she went to Tucurui.’ (70) (A) [éne 2sg i-nöpo-ram o] i-ha-pôtar-i 3-hit-if 3-gO-FUT-OBL.TOP ‘I f you hit him, he will go away.’ For a fuller discussion, see Jensen (1987 and 1998b). 5 Tupi-Guarani 9.3 157 Serial verb constructions When two or m ore actions are perceived as p art o f a whole and the subject o f both is the same, they generally occur in a serial verb construction.^’ If the dependent serial verb is transitive (71) o r active intransitive (72), it receives a suffix having three allom orphs: *-áJ3o (with vowel-final stems), *-a (with consonant-final stem s) and *-fa (with stems ending in a diphthong VJ). If it is an inactive intransitive verb (73), it receives a different suffix, w ith allom orphs *-ram o and *-awo. The transitive verb com bines with person markers from set 2, referring to O, or with a noun. The intransitive dependent verb receives coreferential person m arkers since its S is the same as that o f the independent verb. Lei te ( 1987) cites the following examples from Tapirapé: (71) w y rà ? i a ra - p y ÿ k i-J o k â -w o i-? 6 -w o b ird lex c l-c a tch 3-kill-SER.v 3-eat-SER.v ‘W e c a u g h t th e b ird , k illed it a n d a te i t.’ (72) â-Jaô k w e-yytâp-a Is g - b a th e Isg.C O R E F -sw im -S E R .v ‘I’ll b a t h e a n d I ’ll s w im .’ (73) J e -k an e?6 w e -ty ? â -r a m ô Isg-tired Isg.cOREF-hungry-SER.v ‘I am tired and hungry.’ In some languages the coreferential prefixes have been replaced by prefixes from set 1 (Jensen 1990) and the serial verb suffix has been deleted. For example, a -jív y á -jo ‘I returned, com ing’ from the A m apari dialect o f Wayampi. 9.4 Temporal subordinate clause constructions Verbs o f tem poral subordinate clauses are marked according to an ergative-absolutive system (by set 2 markers or by a noun immediately preceding the verb) and occur at the end o f the clause, followed by the subordinating m orphem e *-ire ~ * -(ri)re ‘after’ o r * -V m V ~ - r V m V % w hen’. What I am calling a dependent serial verb has been referred to as a dependent verb and also (in Portuguese) a gerund. In Portuguese this term refers to what in English would be a present participle, with an adjectival rather than nominal function. In traditional English gram m ar this term implies that the construction occurs as a noun in contrast to a present participle. According to English terminology, I do not see the term ‘gerund’ to be appro­ priate, although there is some similarity to certain nominalizing suffixes {-a, -to) and to the attributivesuffix (-ra/no,-a/no). ■ 158 Cheryl Jensen (74) (A) i-ha-ram o o ro -karo-p6ta 3-go-when le x c l-e a t-F U T ‘W hen he goes, we will eat.’ (75) (A) se-n6po-ram o ere-pdka Isg-hit-when 2sg-laugh ‘When he hit me (with an instrum ent), you laughed.’ I f the referent o f the person m arker (i.e. S^, o r O) is identical w ith the subject of the independent clause, coreferential m arkers may be used, depending on the extent to which these are used in a particular language. (76) (A) w e-n6po-ram o a-h a-p ö ta (ise) Is g .e o R E F - h it- if Isg-gO -F U T Isg ‘If he hits me (with an instrum ent), I will go away.’ In some languages the person-m arking on subordinate verbs has become like that o f independent verbs, as in Mbyá: (77) a-poranû ram o o-m opái Isg-ask when 3-answer ‘Wlien Í asked he answ ered.’ 9.5 Valency-changing devices 9.5.1 Causatives The Tupi-G uarani languages have three types o f causatives: *mo-, which combines w ith intransitive verbs, nouns and even nom inal suffixes, resulting in a transitive verb (78); * (e)ro-, a com itative causative, which com bines with active intransitive verbs, resulting in a transitive verb (79); and *-ukár, which com bines with transi­ tive verbs, resulting in a ditransitive verb (80). (78)(P, Mb) mo-?é (Mb) mo-akii (P, Kb) mo-?i ‘CAUS'Say’ ‘t e a c h ’ ‘C A U S-hot’ ‘(to) heat’ ‘C A U S -snm ir ‘(to) pound, crush’ (79) (W) ero-?á ‘C A U S -fair ‘to fall and c a u s e ( 0 ) to f a ll’ (80) (W) ju k á-ukí ‘kill-CAUS’ ‘to cause to k ill (O )’ In (80) the original O continues to be O and the original A becomes an indirect object. 5 Tupi-Guarani 9.5.2 159 Detransitivizers Two m orphem es, *je- ‘reflexive’ and * j o ‘reciprocal’, may occur with transitive verbs, resulting in an intransitive verb, as In the following examples from A surini o f Tocantins. (81) 9.5.3 o -se-nöpo ‘He hits him self (with an instrum ent).’ o-so-n<5po ‘They hit each other.’ Object incorporation A direct object may be incorporated in a transitive verb construction. If this object does n o t require a possessor, as in example (82), the verb is detransitivized. I f it requires a possessor, the possessor becomes the direct object, as in exam ple (83).“* High saliency o f an object precludes its incorporation. (82) (M ) a-y-?d 1sg-water-eat ‘I drink-w ater.’ (83) (M ) a-i-po-k”á lsg-3-hand-tie ‘I tied his hands.’(lit. ‘I hand-tied him ’) T he incorporation o f the generic m orphem es *poro ‘people’ o r *m a?e ‘thing’ also results in an active intransitive verb, as in examples (84) and (85) from Chiriguano. (84) ■ a-poro-m 6e Isg-people-teach ‘I teach.’ (85) a-m ae-juka Isg-thing-kill ‘I h u n t.’ 10 NOM INALIZATIONS Adverbs and postpositional phrases are nominalized by a reflex of *-tsw éror *-nwár. (86) (K b) kope-w át ‘the here ones’ (W J) arje'íé-w a ‘the now ones’ (K b) y tu pe-w át ‘the one at the waterfall’ In the first case the subject A becomes S^, in the second case it remains A and the posses­ sor is raised to O. 160 CherylJensen Verbs are norninalized to indicate the agent (reflex o f * -tsá r~ -ä r ~ -tái% circum­ stance { * -ts á p ~ -á ^ '~ -tá p ), and, in the case o f transitive verbs, object (*eini-) or patient (*-pfr). (87) (WA) e-m o?é-ar Isg-teach-AG.NMLZR (88) (G j) ‘m y te a c h e r’ i-z ù k a -h à w 3-kill-ciRCUMSTANCE (89) (G u) ‘his being killed’ jan e-r-em i-ápo lin cl-L iN K -o .N M L Z R -m ak e (90) (K b) ‘o u r w o r k m a n s h ip ’ i-juka-pÿt 3-kill-PATiENT.NMLZR ‘the killed o n e’ The relativizer *^a?é (or its variation ma?ê} occurs in m ost languages, though the extent to which it is used varies. In some languages, like O uajajára and Kayabi, it only occurs with intransitive (active and stative) verbs, the referent being S (S^, and Sg), as in example (91). (91) (Gj) o-ho m a?é ‘the one who w ent’ (O j) i-m à?eahÿ m a?é (K b) i-mS m a?é ‘the sick one’ ‘the m arried (husbanded) one’ In other languages, such as G uarayu, M byá, Tupinam bá and Wayampi, it occurs with transitive verbs as well, as in the two examples in (92), where the referents are A and O, respectively. (92) (M ) Je-r-eraá Pa?e ‘the one that took m e’ (M ) o ro-?u Pa?é ‘(that) which we are eating’ In Wayampi, the indirect object may also be relativized, as in example (93). (93) (W) a-m é?ê i-jupe ma?e 3-tO RELZR ‘the one to whom I gave it* Isg-give 11 SEMANTICS In the T upi-G uarani languages the centre o f the em otions is considered to be the liver {*py?á). A num ber o f constructions, all containing the *py7á morpheme, appear in various languages to com municate feelings, as illustrated by examples (94)^98): 1161 5 Tupi-Guarani (94) (G j) z e -m u -p y ? á (REFL-CAUS-Iiver) ‘to d o u b t’ (95) (M ) je - m o -p y ? á (REFL-CAUS-Iiver) ‘to p lan , reso lv e ’ p y ? a -g u a x u (liver-big) ‘to be c o u ra g e o u s ’ p y ? a -p o ri (liver-goo d ) ‘to b e h a p p y ’ (96) (P) py?a-m o-tÿP (liver-CAUS-tighten) ‘to b e h appy, a m u s e d ’ (97) (U ) p y ? á kanim (liver lose) ‘to forget’ (98) (WA) p y ? á r y rÿ i (liver tre m b le ) ‘to be a fra id ’ py?á-hu (liver-big) ‘to be u n a fr a id ’ p y ? á k a tii (liver g o o d ) ‘to love’ p y ?á m o n d (liver send) 'to lo n g fo r’ ji-p y ? a - m o rjé ta (REFL-liver-speak) ‘to th in k ’ p y ? a -k á tu (liver-goo d ) ‘to be a t e a se ’ I t is also possible that the m orphem e p y which occurs in some words is also derived from *py?d, such as: p y a tä (liver-hard?) ‘strong, diligent’ in Tupinam bá, an d p y p é (Hver-at?) ‘in, inside o f ’ in several languages. T he concept o f pairs (irO ‘p air’) is basic to the numerical systems o f several TupiG u aran i languages, as can be seen in the examples (99)-(102). (99) (W) iro *(to be) even’ níröi ‘(to be) odd’ (lit. ‘n o t even’) iröte ‘four’ iröiröte ‘six o r eight’ (100) (A) nairoihi ‘three’ (lit. ‘n o t even’) (101) (Gj) n à?irû z ‘three’ (lit. ‘not even’) (102) (M ) irü ‘p artn er’ je-m o-irû ‘to be a p artner’ iruny ‘fo u r’ BIBLIOGRAPHY Anchieta, J. de. 1595. Arie de gramm atka da lingoa mais iisada na costa do Brasil. Coimbra; Antonio de Mariz. (There are various reproductions o f this work, including one by the Biblioteca Nacional, Rio de Janeiro, 1933. and one by E ditera Anchieta, Sao Paulo, 1946.) Bendor-Samuel, D. 1972. Hierarchical structures in Guajajára. N orm an: SIL o f the University of Oklahoma. 162 CherylJensen Betts, L. V. 1981. Dicionmio Parintintin-Portuguis Português-Parinüntin. Brasilia; SIL. Cabral, A. S. A. C. 1995. ‘Contact-induced language change in the Western Amazon; the nongenetic origin of the Kokáma language’, Ph.D. thesis. University o f Pittsburgh. Dietrich, W. 1986. Elidioma Chirigtiano. M adrid; Institute de Cooperacion Iberoamericana. 1990a. ‘Chiriguano and Guarayo word form ation’, pp. 293-320 o f Amazonian linguistics: studies in Lowland South American languages, ed. Doris L. Payne. Austin; University of Texas Press, 1990b. More evidence for an internal classification o f Tupi-Guarmi languages. Berlin; Gebr. M ann Verlag. 1993. ‘Mbyá, guarani criollo y castellano; el contacto de las tres Icnguas estudiado en un grupo mbyâ en Missiones'. Signa and Seña: revista del Instituto de Linguistica, 3.55-71. 1995a. ‘Elespañol del Paraguay en contacto con el guarani’, pp. 203-16 o f Lengttas en con­ tacto en Hispanoamérka: nuevos enfoques, ed. by K.. Zimmerman. Vervuert; Bibliotheca Ibero-Americana. 1995b. ‘La importancia de los diccionarios guaranies de M ontoya para el estudio comparativo de las lenguas tupi-guaranies de hoy', Ainerindia: Revue d’ethnolinguistique amérin­ dienne, 19/20.287-99. Dobson, R. 1973. ‘N otas sobre substantivos do Kayabi’, Série Linguistica, 1.30-56. Brasilia; SÍL. 1988. Aspectos da lingua Kayabi, Série Linguistica, 12. Brasilia: SIL. Dooley, R. A. 1982. VocabuUirio do Guarani. Brasilia: StL. Grimes, B. E 1996. Ethnologue: languages o f the world. 13th edition. Dallas: SIL. Harrison, C. 1986. ‘Verb prominence, verb initialness, ergativity and typological disharmony in G uajajára’, pp. 407-39 of Handbook o f Amazonian languages, vol. I, ed. D. C. Derbyshire and G. K. Pullum. Berlin; M outon de Gruyter. Hoeller. A. 1932. Guarayo-Deutsches Wörterbuch. Guarayos e Hall in Tirol; Verlag der Missionsprokura der P. P. Franziskaner. Jensen, C. 1987. ‘Object-preflx incorporation in Proto-Tiipi-Guarani verbs’. Language Sciences, 9.45-55. 1989. 0 desenvohimento Mstorico da lingua Wayampi. Campinas: Editora da Universidade Estaduai de Campinas. 1990. ‘Cross-referencing changes in some Tupi-Guarani languages’, pp. 117-58 of Amazonian linguistics: studies in lowland South American languages, ed. Doris L. Payne. Austin; University o f Texas Press. 1998a. ‘The use of corefercntial and reflexive markers in Tupi-G uarani languages’, Journal o f Amazonian Languages. 1(2). 1 ^ 9 . 1998b. ‘Comparative Tupi-Guarani m orphosyntax’, pp. 487-618 o f Handbook o f Amazonian languages, vol. JV, ed. D. C, Derbyshire and G. K. Pullum. Berlin; Mouton de Gruyter. Juliâo. M. R. S. 1993. ‘A lingua dos Indios do Rio Cairari’, MA thesis, Universidade Federal do Pará. Kakumasu, J. Y. 1986. ‘U rubu-K aapor’, pp. 326-403 o f Handbook o f Amazonian Languages, vol. I, ed. D. C. Derbyshire and G. K. Pullum. Berlin; M outon de Gruyter. Kakumasu, J. Y. and Kakum asu, K. 1988. Dicionárío por tôpicos: Unihü-Kaapor-Português, Brasilia; Coedition Fundaçào Nacional do Índio/SIL. Leite, Y. 1911. Aspectos dafonologia e morfofonologia Tapirapé. Lingüistica, 8 . Rio de Janeiro: Museu Nacional. 1987. ‘Referential hierarchy and Tapirape split marking systems’, paper presented at the Working Conference on Amazonian Languages, Eugene, Oregon. Lemle, M. 1971. ‘Internal classification o f the T upi-Guarani linguistic family’, pp. 107-29 o f Ttipi studies 1, ed. D. Bendor-Samuel. N orm an: SIL. 5 Tupi-Guarani 163 Mcllo, A, A. S. 1992. ‘Estudo com parative do léxico da familia lingüisticaTupi-Guarani’, MA thesis, Universidade Federal de Santa Catarina. Newton, D. 1978. ‘Guarayu discourse’, pp. 252-68 o f Work papers o f the SÍL, ed. U. W iesemann eí ai, Riberaita (Bolivia): SIL. Nicholson, V. 1978. Aspecíos da lingim Asiirinl. Brasilia: SIL. Rodrigues, A. D. 1952. ‘Anáiise m orfologica de um texto Tupi’, Logos, 7.15. 1958. ‘Classification of Tupi-G uarani', UAL, 24.231-4. 1984/5. ‘Relaçôes internas na familia lingüistica tupí-guaraní’, Revista de Anlropologla, 27/28.33-53. 1986a. Lingtias brasiteirm: para o conhecimento das Hngtias indigenas. Sào Paulo: Ediçôes Loyola. 1986b. ‘Silêncio, pausa e nasalizaçào', pp. 153-9 o f Anais do 8" Enconiro N aciom t de Lingüistica. Rio de Janeiro. 1996a. ‘Panoram a das linguas indigenas da Am azônia’, paper presented al the conference. As Linguas Indigenas da Am azônia na Ciênciae nas Sociedades, organized by the Museu Paraense Emilio Goeldi. 1996b. ‘A s linguas gérais sui-americanas’, Papia. Revista de Crioulos de Base Ibêrica, 4{2).6-18. Rodrigues, A. D. and Dietrich, W. 1997. ‘On the linguistic relationship between Mawé and Tupí-G uaraní’, Diachronica, 14.265-304. Ruiz de M ontoya, A. 1639. Tesoro de la lengua Guarani. M adrid: Juan Sanchez. (There are various editions of this, including one edited by P. Restivo and published in 1892 as Linguae Guarani Grammatica Hispanice, published in Stuttgart by Guilielmi Kohlhammar.) Schuchard, B. 1979. Ñande Ñ l: Gramálica guarani para castellano hahíanles, Santa Cruz de la Sierra: Ayuda para el Campesino del Oriente Boliviano / Centro Boliviano de Investigacion y Accion Educativas. Seki, L. 1990. ‘Kam aiurá (Tupi-Guarani) as an active-stative language', pp. 367-91 of Amazonian linguistics: studies in lowland Smitii American languages, ed. Doris L. Payne. Austin: University o f Texas Press. Soares, M. E and Leite, Y. 1991. ‘Vowel shift in the Tupi-G uarani language family: a typo­ logical approach’, pp. 36-53 o f Language change in South American Indian languages, ed. M. R. Key. Philadelphia: University o f Pennsylvania Press. Map 5 Macro-Jê languages with approximate locations 6 Macro-Jê A R Y O N D. R O D R IG U E S 1 H IS T O R IC A L SU R V EY M acro-Jê is the nam e that was proposed about fifty years ago by M ason (1950; 287) for a large array o f South American languages (all spoken in Brazil) thought to be related to the Jê linguistic fam ily W. Schmidt (1926: 234-8) had earlier employed the name Ges-Tapuya, and Loukotka (1944 [1942]: 2-6) the nam e Tapuya-zé in the same sense. The possible genetic relationship am ong the many languages ascribed to the Macro-Jê stock is a working hypothesis whose details have varied according to different scholars. Loukotka included in it the following eight ‘language families’: zé (Jê), Opaie (Ofayé), Kaingán (KLaingáng), Coroado (Purí), M asakali (M axakalí), Patasá (Pataxö), Botokudo (K renák) and K am akan (K am aká) (names or spellings added in parentheses are those now currently used by Brazilian linguists and anthro­ pologists and adopted in the present chapter). N im uendajù (M étraux and N im uendajù 1946: 542; N im uendajù 1945 [1980]) considered M alalí an independent linguistic family, whereas M ason (1950) added M alali and Coropo (K oropô) to Macro-Jê, but took out Ofayé as well as laté (Yatê). The latter had already been sep­ arated from the stock by Loukotka (1942). Later, on the basis of good com parative work, Davis (1966) dem onstrated that Kaingáng is truly a member o f the Jê genetic family and not another family in itself. In another paper, Davis (1968) gave evidence of regular phonological correspondences between Jê and M axakali as well as between Jê and K arajá and mentioned possible wider relationships of M acro-Jê with Bororo, Tupi, and Fulniô (Yatê). As for Borôro, G uérios (1939) presented as indica­ tive of genetic relationship the similarities he had found between Eastern Borôro and two N orthern Jê languages, Tim bira (‘Merrime’ = Canela) and Kayapô. G udschinsky (1971), com paring Ofayé with Davis’ reconstruction o f proto-Jê, showed that it is more likely to be a member of M acro-Jê. Boswood (1973) gave some lexical evidence in favour o f the inclusion o f R ikbaktsá in this stock. Rodrigues (1986) included in it K arin and G uato, but dealt with Pataxô as a m em ber o f the M axakali family, the same being true of M alali, which had already been put in this family (and not as a main branch of M acro-Jê) by Loukotka. For Greenberg (1987), all the aforem en­ tioned languages or language families (except for K ariri) belong to M acro-Jê, as do 166 Aryan D. Rodrigues Chiquito, Oti and Yabuti (Jabuti). A lthough Greenberg (1987: 86) states that ‘these three languages are as validly Macro-Ge as the others’, the meagre data he presents for Oti and Yabuti do not substantiate his claim. Kaufman, in his recent revision of the classification o f South American languages (1990, 1994), left out of M acro-Jê these two languages as well as Kariri, but retained Chiquito. In the present chapter Karirl, but not Chiquito, is considered a possible member o f Macro-Jê. A good part of the languages involved in the M acro-Jê hypothesis are already dead and most of them have been very poorly documented. Thus it is very difficult to work out their relationships with the best-known members of the stock. Table 6.1 lists the language families and the single languages that will be considered here as possible members o f Macro-Jè (for which there is some docum entation, albeit very scarce in some cases). For some selected lexical and phonological evidence of the consistency of the whole group see §7 below. Table 6.1 also gives information on the locations of the languages and the number of speakers. Names of dead languages are preceded by t. The approximate locations o f the extant M acro-Jê languages are shown on map 5. 2 D IS T R IB U T IO N Although several M acro-Jê languages are spoken in Brazilian Amazonia, the geo­ graphical distribution o f this linguistic stock is rather circum-Amazonian, encircling Amazonia on its eastern and southern sides. M ost members o f the Macro-Jê linguis­ tic stock have been spoken in eastern and northeastern Brazil, but a few language families are found in central and southwestern Brazil. With the exception o f O tuke of the Borôro family, spoken west o f the Paraguay river in Bolivia, the whole MacroJê stock is found entirely in what is today Brazilian territory. In the sixteenth century, when the Europeans (mainly Portuguese and French) started trading or settling on the Atlantic coast o f S outh America, most o f this coast was occupied by Indians speaking languages o f the Tupi-G uarani family. In some places people speaking M acro-Jê languages used to go to the seashore seasonally in order to gather seafood. This was probably the case with the M arom om im or G uarulho on the coast of present-day S io Paulo State and with the W aitaká on the coast o f northern Rio de Janeiro and Espirito Santo, but it is likely that many other peoples would have done the same. Both the M arom om im and W aitaká spoke languages that probably belonged to the Puri family of M acro-Jê, but both languages became extinct and no record of them has been preserved or, at least, has not so far been found. O f the lan­ guage o f the M arom om im there is clear historical inform ation that a gram m ar and a dictionary as well as a Christian doctrine, were written in the sixteenth century by the Jesuit Manuel Viegas in cooperation with Joseph de A nchieta, the linguistically skilled au th o r o f the first gram m ar o f the T upi-Guaran! language Tupinambá. From a geographical point of view the M acro-Jê languages may be divided into 6 Macro-Jê 167 Table 6.1 M acro-Jê languages I Je family a North-eastern }ê 1 tJaiko (south-eastern Piaui) b N orthern Jê 1 Timbtra (including C anela Ram kokamekra, Canela Apanyekra, Gaviâo Piokobjé, Gaviào Parakatejé, Krinkatí, K rahô, Krenjé; M aranhSo, Pará. Tocantins; 2,800) 2 Apinajé (northern Tocantins; 720) 3 Kayapö (including A’ukré, Gorotire, K araraô, Kikretum , Kokrainiôro, Kubenkrankén, M enkrangnoti, M entuktîre, Xikrin; eastern M ato Grosso, south­ eastern Pará; 5,000) 4 Panará(K reti-akaröre) (Ârea Indîgena Panará, northern M ato Grosso and south-western Pará; 160) 5 Suyá (including Tapayuna; Xingu Indigenous Park in M ato Grosso; 213 S., 58 T.) c Central Jê 1 Xavánte (south-eastern M ato Grosso, formerly western and northern Goiás; 9,000) 2 Xerénte (Tocantins; 1,530) 3 t? X akriabà (M inas Gerais; 5,700 ethnic, probably no speakers) 4 t Akroá (eastern Goiás, southern M aranhâo) d Southern K 1 Kaingáng (including Sao Paulo K., Parana K., Central K., South-western K,, and South-eastern K.; S io Paulo, Parana, Santa Catarina, Rio Grande do Sul; 20,000) 2 Xokléng (Santa Catarina; 1.650) 3 t Ingain (north-eastern Argentina, south-eastern Paraguay) II Kam akä family 1 t Kam akà (south-eastern Bahia, north-eastern Espirito Santo) '2 t Moiigoyo (south-eastern Bahia) 3 t Menién (south-eastern Bahia) 4 t Kotoxo (south-eastern Bahia) 5 t M asakará (north-eastern Bahia) III M axakali family 1 Maxakali (north-eastern M inas Gerais, northern Espirito Santo; 854) 2 t Kapoxd (including Kumanaxo and Panháme; north-eastern M inas G erais and south-eastern Bahia) 3 t M onoxo (north-eastern M inas Gerais and south-eastern Bahia) 4 t Makoni (north-eastern M inas Gerais) 5 t M alali (north-eastern Minas Gerais) 6 t Pataxo (including H âhàhàe; south-eastern Bahia; 4,600 ethnic) IV Krenák family 1 SCrenák (formerly called Botocudo, including Nakrehé, Nakpié, Naknyanùk, Nakyapmä, Nyepnyep, Etwet, M inyâyirùn, Yiporôk, Pojitxá, Potén, Krekmíin, Bakuén, Aranä; north-eastern M inas G erais and northern and central Espirito Santo, formerly also south-eastern Bahia; some families now also in central Sâo Paulo; ! 00 ethnic, about iO speakers) 2 t G uerén (south-eastern Bahia) V Puri family I t Puri (Espirito Santo, Rio de Janeiro, north-eastern Saô Paulo, south-eastern M inas Gerais) 168 Aryon D. Rodrigues Table 6.1 (cont.) t Koropô (Espirito Santo) 3 t Coroado (Espirito Santo) Kariri family 1 t Kipeá (also known as Kiriri; north-eastern Bahia and Sergipe; 1,830 ethnic) 2 t Dzubukuá (on the islands of the Sao Francisco river, in northern Bahia) 3 t Sabuyá or Sapoyá (central Bahia) 4 t Kamurii or Pedra Branca (eastern Bahia) Yatê family 1 Yatê (the people are named Fulniô, formerly known as Carnijo; Pernambuco; 3,000) Karajá family I Karajá (including Southern K., Northern K., Javaé and Xambiwá; eastern Mato Grosso, western Tocantins; 2,900) Ofayé family 1 Ofayé (including Ivinheima O. and Vacaria O.; also known as Opayé and OfayéXavánte; eastern Mato Grosso do Sul; 87 ethnic, about 25 speakers) Bororo family 1 Eastern Bororo (southern Mato Grosso; 1,072 ethnic, some of which no longerspeak Bororo) 2 t Western Bororo (eastern Mato Grosso) 3 Umutina (western Mato Grosso; 100 ethnic, I speaker) 4 t Otiike (including Kovare[ka] and Kurumina[ka]; eastern Bolivia) Guato 1 Guato (south-western Mato Grosso; 380 ethnic, only about 5 speakers) Rikbaktsá family 1 Rikbaktsá (northern Mato Grosso; 990) 2 VI VII VIII IX X XI XII eastern, central and western, extending diagonally across the South American low­ lands from the eastern and north-eastern A tlantic coast to the upper Paraguay river. The central group com prises the K arajá a nd Jê families, the latter spreading north­ wards and southw ards along the savanna (cerrados and campos) plains to Piaui and M aranhâo in the north and to Rio G rande do Sul in the south. K arajá extends also on a n o rth -so u th axis, but is restricted to the valley o f the A raguaia river, includ­ ing Bananal island. Eastern M acro-Jê languages are those belonging to the Puri, K renák, M axakali, K am akà, Kariri and Yatê families. The western families are Ofayé, Bororo, R ikbaktsá and G uato. The distribution o f the eastern M acro-Jê language families extending from eastern to north-eastern Brazil was as follows (taking the modern Brazilian states and some main rivers as geographical references). Puri was along the whole extent o f the P araiba do Sul river and spreading northw ards until the D oce river, in n orth-eastern Sao Paulo, R io de Janeiro, south-eastern M inas Gerais, and southern Espirito Santo. K renák extended from the Doce river to the Paraguaçu river in north-eastern M inas Gerais, central and northern Espirito 6 M acro-Jê 169 Santo, and south-eastern Bahia. M axakali went from the Sapucai and Sâo M ateus rivers to the Jequitinhonha river in eastern M inas Gerais, northern Espirito Santo and south-eastern Bahia. K am akâ was between the Jequitinhonha and C ontas rivers in south-eastern Bahia. K ariri was located m ainly between the Itapicuru and the m iddle an d lower Sâo Francisco river, in central and north-eastern B ahia and southern Sergipe, and with some extensions northw ards and southwards, probably due to the Portuguese invasion o f their territory in the seventeenth century. Ÿatê was no rth o f the lower Sâo Francisco, in eastern Pernam buco. O f the western families, B orôro spread from the sources o f the A raguaia river into eastern Bolivia, and G uato was found on the upper Paraguay, im mediately to the southw est o f Bororo. Ofayé occupied a stretch o f the upper P araná river from the m outh o f the Tietê downwards, passing the m outh o f the P aranapanem a and extended to the west to the headwaters o f the right-bank tributaries of the P araná, reaching also the headwaters o f some tributaries o f the Paraguay river. R ikbaktsá is located far to the north, on the upper Juruena river, one of the headwaters o f the Tapajôs. R ik b ak tsá is the only language of the M acro-Jê linguistic family to be found exclusively in A m azonia. However, some peoples speaking languages o f the Jê family also live in A m azonia, such as the Panará, the Suyá, the Tapayuna and the K ayapô. It is likely that all o f the latter entered A m azonia coming from the east under the pressure o f the Portuguese invasion o f the savannas in central Brazil. In the eighteenth century the P anará m ainly lived around the headwaters of the P araná river, such as the Paranaiba and the G rande, in present-day southern G oiás, south­ western M inas G erais and northw estern Sâo Paulo, where they came to be known as K ayapô. A fter their disappearance due to the w ar waged against them by the Portuguese, this nam e was applied to another Jê people, whose language is sim ilar to, but n o t the same as, that o f the Panará. Very recently (1968) a large population o f P an ará was discovered on the Peixoto de Azevedo river, another headw ater of the Tapajôs, where they h ad taken refuge and m anaged to live for m ore th a n two centuries in freedom and peace. Shortly after having been forced into a new contact with white people in the early 1970s, the P anará population was drastically reduced by epidemics. The survivors were transported to the X ingu Indigenous P ark where they have since lived, but they are now m anaging to get back to a p a rt o f their form er territory. The m odem K ayapô moved into eastern A m azonia during the first half o f th e twentieth century, b u t they were preceded in this move by the Suyá, who were already on the U pper X ingu in the second h alf o f the nineteenth century. 3 LINGUISTIC SCHOLARSHIP A significant num ber of M acro-Jê languages became extinct as a consequence o f the European settlem ent in Brazil. T he worst-hit language families were those located in 170 Aryon D. Rodrigues eastern Brazil, since they were m ost affected by the Portuguese gold-seeking and stave-hunting expeditions. Although the very first victims o f the European expan­ sion were the Tupi-Guarani-speaking Tupinam bá, who during the sixteenth century were wiped out on large stretches of the A tlantic coast, some M acro-Jê-speaking peoples became the objects o f Portuguese attacks around this time - for example, th e Aimoré in southeastern Bahia (whose language probably belonged to the Krenàk family). A t the beginning o f the nineteenth century there were still small groups of , Indians in eastern Brazil who spoke languages of all the families here ascribed to M acro-Jê, but many other languages were already extinct and we do not know whether they were indeed M acro-Jê or not. The first years of the nineteenth century were characterized by the beginning o f the study o f Brazilian natural history by European scientists, mainly G erm ans and French. M any botanists, zoologists and geologists crossed the country gathering samples o f the flora, the fauna and the soils and also gathered samples o f the languages spoken by the Indians they met on their way. A t that time there were no specialists in the study o f languages and the model o f language docum entation was the collection o f short com parative lists o f words, such as those published io the eighteenth century by Adelung and by Hervás. The scientific curiosity o f the naturalists has saved from complete oblivion scores of words o f many languages that, during the following years, became extinct. All the languages o f the Puri, K am akà and K ariri families are now dead. Two lan­ guages of the K ariri family are quite well known thanks to the work o f seveoteenthcentury missionaries. A n Italian Jesuit published a fine gram m ar and a catechism in K ipeá (M am iani 1698, 1699) and a French Capuchin published a catechism in D zubukuá (B em ardo de N antes 1709). K nowledge o f the Puri languages is limited to the scanty data in lists gathered by European naturalists and by a Brazilian engi­ neer (for the linguistic data see L oukotka 1937). The K am akà language was the last of the family w ith the same nam e to disappear; it was still possible to elicit words and phrases from the last speakers in the 1930s and 1940s (Guérios 1945, N im uendajù and G uérios 1948; for earlier data see Loukotka 1932). For the K renâk family there are today no more than ten speakers representing tw o or three dialects of th e one language (inform ation on published and unpublished sources in Seki 1990; see also Silva 1986 and Sebestyén 1981). Some o f these, speaking the N akrehé dialect, live far from their original home, in the interior of the State of Sào Paulo, to where they were deported by the Brazilian adm inistration about forty yeare ago. The M axakali language is the only surviving m em ber o f the M axakali family, the other five languages now being dead (recent studies are Pereira 1992; Popovich 1967,1971, 1985, 1986; Gudschinsky, Popovich and Popovich 1970; Rodrigues 1981; Wetzels 1996; Wetzels and Sluyters 1996; for earlier data see Loukotka 1931 and 1939), Yatê .is the only indigenous language to survive in northeastern Brazil, In a region where there are rem nants o f several other indigenous peoples who now speak only 6 M acro-Jê 171 Portuguese. It is not known w hether the extinct languages o f some o f these peoples belonged to the same family as Yatê o r even w hether they were members o f the M acro-Jê stock (studies on Yatê are Lapenda 1968; Pinto 1956:265-76 and passim; M eland 1968; M eland an d M eland 1967, 1968; Barbosa 1991), T h e languages o f the Jê family were better preserved due to the inland location o f th eir speakers. A pparently only one main branch of the family disappeared com ­ pletely, namely Jaiko, w hose living area in the backlands of Piaui was one o f the first to be used by the Portuguese for intensive cattle-raising. The other three sub­ groups still have some healthy languages, despite suffering heavy population losses. For Jaikô the only source is a small word list gathered and published by Von M artiu s (1867: 143); for northern Jê the main contributions are Popjes and Popjes (1986), Souza ( 1989) (Timbira); Callow (1962), H am ( 1961,1967), H am , Waller and K oopm an (1979) (A pinajé); S tout and Thom son (1974a,b), Thom son and S tout (1974), Jefferson (1989), M . de N. de O. F. Borges (1995) (Kayapo); D ourado (1990, 1993a,b) (Panará); G uedes (1993), Santos (1997) (Suyá); for central Jê: M attos (1973) (Xerénte); Hall (1979), H all, M cLeod and Mitchell (1987), M cLeod (1974), M cLeod an d M itchell (1977), Lachnitt (1987, 1988) (Xavánte); for southern Jê: G uérios (1942), M ullen (1965, 1966), W iesemann (1971, 1972, 1978), Rodrigues and C avalcante (1982), Cavalcante (1987), Teixeira (1988) (Kaingáng); H enry (1935, 1948), U rb an (1985), Bublitz (1994) (Xokléng), T h e m ain literature on K arajá com prises Ehrenreich (1894), K u n ik e(1 9 1 6 ,1919), F ortune (1973), Fortune and Fortune (1975), M aia (1986), Cavalcante (1992), R ibeiro (1996), Borges (1997). T he only analytical essay on Ofayé is Gudschinsky (1974), which makes reference to previous sources. F or the Bororo family there is a gram m atical essay and a m onum ental encyclopaedic dictionary plus the text collec­ tions by the members o f the Salesian Mission (Colbacchini 1925; Colbacchini and A lbisetti 1942, Albisetti and Venturelli 1962, 1969, 1976), as well as Crowell’s gram m ar (1979), all for Eastern Bororo. For U m utina see M. Schm idt (1941), Schultz (1952) and Lima (1995); Rodrigues (1962) provides a com parison o f U m utina with Eastern Bororo. G uato was analysed by Palácio (1984, 1986), who gives inform ation on previous data. For R ikbaktsá see Boswood (1974a,b, 1978) an d Trem aine (1981). 4 PH ONO LO GY 4.1 Vocalic systems A s in o th er language fam ilies o f Lowland South America, a very com m on feature in m any languages of, the M acro-Jê stock is the presence o f phonologically con­ trastive nasal vowels. O ften it is the nasal vowels that condition the variation o f the 172 Aryan D. Rodrigues Table 6.2 Vowel system o f the Paraná dialect o f Kaingáng ( after fViesemann 1972} oral nasal front unrounded central unrounded rounded i e e i s a u 0 0 front unrounded central unrounded Q Î Ê back rounded 5 â Table 6.3 Vowel system o f Apinaje (after H am 1967) front unrounded central unrounded rounded front unrounded central unrounded back rounded consonants an d n o t the reverse (as happens in other languages). In general the num ber o f nasal vowels is less than the num ber o f oral ones. The Paraná dialect o f K aingáng (Jè), for instance, has nine oral and five nasal vowels, as may be seen in table 6.2. In this language the low central nasal vowel oscillates from rounded back [Ô] to unrounded central [â]. All the languages o f the Jê family have vowel systems as large as that o f K aingáng or with one or two more vowels. Apinaje, which dis­ tinguishes four instead o f three degrees o f height for central unrounded, has ten oral and seven nasal vowels as shown in table 6.3. The system o f nine o r ten oral vowels and a lesser num ber o f nasal vowels is typical o f the languages o f the Jê family; Davis (1966) reconstructed a system of nine oral and six nasal vowels for proto-Jé. A system com parable in the num ber o f oral vowels is found in the M akii family (see chapter 9). The languages o f the other families of M acro-Jê vary between nine and five oral vowels, b u t in general preserve the distinction o f three tongue positions (front, central and back). Some o f them have contrasting length. K arajá has nine oral vowels like K aingáng, but has only two nasal vowels, 3 and 5. Phonetically there is also [â], which is an autom atic real­ ization o f the phonem e /a/ when it either stands at the beginning of a word or is pre­ ceded by /h/ o r by a voiced stop, as in the following examples: adi [adi] ‘your 173 6 M acro-Jê Table 6.4 Vowel system o f Kipeá oral nasal front central back i e æ i u 0 a front central back ë X â 5 a Table 6.5 Vowel system o f M axakali oral nasal front central i e i back front central Î 0 Î ë back 5 a â Table 6.6 Vowel system o f O fayé oral nasal short long short long front central back front centra! back front central back front central back i e £ i: e; e: Î i u 0 a i: a: u; o: Î , Î: Î: Ô â 5: à: m other’, a 0 /[äöij ‘grass’, habit [habu] ‘m an’, d m [ohà] ‘arm adillo’, bahadii [màhàdu] ‘group’, dadi [nâdi] ‘my m o th er’, oOada [oOana] ‘face’. K ipeá, o f the K ariri family, has seven oral and five nasal vowels, as shown in table 6.4 {after Azevedo 1965), whereas M axakali has only five in each category, as may be seen in table 6.5 (after O udschinsky, Popovich and Popovich 1970). A ccording to G udschinsky (1974), Ofayé has seven oral and four nasal short vowels and as many long vowels (see table 6.6, after G udschinsky 1974), G u ato has eight oral an d five nasal (table 6.7, after Palácio 1984), and R ikbaktsá has six oral and six nasal vowels (table 6.8, after Boswood 1973). Yatê and B orôro have no nasal vowels and have only seven oral vowels each, but 174 Aryan D. Rodrigues Table 6.7 Vowel system o f Guatô oral nasal front central back front central back i e Í u 0 Ï ê I â Û a 0 Table 6.8 Vawel system o f R ikbakísá oral nasal front central back front central back i e i a u Î ê Î à Û Ô 0 Table 6.9 Vmvel sysiem o f Yatê front central back i e c a u o 0 Table 6.10 VoweÍ system o f Borôro front central back i e * u 0 3 a b oth partially preserve the three-colum n distinction as well as the three-row distinc­ tion o f the fuller M acro-Jê systems, as may be seen in table 6.9 (after M eland and M eland 1967) and table 6.10 (after Crowell 1979). 4.2 Consonantal systems The consonantal systems o f M acro-Jê languages are o f medium size. O nly Yatê has slightly more than twenty consonants incorporating a full set of aspirated and unas- 175 6 M acro-Jê Table 6.11 Consonantal system o f Yatê (after M eland and M eland 1967j voiceless unaspirated ' stop voiceless aspirated stop voiced stop voiceless fricative voiced fricative nasal approximant labial dental P P"* alveolar alveo-palatal velar t ts tJ k t» ts" d f s z m n I w glottal tf d3 J h i Table 6.12 Consonantal system o f Kipeá ( after Azevedo 1965) voiceless stop voiced stop fricative nasal approximant labial dental alveolar p b t d ts dz s m w n r velar glottal tj Ü3 alveo-palatal k g 7 Ji j 0 h Table 6.13 Consonantal system o f Giiatô ( after P aládo 1984) voiceless stop voiced stop fricative nasal approximant labial dental alveo-palatal velar labio-velar P b f m w t d tf d3 k g k»' g" glottal h n r j pirated voiceless stopa Besides these it has two voiced stops and five fricatives, two nasals and three approxim ants, as shown in table 6.11. K ipeá (K ariri family), Yatê’s neighbour to the south, h ad no aspirated stops, but h ad a full series o f voiced stops and also a glottal stop, only two fricatives, but four nasals and three approximants. N ote th a t Yatê has a lateral approxim ant, whereas K ipeá has a central flap, like m ost M acro-Jê languages (see table 6.12). T he consonantal system o f G uato resembles those o f Yatê and K ipeá, b u t has no series o f aspirated stops an d no alveolar stops o r fricatives; it adds to its inventory two labio-velar stops, as m ay be seen in table 6.13. Like Kipeá and G uatô, Borôro 176 Aryan D. Rodrigues Table 6.14 Consonantal system o f Borôro ( after Crowell 1979) voiceless stop voiced stop nasal approximant labial dental alveo-palatal velat p b m w t d n f t/ d3 k g j Table 6.15 Consonantal system o f M axakali (based on Gudschinsky, Popovich and Popovich 1970) voiceless stop voiced stop fricative labial dental alveo-palatal velar glottal p b t d tj j k g ? h Table 6 .16 Consonantal system o f Karajá ( after Fortune 1973) labial voiceless stop voiced stop voiced implosive voiceless fricative lateral approximant b w dental alveo-palatal velar tJ d3 k d rf 0 1 r ; glottal h has a series of voiced stops parallel to tiie voiceless ones, but it distinguishes only four points o f articulation and has no fricatives at all, as shown in table 6.14. M axakali has a rem arkably reduced inventory of ten consonantal phonemes, and also ten vocalic phonem es. It has a full series o f voiced and o f voiceless stops, each in four points o f articulation, and two glottal sounds, a stop and a fricative, as may be seen in table 6.15. The consonantal system o f K arajá is also rem arkable n o t only for its reduced inventory o f twelve phonem es, but also for its concentration on dental and alveopalatal sounds and its great dissymmetry. Table 6.16 displays this system. The voiced s to fs /b/ and /d/ have fully nasal allophones, fm] and [n], occurring before nasal vowels. A nother phonological peculiarity o f K arajá is the systematic differentiation o f 177 6 Macro-Jê Table 6.17 Differences between men's speech and womens speech in Karajá (data fro m Fortune 1973 and Borges 1997) women’s speech men’s speech meaning karfu kihi i/ikura kaluko karihokDcfekre fit/oko karit|akre kabkuda (faki hawako beraku itfaroôa adöda kawaru kabe dôbïku od'u ihi i/iura oluo arihood’ekre ritjoo ariakre oloda d’ii hawo bero idsaroöa aöda awaru abe dôbîu ‘turtle’ ‘wind’ ‘necklace’ ‘labret’ ‘I will borrow’ ‘doll’ ‘I will walk’ ‘feather’ ‘bark’ ‘canoe’ ‘river’ ‘dog’ ‘thing’ ‘horse’ (from Portuguese cavalo) ‘coffee’ (from Portuguese café) ‘Sunday’ (froin Portuguese domingo) the phonological shape o f words between men’s speech and women’s speech (Ehrenreich 1894, K rause 1911, Kunike 1916, F ortune and Fortune 1975, Borges 1997). M en’s speech regularly lacks the velar stops present in the speech o f women, as well as the instances o f the voiceless alveo-palatal affricate th a t are historically derived from velar stops palatalized by a preceding i. As a consequence o f the drop­ ping o f the velar and alveo-paiatal consonants, several vowel contractions may result, thus m aking the shape o f the words and sentences uttered by men m ore dis­ tinct from those pronounced by women. Even borrow ings from Portuguese are subject to the dropping o f the velar stop. Table 6.17 presents som e examples o f women’s an d m en’s speech in K arajá. It is not easy to conceive of a situation in which this gender-based systematic differentiation between the phonological shape o f the words might have originated. Perhaps in the past the women o f one dialectal group o f K arajá could have been sub­ jugated by warriors speaking another language, say one lacking velar stops but having glottal stops. Such warriors could have killed all K arajá male adults, taken their place as husbands and learned the K arajá language from their new wives, but substituted their glottal stops for the velar stops o f the women. This bad pronunciation by the new masters of the group would then have been maintained through the following generations and spread to other dialectal groups. In the course o f time, the articula­ tion o f the glottal stop would have weakened and finally disappeared, giving place to 178 Aryan D. Rodrigues Table 6.18 Consonantal phonemes o f Xaváme (after M cL eod 1974) voiceless stop voiced stop approximant labial dental alveo-palatal p b w t d r tj d3 glottal ? h vowel sequences and contractions (as shown in table 6.16, the K arajá language has no glottal stop phoneme). It happens that the neighbouring language Xavánte has historically undergone the systematic change o f velar consonants into glottal stops (compare Xerénte dakrà with Xavánte da?rä ‘somebody’s head’, Xe. d a h m with Xa. da ?wa ‘somebody’s to o th ’, and Xe. kuza w ith Xa. Piirfja ‘fire’). This language, which does not show a similar difference correlated with the sex of the speakers, may well be the source of the difference between K arajâ men’s and women’s speech (at least the main difference - the dropping of velar consonants in men’s speech). Table 6.18 shows the consonantal system o f Xaváiite, in which there is no velar phoneme. This is the only M acro-Jê language lacking velars. The consonantal systems o f the other languages o f the Jê family are in general simple, with a series o f four voiceless stops (labial, dental, alveo-palatal and velar) and another o f four corresponding nasals, and three approxim ants; Davis (1966) reconstructed such a system for proto-Jê (the approxim ants being »v, r and the latter being an arbitrary sytnbol for a phonetically undefined proto-phoneme). Some languages also include a glottal stop and/or a glottal fricative. T im bira has an aspirated velar stop contrasting with the plain one. K aingáng adds to this system a bilabial fricative, that is voiceless (<p) in m ost dialects, but is voiced ( f i in the dialect o f Sào Paulo. Xokléng, which is more sim ilar to Kaingáng, has a voiced interden­ tal fricative corresponding etymologically to the bilabial fricative o f Kaingáng, e.g. K aingáng (pi, Xokléng <5#‘seed’; K ( p a ,X â J ‘bitter’; K p a ^ , ‘to take the breast’, X p e ih , ‘to give the b reast’; K # 9 r, X del ‘skin’. X okléng also has a lateral approxi­ m ant where K aingáng has a central flap, as in the last example and in the follow­ ing: K râ, X la ‘sun’; K h ip /i, X kiipli ‘w hite’; K krl, X kl£ ‘head’. The Jê language with the m ost com plex system is K ayapô (table 6.19), which has a complete series o f voiced stops, contrasting with the voiceless ones and the nasals, as may be seen in the following examples: p i ‘w ood’, b î ‘to kill’, m l ‘alligator’. Some M acro-Jê languages have intem aily complex phonem es - that is to say, phonem es th at are realized by a sequence o f sounds (or, in physiological terms, produced by a sequence o f articulatory movements). The Southern Jê language K aingángis one of the most striking cases. In this language nasal conso- 179 6 M acro-Jê Table 6.19 Consonantal system o f Kayapá ( after S tout and Thomson 1974a) voiceless stop voiced stop nasal approximant labial dental alveo-palata! p b m w t d n r tj d3 velar k g i) glottal ? j nantal phonem es are pronounced in up to three phases of articulation, [- n a s a l|+ n a s a l ] [ —nasal] (Rodrigues and Cavalcante 1982, Cavalcante 1987; cf. A nderson 1974, 1976). T he word ka n cr ‘sm ooth’ is pronounced [kadnderc]: between oral vowels the nasal phonem e n starts as an oral stop [d], changes to the nasal [n], and changes back to an oral [d]. T he inner phase is fully nasal, whereas the first and third phases are assimilated to the oral vowels bordering the conso­ nant. If one o f the vowels is nasal, only the phase contiguous to the oral vowel is oral: ka n c ‘eye’ is pronounced [kadnê], whereas e /n e 'blue sky’ is [?êmbe]. If both vowels are nasal, the consonant is fully nasal; c m í‘bread’ [?èm î], ^ á n í‘b ack’ [pânî]. Pause or silence before and after the nasal consonant has the sam e effect as a nasal vowel; ncn ‘thing’ [ndedn], nlm ‘to give a long object’ [nïm], but word boundaries inside an utterance are no obstacle for the assim ilation o f the lirst and third phases o f the nasal consonants; no [ndo] ‘arrow ’, ti no [tidndo] ‘his arrow ’, no nïm [ndodnTm] ‘to give arrow s’. T he first example above, kaner, illustrates another com plex phoneme. The phonem e r starts as a vowel at the beginning o f a word, as in ra [ara] ‘thin’, rä [ârà] ‘sun’; and ends as a vowel with the same features as the vowel th at precedes it at the w ord end, as in t o r ['kara] ‘all’, ['(fiara] ‘full’, (per ['<|»Êrê] ‘feather, wing’. T he approxim ants w and j show the same behaviour as r, but only in final position: tjw ['tow aj ‘broken’, täw ['táw ä] ‘covering’, tej ['tejc] ‘long’, rD /[3'f3jo] ‘ripe’, n'ÿ['w iji] ‘bow ’, ja ja r i ja 'ja r a ] ‘h o o k ’. A n o th er notorious case o f complex phonem es is M axakali, whose stop conso­ nants may be realized w ith fully vocalic phases and, according to the phonological environm ent, may actualize only the vocalic phase, as in pipkip [pipkiep''] ‘nail’, tja k [tja'i] ‘to cu t’, b a ttik gäj [m baâtïï gâjJ ‘angry to a d ’, tapet [tapEat] ‘paper’ (p is realized as [ëp’’] in the first example, k as p] in the second and third, b as [mb] in the third, and t as [5] in the third and as [at] in the fourth) (see G udschinsky, Popovich and Popovich 1970 for the details o f the allophonic variation o f M axakali conso­ nants, Rodrigues 1981 for the nasalization o f the consonants, and Wetzels 1996 for an autosegm ental interpretation o f the data). 180 Aryan D. Rodrigues C onsonantal clusters in m ost M acro-Jê languages are limited to the com bination o f grave (labial and velar) stops followed by a central or a lateral approxiniant, as in K ipeá/jri' ‘blood’, kru ‘tail’, or in the following K aingáng examples: pro ‘to bite’, m rflr ‘vine’, k r c ‘burrow ’, grcn ‘to dance*. A n exception is Yatê, which presents a large variety o f sequences o f two, three and four consonants - for example: kwtélja ‘rotten’, ehklendonklja ‘car’, kéjkja ‘hearing’, èé'ohifkja ‘admiring*, è tftk ilkja ‘bad odour’, èskwa ‘pulling o u t’, è ftfô k a ‘cleaning’, êtJTàése ‘blanket’, ki'lxif'a ‘nose’, kfælitésa ‘toy’, fytija ‘star’, hie ‘now ’, fi'ánhan ‘alone’, dmàhného ‘beautiful’, tf'lm k a ‘tree’, ujow a ‘louse’, knáfnlka ‘advising’, km áskja ‘asking’, ‘knife’, fn m ja ‘w ind’, ktsàléne ‘message’, fd ja k a ‘snake’, fn á h tfo ‘pepper’ (M eland and M eland 1967). A nother exception is th e Central Jê language Xerénte, which has developed very complex consonantal clusters: psedi ‘it is good’, b d sd i ‘path’, k d ë [knê] ‘stone’, the ‘Ash’, kbure ‘all’, sblka [smlka] ‘ashes’, sha ‘to cut’, srä ‘hill’, zraku ‘to the other river b an k ’, rbcta ‘iron’, krsasari ‘to h u n t’, krdbröhdä [krnmrônunô] ‘to live’, bä to tk tik rc ‘he got thin’ (M attos 1973). 4.3 Tone Two M acro-Jê languages have been described as being tonal, Yatê in the Brazilian northeast and G u ato in the Brazilian southwest. Yatê has two distinctive levels o f tone, but word-finally the tone tends to be middle; according to Meland and M eland (1967) there is m orphophonological ‘tone perturbation’. Examples o f words contrasting only in tone are (tones marked as follows: á high, a middle, à low): ijkja ‘wide’, ijkja ‘fast’; iisija ‘a kind o f pigeon’, usija ‘centipede’; idnika ‘denouncing’, idnika ‘tasting’; è j i ‘to suck’, è f i ‘to scratch’. G uato (Palácio 1986) also has only two distinctive levels o f tone: m ôtl ‘piranha’, n w í/'w o rd ’; m ttl'm a t’, ni/Æi‘p o t’; màbà ‘foot’, màhô ‘tobacco’, màbô ‘dove’. A lthough none o f the bestanalysed Jê languages have been reported to be tonal, it is said that K rahô (Timbira) speakers may com municate at distance by means o f ‘sentence whistlihg’ (Sueli M. de Souza, p. c.). 5 MORPHOLOGY M acro-Jê languages are agglutinating and mildly synthetic; they com bine elements o f head- and o f d ependent marking. 5.1 Inflection for contiguity o f a determiner The m orphology o f m ost M acro-Jê languages is n o t very complex. Word classes are distinguished more by syntactic strategies than by affixation. A pervasive inftec- 181 6 Macro-Jê tional device in M acro-Jê languages is the m arking o f the head o f a nom inal, verbal or postpositional phrase for textual contiguity (C N T) o r non-contiguity (N C N T ) o f the determ iner (or dependent), as in examples (1), (2) and (3) from Panará (D ourado, p. c.), in (4) and (S) from Tim bira (Popjes and Popjes 1986), and in (6) and (7) from Ofayé (G udschinsky 1974: 210, 194). W hen the possessor is stated we find the C N T marker, and w hen it is absent the N C N T marker. (1) soti j-akoa anim al CNT*mouth ‘the m outh o f the anim al’ (2) S3ti j-Ôt3 anim al CNT-tongue ‘the ton gu e o f th e a n im al’ (3) s-oto s-ak oa am â NCNT-tongue NCNT-mouth in ‘Its ton gu e is in its m o u th .’ (4) i te p î.c o 1sg ERG.PAST . tree.fruit j-û?k''3r CNT-buy ‘I bought fruit.’ (5) i te h-fl?k''or Isg ERG.PAST NCNT-buy ‘1 b ough t i t ’ (6) p ik itie n |- e n j i h caim an CNT-heart ‘the heart o f the caim an ’ (7) h- - ■ e n /ih NCNT- heart ‘its h eart’ This device must be a very o ld one in the M acro-Jê stock. I t is also present in the A m azonian languages o f the Tupi family as well in those o f the C arib family and may well be an ancient areal feature. But in some M acro-Jê languages it is only a relic o f a disappearing system, restricted to very few ‘irregular’ words, as in K aingáng (P arana dialect): (8) ?iji - j-3i) j-apS Isg ■CNT-father ■ CNT-field ‘my father’s field’ (9) NCNT-field ‘in the field’ in 182 Aryan D. Rodrigues Table 6.20 Possessive paradigm o f a Yati noun ( ‘fa th e r’) Isg 2sg 3sg Ipl 2pl 3pl 3 REFL 3 IN D E F i-tfe a-tfe e-tfe ja-tfe wa-tfe t^'a-tfe sa-tfe se-tfe ‘my father’ ‘your father' ‘his/her father’ ‘our father’ ‘your father’ ‘their father’ ‘his/her/their own father' ‘somebody’s father, a father’ or in Karaja: (10) habu 1-awo (11)h-avvo m a a CNT-Canoe NCNT-canoe 'th e m an’s canoe’ ‘the canoe’ In Kipeá the prefixes m arking contiguity o f the determ iner are restricted to occurring only after the personal pronouns: hi dz-ebaja ‘my flngernail’, e ds-ebaja ‘your fingernail’. Borôro has reflexes o f them in the allom orphy o f its personm arking-paradigm s (Rodrigues 1993), as in i t-o (I C N T -tooth) ‘my tooth’, 0-0 (N C N T -tooth) ‘his to o th ’. They are not found in G uato and Yatê. 5.2 Inflection for possession Some M acro-Jê languages inflect nouns for the person o f the possessor, for example Yatê, which has a set o f eight personal prefixes, as shown in table 6.20. However, m any other languages use clitidzed pronouns instead o f prefixes for marking per­ sonal possession, e.g. K atngáng ?iji rcrjre ‘my brother’, B orôro p a mana ‘our (incl.) elder brother’, K arajá wa m airehc ‘my m achete’. Several languages have forms inflected with personal prefixes only for some persons, m ost frequently for 3rd reflexive and for 1st inclusive, as K ipeá di-padzu ‘his/her own father’, ku-padm a ‘ou r (incl.) father’ (a is a pluralizer), d-amhe ‘his/lier own paym ent’, k-am be a ‘our (in d .) paym ent’, but e J-ambe a (you CN T-paym ent PL) ‘your (pi.) paym ent’. Ttm bira has no reflexive prefix and has a prefix only for the 1st plural inclusive:p a ­ nto ‘o u r (incl.) eyes’, but ; 0-to (I CNT-eye) ‘my eyes’, a ehto (you CNT-eye) ‘your (sg.) eyes’, rop et-to (dog CNT-eye) ‘the eyes o f the dog’. K arajá, on the other hand, only has a prefix for the reflexive 3rd person: cfa-9c ‘his/her/their own m other’. Both Kipeá and Tim bira, and likewise K arajá and M axakali, m ake reference to a (non­ 6 M acro-Jê 183 reflexive) third person possessor only by means o f the m arkers of contiguity or non­ contiguity, e.g. K ipeá s-anrbe (N CN T -paym ent) ‘his/her paym ent’, T im bira i-nto (N C N T-eye) ‘his/her/their eyes’. G uatô has prefixes for all the persons it distin­ guishes, but for the 1st sg. it also has a suffix that occurs simultaneously with the corresponding prefix: a -tim ~ ru ‘my son’, but e-tára ‘his son’, b i-tám ‘their son’. (The alienable/inalienable distinction is discussed in §6.3.) 5.3 Number Plurality o f the noun is not m orphologically expressed in the Jê family, b u t is diversely manifested in other families o f the M acro-Jê stock. Bororo, for instance, has several ways o f pluralizing nouns. O ne o f these applies mainly to nouns refer­ ring to anim als and consists in adding the suffix -e to the nouns, often with the drop­ ping o f the final vowel o f the stem, e.g. djom o ‘o tter’, pi. djom oe; apaga ‘small an teater’, pi. a p 3 g x ; inert ‘sting ray’, pi. m e r k \ apada, ‘tucanuçu, a kind o f toucan’, pL apsde, ka ra ‘fish’, pi. kare, pa i ‘howler monkey’, pi. pae; orari ‘painted catfish’, pi. omre. Some kinship term s take the suffix -mage, e.g. mana ‘elder bro th e r’, pi. manainage; wagedo ‘son-in-law ’, pi. wagedomage; ta rfje ‘your (pi.) m o th er’, ta d y m a g e ‘your (pi.) m others’. N om inalizations ending i n -wi, as well as those ending in -epa, when referring to anim ates, take the suffix -ge in the plural: w sewi ‘one from here’, waewige pi. ‘the ones from here’; uiurewi ‘the one who w ent’, uturewige ‘the ones who w ent’; d jom bokuntepa ‘doctor, nurse’, pi. dsombokuruepage. T he m ost universal pluralizing suffix is -doge: bsiga ‘bow ’, pi. bsigadoge; arigao ‘dog’, pi. artgaodoge; kaiamo ‘X avánte Indian’, pi. kaiamodoge; paga ‘creek’, pi. pagadoge. B ororo also has a singulative suffix -di, which is added to nouns with a collective m eaning for expressing one individual: ime ‘men’, sg. imedi ‘a m an’; ipare ‘young m en’, iparedi'a.yo%mg m an’; arem e‘w omen’, sg. arerff‘a woman’; l>a/-ae‘w hite men’, sg. baraedi ‘one white m a n ’; kaiamodoge ‘the X avánte Indians’, kaiamodogedi ‘one X avánte’. Some languages, although they have no m arker o f pluralization on nouns, have plural pronouns or personal prefixes on the verb for agreement with plural subjects and som e o f them also for plural objects. In the Jê family one such language is K aingáng (P araná dialect), which has personal pronouns for 3rd person plural mas­ culine and feminine: ?ag ‘they (m asc.)’, (pari ‘they (fem .)'. K aingáng also has plural verbs for agreeing with plural S or O, even though the nouns are not m arked for number. In this language there are several m orphological devices for m arking plu­ rality on verbs: prefixation, infixation, reduplication, a com bination o f two o f these with or w ithout ablaut, and also suppletion. See table 6.21 for som e examples from the P aran á dialect. 184 Aryan D. Rodrigues Table 6.21 Number-marking in K aingm g (Paraná dialect) (from Cavalcanie 1987) SG (S or O) PL (S or O) gloss kigjam kömköm ‘to ‘to ‘to ‘to ‘to kogungun ‘to w ith er’ m fijim raji nlr)ni ‘to ‘to ‘to ‘to 1 p ra kipra 2 «^änän <jíägnän 3 4 5 6 7 8 kigcji} # kajäm köm koQun mraii nT 9 we 10 r l wigwe Oe bite’ use almost all’ weave’ pay, to buy' dig’ break’ sit’ see’ go in’ In table 6.21 the following devices are exemplified: (1) prefixation o f ki-, (2) infix­ ation o f -p-, (3) prefixatioii and infixation, (4) inflxation and ablaut (5) redu­ plication o f a monosyllabic verb, (6) reduplication o f a disyllabic verb, (7) reduplication and ablaut, (8) reduplication and infixation, (9) reduplication, infixation and ablaut (10) suppletion. A lthough m ost K aingáng verbs are invariable, there are ab o u t 150 that have a plural form in one o f the patterns illus­ trated in table 6.21. Even a verb that is probably a loan from Portuguese,/»a/a ‘to prom enade’ (Portuguesepassear [pasja]), pluralizes in one o f these patterns, namely p i^ fa (infixation and ablaut). Xavánte (C entral Jé) also expresses the num ber distinction o f nom inal arguments (S and O) by means o f verbal agreement, b u t it systematically distinguishes three numbers, singular, dual and plural. A bout a score o f verbs have thi«e diiierent stems, one for each num ber o f S or O, such as sg. wara, du. a tfa b rö (i), pi. tfitja?re ‘to ru n ’; sg. n»i, du. a jm a tfitfi, pi. ajhutu ‘to arrive’; sg. w i'frij, du. pari, pi. tfibrö ‘to kill’; sg. b e(i), du. wahdzu(ri), pi. tabra ‘to throvi^’ (some stems have two allom orphs, one o f them short, the other long, vsfith the extra syllable put in parenthe­ ses in the examples above). In m ost verbs, however, the num ber distinction is made by means o f a complex interplay of particles in the verb phrase. 5.4 Noun classification The category o f n oun classes is not typical o f M acro-Je; these are only reported for the languages o f the K ariri family. There are twelve prefixes that are attached to quantifiers and descriptive adjectives o f dim ension, consistency and colour, accord- 185 6 Macro-Jê Table 6.22 Class agreement in quantifying and qualifying phrases in Kipeá ‘long’ ‘bright’ ‘one’ ‘m any’ spherical objects ii/e ‘sun, day’ conical objects udza ‘Icnife’ sinuous objects wo ‘snake’ convex objects pika ‘stool’ u/e kro-Ji uje kro-dzodzo kro-bihe uJe kro-Jo uje udza ja-/i udza ja-dzodzo ja-bihe udza ja-jo udza wo ho-/i wo ho-dzodzo ho-bihe wo ho-jo wo pika b e-|i pika be-dzodzo be-bihe pika be-jo pika ing to the shape o f the n o u n ’s referents. T he nouns themselves have n o overt m arking for class. According to M araiani (1699), the twelve prefixes in K ipeá arc the following: be- for hills, dishes, stools, foreheads, etc.; kro- for birds, stones, stars and ro u n d objects, such as beads, fruit, eyes, etc.; kru- for liquids and rivers; eprufor clusters and bunches; he- for sticks, legs and w ooden objects; ho-, hoi- for ropes, vines, threads, snakes; ja - for iron objects, bones and pointed things; mu-, muifor edible roots; nu- for holes, wells, m ouths, fields, valleys, fenced spaces; ro- for clothes, fabric and furs; woro- for roads, conversations, speeches, stories; bufor houses, arrows, containers, corn-cobs and living beings (except birds), as well as for any other nouns n o t specified for the other prefixes. T h e classifier prefix occurs with adjectives (which follow the head noun) and with quantifiers (which precede it), as exemplified in table 6.22. T here is a masculine/feminine gender distinction in 3rd person pronouns in the Southern Jê languages (i.e. K aingáng and X okléng). R ikbaktsá (W iesemann 1986; 361) and Y atê(L apenda 1968:91) have a gender distinction in all three persons, and the latter language marks masculine and feminine genders on-adjectives (see §6.6), as well as on dem onstratives (M eland 1968: 18, 19b). 5.5 Agreement marking on the ferb M any M acro-Jê languages have no agreement m arkers on the verb. Some o f them, however, do m ark the subject (most often, if it is third person) in agreem ent with the corresponding nom inal phrase in the clause. K aingáng (Jê) exemplifies the com ­ plete absence of markers in (12) and (13), whereas K ipeá (K ariri) shows 3rd person an d 1st inclusive agreement in (14) and (15): (12) ?iji rco re wï je Isg brother S stand ‘My brother is standing.’ 186 Aryan D. Rodrigues (13) mît) WÏ kàijrcji tàji tî ja g u ar A tayra kill a sp e c t ‘T he jaguar killed a tayra ( r barbara).’ (14) more si-te soon karai 3-corne white.m an T h e white m an comes soon.’ (15) ku-te di Ipl.incl-com e fu t ‘We (incl.) will come.’ Some languages use a m arker on the verb only if the subject nom inal phrase does not immediately precede an intransitive verb, for example M axakali in (16) and ( 17) (Pereira 1992: 83): (16) pit Jap t/ip e p (17) duck arrive ? î-tjlp e p S-arrive ‘The duck arrives.’ p itja p duck ‘T he duck arrives.’ The same marking occurs in M axakali for the direct object o f transitive verbs, as in (18) and (19) (Pereira 1992: 88), and is fundam entally the same device as that shown in §5.1 above as the marking o f non-contiguity versus contiguity o f a deter­ miner. (18) tik te man erg ? ï-tj’it k ip i? ik 3-sharpen axe ‘The man sharpens the axe.’ (19) tik te m an ERG k ip ik t jit axe sharpen ‘T he m an sharpens the axe.’ A n example with transitive verbs in T im bira (Popjes and Popjes 1986; 163) is (20): (20) jako te po pupun ne i?-kuran Jaco ERG.PAST deer see and 3-kill ‘Jacô saw the deer and killed it.’ Yatê, K arajá and G uato have more com plex verb m orphologies and manifest verb agreement n o t only with the subject, b u t also with the object. G uato is partic­ ularly complex; it has n o t only personal prefixes like other M acro-Jê languages, but also persona! suffixes. M oreover, G uato verb m orphology is a com pound o f ‘three different inflectional patterns: one for first person singular, a neutralization of 187 6 Macro-Jê Table 6.23 Personal affixes in the Gtiato verb Isg -jo -jo -jo 2 s g /p l g^a- -h e -h e 3sg Idu.incl 1non-sg 3pl E - ~ i- -0 gadsa- gad ja be- be- -0 gedse- 0- systems; an o th er for second and third persons singular, which follows an ergative/absolutive system; an d still a third for plural, which behaves according to a nominative/accusative system ’ (Palácio 1986: 369). Palácio’s chart 6 is reproduced here as table 6,23. (N ote that Inon-singular covers I dual exclusive, 1 plural inclu­ sive and 1 plural exclusive.) Examples o f this person m arking system in a transitive verb are the following: (21) na-g'*a-bagâki-(j)o (22) n(a)-e-bagáki-he iN D ic-2A -h it-lsgO iNDiC -3sgA -hit-20 ‘You hit m e.’ ‘H e hits yo u .’ It is probable that the object prefixes g s- and d je - are the result o f a co n trac­ tion o f the expected sequences c-ga- and e-dja- (cf. Palácio 1984: 68; 1986: 370). T h ird person prefixes are obligatory even when the corresponding nom inal phrases are fully expressed in the clause, as in (23); therefore they are truly agree­ m en t m arkers. (23) n(a)-e-bagáki -0 go-dé g -ô h ad ja iNDiC-3sgA-hit-3sgO D E T -m a n D E T -w o m an ‘The m an hits the w o m a n .’ 6 SY N TA X 6.1 Constituent order in declarative sentences T h e languages o f the Jê family have SV and AOV as the m ost frequent constituent orders in their declarative clauses. Examples (24) and (25) from the C anela dialect o f T im bira illustrate these orders for one-argum ent and tw o-argum ent verbs respec­ tively. 188 Aryon D. Rodrigues (24) kapi JA pir Capi climb ‘Capi climbed.’ (25) i Isg k'’ra te rap ti pupun so n ERG.PAST ja g u a r see ‘My s o n saw a j a g u a r . ’ The same pattern also prevails in M axakali, K renák, Yatê, K arajá, Borôro and Rikbaktsá. D ata on K am akâ, Puri and Ofayé are so limited that no statem ent on their clause syntax is possible. This is an irremediable situation for the languages o f the first two families, which are now dead, but it is not so for Ofayé, from whose last survivors (about a score) linguistic data could still be recorded. K ariri an d G uatô represent strong departures from the pattern illustrated above. In both language families clauses are norm ally verb-initial. In K ipeá o f the K ariri family and in G u ato, clauses with one-argum ent verbs show the same order VS, but those w ith tw o-argum ent verbs differ in the relative position o f the arguments. In K ipeá (see examples 26 and 27) we get VOA (where the A is m arked by the ergative preposition no). In G uato we find VAO, as shown in (28) and (29): (26) si-te karai 3-come white.man ‘The white man comes.’ (27) si-pa kradzo no karai 3-kill cow ERG white.man ‘The white man kills the cow.’ (28) na-kini g-eti iNDic-sIeep DET-boy 'T h e boy sleeps.’ (2 9 ) m a-e-ro g-épagu g-éki iMPERFV-3-eat DET-jaguar DET-rabbit ‘The jaguar ate the rabbit.’ 6.2 Adpositional phrases All M acro-Jê languages except K ariri and G uatô have postpositions and not pre­ positions, as may be seen in the following examples. 189 6 Macro-Jê Kayapé (30) pur kâm ba a ni pijo re garden in Isg you to fruit pick ‘I pick up fruits for you in the garden.’ M axakaii (31) piticnag bird ?ï-cip 3-be m ïin tree ti on ‘The bird is on the tree.’ T he K ariri languages an d G uato, which have verb-initial clauses, have preposi­ tions and n o t postpositions: Kipeá (32) Guato pa kri d o kill udza PERFV w i t h (33) knife jo in ‘It was killed w i t h a k n if e .’ go-dsékî D E T -river ‘in t h e r i v e r ’ If an adposition in K ipeá relates to a noun, it precedes it as in (32); if it relates to a prorîoun, this may be prefixed to the adposition. (34) i-jia kfi ku-bo 3-die PERFV Ipl.incl-for ‘He died for us.’ (35) di give kri do PERFVto i-de 3-m other ‘He has given to his m other.’ (36) di g iv e kri PERFV i-djoho 3-to ‘He has given to her.’ (37) n io m ake k ri no k arai perfv by w h ite .m a n ‘It w a s m a d e b y t h e w h ite m a n . ’ (38) n io m ake k ri perfv e you na by ‘It w a s m a d e b y y o u . ’ As s e e n in t h e l a s t four e x a m p le s , s o m e a d p o s i t i o n s h a v e d iffe re n t a l l o m o r p h s when they follow a pronoun o r pronom inal prefix. 190 Aryon D. Rodrigues 6.3 Genitive plirases With the exception o f K ariri and G uato, the M acro-Jê languages have the posses­ sor preceding the possessed noun (with no overt m arker o f possession): Timbira (39) M axakali kapi k^ra Capi child (40)?i tit Jik p o t 3 ‘C ap i’s child’ m other grave ‘his m other’s grave’ G uato and K ariri are again the languages that behave differently, having an inverted order for their genitive phrases. In these two M acro-Jê families the heads o f noun phrases precede their determ iners (or dependents). If the relation between head and determ iner is one o f possession, the head is inflected for 3rd person (or non-contiguous) possession as shown in the following examples from Guato: (41) i-pána g -á k ^ o 3-tail ‘th e D E T -m o n k e y tail o f t h e m o n k e y ’ ( lit. ‘its ta il t h e m o n k e y ’) (42) e-tá ra a-dùnihi-ru 3-son 1-brother-1 ‘the son o f my brother’ (lit. ‘his son my brother’) N ote that Kipeá has the reverse order in com pounds: (43) s-era ware 3-house priest (44) boro-po arm-eye ‘the house of the priest’ ‘elbow’ Many languages o f the M acro-Jê stock distinguish alienable from inalienable possession. In general, alienable possession is expressed by m eans o f an inalienable generic noun, m eaning ‘thing’ or ‘belongings’ or ‘possession’. In some languages there are two or more generic nouns, distinguishing classes o f possessable things. The Jê languages Tim bira, Kayapô and Panará have only one generic form, -Ö, as in examples (45) and (46) from the Canela dialect o f Tim bira. (45) kapi j-o Capi CNT-POSSESSION field ‘C api’s field’ pur 191 6 M acro-Jê (46) h-ô wapo NCNT-POSSESSION kn ife ‘his knife’ B ororo has two generic nouns for expressing ah'enabie possession, -aku ‘dom es­ ticated anim al’ and ~o ‘thing’, as illustrated by examples (47) and (48). (47) 0-aku kogariga 3 - dom estic .a n im a l chicken ‘his chicken’ (lit. ‘his-dom esticated-anim al chicken’) (48) 0-0 tori 3-thing stone ‘his stone’ (lit. ‘his-thing stone’) In Kipea, of the Kariri family (see §5.4), there are twelve generic nouns which refer mainly to as many different classes of alienably possessable objects, including animals, according to the way in which they were acquired. The generic noun is pos­ sessed in the same way as any other noun and is linked to the specific noun of the pos­ sessed object by means of the preposition do. Three generic nouns are illustrated here: (49) hi er)ki Isg D O M E ST IC . A N IM A L do kradzo of COW ‘my (raised) cow ’ (50) dz-uapru do 1-game m urawo of wild.pig ‘my (hunted) wild pig’ (51) dz-uito do 1-find of udza knife ‘m y (found) k n ife’ A ccording to the way in which its referent was acquired, a noun may occur with different generic nouns, as in examples (52)-(54). (52) hi Isg egki D O M E S T IC .A N IM A L do sabuka of chicken ‘my (raised) chicken’ (53) dz-ukisi 1-portion do of sabuka chicken ‘my chicken (received in a partition o f goods)’ 192 A rfo n D. Rodrigues (54) dz-uba do sabuka 1-present of chicken ‘my chicken (received as a present)’ With the exception o f egki ‘dom esticated anim al’, all possessable generic names o f K ipeá begin w ith this «- is probably a morphem e cognate with Bororo o ‘dom esticated anim al’, Tim bira -o ‘belongings’, M axakali ?<5 ~ ?öi; ~ jo g ‘belong­ ings’, and probably also K arajá ôhô ‘domesticated anim al’. 6 .4 Demonstrative phrases D em onstratives follow the head noun in some languages and precede it in others. Canela (o f the Jê family) and K renák illustrate the first situation, as in example (55), whereas K ipeá o f the K ariri family, B ororo and G uato show the reverse, as in example (56). Canela (55) Kipeá ra p ita dog this (56) ‘this dog’ 6.5 igi era this house ‘this house’ INumeral phrases In the K ariri and G u atô families numerals precede the head nouo, whereas in the o th er language families they norm ally follow it, as in examples (57)-(61). In the K ariri languages the numerals carry a classifying prefix in agreement with the class o f the phrase head (see §5.4 above), as in (57), but they take no prefix if the head noun is unclassified, as in (58). In Yatê the numeral for ‘one’ forms a com pound with the noun, and this com pound is marked for gender as in (60). Kipeá (57) b u -b ih e eru m u C L -o n e squash (58) (59) tupa one god ‘one god’ ‘o n e s q u a s h ’ Guatô bihe Yatê tjù m u i-tana three 3-root ‘three roots’ (60) tjá ji-ft'* a - n e w o m a n -o n e -F E M ‘o n e w o m a n ’ 193 6 M acro-Jê Borôro (61) u-re kogariga-re m it3-d» m aki in-ai 3sg-NEUT chicken-NEUT one-NMLZR give 1-to ‘He gave me on e chick en .’ 6.6 Adjectival phrases In M acro-Jê languages the adjective follows the noun. This is true of the languages with constituent order AOV as well as o f those w ith orders VOA and VAO. Only in the YaEê family is there gender agreem ent o f the adjective with the noun, see (62) and (63). In the Kariri family the small set o f adjectives o f dim ension, colour, and qu antity shows class agreem ent w ith the head nouns, as in (64)-(66) from K ipeá (see §5.4 above). In G uato the adjective takes the sam e m arker for determ ination as its head noun, as in (67). (62) (63) itsk a kaka -0 man good-MASC ‘a good m an’ (64) kro-jo CL-inany u je (65) sun udza knife ja-Ji CL-long ‘a long knife’ udza ja-n e knife CL-sharp (67) g-ô d â DET-basket ‘a sharp knife’ 6.7 kaka-ne good-FEM *a go o d w om an’ ‘many days’ (66) t/áji w om an g-itavi DET-heavy ‘the heavy basket’ Ergativity There are some techniques for linking S and O that are universal and do n o t indi­ cate th at a language in which they occur has ergative properties. For instance, if a verb m arks the num ber o f a core argum ent this will always be S in an intransitive and O in a transitive clause, as reported for K aingáng and X avánte in §5.3. There is ergativity in Jê languages where an A N P can, at least in some contexts, take a preposition that appears to have ergative function, as in the following exam ple from Timbira: (68) i ts Isg ra p ERG + PAST dog kak''w ln beat ‘I (recently) beat the dog.’ 194 Aryan D. Rodrigues In M axakali any transitive sentence has its subject m arked by the ergative post­ position te, as in examples (69) and (70). (69) tik te man k ip ik t/it erg axe sharpen ‘T he m an sharpens the axe.’ (70) k ip ik te axe ERG mTmka?ok wood h ard mep cut 'T he axe cuts hard w ood.’ Kipeá (K ariri family) has VS, VOA constituent order with the A argum ent m arked by an ergative preposition no: (71) peho i-wo flow 3-way dzu m o water imera in field ‘The river flowed on the fields.’ (72) so bite hietsâ Isg no wo ERG snake 'A snake bit me.’ This now-extinct language was described by Father M am iani (1699) who struggled to fit it into a classical gram matical framework. H e characterized all transitive verbs as inherently ‘passive’ (with no corresponding active form). But, besides the semantics o f the verbs, there is a m orphosyntactic device that reveals the ergative character o f M am iani’s ‘passive’ verbs. Every Kipeá verb may derive a nom inalization o f its absoliitive (S o r O) argum ent (th e ‘subject’ according to M am iani), by m eans o f the absolutive prefix di- ~ d- and the nominalizing suffix -ri, as in the following examples: (73) (75) di-te-ri (74) di-pa-ri ABS-COme-NMUR ABS-kill-NMLZR ‘the o n e who comes’ ‘the one th a t was killed’ udza di-di-ri no ware knife ABS-give-NMLZR by priest ‘the k nife that was given by the priest’ Only M am iani’s ‘passive’ verbs, however, have a second nom inalization, referring to the agent o f the action, therefore the A argum ent. This nom inalization takes the ergative prefix du- and the same nom inalizing suffix -n, as in the following exam ­ ples: 195 6 Macro-Jê (76) w are aïc du-di-ri uu-ui-ii udza uuLü priest ERO-give-NMLZR knife ‘the priest w h o gave the k n ife’ (77) koho du-nio-ri arâkie that ERG-make-NMLZR sky ‘he was who m ade the sky’ T here is further discussion o f the ergative c haracter o f K ipeá in Larsen (1984), who also indicates an S/O syntactic pivot for subordinate clauses. (Larsen’s discus­ sion also suggests a class o f ‘extended intransitive’ verbs in Kipeá.) 6.8 Valency-changing processes 6.8.1 Reflexives and reciprocals Both m orphological and syntactic devices are found in the Macro-Jê languages for reflexives and, in some cases, also for reciprocals (n o t every language clearly distin­ guishes these two processes). M orphological devices may be seen as the derivation o f an intransitive verb from a transitive one, whereas syntactic devices involve a reflexive o r reciprocal use o f a generic pronoun (and m aintain transitivity). This Is the case with languages o f the Jê family, such as T im bira (Canela dialect) with the generic reflexive object amji and the reciprocal ajpën: (78) i te Isg PAST amji pitAf REFL defend ‘I defended myself.’ (79) ku-te am ji m a 3-erO-Past r e f l for pï jak**ep wood cut ‘H e c u t wood for him se lf’ (80) . . jako me kapi aipen mS pT jak^’ep Jaco and Capi for wood cut recip ■Jac6 and Capi cut w ood for each other.’ Xavánte, a Central Jê language, has a reflexive prefix tsi-: ?mada ‘to look a t’, tsiîm ada ‘to look at oneself, to take care of oneself’. Analogously, Yatê has a prefix sa< 6.8.2 ‘to cu t’, sa-t^uli ‘to cut oneself’. Causativization Some languages in the M acro-Jê stock have m orphological devices for the form a­ tion o f causative verbs, whereas other languages have only syntactic m eans of 196 Aryan D. Rodrigues causativization, involving verbs such as ‘to m ake’ or ‘to cause’. Morphological devices consist essentially in deriving a transitive stem from an intransitive one. Among the languages that have m orphological derivation o f causatives, some can causativize only intransitive verbs while in others they can also derive transitive verbal stems from nominals. The Jê language T im bira causativizes syntactically by m eans o f the verb tohon ‘to make’, to which the causativized verb is linked by the particle m (cf Popjes and Popjes 1986; 142-3): (81) kapi Capi te i ERG.PAST jo t ISG Sleep na i to na Isg make ‘Capi m ade m e sleep.’ (82) i te Isg ERG.PAST i pro Isg j-Apen wfife na CNT-work na e-tan NCNT-make ‘I m ade my wife work,’ A n interesting example o f a syntactic causative involves Borôro. In this language there are two causative morphemes, d a ‘to cause’ (simple causative) and g 3 ‘to cause to begin’ (inceptive causative): (83) a re b oe e ridiw a-da you NEUT people they know-CAUS ‘You caused people to know.’ (84) a-re boe e rid iw a -g 3 you-NEUT people they know-iNCEFnvE.caus ‘You caused people to begin to know.’ d3 and g3 are n o t suffixes as examples (83) and (84) might suggest. They are clitics th a t occur after intransitive verbs (as above), but also after the subject o f causitivized transitive verbs, as in (85); (85) i re a-d3 bola barigu iraedi ae Isg NEUT you-CAUS ball throw m an to ‘I caused y ou to throw the ball to the m a n .’ Looking now a t m orphological devices, some languages use prefixation and others exhibit suffixation. K ipeá (Karirí) has a causative prefix m i- added to intran­ sitive verbs; pere ‘to go out’, mi-pere ‘to cause to go o u t’; te ‘to com e’, m i-te ‘to bring’. K aingáng (Jê) has a suffix -n which occurs with both intransitive and tra n ­ sitive verbs: f i ‘to be old’, fi-n ‘to cause to be old’; rä ‘to be ripe’, rä-n ‘to make ripe’; we ‘to see’, we-n ‘to show ’. 197 6 M acro-Jê 6 .9 S w itch -referen ce Some M acro-Jê languages distinguish coordinate clauses with the same subject from those with different subjects. In K ipeá this distinction is m ade by using different prefixes on the verb o f the coordinated clause, When the subject o f this clause has the same reference as the subject o f the first one, it is m arked by di-ldwhereas the non-contiguity m arker si-ls-li- is used for subjects differing from that o f the first clause (M am iani 1698: 61,49): (86) doro then si-te 3-come bo aràkje m o from heaven in tsöho do di-nja nodehe people for SS-die also rada earth do di-wi do for SS-become to ‘Then he came from th e heavens to the earth in order to becom e people and also to die.’ (87) mo A dam si-pei-kri A dam NCNT-take.out-PERFV s-unu-te in NCNT-sleep-NMLZR i-mesu NCKr-rib no tu p â i-bo bo si-nio i-bujêwoho Eva ERG G od NCNT-from for NCNT-make NCNT-body Eve ‘D uring A dam ’s sleep G od took out bis rib for making Eve’s body.’ A n o th er switch-reference strategy is found in som e Jê languages an d in M axakali. Different connective particles o r conjunctions distinguish clauses with different subjects from those with the sam e subject, as shown in examples (88) and (89) for the Canela dialect o f T im bira (Popjes and Popjes 1986: 147) as well as in (90) an d (91) for M axakali (Popovich 1986: 355). (88) kapi Capi te ERG.PAST po kuran ns ke deer kill and 3.SS ha fut ku-k’’u 3-eat ‘Capi killed a deer and will eat it.’ (89) kapi apu ajkahu ma hitsi apu no Capi CONT run and.O S his.wife cont lie.down ne tjor and.SS sleep ‘Capi is running and his wife is lying dow n and sleeping,’ (90) ?î-m ôg t+ 3-go and.SS ?-nîn 3-come ‘He went and returned.’ 198 Aryan D. Rodrigues (91) ?ï-môf) ha 3-go and.D S ?-nîn 3-come ‘H6j went and hCj returned.’ Detailed studies o f M acro-Jê languages are rather scarce, so th a t no fully system­ atic coverage o f their convergent and divergent m orphological and syntactic prop­ erties can be undertaken. The above survey o f selected aspects o f some o f the better-know n languages may give an idea o f the kind o f structural phenom ena m anifested in this huge language group and vi^ill hopefully stim ulate research in the languages that are still alive, m ost o f which have been - an d still are - exposed to strong adverse sociocultural pressures a n d are seriously endangered. 7 ABOUT THE CONSISTENCY OF MACRO-JË AS A GENETIC GROUP 7.1 A brief appraisal o f the grammatical affinity T he gram m atical survey in §5 and §6 above, even though rather superficial, shows th at there are striking typological similarities am ong all the twelve branches o f the M acro-Jê stock. It shows also that two branches, K ariri and G u atô , neatly diverge in syntactic typology from the other ten branches; whereas the latter are predom i­ nantly verb-final and postpositional, with genitive-noun phrases, Kariri and G uatô are both verb-initial and prepositional, with noun-genitive phrases. It is rem arkable that Kariri and G uatô are not geographical neighbours but, on the contrary, Kariri lies in the northeast o f the M acro-Jê area, on the lower S io Francisco river, whereas G u atô is found in the extreme southwest o f the area, on the Paraguay river. As yet it is to o early to decide whether these constitute cases o f m arginal conservatism in a large area o f linguistic spread or whether K ariri and G uatô may be viewed as inde­ pendent examples o f the same sort o f typological rearrangem ent. To judge from internal evidence in Kariri, the rearrangem ent hypothesis is m ore likely for this family, since its languages show residual traces o f the typological alignment prevail­ ing in other M acro-Jê languages (e. g., postpositional use o f the prepositions when their objects are pronom inalized, genitive-noun order in com pound words, see §6.2 an d §6.3 above). 7.2 Phonological equations The Macro-Jê stock o f languages has so far been a mere hypothesis, or rather a bundle o f hypotheses, about a possible remote common origin for the languages involved. The spatial distribution o f its members is very large and the lexical 199 6 M acro-Jê differentiation am ong them is too deep for delivering an easy-to-discover series of possible cognate words. A bout ten years ago the present author stated that the words for ‘fo o t’ were the only series of probable cognates he had found across all the twelve families o f M acro-Jê (Rodrigues 1986: 50ff.). A lthough very few new lexical contri­ butions have appeared in the meantime (only two small dictionaries of Xavánte; Hall, M cLeod and Mitchell 1987 and Lachnitt 1987), a closer scrutiny of the available data has delivered more lexical correspondences and more phonological equations across the putative Macro-Jê stock, A sample o f these correspondences and equations is pre­ sented here in order to enable the reader to have an idea o f the likelihood of the genetic relationship of the languages involved. Table 6.24 displays the phonological correspondences and the num bers for the examples presented following the table. T ab le 6.24 Phonological correspondences among the M acro-Jê language fam ilies proto- Exemplifying MJ lexical items Language families 1 p m/p m k/k'’ kr or r 3 Ö a a ä II w w tn k her gr r ,0 Ô a a III IV P m/p m k/tj P P m m k/tJ k/t/ kr or or r Pt kt 0 Ô a a 0 a 0 a olo ë 1 V à VII VIÍI IX f w P b p/w m b b k k" r kit kir kr k"1 0 JP r r/d 1 o o o D Ô u/o 0 0 e a a e a e/a o/i e e a a/a VI b X b w m k XI XII 1.2 ,1 2 , 34 *P 3, 5 ,1 6 ,1 7 ,2 1 ,2 6 *mp 37, 35,36 *m 7, 1 4,19,31,33 k k *k h ar *kr ■ 15,25 k ir k" k ar *t)r 8 , 10,30 r 4 ,9 , 2 0 ,2 2 .2 6 .2 7 , 29 *r r r 1 ,4 ,1 1 ,3 4 0 *0 0 u/o u 7, 13,28, 32 *0 2 1 ,2 3 ,3 8 ,3 9 £ a a *a i *0 1, 12 3 i 24, 36 e e P P p arm ; la pa, Ib pa, Ic pa. Id pc, IV po, VI bo, VII f e ‘arm pit’, IX p c, XI p6 2 arrow: Ic po. Id puj}, II waj, U\ poj, V pan, pun, VI btij-ku, V lll wihi, X bëiga ‘h ow ’< *hëj-ika ‘*arrow -bow ’ 3 axe: Ib krÄ -m sji, Id 4 back: VI worn, VIII boro, IX -hor, X p o r i 5 caiman: lb mT-ii, II wcj-e. III m a?Sji, mai, X wai m eg. III ki-p ik, IV kra-pok, V kra-mar) 6 come: Ib fe, Ic dê, Id « ‘to go’, II nl. III nîn, IV ne, V «i, VI te, VII tfi 7 drink: lb k ö m , là ô . III lfo?op, tfo m , IV tjop, V some, VII kf'o, V III Ö, X ku, XI á k(, X II ku 8 dry: Ib gra, Ic ?f€, VI kra, X kirew ë 9 eat: Id rog ‘to swallow’, VI do, V III ro ‘to eat meat’, IX rö , XI ro 200 Aryan D. Rodrigues 10 egg: Ib gre, Ic ?re. Id kre, II sa-kre, III kir, V III di, IX kite, X I X II kare 11 12 eye: Ib no, to, Ic to, II to, VII t"o, V III rue foot: la peno, Ib par, Ic para. Id pen, II wade. III pata, IV po, VI bi, b ir i; y i lfe -h e ,fe t-, VIII wa, IX par, X hire, X I àbb, X II p k i 13 give: Ib g o , Ic tsô. III horn, IV -Op, hum, VII ko, V III ô, IX no 14 hair: la fe , Ib ki, làî, II ke. III tje, IV ke, V ke, tfe, XI k i 15 head; la krâ-, Ib krÂ, Ic ?ra. Id krT, II hero, xaro, IV krcn, V III ra, IX 16 hear: Ib m a, Ic wa-pa. Id m ëg, IV paw, IX paj, X m ea-ridi 17 honey: Ib m eji. Id m Sg, III par], IV par), IX pik , X I pagua kite, X II -hara- 18 I: Ib i, Ic ?i. Id ?iji, III ?ik, VI hi, VII i, X i, X II ik 19 in :I d /ti, V ll/te .V I IU ^ X g i- lO leaf; la arâ-tife, II ere, VI ærâ, X ari ‘leaf’, aro ‘small leaves’ 21 liver: Ib ma, Ic pa. Id tà-m ë. III ta-ma-gaJ, V III ha, IX pa, XI pe 22 long; Ib ri. Id rira ‘to push’, II roro, IV ron, V III rehe, IX ra, X raire, X II (ze-ze) 23 maize; V m aki, m aeki, VI masiki/masitfi, VII m á ltfí, V III maki, XII na tfi 24 m other; la na, Ib nÄ, Ic dâ. Id ni. III të, te, ta, V tä, VI de, VII sa, VIII na-di, X tfe, XII Je 25 mount; Id krî, II heri, kece, III jiî-ktij 26 night: le ta r a , II hwera, V meri, IX we:r 27 penis/male: la rë, VI ræ ‘maie’ 28 possession; Ib ô, III 5, ôr),jôri, VI u-, X o 29 short: Id n ir, VII lulija, X ro-gu 30 sing/dance; Ib grc, le ?rë, ‘dance’. Id ^ren ‘dance’, II gre. III ktej, IV 31 skin/bark; Ib kA, le hg, II ka, III tfaj, kaj, IV kat, VII k ^ à -tfa , IX ha, Ori, V ogre, VII k^læ -tf'a , V III 9c, IX kirih, XII kari ‘to dance’ X -k a 32 sleep: la rjô, Ib r)ôr (nô, n ô r ‘to lie’), le jô d ô (dô ‘to lie’), Id nör, II jundû, hondô, III jiô n , hôn, ?ôn, VI unit, V III rô, IX no, noro ‘to sit’, X nudu, X II uni 33 stone|: Ib kên, k''en, le ?ëdê, II kéa, VI kro, IX ketch 34 stone 2: Id po, VII fàw a, X I àfô ‘ground’ 35 sun; Ib m it, le bodo, VI bati ‘star’, VII fe tfa , X meri 36 to: Ib m a, le bâ. Id m a, VII ma, V III b3 37 walk/go; Ib m ô/m ôr, le bôlbôri. Id m û ‘go pl.’, II mat], m an, III m ô g ‘to g o ’, IV mûr], m û, V mu, mon ‘to go’, VI wo, X meru 6 Macro-Jê 201 38 we (incl.); I b pa-, le wa-, II pa-fiike, V III wa, X pa 39 you (sg.); Ib a-, l e ?a. Id ?â, II a, III ?â, IV a, VI e, VII a, V III a, IX e, X a, XII a The proto-phonem es (marked with an asterisk) to the right o f each series of phonologieal correspondences in table 6.24 are very tentative. There rem ain many inconsistencies in the correspondence o f vowels and consonants across the twelve M acro-Jê families, and the relative scarcity o f d ata m akes it very difficult to find examples th a t would substantiate each series o f possible cognate words. A nother difficulty, well known in historical linguistics, com es from the fact that m ost com ­ parable words or m orphem es in these languages are very short ones, consisting in general o f only one o r two syllables and therefore w ithout the phonological redun­ dancy th at makes a genetic connection really plausible. However, there is a reason­ able consistency in a good part o f the series o f correspondences that go across the whole M acro-Jê stock. M oreover, m ost o f the m eanings in these series belong to those sections o f the lexicon (universal concepts) th a t are less exposed to cultural influences an d , therefore, are less likely to be loans from one language to another. These factors m ake us reasonably confident that the M acro-Jê hypothesis - namely the hypothesis o f a com mon genetic origin for the M acro-Jê languages - may be proved correct in the future - if not for all the language families included here, at least for m ost o f them. B IB L IO G R A P H Y Albisetti, C. and Venturelli, J. 1962. Enciclopédia Bororo, vol. I; Vocabulários e etnografia. C am po Grande: Museu Regional Dom Bosco. 1969. Enciclopédia Bororo, vol. II: Lendas e antropônimos. Cam po Grande: M useu Regional Dom Bosco. 1976. Enciclopédia Bororo, vol. Ill, parte I: Textos dos cantos de caça e pesca. Campo Grande: Museu Regional Dom Bosco. Anderson, S. R. 1974. The organization o f phonology. New York: Academic Press, 1976. ‘Nasality and the internal structure of segments’, Language, 52.326-44. Azevedo, G. M. C. 1965. ‘Lingua Kariri: descriçào do dialeto Kipeá’, MA thesis, Universidade de Brasilia. Barbosa, E. A. 1991. ‘Aspectos fonologicos da lingua Yatê’, M A thesis, Universidade de Brasilia. Bernardo de Nantes, R. P., Fr. 1709. Katecismo indico da lingua Kariris. Lisbon: Valentim da Costa. (Facsimile reproduction by J. Platzmann under the title Catecismo da lingna Kariris, Leipzig: Teubner, 1896.) Borges, M. de N. de O. F. 1995. ‘A spectos da morfossintaxe do sintagma nominal na lingua K ayapo’, MA thesis, Universidade de Brasilia. Borges, M. V. 1997. ‘As falas ferainina e masculina no K arajá’, MA thesis, Universidade Federal de Goiás. 202 Aryan D. Rodrigues Boswood, J. 1973. ‘Evidèncias para a inclusào do A ripaktsá no filo M acro-Jê’, pp. 67-78 of Série Lingiilstka 1, ed. L. 1. Bridgeraan. Brasilia: SIL. 1974a. Ä lgum as funçôes de participante nas oraçSes Rikbaktsa’, pp. 7-33 o f Série Lingüistka 3, ed. L. I. Bridgetnan. Brasilia: SIL. 1974b. ‘Citaçôes no discurso narrative da lingua Rikbaktsa’, pp. 99-129 of Série LingBistka 3, ed. L. I. Bridgeman. Brasilia: SIL. 1978. Querfalar a lingua dos Canoeiros? Rikbaktsa em 26 liçdes. Brasilia: SIL. Biiblitz, T. 1994. Ä nálise fonolôgica preliminar da lingua Xokléng', M A thesis, Universidade de Brasilia. CaUow, 5. C. 1962. ‘The Apinayéianguage: phonology and grammar’, Ph.D. thesis, University o f London. Cavalcante, M. P. 1987. ‘Fonologia e morfologia da lingua Kaingáng: o dialeto de Sào Paulo comparado com o do Paraná’, Ph.D. thesis, Universidade Estadual de Campinas. 1992. ‘Fonologia do ICarajá’, Revism do Miiseu A nm pologko (Universidade Federal de Goiás), 1.63-75. Colbacchini. A. 1925. I Bororos orienlali 'Orarinnigiidoge’ del M atto Gro.iso (Brasik), Contrihuti Scientifici delk Mission! Saksume del Venerabik Don Bosco, 1. Torino: Societa Editrice Internazionale. Colbftcchini. A. and Albisetti, C. 1942. Os Bororos orientals Orwimogodogue do Planalto Orkntal de M ala Grosso. S8 o Paulo: Com panhia E ditora Nacional. Crowell, T. 1977. ‘The phonology o f Borôro verb, postposition, and noun paradigms’, Arqiiivos de Anatomla e Antropologia (Instituto de Antropologia Prof. Souza Marques, Rio de Janeiro), 2.157-78. 1979. ‘A gram m ar o f Bororo’, Ph.D. thesis, Cornell University. Davis, I, 1966. ‘Comparative Jê phonology’, Estudos Lingiiislicos: Revisla Brasileim de Lingiiistka Teárica e Aplkada, 1: 2.10-24. 1968. ‘Some Macro-Je relationships', UAL, 34.42-7. Dourado, L. G. 1990. ‘Estudo preliminar da fonèmica Panará’, MA thesis, Universidade de BrasiHa. 1993a. ‘Classiflcadores de nomes em Panará’, pp. 387 4 4 o f Lingiilstica indigena e ediicaçcio na America Latina, ed. L. Seki. Campinas; Editora da Unicamp. 1993b, ‘Tendèncias tnorfofonêmicas em Panará; um a proposta de análise’, Bolelim do Museu Paraense Emillo Goeldi, Série Antropologia, 9.199-208. Ehrenreich. P. 1894. ‘Materialien zur Sprachenkunde Brasiliens, I: die Sprache der Caraya (Goyaz)', Zeiíschnfi fiir Ethnologie, 26.20-37, 49-60. Fortune, D. 1973. ‘Gram ática Karajá: um estudo preliminar em forma transform acional’, pp. 101-61 o f Série Lingíiistica 1, ed. L. L Bridgeman. BrasiHa; SIL. Fortime, D. and Fortune, G. 1975. ‘Karajà men’s-women’s speech differences with social cor­ relates’, Arquh'os de Anatomia e (Instituto de Antropologia Professor Souza Marques, Rio de Janeiro), 1.111-24. Greenberg. J. H. 1987. Language in the Americas. Stiuiford; .Stanford University Press. Gudschinsky, S. C. 197L ‘Ofaié-Xavánte, a Jê Sang\!age', pp. 1-16 of Esiudos sôhre lingms e atltiiras indlgenas, ed. S. C. Gudschinsky. Brasilia: SIL. 1974. ‘Fragmentos de Ofaié', pp. 177-249 o f Série Lingiilstica 3, ed. L. I. Bridgeraan. Brasilia: SIL. Gudschinsky, S. € ., Popovich, H. and Popovich, F. 1970. ‘Native reaction and phonetic simi­ larity in M axakali’, Language, 46.77-88. Ouedes, M. 1993. ‘S íw |a m êkapérêra. Stiyá: a lingua da gente, um estudo fonologico e gramatical', Ph.D. thesis, Universidade Estadual de Campinas. 6 Macro-Jê 203 Guérios, R. F. M. 1939. ‘0 nexo lingüistico B ororo-M errim e-C aiapé’, Revista do Circula de Estudos Bandeirantes, 2.61-74. 1942. ‘Estudos sobre a lingua Caingangue*. Arquiws do Museu Parameme, 231-111. 1945. ‘Esludos sobre a lingua C am acà’, Arqtiivos do Museu Paranaense, 4.291-320. Hall. J. 1979. Os sistemas fonotôgkos e grâflcos Xamnte e Ponuguês (anâlise contmsliva), Ensaios Lingüisticos, 4. Brasilia: SIL. Hall, J., M cLeod, R. A. and Mitchell, V. 1987. Pequeno dicionârio Xavânie-Portugtiês. Português-Xawme. Brasilia: SIL. H am , P. 1961. Âpinayé grattirnar', Arquivo Lingiüstico, 108. Brasilia: SIL. 1967. ‘Morfofonêmica Apinayé’, pp. 123-6 o f Ätas do Simpásio sobre a Biota Amazônica, vol. Il, ed. H. Lent. Rio de Janeiro; Conselho Nacional de Pesquisas. Ham , P., Waller, H. and Koopman, L. 1979. Aspeclos da lingua Apinayé. Brasilia: SIL. Henry, J, 1935. Ä Kaingàng text’, UAL, 8.172-218. 1948. ’The Kaingàng language’, UA L, 14.194-204, Jefferson, K. 1989. ‘Gram âtica pedagôgica Kayapô, p a tte 1’, Arquivo Lingüîstico, 186. Brasilia: Instituto Lingüîstico. Kaufm an, T. 1990. ‘Language history in South America: what we know and how to know m ore', pp. 13-73 o f Amazonian linguistics: studies in Lowland South American languages, ed. D. L. Payne. Austin: University o f Texas Press. 1994. ‘The native languages of South America’, pp. 46-76 and maps pp. 14-25 o f Atlas o f the world's languages, ed. R. E. Asher and C. Moseley. London: Routledge. Krause, F. 1911. /n den Wildnissen Brasiliens: Bericht m d Ergebnisse der Leipziger AraguayaExpedition. Leipzig. Kunike, H. 1916. ‘Beitriige zur Phonetik der Karajá-Sprache (Brasilien)’, Internationales Archiv fiir Ethnographie, 23.147-82. 1919. ‘Die Phonetik der Karaiá-Sprache nach linguistischen Principien’, Journal de la Société des Américanistes, n. s., 11.139-81. Lachnitt, G. 1987, Momnhltri'ubumro: a'uwe mreme - waradzu mreme Dicionârio xam nteportugtiis. Campo Grande: Missâo Salesiana de M ato Grosso. 1988. Damreme'uwaimramidzé: estudos sislemálicos e comparativos de graminica Xavante. C am po Grande: Missâo Salesiana de M ato Grosso. Lapenda, G. 1968. Estrutiira da lingua lati. Recife: Imprensa Universitária d a Universidade Federal de Pernannbuco. Larsen, T. W. 1984. ‘Case marking and subjecthood in Kipeá, K irirí’, Berkeley Linguistic Society Proceedings, 10.189-205. Lima, S. T. P. 1995. ‘A lingua Um utina; “um sopro de vida” ’, M A thesis, Universidade Federal de Pernambuco. Loukotka, C. 1931. ‘La familia lingiiistica M asakalí’, Revista del instituto de Elnologia de la Universidad Nacional de Tucumán, 2.21-47. 1932. ‘La familia lingiiistica K am akan del Brasil', Revista del Instituto de Etnologia de la Universidad Nacional de Tuaimàn, 2.493-524, 1937. ‘La familia lingiiistica Coroado’, Journal de la Société des Américanistes de Paris, n.s., 29.157-214. 1939, ‘A lingua dos Patachos’, Revista do Arquivo Municipal de SâoPauh, 55.5-15. 1944 [1942). ‘Klassifikation der südamerikanischen Sprachen’, Zeitschrift fü r Ethnologie, 74,1-69. 1955. ‘Les indiens B otocudo et leur langue’. Lingua Pasnaniensis, 5.112-35. M aia, M. A. R. 1986. ‘Aspectos tipolôgicos da lingua Javaé’, M A thesis, Universidade Federal do Rio de Janeiro. 204 Aryan D. Rodrigues Mamiani, L. V. 1698. Catecismo da doutrina christâa na lingua brasilica da naçâo Kiriri. Lisbon: Miguel Deslandes. (Facsimile edited by R, Garcia under the title Catecismo Kiriri, Rio de Janeiro: Imprensa Nacional, 1942.) 1699. Arte de grammatica da lingua brasilica da naçam Kiriri. Lisbon: Miguel Deslandes. (2nd edition, Rio de Janeiro; Bibliotheca Nacional do Rio de Janeiro, 1877.) Martius, K. F. P. von. 1867. Wörtersammlung Brasilianischer Sprachen. Glossaria linguarum Brasiliensiiim. Glossarios de di versas lingoas e dialectos, que fallao os Indios do imperio do Brazil. Beiträge zur Ethnographie und Sprachenkunde Amerikas zumal Brasiliens, IL Leipzig: Friedrich Fleischer. M ason,J. A. 1950,‘The languages of South American Indians’, pp. 157-317of H andbookof South American Indians, vol. VI, ed. J. H. Steward. Bureau o f American Ethnology, Bulletin 143. Washington, D.C.: Smithsonian Institution. Mattos, R. de. 1973. ‘Fonêmica Xerénte’, pp. 79-100 o f Série Lingiustica 1, ed. L. I. Bridgeraan. Brasilia: SIL. McLeod, R. 1974. ‘Fonemas Xavánte’, pp. 131-52 o f Série Lingiiistica 3, ed. L. I. Bridgeman. Brasilia: SIL. McLeod, R. and Mitchell, V. 1977. Aspectos da lingua Xavánle. Brasilia: SIL. M eland, D. 1968. ‘Fulniô grammar’, Arquivo Litigüístico, 26. Brasilia: Insthuto Lingüístico. Meland, D. and M eland, D. 1967. ‘Fulniô (Yahthe) phonology statement’, Arquivo Lingüístico, 25. Brasilia: Institute Lingüístico. 1968. ‘Word and m orpheme list of the Fulniô Indian language (Yahthe)’ in the archives of the SIL, Cuiabá, M ato Grosso, Brazil. Ms. M étraux, A. and Nim uendaju, C. 1946. ‘The M ashacali, Patasho, and M alali linguistic fam­ ilies’, pp. 541-5 o f Handbook o f South American Indians, vol. I, ed. J. H. Steward. Bureau o f American Ethnology, Bulletin 143. W ashington, D.C.: Smithsonian Institution. Mullen, P. 1965. ‘Kaingáng verbs’, Arquivo LingUistico, 161. Brasilia: SIL. 1966. ‘Verb pluralization in Kaingáng’, Arquivo LingUistico, 159, Brasilia: SIL, Nimuendaju, C. 1945 [1980]. Mapa etno-histôrico do Brasil e regiöes adjacentes. Rio de Janeiro: Fundaçào Instituto Brasileiro de Geografîa e Estatistica. (Completed 1945, pub­ lished 1980.) Nimuendaju, C. and Guérios, R. F. M. 1948. ‘Cartas etnolingüísticas’. Revista do Museu PauUsta, n. s., 2.207^1. Palácio, A. P. 1984. ‘G uato, a lingua dos indios canoeiros do rio Paraguai’, Ph.D. thesis, Universidade Estadual de Campinas. 1986. ‘Aspects o f the m orphology o f G uato’, pp. 363-72 of Language in global perspective, ed. B. F. Elson. Dallas: SIL. Pereira. D. G. 1992. ‘Alguns aspectos gramaticais da lingua M axakali’, MA thesis, Universidade Federal de Minas Gerais, Belo Horizonte. Pinto, E. 1956. Etnologia brasileira (Fulniô - os liltimos Tapuias), Biblioteca Pedagogica Brasileira, Série 5a., Brasiliana, 285. Sao Paulo: Companhia Editora Nacional. Popjes, J and Popjes, J. 1986. ‘Canela-Krahô’, pp. 128-99 o f Handbook o f Amazonian lan­ guages, vol. 1, ed. D. C. Derbyshire and G. K. Pullum. Berlin: M outon de Gruyter. Popovich, H. 1967. ‘Large grammatical units and the space-time setting in M axakali’, pp. 195-9 of Alas do Simposio sobre a Biota Amazônica, vol. II, ed. H. Lent. Rio de Janeiro: Conselho Nacional de Pesquisas. 1971. ‘The Sun and the M oon, a Maxakali text’, pp. 29-59 of Estudos sobre Linguas e Culturas Indigenas, ed. S. C. Gudschinsky. Brasilia: SIL. 1985. ‘Discourse phonology of Maxakali: a multilevel, multiunit approach’, MA thesis. University of Texas at Arlington. 6 M acro-Jê 205 1986. ‘The nominal reference system of M axakalí’, pp. 351-8 of Pronominal systems, ed. U. Wiesemann. Tübingen: G unther N arr Verlag. Ribeiro, E. R. 1996. ‘M orfologia do verbo K arajá’, M A thesis, Universidade Federal de Goiás. Rodrigues, A. D. 1962. ‘Com paraçâo das linguas U m utina e Borôro’ in H. Schultz, ‘Informaçôes etnográficas sobre os Um utina’, Revista do Museu Paulista, n. s., 13.100-5. 1981. ‘Nasalizaçâo e fronteira de palavra em M axakali’, pp. 305-11 of VEncontro Nacional de Lingüística, vol. II. Rio de Janeiro: Pontificia Universidade Catolica do Rio de Janeiro. 1986. Linguas brasiteiras: para o conhecimento das linguas indigenas. Sao Paulo: Ediçôes Loyola. 1993. ‘U m a hipotese sobre flexào de pessoa em Borôro’, p. 509 o f Anais da 45a. Reuniäo Annal da Sociedade Brasiteira para o Progressa da Ciência, vol. II. Recife: Sociedade Brasileira para o Progresso da Ciência. Rodrigues, A. D. and Cavalcante, M. P. 1982. ‘A ssimilaçào intrassegmental em Kaingâng’, Ciência e Cullura 34.7, Suplemento: 34a. ‘Reuniäo Anual da Sociedade Brasileira para o Progresso da Ciência. Säo Paulo', 198. Rodrigues, A. D. and D ourado, L. G. 1993. ‘Panará: identificaçâo lingüistica dos KrenAkarore com os Cayapo do Sul’, p. 505 of Anais da 4Sa. Reuniâo Anual da Sociedade Brasileira para o Progresso da Ciência, vol. II. Recife: Sociedade Brasileira para o Progresso da Ciência. Santos, L. C. 1997. ‘Descriçâo de aspectos morfossintâticos da lingua Suyâ/Kïsêdjê (Jê)’, Ph.D. thesis, Universidade Federal de Santa Catarina. Schmidt, M. 1941. ‘Los Barbados o Umutinas en M atto Grosso (Brasil)’, Revista de la Sociedad Cientifica del Paraguay, 5.1-51. Schmidt, W. 1926. Die Sprachfamilien und Sprachenkreise der Erde. Heidelberg: Carl Winter. Schultz, H. 1952. ‘Vocabulário dos indios Um utina’, Journal de la Société des Américanistes de Paris, n. s., 41.81-137. Sebestyén, E. 1981. ‘H. H. M anizer’s Botocudo folklore texts’, Arles Populares, 7.140-63. Seki, L. 1990. ‘Apontamentos para uma bibliografia da lingua Botocudo/Borum’, Cadernos ■de Estudos Lingiiislicos, 18.115-42. Silva, T. C. 1986. ‘Descriçâo fonética e anâlise de alguns aspectos fonologicos da lingua K renák’, MA thesis, Universidade Federal de M inas Gerais. Souza, S. M. de. 1989. ‘O sistema de referência pessoal da lingua K rahô’, M A thesis, Universidade Federal de Goiàs. Stout, M. and Thomson, R. 1974a. ‘Fonêmica Txukuham ëi (Kayapô)’, pp. 153-76 o f Série Lingüistica 3, ed. L. I, Bridgeman. Brasilia: SIL. 1974b. ‘M odalidade em Kayapo’, pp. 69-97 of Série Lingüistica 3, ed. L. I. Bridgeman. Brasilia: SIL. Teixeira, J. B. 1988. ‘Contribuiçâo para a fonologia do dialeto Kaingâng de N onoai’, MA thesis, Universidade Estadual de Campinas. Thom son, R. and Stout, M. 1974. ‘Elementos proposicionais em oraçôes Kayapô’, pp. 35-67 of Série Lingüistica 3, ed. L. I. Bridgeman. Brasilia: SIL. Tremaine, S. 1981. ‘Rikbaktsa narrative discourse’, Arquivo Lingüistico, 162. Brasilia: SIL. Urban, G. 1985. ‘Ergativity and accusativity in Shokleng (Gê)’, U A L, 51.164-87. Viana, A. M. S. 1995. ‘A expressào de atributo na lingua K arajá’, MA thesis, Universidade de Brasilia. Wetzels, L. 1996. ‘Oclusivas intrusivas em M axacali’, pp. 85-102 of Estudos fonolôgicos das linguas indigenas brasileiras, ed. L. Wetzels. Rio de Janeiro: Editora Universidade Federal de Rio de Janeiro. Wetzels, L. and SIuyters, W. 1996. ‘Formaçào de raiz, formaçâo de glide e decrowding fonético 206 A i y m D. Rodrigues em Maxacalf’, pp. 103-49 o f Estucbs fonotágicos das llnguas indígenas brmikiras, ed. L. Welzels. Rio de Janeiro: Editora Universidade Federal de Rio de Janeiro. Wiesemann, U, 1971. Dkionárío Kaingàng-Ponugiiês, Portugiiês-Kairigàng. Brasilia; SIL. 1972. Ûie phonologische undgrmtmatische Siruktur der Kaingàng-Sprache. The Hague: M outon. 1978. ‘Os dialetos da lingua Kaingáng e Xokléng’, Árqmvos de Am tom ia e Anlropohgia (Institiito de Antropologia Prof. S o u a Marques, Rio de Janeiro), 3.Î97-217. 1986. ‘The pronoun systems o f some Jê and M acro-Jê languages’, pp. 359-80 o f Pronominal systems, ed. U. Wiesemann. Tiibingen: Gunther N arr Verlag. 7 Tucano JA N E T B A R N E S 1 in t r o d u c t io n ' A listing o f the various languages in the T ucano language family that are currently in use first requires a decision as to which languages are sufficiently distinct to w arrant a separate listing. T hus it may be n oted that soine entries in table 7.1 have two names. B arasano (also known as Panera) and Taiw ano (also known as Eduria) differ mainly in pitch-stress on words (Jones and Jones 1991:2) and so are grouped together. R etuara and Tanim uca differ mainly in a few lexical items (Strom 1992: 1 ). A lthough [ consider B ará to be distinct from W aim aja, there is not sufficient data on Bará to list it separately. Pisam ira would be included if there were sufficient data available. Waltz and W heeler (1972: 128) group Pisam ira with Tuyuca and Bará under the nam e Pápiwa, and indeed a taped word list th a t a Tuyuca lady and I lis­ tened to indicates that it has m uch in com m on with Tuyuca. T he Eastern and Central Tucano languages are spoken in the northw estern A m azon Basin in the state o f Vaupés, Colom bia, and in the state o f A mazonas, Brazil. T he Western Tucano languages are spoken in southw estern C olom bia along the Putum ayo and C aquetá rivers, and along the Putum ayo and N apo rivers in E cu ad o r and Peru, (n table 7.1, the letters E, W and C represent Eastern, Western and C entral. Tlie subgroupings in table 7.1 are taken frorn Waltz and Wheeler (! 972), and are based on phonological and lexical affinities between the Tucano lan­ guages. M alone (1987), on the basis o f her study o f the development o f the current languages from the proto-language, has the sam e groupings, with the exception that she pulls £ 6 and E7 out o f the Eastern category and puts them in a fourth m ajor category, tentatively also including El 1 and £12 in that category. The approxim ate locations o f the Tucano languages are shown on m ap 6. ' I am grateful to the following linguists for answering my many questions relative to the lan­ guages they are studying: M ark and Paula Bostrom (Tatuyo), Bev Brandrup (Siriano), Ronald Mcizger (Carapana), Jim and M arion Miller (Desano), Joel and Nancy Stolte (Waimaja), N athan and Carolyn Waltz (Wanano and Piratapuyo). I am also grateful to Terrell Malone for reading the draft and making suggestions based on her study o f protoTucano. 208 Janet Barnes The Tucano groups are rapidly becom ing accultiirated to m any aspects o f the d om inant Latin culture, yet m any o f the people in these groups still carry ou t their daily life m uch as th eir ancestors did: using slash-and-burn agriculture, fishing, an d eating casabe bread with a fish sauce prepared with h o t peppers. Individual family h ouses w ith corrugated zinc roofs are replacing the ‘m aloca’, a large com ­ m unal house typically shared by a m an and his sons and their families. In general, the Eastern Tucano language groups are exogamous w ith respect to language, and individuals continue to speate their own (i.e., their father’s) laoguage, as long as those they are speaking to understand it. Thus, children grow up knowing both their m other’s and their father’s language, as well as having at least a passive competence in the languages o f the other women in the village o r longhouse. For more inform ation on this unique m ultilingual situation, see chapter 14 o f this volume as well as Sorensen 1967. Word lists taken in the 1960s and before indicate a very high degree o f shared vocabulary between the languages. M ore recent studies o f verb m orphology and discourse gram m ar indicate th a t the actual degree o f intelligibility may not be as high as the lists o f shared vocabulary might indicate (see West 1977). 7 Tucano 209 Table 7.1 The T ucam language fa m ily (B, Brazil; C, Colom bia; E, Ecuador; P, Peru) W ESTERN Western north W 1 Koreguaje 2,0 0 0 W2 Secoya 400 Rio Putumayo (1) W3 Siona 300 Rio Putumayo (C, E) Western south W4 Orejon 200-.300 Rios Napo, Ampi Yacu and Algodon (P) CE N TR A L Cl Cubeo 6,0 0 0 Rios Vaupés, Cuduyari, Querari and Piraboton (B ,C ) C2 300 Rios Guacayá, Miriti, Oiyaká and Apaporis (C) Tanimuca/Retuarä Rios O rteguaza and Caquetá (C) EA STERN Eastern north E 1 Piratapuya 1,100 Rio Papuri (B, C) E2 Tucano 4,100^,600, Rio Papuri and Cano Paca (B, C) E3 W anano 1,100 Rio Vaupés (B, C) Eastern central E4 BaríiAVaimajä 500-600 Caños Colorado, Frijol, Lobo, Inanibii and Yapü, Rio Tiquié (B, C) Cafio TÍ, Rios Piraparaná, Papuri and Vaupés (C) E5 C arapana 600 E6 Desano 1,00 0 Rios Papuri and Vaupés (C) E7 Siriano 250-300 Rios Paca and Vina (C) Rios Piraparsná and Papuri, Cano Yapil (C) E8 Taiuyo 350 E9 Tuyuca 725 Rios Papuri and Tiquié, Caño Inam bù (B, C) ElO Yuruti 200-250 Caños Paca and Ti (C) Eastern south E ll Barasano/Taiwano 250 Rio Piraparanà (C) E12 350 Rios Comeña, Piraparaná and Apaporis (C) 2 M acuna PHONOLOGY The proto-consonants according to M alone (1987: 7) are given in table 7.2. She posits nine consonant phonem es plus suprasegm entals o f nasality, tone and/or stress, an d glottal stop. T h e six-proto-vowel system is given in table 7.3. 210 Janet B arm s Table 7.2 Proto-consonants voiceless stop voiced stop voiceless sibilant voiced semi-vowel bilabial alveolar velar *b ♦t ♦d *s *g *k •w Table 7,3 Proto-vowels fro n t c e n tra l high low The present-day languages all m aintain the voiœd/voiceiess contrast in the stops. Koreguaje has nasals rather than voiced stops plus both unaspirated and aspirated voiceless stops. W anano has three sets o f stops: voiced, voiceless unaspirated and voiceless aspirated. O rejon, in addition to voiced bilabial and alveolar stops, has voiced implosive stops at those points o f articulation. Secoya and Siona have a com ­ bination o f nasals and voiced stops in con trast to the voiceless stops. 2.1 Syllable structure The basic syllable structure for the Tucano languages is (C)V. For those languages with a glottal stop, there is an additional syllable type: (C)V?. W l-3 and C2 have diphthongs, resulting in the additional syllable type (C)VV for those languages. The other languages have gem inate vowels which are generally analysed as a sequence o f two syllables - CV.V - due to accent placement. Examples from Tiiyuca and Tucano include; ( 1) 2.2 dii ‘blood’ d ii ‘clay, m eat’ (Tuyuca) dii ‘blood’ d i? i ‘clay, m eat’ (Tucano) Segmental phonology All o f the Tucano languages, with the exception o f Tanimiica, have a six-vowel system which is the same as the proto-vowel system given in table 7.3. Tanimuca has 7 Tucano 211 a five-vowel system, lacking the high central unrounded vowel !\l} T he languages E4, E5, E 7 -Í2 , have the consonant inventory given in table 7.2 plus /r/ and /h/, except that E4 and E8 lack /s/ and E l l lacks /p/. E l and E6 also have the consonant inventory given in table 7.2 plus /r/, /h/, but they also have a glottal stop. T he glottal stop occurs not only in E l and E6, but also in E2, E3, C2 and all the Western languages. In W1 the glottal stop is very strong. In W2 and W3 it is very weak. There is no inform ation available on W4. T he glottal stop o f C2 is strong, as it is in E l , E3 and E6. In E2 its eBect is to lengthen and laryngealize the vowel that precedes it. Siona (W3) and Koreguaje (W l) each have eighteen consonant phonemes. W anano (E3) has sixteen. T he rest o f the languages have between ten and twelve consonant phonem es as follows: E l , E2, E6, W2 and W4 have twelve; C l, C2, E5, E7, E9, ElO and E12 have eleven; and E4, E8 and E ll have ten consonant phonemes. In the Eastern and C entral languages /b, d, g, r, w, j. hi are realized as Im , n, tj, r, w, ji, h / in nasal morphemes. T he flap h i is realized as a lateral flap in som e environments. The glide /j/ varies from a nonsyllabic vowel to an alveo-palatal affricate [d3] in oral morphemes, and from a nasal glide to a nasal in nasal morphemes. 2.3 Nasalization In the Centra! and Eastern T ucano languages, nasalization is an autosegm ent asso­ ciated with the morphem e (see Barnes 1996), M orphem es are m arked in the lexicon as nasal, oral, or are unm arked. N asalization spreads progressively in all the lan­ guages to unm arked morphemes. In Barasano, D esano and Siriano nasalization also spreads regressively, aflecting only a few specific m orphem es (see M iller forth­ com ing). A typical example o f nasalization spreading through an unm arked m or­ phem e is given in the following example from Tuyuca, using the unm arked im perative suffix -ja. (2) bia-ja [biajá] close-iMPER ‘close (it)’ ^ Malone, through her study o f proto-Tucano (p.c.), has concluded that Tanlmuca resulted from a group o f Arawakan speakers who learned a Tucano language. This would help explain why Tanimuca has only the five vowels found in Arawakan languages. In addition, she has said that it can be useful to include Tanimuca when comparing Tucano morphemes, but it is not very useful to include Tanimuca when doing a comparative study o f the gram ­ mars o f Tucano languages. N ote the number o f instances in the gram m ar section where Tanimuca differs from all the other languages therein described. 212 Janet Barnes (3) pô5-ja [pôôjià] open-iMPER ‘open (it)’ The inventory o f consonants through which nasalization spreads varies from lan­ guage to language, but m ost of the languages include the liquids j, h/ and the flap /r/ in their inventory, if they have suffixes that begin with those consonants. 2.4 Accent The Tucano languages have accent, or pitch-accent systems in which there is high pitch vs low pitch. High pitch is shown by ('), and low pitch is unmarked. High pitch is generally associated with accent. Some o f the languages, such as Desano, Tuyuca, Yuruti and Cubeo, have one accented syllable per phonological word. Several of the languages, am ong them Barasano, C arapana, M acuna, Siriano, Tucano and W aimaja, may have two or m ore contiguous high-pitched syllables, all o f these syl­ lables being equally accented. In Secoya, words o f four or m ore syllables have accents on alternating syllables. Examples o f these are: — one accented syllable per phonological word in Cubeo (Salser, Morse and Ferguson, 1986: 3): (4) — (5) — h a tiö ö o ‘she is co o k in g ’ multiple high pitch in Barasano (Jones and Jones 1991: 11-12):^ ságárö ‘thigh’ wasöbí ‘robe’ accent on alternating syllables in Secoya (Johnson and Levinsohn 1990: 19): (6) 3 de?w ádo'tá?adé ‘in the place where he fixed it’ M O R PH O LO G Y The Tucano languages are almost entirely suffixing languages, with a strong ten­ dency towards agglutination. One exception is the prefix ka-, the specific nominal referent, possibly borrow ed from Arawak, that occurs on nouns in E4, E5 and E8 (see Metzger 1998). Tanim uca/Retuara have prefixes that are ‘cliticized subject pro­ n o uns’, and also one ‘non-hum an object pronoun’ (Strom 1992: 5). Koreguaje has two prefixes that occur on verbs. The first is ape- ‘just beginning to ’. The second is ’ Gomez-Imbert (1995) also analyses Barasano as having a pitch-accent system. 7 Tucatio 213 ka- an d is m ore difficult to define. When ka- is prefixed to stative verbs, it indicates the process involved in arriving at that state; and, when it is prefixed to active verbs, it indicates an additional and sim ilar action (C ook and Criswell 1993; 73-4 plus p.c.). 3.1 T h e verb The agglutinative tendencies o f the Tucano languages are best exemplified in verbs. Independent verbs are minimally com posed o f a verb root plus an ending which indicates person, number, tense and evidential category. Between the stem and the ending, a num ber of aspect and m odality m orphem es may occur, as well as miscel­ laneous suffixes indicating passive voice, causation and m otion. Verb stems may be transitive, active intransitive, stative intransitive, negative, quotative or qualitative. M ost o f the languages have a suffix that distinguishes recent from distant past. Serial verb constructions are com m on, and often the m eaning o f the construction is evident from the m eanings o f the stems. N ote the following examples: ‘to h it’ plus ‘to find’ is ‘embrace, patting on the back’; ‘to hit’ plus ‘to take’ is ‘to grab, or snatch’; ‘to h it’ plus ‘to fold’ is ‘to split’; etc. However, some com pounds are not quite so transparent, so th at while ‘to w alk’ plus ‘to know /ability’ is ‘to be able to w alk’, the com pound ‘to h ear’ plus ‘to know /ability’ unexpectedly results in ‘to understand’. 3.1.1 E v id e n tia ls The evidential morphem es indicate either how speakers received their inform ation or how closely they identify with what they relate. The evidentials also include person, num ber and tense inform ation. In Tanim uca the three suffixes that give evidential inform ation are optional. These optional suffixes indicate: (1) strictly auditory inform ation, (2) assumed inform ation an d (3) secondhand inform ation (Strom 1992: 90-1). The suffix that indicates th a t the inform ation is strictly auditory is consistently used. The other two are quite optional. Thus, the gram m ar used by Tanim uca speakers does n o t neces­ sarily convey how they obtained their inform ation (Strom p.c.). O ne o f the simplest evidential systems is found in Koreguaje. If speakers were present for the event or state they are speaking of, there are no special markers. If they o btained the inform ation from another source, they include an auxiliary verb which indicates that they are n o t giving a first person account. If they assum e that the assertion is/was true, they use an auxiliary verb th a t indicates probability (C ook and Criswell 1993:86-7). Tuyuca is one o f the systems that has five evidentials. These evidentials are real­ ized as verb suffixes, some o f which are single syllables, and contain the following 214 Janet Barnes inform ation: person, number, gender, tense, and evidentiality. T he five suffix sets convey the following evidential inform ation: (1) the speaker saw the event o r state, (2) the speaker h eard, tasted, smelt or felt the event or state, (3) the speaker saw the results of the event o r state, (4) the speaker heard about the event or state from someone else and (5) the speaker assumes that the event o r state occurred (Barnes 1984: 257). The fourth evidential (secondhand inform ation) only occurs in the past tense, as the speaker necessarily heard the inform ation prior to relating it. The recent past morphem e is inserted when the inform ation was just heard, or was heard recently For example, if a Tuyuca girl uses an interm ediary to express her wishes, the intermediary will use the fourth evidential to say ‘She wants to look at that mag­ azine’, and it will literally be, Tt was reported to me that she wants to look at that magazine.’ The fifth evidential (assumed inform ation) is used when there is no reason to assume that an event did not occur, or is not occurring. I f speakers really have no idea as to whether or not an event occurred, they will use the assumed evi­ dential, and will preface their statements w ith the word /öba/, which indicates that they are no t at all sure. T he future tense does n o t employ evidentials. M alone (1988) has dem onstrated that person, num ber and evidential inform a­ tion were represented by separate m orphem es in proto-Tuyuca. In some o f the lan­ guages there still exists a string o f morphemes, while in others the string has been collapsed into a single syllable. In the Western north languages, tense suffixes distinguish between masculine sin­ gular, feminine singular, and plural. In the Western south language, tense suffixes distinguish between third masculine singular, third feminine singular, first aod second singular, and first and second plural. In some conjugations, singular and plural in first and second persons is represented by the sam e suffix. The Central lan­ guages distinguish between masculine singular, fem inine singular, anim ate plural, and inanimate in their person/num ber markers. In the Eastern languages, with the exception o f P iratapuya and W anano, the third person evidentials distinguish am ong m asculine singular, feminine singular, and plural. (N ote the same distinction in third person pronouns in table 7,5.) In many o f the Eastern languages, first and second persons, singular a nd plural, plus inaninw te are all represented by the same evidential, while in others there are person and num ber markers accom panying the evidentials in non-third-person verbs. See table 7.4 (from West 1980: 28) for a typical example. 3 .1 .2 A sp e c t In m ost o f the languages, aspect is indicated by means o f suffixes to the verb stem, and may include; iterative, completive, anticipatory and inchoative. 215 7 Tucano Table 7.4 Tucano ( E 2) distant past visual evidentials PL SG -wi 1/2/INAN M ASC -W Î FEM -W Ô -wa 3 Iterative iit Barasano (7) (Jones and Jones 1991; 101) g a h e -rîb î bota-ri kêà-kudi-k a-bâ ïd à other-day post-PL chop-iTERATivE-far.past-EViD:visUAL 3pl g a h ir îb ï botari k ë ik u d ik a b â Id a ‘The next day they w ent from place to place chopping down posts (for the new house).’ Completive in Tucano (8) (West 1980: 56) di-to ha-?a Say-COMPL-EVIO:PRES. VISUAL d îto h a ? a ‘1 already said (it).’ Anticipatory in Desano (9) (M iller forthcoming: § 5.5) o-ge árl-ju-ke here-LOC be-ANTICIPATORY-lMPER ôge áríjiike ‘Stay here in anticipation (of our arrival)!’ Inchoative in Koregm je (10) (C ook and Criswell 1993: 66) w ai-pi?ra-sô aso-mc hit-lNCHOATIVE-INTENSIFlER REPORTED-PL w aip iírasö asom ë ‘They began to hit (them ).’ 216 Janet Barnes The progressive and perfect aspects are generally achieved through verb com­ pounding. The main verb is in a participial form, and is followed by an auxiliary verb w ith a tense/person/evidential suffix, 3.1.3 M ood and modality Imperative and interrogative markers replace the evidential endings, com ing at the end o f the verb. Imperative in Barasano (11) (Jones and Jones 1991: 76) ji-re goti-ja ISG-SPEC tell-PRES.IMPER bï 2SG jire gotija b î ‘Tell me!’ Interrogative in Tiicano (12) (West 1980:40) di-ri be-PAST.INTERROG d îrî ‘Were you (there)'?’ (used for all persons) O ther modality indicators are suffixes which follow the verb root and precede an evi­ dential, imperative o r interrogative ending. These mood indicators include: negative, probability/conditional, contraexpectation, desiderative, ability and emphatic. Negative and probabiliiylconditional in Tuyuca (13) jâ b ïk a okô peá-ri-atä y e s te r d a y w a te r fall-NEG-DEPENDENT.CLAUSE bîâ-ja-w ii h ïî-b o -a-h ïju 2PL-P0SSV-CL:BUILDING burn-COND-RECENT-EVID;PAST.ASSUMED jâ b ïk a o ké peàriatâ, bïàjaw i h tîb o a h ïju ‘If it had not rained yesterday, your house probably w ould have burned.’ Contraexpectation and desiderative in Carapatm (14) (M etzger 1981: 83) ji ii-koa-ga-M -wi ISG smoke-COMPL-DES10-CONTRAEXPECTATION-EVID:PAST. VISUAL ji ukoagabîw ï ‘I w anted to sm oke (but it was n o t possible).’ 7 Tucano 217 Table 7.5 Tucano (E 2 ) pronouns SG PL »sa EXCL 1 j« i bâiî INCL 2 Wλ MASC k{ FEM k6 bïsa dá 3 Ability in Tuyuca (15) h ô a-b àsî -ri -ga Write-AB1L1TY-NEG-EVID:PRES.N0NV1SUAL hôabâsïriga i do not know how to write.’ Emphatic in Yuruti (16) (K inch and de K inch 1992:40) kiii-eri-bôkà-ji bc.afraid-N EG -EM PH A TIC -IM PER kuieribôkâja ‘D on’t be afraid!’ 3.2 Pronouns 3 .2 .1 Personal pronouns The singular pronouns distinguish between first person, second person, third person m asculine and third person feminine. T h e plural pronouns distinguish between first person exclusive, first person inclusive, second person and third person. The sam e set o f pronouns is used for both subject and object pronouns. The Tucano pronouns presented in table 7.5 (West 1980; 177-8) are phonologically rep­ resentative o f present-day forms. Tuyuca (17) jii Is o k ti-re ïjâ -w i 3SO .M A SC -SPEC See-EVID:PAST. VISUAL jif k îîrê Î já w í i saw him .’ 218 Janet Barnes 3.2.2 Possessive pronouns In six o f the languages: C2, E l and W l-4 , subject pronouns are used as possessive pronouns. The rest o f the languages suffix a genitive, generally -Ja ‘singular’ or -je ‘plural’, to the possessive pronoun** and then the item possessed follows. The item possessed is expressed by a classifier or a noun. However, if the noun is a kinship term , the genitive is n o t used (see Jones and Jones 1991: 61-3 for a description of a typical Tucano possessive system). Koreguaje (18) (C ook and Criswell 1993:45) d 3Í?i w a?thi Iso knife d j i î i w a?th i ‘my knife’ Carapana (19) (M etzger 1981:184) ko-ja-wi 3SG. FEM -GEN-CL:C YUNDRICAL kojawi ‘her cylindrical object’ Barasano (20) (Jones a nd Jones 1991: 61 ) Î Jlkï 3SG.MASC grandfather Î jî k î ‘his g randfather’ 3.3 Classifiers The Tucano languages are prim arily numeral classifier languages. Classifiers in the Tucano languages are suffixes that are subdivided into a small set o f anim ate clas­ sifiers (which also function as nominalizers) and a larger set o f inanim ate classifi­ ers. Each classifier denotes salient characteristics o f the noun(s) associated w ith it. N ouns which are n o t associated with any classifier function as self-classifier suffixes (or ‘repeaters’, see A ikhenvald 1994: 420-2), for example, nam es for some body parts. In all o f the Eastern languages plus the Central language Cubeo, classifiers are found suffixed to numerals, nouns, dem onstrative adjectives, possessives and '• W anano uses shortened forms o f the subject pronouns as possessives (see Waltz 1997). 7 Tucano 219 verbs (to form nom inalized verbs). T he other C entral language Tanimuca does not suffix classifiers to possessives. In Koreguaje, classifiers are found suffixed to the same categories as the Eastern languages. Full inform ation is n o t available on the rest o f the Westero languages. O ne o f the differences between the Western and the Eastern languages regarding classifiers is that in the Western languages nouns with a specific referent are obligatorily suffixed by classifiers, although nouns that denote a general category are n o t suffixed. O f the 4 Western Tucano languages, Secoya has 17 classifiers, Siona has 20 and Koreguaje has 28. N o inform ation is available on Orejon. O f the 2 Central Tucano languages, Tanim uca has at least 21 classifiers. Cubeo has aro u n d Î00. The classifiers denote form o r function (Strom 1992: i 1, and Salser et al.: 16). Eastern Tucano languages have anywhere from 50 to 140 classifiers. T he salient characteristics denoted by these classifiers include n o t only the three dimensions, but also collection and arrangem ent. There are rarely heard classifiers, such as one denoting bark that does n o t cling closely to the tree. By extension, this classifier is used for such things as baggy pants and plywood that has become wet so th a t the sheets have separated one from another. (F or listings o f classifier categories, see Barnes 1990 and Jones and Jones 1991: 50-6.) Classifiers in Secoya (21) (Johnson and Levinsohn 1990: 41) m ta s 5 k i-j£ -p i jo -w i cedar tre e -CL:wiTH.ROOTS-iNST c a n o e -CL:coNTAiNER m ia s ô k ijîp i jo w i ‘a cedar canoe’ Classijiers in Carapana (22) (M etzger 1981: 154) îd ë -jô -ri k a -îb îi-ri-jô -ri c h o n ta - CL:PALM-PL SPEClPlC.NOMlNAL.REFERENT-high-NMLZRc l : pa l m - pl ïd ë jô rî k a îb îà r ljô r ï ‘ch onta palms, specifically the tall ones’ 3.4 Specificity marker The specificity m arker may occur on any nonverb word or construction, and in most o f the languages it is obligatory on specific direct objects, experiencers and benefactees. In some o f the other languages, it m arks specific items in the discourse, even 220 Janet Barnes occurring on subjects in Tanimuca (Strom 1992: 7-8). It also occurs on location words (and occasionally on time words) to indicate that the location being referred to will have further significance in the discourse. (See Cook and Levinsohn 1985: 104-8 for a description o f a typical Tucano specificity marker.) In all o f the Eastern languages, plus C l and W2, the specificity m arker is the suffix -re. In the remaining languages its shape is -Ce, where C is an alveolar consonant. Objects o f the verb which refer to non-specific entities, such as ‘cattle’ (as opposed to ‘that cow ’), or ‘a wife’ (as opposed to ‘his wife’), are not m arked with the specificity marker in any of the languages. In some instances, this results in noun incorporation. C onsider the following examples: Tuyuca (23) b id t- k k i b âd i-b àd î husband-have husband-not.have ‘have a husband’ ‘not have a husband’ diká-kiti fruit-have w apa-kiti value-have ‘have (bear) fruit’ ‘have value’ kape-bâdï d ît î- ta eyes-not.have scales-cut be blind (lit. ‘not have eyes’) ‘cut scales off a fish’ táa-se grass-gather dije-kö eggs-put ‘pull up weeds’ ‘lay eggs’ Objects o f the verb which are categorized a s specific include personal and dem on­ strative pronouns, proper nam es and other non-general expressions. These expres­ sions in many o f the languages require the specificity m arker -re. F orexam ple, there would be a distinction between visiting one’s father as opposed to the norm al activ­ ity o f ‘father-visiting’. In the form er instance, -re would be suffixed to ‘father’, whereas in the latter it would not. Yiiniti (24) (Kinch and K inch 1992; 4) k îh à bie-re 3pl th in e s -sPEC k îh à -re 3 p l-sp ec k f h t biere k ih à rë d îd îw â ‘They gave the things to them .’ dïdi-w a give-EViD:PAST.visuAL 7 Tucano 3.5 221 Nouns The plural suffix for inanim ate nouns in the Eastern languages is -ri, and in the Central an d Western languages it is -a or -bä. N um ber-gender suffixes for anim ate nouns are: - ( Q # f o r masculine singular, ~(C)o for fem inine singular and, in general, -al-ä, -ral-rä, or -na for plural, although there are a num ber o f irregular plural suffixes for anim ate nouns. C2 does not distinguish between singular and plural for non-hum an/inanim ate nouns (Strom p.c.). The shape o f the num ber-gender suffixes is related to the personal pronouns. There is a set o f anim ate nouns in which the m em bers tend to occur in groups, such as gnats and bees. In the Eastern lan­ guages the base form o f these nouns is plural. To refer to ju st one gnat o r bee, a singularizer suffix is needed. T h a t suffix in some o f the languages is - b i and in others it is - w t. Inanim ate nouns which refer to entities that m ay occur singularly or in bunches, such as potatoes, firewood, needles, etc., have a general form that does not indicate quantity. To indicate ‘one potato’ or ‘several needles’, either the singular or the plural classifier appropriate to that noun is suffixed to the general form. When indicating a specific quantity (generally one to five items), the plural o f the noun or classifier is n o t used until referring to four or more items. The locative in all but four o f the Eastern languages is -pi. E l l and E l 2 substi­ tute /h/ where the o ther Eastern languages have /p/, so th a t their locative is - h i . The locative in E6 and E7 is -ge. The C entral and Western languages have between one and three locatives, which are not interchangeable. In the following example, note th at the locative may co-occur with the specificity marker. Yuruti (25) (K in c h a n d de K inch 1992: 13) w i-pi-re dî-wi house-LOC-SPEC be/live-EViD:i>AST.visuAL w iptre diw î 'W e lived in the longhouse.’ 3.6 Adjectives The class o f adjectives in Tucano languages includes such categories as d em onstra­ tive adjectives, num erals a nd som e quantifiers. Except in C2 which has descriptive adjectives, descriptive terms are generally stative verbs, though some descriptive terms, such as ‘old’ and 'young’, take the same suffixes as nouns. For example, in Tuyuca, the following descriptive terms are verbs: good, bad, heavy, wide, tall, deep, sm ooth, sharp, big, difficult, etc., and the colours - dark, light, nature (greee-blue). 222 Janet Barnes contrast (red-orange). These verbs do not take the full range o f v e rb suffixes, b u t do function as verbs, as the following examples illustrate. Tiiyuca (26a) d ïk î-ri-k u b e.h e av y -N E G -E VID: PRES. ASSUMED d ï k îr ik u ‘It is not heavy.’ (26b) basoki deb5-ro îb îà -k i p e rso n be.bigger.than-A D V z be.tail-EVtD:PRES.ASsuMED basoki dêbôrô îb îâ k i ‘H e (a gorilla) is taller than a person.’ 3.7 Negation Verbs are negated by m eans o f a negative suffix which precedes the evidential (or person-tense) ending. N ouns can be negated by suffixing a different negative m or­ pheme to them. There are generally two negative verbs; ‘to not have’ and ‘to not be’. Concepts such as ‘never’, ‘no one’and ‘nothing’ are encoded w ithout the use of spe­ cific words for these concepts (Barnes 1994: 337-9). N either is there a negative response word.^ Rather, the negative response to ‘Will you go?’ is a full clause ‘I will n o t go.’ 3.8 Nominalized verbs N ominalized verbs in the Tucano languages are used where other languages use rel­ ative clauses.* N ote the following examples which are inflected for tense and mood: Bamsano (27) (Jones and Jones 1991:43) bue-go buego Study-SG.FEM ‘she w h o s tu d ie s ’ * The Western language Koreguaje and the Central languages do have a negative response word. In Koreguaje it is päam i The negative response words in the Central languages are similar to negative response words in one or more Arawak languages. In Cubeo it is bi, and in Tanimuca/Retuarà it is herd?d. ‘ Tanimoca/Reluara has both relative clauses and nominalized verbs. 7 Tucano 223 bue-ka-ko 3.9 study-PAST-SG.FEM buekako ‘she w h o s tu d ie d ’ bue-ro-ko bueroko Study-FUT-SG.FEM ‘sh e w h o will s tu d y ’ Switch-reference In the Eastern languages, if the subject o f a dependent clause is the sam e as the subject o f th e clause following it, there is a set o f suffixes, often in d ic a tin g person and number, th a t is used o n the verb o f the dependent clause. However, there is on ly one suffix to indicate that the subject o f the dependent clause is different from the subject o f the following clause. The set o f same-subject markers, in many o f the lan­ guages, is identical to the set o f nominalizers. In other languages the sets differ in accent placem ent only. If, in a dependent clause, no suffix is added to the verb stem, it is understood that the subject o f that clause is the sam e as the subject o f the fol­ lowing clause. This type o f construction is typically used when describing a series o f events th a t took place one right after another. (F or a discussion o f a typical Eastern Tucano switch-reference system, see Longacre 1983.) Different subject in Wanano (28) (Longacre 1983; 202) to th is w a ? a -tj'i tid â th u a ta s i w hen.going-D S th ey w o n ’t (r e tu m ) to w a ? a tji tidâ thuatasi ‘When he goes, they won’t return.’ Sam e subject in Wanano (29) (Longacre 1983: 202) tiro w a?a-ro thuatasi 3 m asc w h en .goin g-ss w on’t(retum ) tiro w a?aro thuatasi ‘When he goes, he won’t return.’ Different subject in Tuyuca (30) b îî 2SG wáa-ri gO-DS w áa-ida gO-SG.MASC.FUT b î î w áari, w áaida ‘If you go, I will go.’ 224 Janet Barnes Sam e subject in Tuyuca (31) k îl àjû-ro 3SG.MASC gOod-ADVZ k â d î-g i uti-ri-ki sleep-SS cry-NEG-EVID;PRES.ASSUMED k îi i j u r ö k àd ïg l, utíriki ‘If he sleeps well, he does not cry.’ The Western north languages utilize same-subject and different-subject markers between sentences. Typically, the verb o f the preceding sentence is repeated, and suffix(es) are added to indicate same subject o r different subject. (F or examples, see Wheeler 1987: 179-80.) 4 Syntax The Tucano languages are typically AOV, SV languages, with variations in word order d ue to discourse considerations. In some o f the languages, if the actor is an anim ate noun, w hether expressed o r understood, the corresponding pronoun is m andatory just prior to the verb. Thus, these languages have a large percentage of sentences that are OAV, with the A argum ent being a pronoun. T he two Eastern south languages, Barasano and M acuna, are OV’ and OVA respectively, and the Western language Koreguaje is VAO. Typically the object is unexpressed if understood, so that it is com m on to hear a com m and such as ‘Close!’, rather than ‘Close it!’Also, statements such as: ‘I already to ld ’, rather than ‘I already told it to her’ are com m on. Sentences with three partic­ ipants explicitly stated, such as actor, direct object and indirect object, are not com ­ monly heard. When they do occur, the direct object precedes the indirect object (see example 24). In some o f the Eastern languages, such as Tuyuca, utterances generally involve m any short sentences, w ith a marked lack o f hypotaxis. Others, such as Taiwano, tend to have sentences with long series o f dependent clauses. Typically am ong speakers o f the Tucano languages, listeners will respond to a speaker by uttering an affirming ‘H m m ’, repeating the final verb, o r making a ques­ tion o f the final verb, as if to say, *Is that so?’ 4.1 Time and location There is flexibility within the sentence regarding the order of words or phrases denoting time and location. In general, expressions o f time precede the verb and Barasano is described in Jones and Jones (1991:2) as ‘OV, with the subject generally foHowing the verb’. One o f the authors, through personal communication, has indicated that further studies may indicate that the language is AOV. Göm ez-Imbert (1988; 99) considers Barasana to be OVA. 7 Tucano 225 may follow a subject pronoun. Expressions o f location generally follow expressions o f time, and often follow the verb. Expressions o f location may also signal para­ graph boundaries (Waltz 1975: 50). 4.2 Noun phrase N oun phrases are no t com monly used in m ost Tucano languages. If, in a given sen­ tence, the context makes it clear that a canoe is being discussed, then to state th a t it is big requires only that the classifier referring to canoes be suffixed to the word ‘big’. Phrases such as ‘My big, old, beat-up canoe’ are n o t often heard in most o f the lan­ guages. Generally ju st one descriptive word will occur in a given sentence. But if all o f the descriptive concepts are new to the context, and necessary for full under­ standing, then each descriptive term will occur with the appropriate classifier attached to it. The following example is taken from a text about a DC-3 aeroplane th at landed upriver. The ‘hollow’ classifier, which refers to the aeroplane, is also that used for canoes and reed-like plants which grow w ith hollow centres. Tuyuca (32) tii-w i pai-ri-w i-bëdà t h a t - CL:HOLLOW w e k i-a big-SG .N M LZR -CL:HOLLOW-lNST p ia - r â Cattle-PL twO-PL.ANIM tik o k o - w a send-EVID-.PAST.VISUAL tiiw f pairiw ibëdâ w ek iâ p iarâ tik ák o w a ‘Two bulls were sent in (by means of) that big plane.’ B IB L IO G R A P H Y Aikhenvald. A. Y. 1994. ‘Classifiers in Tariana’, Anthropological Linguistics, 36.407-65. Barnes, J. 1984. ‘Evidentials in the Tuyuca verb’, UA L, 50.255-71. 1990. ‘Classifiers in Tuyuca’, pp. 273-92 o f Amazonian linguistics: studies in Lowland Soutii American languages, ed. D. L. Payne. Austin: University o f Texas Press. 1994, ‘Tuyuca’, pp. 325-42 o f Typological studies in negation. Typological Studies in Language, 29, ed. P. Kahrel and R. Van Den Berg. Amsterdam: John Benjamins. 1996. ‘A utosegments with three-way lexical contrasts in Tuyuca’, UAL, 62.31-58, Cook, D. M . and Levinsohn, S, H. 1985. ‘Coregiiaje: D om ains o f Focus Markers’, pp. 91-116 o f From phonology to discourse; studies in six Colombian languages, ed. R. M. Brend. Dallas: SIL. Cook, D. M. and Criswell, L. L. 1993. El idioma koreguaje (Tucano Occidental). Colombia: Asociacion Institute Lingiifstico de Verano. Gôm ez-lm bert, E. 1988. ‘Construcciôn verbal en barasana y tBt\iyo', Amerindia, 13.97-108. 1995. (ms.) ‘Extrametricality and postlcxical rules in Barasana (Eastern Tucanoan)’. (Paper presented at the summer meeting of Society for the Study o f the Indigenous Languages of the Americas.) 226 Janet Barnes Huber, R .Q . and Reed, R. B, 1992. Vocabttlario comparatiw:palabras selectaa de knguas indigenas de Colombia. 1st edition. Colombia: Asociaciön Institute Lingiiistico de Verano. Johnson, O. E. and Levinsohn.S. H. Í990. G ram átkasecoya, Cuadernos EtnoHngiiisticos, 11. Ecuador; Institute Lingiiistico de Verano. Jones, W- and Jones, P. 1991. Barasano syntax. Studies in the Languages o f Colombia, 2. Dallas: SIL and University o f Texas at Arlington. Kinch, R. A, and de Kinch, P. 1992. (ms.) ‘El y u ru tf, Esludio prelimimr para el atlas etnolingmsfico colombiano. Colombia; Instituto Caro y Cuervo. Longacre, R. E. 1983. ‘Switch-reference systems in two distinct linguistic areas; Wojokeso (Papua New Guinea) and G uanano (Northern South America)’, pp. 185-207 o f Switchreference and universal grammar, ed. J. Haiman and P. M unro. Amsterdam: John Benjamins. Malone, T. 1987, ‘Proto-Tucanoan and Tucanoan genetic relationships’, Colombia; Instituto Lingiiistico de Verano. Ms. 1988. T h e origin and development of Tuyuca evidentials’, UAL, 54.119^0. Metzger, R. G. 1981. Gramática popular del carapana. Bogota; Institisto Lingiiistico de Verano. 1998. ‘The morpheme KA- o f Carapana (Tucanoan)* (online) available at http://www.sil. org/silewp. Miller, M. 1999. D esam grammar. Studies in the Languages o f Colombia, 6 . Dallas: SIL and University o f Texas at Arlington. Saiser, J. K., Morse, N. and Ferguson, X 1986. ‘Cubeo workpaper’. Colombia; Instituto Lingiiistico de Verano. Ms. Saîser, S. K., Jr. 1971. ‘Cubeo phonemics’, Linguistics, 75.74- 9. Smothermon, 1 R., Sm othermon, J. H. with Frank, P. S. 1995. Bosquejo delmacuna: aspectos de la ciihtira material de los macunas; fonologta; gramática. Colombia; Asociacion Instituto Lingiiistico de Verano. Sorensen. A. P., Jr. 1967. ‘Multilingualism in the Northwest Amazon’, American Anthropologist, 69.670-84. Strom, C. 1992. Retuarä syntax. Studies in the Languages o f Colombia, 3. Dallas: SIL and University o f Texas at Arlington. Velie, D. and Velie, V. 1981. Vocabulario oreján. Serie Lingüistica Peruana, 16, Peru: Instituto Lingiiistico de Verano. Velie Gable, D. 1975. ‘Bosquejo de la fonologia y gramática del idioma orejon (coto)’, Datos Etno-Lingiiisticos, 10. Peru: Instituto Lingiiistico de Verano. Waltz, N. E. and Waltz, C. 1972. ‘Fonologia del guanano’, Sistemas fonolögicos de idiomas colombiunos, 1.29-40. (Columbia: Instituto Lingiiistico de Verano.) Waltz. N. E. 1975, ‘Discourse functions o f G uanano sentence and paragraph’, pp. 21-145 of Studies ill indigenous languages o f Colombia. Panama, and Ecuador, vol. I: Discourse grammar, ed. R, E. Longacre and F. Woods. Dallas: SIL. 1997. ‘A comparison o f W anano and Piratapuyo’. Ms. Waltz, N. E. and Wheeler, A. 1972. ‘Proto Tucanoan’, pp. 119-49 of Comparative studies in Amerindian languages, ed. E. Matteson et al., Janua Linguarum, Series Practica, no. 127. The Hague. M outon. West, B. 1977, ‘Results o f a Tucanoan syntax questionnaire pilot study’, pp. 339-74 o f Studies in indigenous languages o f Colombia, Panama, and Ecuador, vol. HI; Discourse grammar, ed. R. E. Longacre and F. Woods. Dallas: SIL. 1980. Gramáticapopular del tucano. Colombia: Instituto Lingiiistico de Verano. Wheeler, A. 1987. Gantéya Bain: el pueblo siona del rio Putumayo. Colombia. Tomo I. Etnologia, granmtica, textos. Bogota; Instituto Lingiiistico de Verano. 8 Pano E U G E N E E. LO O S 1 in t r o d u c t io n ' Languages o f the Pano family are spoken on the eastern side of the Andes in Peru and nearby regions o f Brazil and Bolivia. They were first recognized to constitute a family, on the basis o f the com parison o f seven languages, by D e la G rasserie {1890). He nam ed it ‘la famille linguistique P ano’ after one o f the languages, called P ano o r W ariapano. T he languages show close similarities, indicating a fairly shallow tim e-depth and recent expansion and split. It has been suggested that the P ano family may be genetically related to the Tacana family, which is discussed in chapter 13 (there is phonological com parison and reconstruction in Key 1968 and G irard 1971). T here are some lexical and some gram m atical similarities, e.g. 2sg pronoun mi- and the forms o f switch-reference markers. However, more work is required to determ ine whether these similarities arc sure indicators o f genetic connection, rather than being due to areal diffusion. Shell (1965/1975) undertook a systematic com parison and reconstruction o f proto-P ano phonology and aspects o f the morphology. She listed more than eighty ‘language’ nam es found in the literature but som e o f these are names o f dialects or clans within a language group, or alternative nam es o f languages. The num ber of distinct languages is probably no m ore than thirty, as shown in table 8.1; some o f these may tu rn out, on further study, to be dialects o f a single language. M any different subgroupings have been suggested in the literature. In table 8 .1 ,1 recognize three subgroups and a num ber o f ungrouped languages; the subgroupings are made, tentatively, on the basis o f shared phonological and m orphological charac­ teristics an d - to a limited extent - shared vocabulary. N ote that several o f the nam es end in -nawa ‘people’ and som e others in the plural m arker -bo. Kensinger (1985) provides a useful overview o f Pano groups w ith bibliography. I owe special thanks to Ivagene Shive for Yaminawa, Eugene and M arie Scott for Sharanawa, Kim Fowler for Yoranawa, Harriet Fields for Matses, and Wayne Gill o f NTMBolivia lor Chimane. 228 Eugene E. Loos One often finds m any lexical roots shared between two languages, although the foffin (5f tlie^uffixes vary, leading to initial lack o f intelligibility. However, this can be overcome through a degree ofeep®9^^K•lí^»B'ôthel^afl■g»age,'Aflep«-few#»^f adjustm ents to differences, a Yaminawa or a Sharanaw a can understand most o f a give-and-take exchange w ith a Yoranawa but gets lost in flowing discourse. The Sharanaw a are reported to be able to understand much o f the Yawanawa. ¥/hen Cashinawas converse with Sharanaw as they are known to resort to a kind o f pidginP ano by suppressing the use o f m ost suffixes, especially those n o t shared between them. N ote that, unless otherw ise indicated, all examples in this chapter are from Capanaw a, the language w ith which I am m ost familiar. Also, allom orphic forms are used in examples w ithout identifying the basic form o f the m orphem e each time. 8 Pam 229 Table 8.1 Languages o f the Pano fam ily The Yaminawa subgroup 1 Yaminawa 500 P, Br 2 Araawaca 2CK) P 3 Cashinawa/Honikoin 500 P, Br 4 Sharanawa/Shanindawa/Chandinawa/lnonawa/M arinawa 300 P 5 Yawanawa 200 Br 6 Chitonawa 35 P 7 Yoranawa/Nawa/Parquenawa 200 P 8 M oronawa 300 Br 9 Mastanawa 100 P The Chacobo subgroup to Chacobo 400 Bo 11 t Arazaire P 12 tA tsa w a c a P 13 t Yamiaka P 14 Katiikina/Camannawa/W aninnawa 300 Br 15 Pacawara 12 Bo The Capanawa subgroup 16 Capanawa/Pahenbakebo 400 P 17 Shipibo/Conibo/Xetebo 8,000 P 18 t Remo Br 19 M arubo400B r 20 t W ariapano/Panobo/Pano P 21 Isconaw a30P 22 t Canamari/Taven'/M atoînahâ Br Ungrouped languages 23 Cashibo/Cacataibo/Comabo lOOP 24 t Cultno Br 25 t K aripuná Br 26 Kaxariri 100 Br 27 M atses/M ayoruna 2,000 P, Br 28 t N okam án Br 29 t Poyanáwa Br 30 t Tutxinawa Br Note: The approximate num ber o f speakers is given lor each language; t indicates a language that is probably no longer spoken. Locations are indicated by P, for Peru; Br, for Brazil; Bo, for Bolivia. (For approximate geographical locations o f each language see map 7.) N ote that in Spanish ‘hu’ would be substituted for ‘w' in these names. 230 Eugene £1 L oos Table 8.2 Proto-Pano consonants voiceless stop voiceless affricate voiceless continuant voiced continuant bilabial alveolar palatal p t ts s r p 2 PHONOLOGY 2,1 Segmental phonology apico-palatal dorso-veUir tj J There are four proto-Pano vowel phonemes: low a, high front unrounded i, high back unrounded i and high back open unrounded o. All have been retained with a high degree o f consistency. M atses now also has a mid unrounded vowel e; in C ashibo the at sequence has developed into e and the sequence aw becomes o. O f the sixteen proto-P ano consonants, the stops and affricates have m ost consis­ tent reflexes in daughter languages. T hough the voiced bilabial fricative is n o t a resonant in phonetic terms, it is here classed with r it because it patterns like r in not being able to serve as the first member o f a consonant cluster and like r it blocks some syllable reduction rules. Possibly P was not a fricative in proto-Pano; it is realized as a voiceless bilabial spirant in the languages o f the Yamiiiawa group. The glottals have been lost in most daughter languages. Both ? a n d h are retained in Amawaca, Chacobo, C apanaw a and Pacanawa, just li in Yawanawa, Shipibo.and W ariapano, and just ?in Cam annawa, The reflexes o f the nasal consonants range from nasals to oral plosives in the Yaminawa group. Sharanaw a and Cashinawa have lost the distinction b e tw e e n /a n d g in some envi­ ronm ents. In Sharanaw a, Yaminawa, Chitonaw a and Yoranawa w, like tends to be realized as a voiceless bilabial spirant. Some Pano languages have been reported to have distinctive tones. However, in the cases that have been closely analysed, tone is predictable from the underlying syllable structure. T hus in C apanaw a the underlying forms /nanewe/ 'p u t on board* and /nanet-w e/ ‘get aboard’ have surface forms [nánéwé] and [nàiiéwé] respectively (where ' indicates high tone and ’ indicates low tone). 231 8 Pano 2.2 (i) Common phonological variations N asal spread Syllable-final m and n reduce to leftward-spreading nasalization on any preceding contiguous string of vowels and semivowels, and when the nasal consonant is followed by a glide (ii>, v. A, ?) the nasalization also spreads rightwards onto contiguous vocalic and semi-vocalic segments. This nasalization rule is missing in Chacobo and to a large extent in Yaminawa, Sharanaw a and Yoranawa where m and n tend to be denasalized, becoming plosives [b] and [dj except when there is a following nasal vowel in the word. Yaminawa (la ) [ado] (b) [ani] ano ana-n ‘m ajás’ (edible rodent, O form) ‘m ajas’ (edible rodent, A form) m a já s - E R G (2a) O ral vowels: h aw ina (b) Leftward spread: (c) Rightward spread: [h iw i foP o] [wiràwî] hawin gopo w iran-w i pUSh-lMPER ‘his (own)’ ‘his house’ ‘Push it over!’ Unlike the continuants s, / , §, in surface structure m and n do not occur in syllable-final position and hence do not serve as the first m em ber o f a consonant cluster e,\cept In Matses. However, nasalized vowels preceding obstruents typically show a transitional phonetic closure which produces an apparent epenthetic [n] or [m] which som e investigators have taken as a neutralization o fn or m. Examples like the following show the seeming displacem ent because o f the pre-consonantal closure o f the oral passage: (3a) [hoJTâijkî?] [k a tim p i? ] k a -ta n -p a n re d -m a k e -c o N J g o - D iS T - s h a ll ‘m a k e s (ii) (b) h o jin -h a -k in it red’ ‘1 s h a l l go away.’ Vowels are elided in a weak position such as in a stem-final unaccented open third syllable. (4a) [nonoraâ] nonom a-n (b) [non^] nonom duck-ERG ‘duck’ (A form ) ‘duck’ (O form) 232 (iii) Eugene £ Loos Syllable-final stops p, t, k, rn, n are elided; n ote that m and n function as stops b u t P and r function as resonants. Chacobo (5) k ap iti > k ap it > [kapi] ‘alligator’ (O form) (iv) h is elided word-internally. The phonem e h occurs only morpheme-initially but morphem e concatenation can m ake it follow an o ther segment word-internally in the underlying form (where it can then be the cause o f the deletion o f a preceding nasal consonant before it is subse­ quently suppressed). (6 a) [h im iiw i] (b ) [h a in i] h im i-h iw i (v) h a -h in i b lo o d -tre e m o u th -w a te r ‘b l o o d w o o d ’ ‘s a l i v a ’ Vowel harm ony In some contexts an a vowel fully assimilates to the vowel o f the preceding syllable. We get assimilation in (7b) but not in (7a). (7a) [ti? k a k l] (b ) [t» ? k k i] ti? k a -k in tiîk a -t-i b re a k -d id ‘He broke b r e a k -R E F L - P R E S it.’ ‘I t b re a k s .’ In the Yamiiiawa group, i assimilates to an i vowel o f the following syllable. Yaminawa (8) [pitsikl] pi-tsikin shoulder-push ‘to push on the shoulder’ (vi) A lternate m ora timing An odd-even syllable-timing characteristic com mon in Pano languages cau.ses phonological modifications such as segment deletion, plosive nasal release, stress assignment and possibly vowel harmony. — In Shipibo the morphem e -ríjäa- changes to -r/jS/- when the syllable j8n is even-numbered from the beginning of the word: 233 8 Pano Shipibo (9a) — (b) ka-riPa-i lo o k - a g a in - iM P E R g o - a g a in - P R E S ‘L ook for it ‘I a g a in î’ a m g o in g a g a in .’ A syllable-final glottal is elided in even-num bered syllables, as in ( I Ob): (10a) — p*na-ri?pi-wi pina-ri?pi-w i (b) pi?-ripi-wi lo o k - a g a in - iM P E R ta k e -a g a io -iM P E R ‘L o o k ‘Take f o r it a g a in ! ’ it a g a in !’ Stress assignm ent often depends on the syllable structure o f the word, but details vary from language to language. In Capanawa, Shipibo, C anam ari and som e others o f the Capanawa subgroup, a closed second syllable is stressed. This stress assignment takes place before syllable-final consonant deletion so th a t some surface open second syl­ lables rem ain stressed: (11a) [ jis is ] (b) [ban] h is is i bari-n ‘a variety o f a n t’ ‘the sun’ (A form) su n -E R G — An invervocalic nasal consonant before an oral vowel takes on a plosive release when it begins an even-num bered syllable: Amawaca (12a) — [hondi] (b) [honî] honi honl-n ‘m an’ (O form) ‘m an’ (A form) A final odd-num bered syllable o f certain verb affixes is elided, e.g. the sufRx -paki- ‘all’ retains its final syllable if that syllable is even-num­ bered: Sharanawa (13) [yoipakitiroki] yoi-paki-tiro-kin tell-all-well-having ‘having told it all well’ — The final syllable o f the suffix is lost when the syllable is odd-num ­ bered: 234 Eugene E. Loos Sharanawa (14) [opakî] o-paki-kin come-down-having ‘having come dow n’ 3.1 Word classes and noun phrase formation The open word classes are Nouns, Verbs, Adjectives and Adverbs; Adverbs are mostly limited to tim e and location words. Pronouns, D emonstratives, Numbers, Postpositions and postpositional Particles form closed sets. N ouns and Adjectives are easily verbalized by adding verb suffixes; if no transitivizing suffix is present the m eaning is ‘to b e . , or ‘to become . , otherwise the m eaning will be ‘to make or cause . . Typically, if the verb is in present tense and is not needed to carry verbal suffixes, b o th verb and tense may be omitted. (!5) mia ?ino-wi you tiger-iMPER ‘You be the tiger!’ (instruction to a child in a game) (16) § o b o -h a -w i h o u s e -m a k e -iM P E R ‘Make (17) a h o u se !* h o jin -k aîin -i red-quickly-PRES ‘He tu rn s red.’ (18) h o jin -h a -w i re d -m a k e -iM P E R ‘M ake it red!’ (i.e. paint it red) AOV and SV are the unm arked orders; pragm atic or discourse cotiditions may change the order o f all constituents but the verb is last unless em phasized constit­ uents are postposed. The Pano languages are agglutinating, entirely suffixing and use postpositions. The num ber o f verb suffixes can exceed 130; many o f them can be concatenated on a single verb. 235 8 Pano 3.1.1 N oun phrase T he noun phrase can consist o f a h ead and a series o f modifiers. There are no def­ inite o r indefinite articles; nouns may be used in the adjective modifier slot: Q uantifier - Possessive - Adjective - H ead-noun - Adjective Num eral For example: (19) hawin gibon his p alm .leaf k a ? ti back veins pono rom ijin small ‘the tittle veins o f the back side o f his palm leaves’ Possessives are formed by appending a possessive suffix to a noun or pronoun. Som e adjectives are by preference preposed to the head, others are postposed, and som e may occur in either position. W hen preposed, adjectives are subject to the phonological rules that shorten them. Preposed adjectives have a delimiting func­ tion: (20a) hoa p ajin i io w e r red ‘the reddish flower’ (b) p a jl hoa red flower ‘a flower that is red’ Two nouns may be com pounded, in which case the com bination is subject to the shortening rules. (21) [maribï] mari-bina aguti-wasp ‘a black w asp’ (a wasp that is black like an aguti) 3 .1 .2 P ro n o u n classe s Any constituent may be fronted to put it in focus, though if a verb is fronted there may be an auxillaÍ 7 verb in finiU pcStion to carry the tense. Pronouns occur in free form for focused A and S, non-focused A an d §, and O functions. B o u n d ^ rm s are used when an oblique case-m arking suffix such as indirect object is added. N ot all P ano languages show the focus distinction but it is found in each o f the major groups. (22) ?ian ‘1’ (free form) ?i?-ki ‘m e-to’ (bound form) 236 Eugene E. Loos 3.1.3 R elative clau ses The head noun is shifted out o f a relative clause o r om itted altogether. (23) ?ani hiwi mebi b ig ‘A 3 .1 .4 tr e e b ran ch te jp a t-a i te ip a h a n in ? is o b ifu rc a te -P R E S fo rk loc m onkey m o n k e y is h i d i n g large tree.’ in th e f o r k o f a b r a n c h t h a t b if u r c a t e s h o n it-i h id e -P R E S fro m a N o u n p h r a s e p lu ra liz a tio n Plural is usually marked only on the verb rather than on an N P in A or S function; the plural on NPs in o th er functions is marked only when it is essential for disam­ biguation. The plural m arker for nominals is -bo, hom ophonous in m any instances with the generic m arker -bo: honi-bo (m an-pl) ‘men’ or (m an-G E N E R IC ) ‘m ankind’. (The suffixes and postpositions are appended to the NP, not just to the head noun.) (24) ?oá there tsa7ot-al'bo his4 s i t- P R E S - p l s e e -iM P E R ‘Look at those (who are) sitting over there!’ 3.2 Transitivity concord All Pano languages are characterized by a distinctive transitivity concord system. Transitivity concord m arking is interrelated with the systems for identifying verb transitivity, switch-reference, sequential action, adverbs, relative clauses, temporal clauses and locative expressions. The usual conditions are: (a) Transitivity is semantically determined: a verb phrase istransitive when it has an O referent that is different from the A referent. The ref­ erent-identifying noun o r pronoun need not be overtly expressed. (25a) (25b) (b) haa-n ta his-i-ki 3-A D ECL See-P R E S -F A C T ‘H e sees haa ta 3S ‘H e D EC L sees [it].’ his-it-i?-ki S e e -R E F L -P R E S -F A C T himself (in a m irror).’ P ano languages show a variety o f split systems, the m arking o f the A and S being affected by focus. 237 8 Pam (c) Subordinate clauses in which the action is semantically prior in time to their main clauses are marked according to the transitivity o f the main clause. (d) Conjoined (sequential clause) verbs are marked by the final verb suffix as transitive (-k î) or intransitive {-/). (e) Some adverbial verb suflixes have selectional variants that m ust match the transitive or intransitive valence o f the clause. (f) Locative phrases that indicate the location o f the A or S require m arking to agree w ith the transitivity o f the verb o f the clause; loca­ tive phrases th a t indicate the location o f the O take no m arking. (g) Reflexive and reciprocal verb constructions are autom atically intransi­ tive because the referents o f A and O are identical. (h) The roots o f com pound verb stems m ust agree in transitivity. If the first is reflexive, the second m ust also be. Thus, transitive maya- ‘to tu rn ’ and transitive paki- ‘to fell’ can both be made intransitive by the reflexive suffixes and com bined to form a com pound verb stem: (26a) m aya-paki-w i tu m - f e ll- iM P E R ‘K nock it down spinning!’ (26b) m ayápakitipijki m aya-t-paki-t-ipi-J-ki tum-REFL-fell-REFL-PAST2-3-FACT ‘H e fell spinning.’ 3.2.1 Switch-reference Pano switch-reference is m arked by suffixes on subordinate verbs that also indicate the tim e o f the subordinate verb relative to the m ain verb. A subordinate verb may precede o r follow its m ain verb. Table 8.3 shows the Capanawa suffixes. There are corresponding forms in m ost other Pano languages. (Shipibo forms are given in L oriot, L auriault and D ay 1993; 54-7.) T h e following sentences exemplify these sufTi.xes: (27) hiwi ?ani bago kiyani bimi t/a?o ?-a§ nam an tree big b a jo tall fruit soft.become-SW a below m i-raka?t-ai-ton ska honon m ira-io n pi-kin earth.on-lie-PRES-SWc then peccary flnd-SW b eat-coNJ ‘The bago tree is huge, very tall. W hen the fruit becomes soft and falls to the ground the peccary finds it and eats it.’ 238 Eugene E. Loos Table 8.3 Switch-reference markers in Capanawa function o f coreferential argument in: verbal siiflix subordinate clause SWa SWb SWc SVVd SWe SW f -a? -jon -ton -a? -no§on -a?bo S A/S SWg SWh -ya -non (28) A/s A/S (must be plural) 1 action of subordinate clause in relation to action of m ain clause S A 0 A/S A/S O before or same time before or same time same time before follows follows A/S differ before o r same time follows | m ai k in ti ki k o b in h a -? a bi ki g in is k a -a j bi?m iki?t-i c la y pot in cook-SWd but rep old.get-SW a c ra c k s-P R E S ‘When (29) A/S O/IO main clause a clay p o t that a n y o n e c o o k s in ?otJ\5 ka?-ipi-j fa r is old, the p o t c ra c k s .’ ho-ribi-?i-ton g o in g -P A S T -S W a c o m e -a g a in -rR E S -S W c ?a?-kin h o -rib i-? i-n ? i? ti h in im a ? -i m e t-c o N J v e ry h ap p y -c o N j min you hawi-ya come-SW g c o n ie - a g a in - P R E S - iN T E R R O O s a y - c o N j b i? t/i-k in b i? tji-k in m e e t-c o N J h in im a -jo n rejoice-SWb ‘H e h a d g o n e a g r e a t d i s t a n c e a n d u p o n h i s r e t u r n s h e m e t h i m w i t h th e w o rd s “ H a v e y o u re tu r n e d ? ” B e c a u se h e c a m e , s h e re jo ic e d a n d w e n t o u t to m e e t h im . S o (30) nonti ?aniha-gon canoe enlarge-SW b she w a s v e ry h a p p y .’ miin pitibo.ri?bi banaw i your food.also plant hano there napo-ya?-no§on pile-FUT-SWe Ä f t e r e n l a r g i n g y o u r c a n o e , p l a n t y o u r f o o d a l s o i n o r d e r t o p i l e i t in th e c a n o e la te r.’ (31) yo?a kobinha-§on bim akanya m i?a pa§a-?i-ton yucca cook-SW b of raw-be-SWc on.the.surface birab4-kin ra m i? o ska turning. Over-CONJ other-side then kobin-non cook-SWli 239 8 Pano 'W hen she cooks yucca (a succulent plant), she turns over the ones on the surface that are raw so th a t they will cook on the other sid e.’ (32) hati?ibi hato yo?i-w i honan-a?bo all them tell-iMPER know -SW f ‘Tell all o f them so th a t they will know.’ 3.2.2 Adverbial suffix concord Some adverbial verb s u ffix e s have pairs of selectional variants according to the tran­ sitive or intransitive valence of the verb; the adverbial suffix must agree with the tran­ sitivity o f the verb. For example, in Capanawa the variants for ‘abruptly, rapidly, completely’ are -kaîin- (intransitive) and -ba?in- (transitive) and the variants for ‘movement towards the speaker' a r e -ki?ran- (intransitive) and -be?nan- (transitive). (33a) m api-ka?in-i go. up-rapidly(i n tr )- conj ‘H e goes up (the bank) rapidly.’ (33b) bi?-ba?in-kin g ra b -ra p id ly (T R )-C O N J ‘H e g r a b s it a b r u p t l y ’ (33c) w ao?i?-kiran-i haw-i cry -co m in g (iN T R )-C O N J c o m e -c O N J ‘He cam e c r y i n g .’ (33d) his-binan-kin bi see-com ing(T R )-coN i but m t?i-yam a-k in touch-NEG-coNJ ‘He saw it as he cam e but didn’t touch it.’ 3.2.3 Locative phrases Locative phrases in sentences w ithout a verb and locative phrases that relate to the O take n o transitivity m arking. Locatíve phrases relating to the source o r position o f the A carry transitive m arking; those that modify S take intransitive m arking. (34a) h an o there ta? decl ki fact ‘There he is.’ (no verb present) (34b) hano there ?ano his-kin m ajás see-cONJ ‘He saw the m ajas (rodent) there.’ (the locative modifies the O) 240 (34c) Eugene E. Loos hano?-ag haw-i there-LOC(iNTR) com e-coNJ ‘He cam e from there.’ (34d) h an o -g o n his-kin th e re -L o c (T R ) see-coNJ ‘He saw it from there.’ (lit. ‘He from there saw it at the other place.’) 3.2 .4 E rg a tiv e m a r k in g ( i) F orm In m ost P ano languages, any syllable-final nasal consonant is realized as nasaliza­ tion on the preceding contiguous vowels and semi-vowels. Hence nasalization is the only clue to the presence o f the ergative m arker -n, but in some o f the languages the nasalization has been lost. In Chacobo and Amawaca, if the verb is in present tense the A or S is autom at­ ically postposed to follow the verb where it is n o t m arked as A , and is then identi­ cal to its ‘citation form’ (i.e. as an answer to ‘W hat is that?’) which is not the same as the O form. Chacobo (35) inaka ‘dog’ (citation ina ‘dog’ (O form) form) Syllable reduction has reduced the word for ‘dog’ above to two syllables. Switching the order of the A and O m akes no difference to the marking: Chacobo (36a) kam anô ina pi-ki tiger dog eat-PAST ‘The tiger ate th e d o g .’ (b) ina kam and dog tiger pi-kieat-PAST ‘The tiger ate th e d o g .’ A t first glance it appears that the reduction o f the final syllable o f the O noun is an O -m arking device, but it is rather simply pre-posing the noun to a co-constituenE that causes the reduction o f the final syllable, as shown when the noun is preposed to an adjective or postposition: Chacobo (37) ina pistia dog small 'sm all dog’ 241 8 P am It is possible that stress, which in Chacobo has not been satisfactorily analysed, is a residue o f A-marking. It h as com m only been observed that in many P ano languages the ergative marker -n an d the markers for possessive, locative and instrum ental are hom ophonous, but in the underlying forms they are different. C hacobo clearly reveals some o f the differences th at are still evident in other P ano languages only in pronom inal and dem onstrative forms. For example, the locative suffix is -no: Chacobo (38a) kô-ki ba?i-no go-PRES path-on ‘He is going on the p ath.’ (b) kô-ki go-PRES jo bo-no house-in ‘He is going hom e.’ M oving the noun with its locative suffix into preverbal position causes the loss o f the final consonant and vowel from ba?i-no ‘on the path’. Chacobo (39) ba?i kö-kiy-a path-(on) go-PROG-lsg ‘(I) am goin g o n the path.’ Similarly in C apanawa the locative -no is preserved when attached to single-syllable forms like the dem onstrative ha- ‘th a t’: ha-no (3-LOC) ‘there’; but when the loca­ tive is affixed to a noun the reduction rules reduce -no to nasalization: gobo-no becomes ÿobô. Likewise the instrum ental -?an and the possessive -na are reduced to nasalization except when attached to a single-syllable form o r when not preposed to the head noun, as in hawina ‘his’. (ii) Function There are a num ber o f kinds o f ergative system in P ano languages. (a) Some, such as W ariapano, have a split system, with nouns having S and O m arked in the same way (absolutive case) and A differently (ergative), but pronouns having S an d A marked in the same way (nominative case) and O differently (accu­ sative). (b) O thers, such as Shipibo, have both noutis and pronouns showing an absolu­ tive (SO) versus ergative (A ) system. (c) C apanaw a has a third type o f organization. H ere em phatic pronouns are like nouns in having ergative suffix -n m arking A function, w ith S and O left unm arked. N on-em phatic pronouns show a different system with -n m arking both S and A functions, while O is left unm arked. The forms o f the 1st person pronoun are: 242 Eugene £ Loos em phatic non-em phatic S ,0 ?»-a O ?i7- A ?i-a-n A,S ?i-n A clause can include ju st one instance o f a non-^m phatic pronoun, as in: (40a) intransitive ra?m a ta ? ?t-n now DECL 1-SA ica-?i gO-PRES ‘I’m going now,’ (40b) transitive kapi ta? ?i-n alligator decl 1-SA h is-i See-PRES ‘I see an alligator.’ A pronoun can be repeated at the beginning of a clause to show th a t it is in focus. As long as it is non-em phatic, a nominative (marked by -«)/accusative system is still followed: for example: (41a) intransitive ?i-n ta ? 1-SA DECL 1-SA ?i-n ka-?i gO-PRES ‘I am going.’ (41b) transitive ?i-n ta ? 1-SA DECL ?i-n pi-?i I-SA eat-PRES *r am eating [it].’ However, if the focused pronoun is in em phatic form, an absolutive/ergative (marked by -ii) system o f m arking is used; for example: (42a) intransitive ?i-a ta ? ?i-n 1-EMPH + SO DECL 1-SA ka-?i gO-PRES ‘I am going.’ (42b) transitive, em phatic pronoun in A function ?i-a-n ta ? 1-e m ph -A decl ‘I am eating [it].’ ?i-n 1-SA pi-?i eat-PRES 243 8 Pam (42c) transitive, em phatic pronoun ?i-a ta I-EMPH + SO DËCL in O function ?onan-i?-ki knOW-PRK-FACT ‘H e knows me.’ Note, in (42c), that a 3rd person non-eniphatic pronoun in A o r S function has zero realization. 3.3 Noun incorporation Pano languages have no prefixes. There is, however, incorporation o f som e noun roots immediately before the verb. Typically, about forty-five nouns may be incor­ porated, m ostly body parts plus things like ‘liquid’ and ‘inside o f’. When incorpo­ rated, their m eaning may be extended, e.g. ‘face’ can now m ean ‘surface o f’. There are occasional examples o f two nouns being incorporated to a single verb. In C ashibo incorporation is no longer productive, there being ju st an archaic residue. Incorporated forms are all monosyllabic. Since they add a syllable to bisyllabic words, the words become subject to the syllable reduction rules. (43) [popàkï] po-pan-kln poyan-pano-kin arm-wash-cONj ‘to wash the arm (o f someone)’ 3.4 Verb formation 3.4.1 Verb roots Verb roots are inherently intransitive (‘to sleep’), reflexive (‘to sit dow n’), transitive (‘to ea t’) or ditransitive (‘to give’). The verb root may be com pound, form ed by jo ining two roots: (44) Jo rö p ak iti Jo ro t-p ak i-t-i floW-fall-REFL-PRES ‘It flows dow n.’ M any Pano languages have a few verbs that take an O but are marked as intransi­ tive, such as C apanaw a pa?i- and kiin- ‘to w ant’, p ifin - ‘to lack’. Intransitives can be m ade transitive by a transitivizing them atic suffix (‘to m ake’, ‘to cause’, ‘to involve w ith’); transitives can be m ade intransitive by a reflexive or reciprocal suffix. 244 Eugene £ . Loos 3.4.2 Suffixes In some languages m ore than 130 verb suflixes are available. The suffixes tend to fall into 9 groups, each group having semantic o r functional features in com m on and sharing a preferential order with respect to the other groups. Beginning closest to the verb stem, the general order is (1) thematics, (2) negation, (3) m otion, (4) aspects, (5) person and tense, (6) m ood and subordinate markers, (7) evidentiais, (8) negation, (9) connectives. Each group contains some suffixes that are semantically m utually exclusive with one another and others not mutually exclusive but which follow a preferential order within the group. The further to the right their occur­ rence in the string o f suffixes, the greater the corresponding semantic scop* regard­ ing other constituents o f the clause. G ro u p 1. Thematics G ro u p 1 suffixes com plete the thematic structure o f the verb, such as morphemes th at signal reflexive, reciprocal, causative, transitivizer, benefactive, detrim ental and dative o f interest. A reflexive is always intransitive. (45) [tam áni?tagki] ta-m a?ni-t-a-§-ki foOt-Switch-RF.FL-PASTl-3- fact ‘He took a step.’ (lit. ‘He switched his foot.’) (46) m ayáp ak itip ijk i m aya-t-paki-t-ipi-J-ki turn-REFL-fell-REFL-PAST2-3-FACT ‘H e fell spinnin g.’ (47) haa ta ? a ? -m i-b i r i? ti-t-a -g -k i 3S DECL 3-REFL-even kill-REFL-PASTl- 3 -f a c t ‘He killed h im s e lf ’ Reciprocals are m arked by the them atic suffix -nan and are always intransitive. In a reciprocal clause the S N P refers to the group o f participants, for example, (48) haa-bo 3-pl rabi two ta his-inan-i-ki DECL see-REClP-PRES-FACT ‘Both o f them are loo k in g at each other.’ (49) haatian-bi ta his-inan-ipi-kana-j-ki there-JNTENSIFIER DECL see-RECIP-PAST2-PL-3-FACT ‘Right th e re they looked at o n e another.’ 245 8 Pam G roup 2. Negatives and strong affirmatives There are two possible locations in the verb suffix string for negatives: group (2) -yama‘n o t’ and group (8) -ma ‘n o t’. The morpheme -yama- follows the thematics, may be interspersed am ong groups (3) and (4), and always precedes the tense o r terminal group. In Shipibo, Capanawa and many other languages, -yama- is used in all forms of finite and subordinate verbs, but nouns, predicate nominals, nominalized verbs and relative clauses are negated by postposing -ma to the tense or to the nom inal. In Yaminawa and Sharanawa -yama- is used only in verbs having a future or incompletive tense or aspect and in the imperative m ood, otherwise -ma is used, postposed to the tense as described above. In some languages -yama- also functions as a verb stem: (50) yam a ta ha-i-ki not DECL AUX-PRES-FACT ‘There aren’t any.’ or ‘There is nothing.’ (51) m ia ta ? ? in y a m a -h a -? ip i-k i you DECL Isg NEG-make-PAST-FACT ‘I m issed y o u .’ Camannawa (52) 5T-yama-wi See-NEG-IMPER ‘D on’t look!’ Isconawa (53a) ?iah I ka-i-m ba gO-PRES-NEG i be.PRES ‘I’m n ot g o in g .’ (53b) ka-yainba-wi gO-NEG-lMPER ‘D on’t go!’ G ro u p 3. Adverbs o f m otion such a s upw ard, dow nw ard, circular, to one side, a lo n g a n o th e r re fe ren t, inverted, direction t o o r aw ay from th e speaker, action done over a d istance. G ro u p 4. A spectual adverbs o f daytim e or night-tim e activity, quickly, desiderative, truly, intensively, first, just, next, perhaps, again. G roup 5. Person m arkers, plural m arkers, tenses, aspects of duration. G roups 5 and 6 are generally final; at least one from either group is generally obligatory on the verb though discourse constraints allow variation. 246 Eugene E. Loos G roup 6. M arkers o f subordinate verbs, nominalization, conditional, permissive, conjunctive, interrogative and imperative. The interrogative indicators vary in form and location; some attach to the verb and others are postposed to the initial m ajor constituent o f the sentence: Shipibo (54) nii-a ki ho-a 2-S INTERROG COrae-PASTl ‘Have you come?’ Capanm m (55) min haw-i-n 2S COme-PRES-lNTERROG ‘Have you come?’ G roup 7. Contains evidentials (e.g. ‘factual’, ‘reported’, ‘assum ed’) and also m ood m arkers and other morphemes that indicate intended illocutionary force, and though they attach as suffixes to the verb some may also be fronted to the begin­ ning o f the sentence where they attach to whatever is focused. In the answer part o f (56) there are two evidentials; -s- ‘visible’ and -ki ‘fact’, and the declarative m ood m arker -ta-, (56) Question: m am a ka?rà? m other w hat.about ‘Where is (your) mother?’ Answer: ? o á -s - ta h aw -i-ki there-VISIBLE-DECL COme-PRES-FACT ‘There she com es, (visibly, in fa c t).’ G roup 8. M utually exclusive with some o f group 7 and contains only the negative -ma. G roup 9. C ontains logical connectives o f various tyjjes, e.g. ‘because’, ‘since’, ‘when’, ‘o f course', ‘indeed’, ‘but’, ‘although’, ‘even though’, ‘for’. They are more loosely bound to the verb than the other groups and are often found written as inde­ pendent particles. The C apanawa connective kin ‘because’ is an example: (57) haa §obo-?o rakat-ai-kj ta ? ?i haw-ipi-ki th a t ho u se -in lie-PRES-because d ecl IS come-PAST2-FAcr ‘I c a m e because she was lying in th a t h o u se (th e h o s p ita l).’ s 247 Pam There may be m ore than one occurrence of affixes in groups ( 1H ^ ) in the sam e verb. F or example, the causative -ma- in tro d u c e an O argum ent to the verb. W hen m ote than one causative occurs, each introduces a new argum ent. In (58) the argum ents are differentiated by subscripts. (58) bi?-m a-m a-w i grasp-CAUS-CAUS-IMPER ‘(You) make (him ,) get (him ,) to get (it^).’ (59) m iin your bene husband kiin-yam a-yam a-w iwant-NEO-NEO-iMPER ‘D o n ot despise your husband!’ (lit, ‘D o n o t not-love your husbandl’) There are generally strong preferences in the order o f suffixes between groups 1 and 4, an d similarly within each group there is a preferential internal order; when the order changes there is a corresponding difference o f scope, the rightm ost suffix having the widest scope. (60) his-m a-ri?bi-w i see-CAUS-again-i m per ‘Show it to him again!’ (61) his-ribi-m a-w i see-again-CAUS-iMPER ‘M ake him take a second look!’ Within groups 5, 6 and 7 the order is fixed, though in some languages the plural m arker precedes certain tenses and follows others. Some Pano languages have four or more past tenses but the future tense set does not always mirror the past tense set in the spans o f time covered. All P ano languages seem to have a large num ber o f verb stems that term inate in -i or -a; -i verbs are intransitive, ‘be/do’; and -a verbs are transitive, ‘d o ’. The form o f these verbs must be lexicalizations o f periphrastic constructions formed on the auxiliary verbs ?i?k- ‘to be, do, say’ and ?a?k- 'to d o ’, e.g. Capanaw a fo/i- ‘to burst’, ‘to crush’. Repetition of p art o f the verb to show duration, intensity o r repetition c include some o f the suffix string (up to and including group 4)attached to the verb stem: (62) han pi-panan ta? 3S eat-w ould decl min h ini-ba?in 2 hini-ba7in-ai release-completely-PRES reiease-com pletely 248 Eugene E. Loos ‘She [the cow] would eat [the grass you offer her] but you always let go o f it (so it falls dow n).’ 3.5 M ood indicators P ano m ood indicators are part o f suffix group 7 and are either (I) indicators o f intended illocutionary force, o r (2) indicators o f speaker attitude towards the addressee o r tow ards the articulation. All modify the whole sentence and tend to occur either a t the end o f the verb affix string o r postposed to the first m ajor con­ stituent o f the sentence. 3.6 Deictics It is com mon to find that, in discourse, explicit N P reference is kept to an absolute minimum. The switch-reference system enables listeners to identify m any referents, and dem onstrative pronouns are sometimes used to point to a referent since there are no definite or indefinite articles. Pronouns may be anaphorically om itted, unless needed for disam biguation. In general th e identification system for the referents o f focus and topic is that, after a referent has been p u t in focus by fronting, it is thereafter referred to by ‘zero’ pronouns (i.e. pronom inal reference to it is om itted) but a topic referent (what the discourse is about), identified a t the beginning o f the discourse and usually n o t specified thereafter, cao be identified by a personal o r possessive p ronoun. In dialogue the dem onstrative pronouns are used to differentiate referents, for example in Capanawa: (63) n iá ‘this’ (near the speaker) ?oá ‘th a t’ (away from speaker) to á ‘th a t’ (near the addressee) haa ‘th a t’ (whatever has been referred to in context) nino ‘here’ (near the speaker) hano ‘there’ (away from the speaker) or (the place ?ono ‘there’ (away from speaker and addressee) ?ohno ‘over there’ (far from speaker and addressee) previously referred to) 8 Pam 249 BIBLIOGRAPHY Abreu, J. C. de. 1914. RS-lxalni-ni-ku-lalinguadoscaxinauâsdarioIbuacû, affluentedoMaru (prefeitura de Tarauacá). Publicacôes da Sociedade Capisiano de Abreu. 2nd edition, Rio de Janeiro. Camargo, E. 1994. ‘Les différents traitements de la personne dans la relation actancielle: l’ex­ emple du caxinaua’, Actances, 8.121-44, Créqui-M ontfort, G. de and Rivet, R 1913. 'Linguistique bolivienne. Les dialectes pano de Bolivie’, Le Mouscon, 14.19-78. d’Ans, A. M. 1970. Maleriales para el estudio del gnipo tingüistko pano. Lima, Peru: Universidad Nacional Mayor de San Marcos. 1972. ‘Repertorios etno-botânico y etnozoolôgico am ahuaca (pano)’, Docum ento de Trabajo, 3. Lima, Peru: Universidad Nacional M ayor de San Marcos. Faust, N. 1973. Lecdones para el aprendizaje del idioma shipibo-conibo. Docum ento de Trabajo, 1. Peru: Institute Lingüístico de Verano. Fields, H. 1973, *Una identificacién preliminar de les sufijos indicadores de referencia en mayo runa’, pp. 283-306 o f Estudiospanos II, Serie Lingûistica Peruana, 11. Peru: Instituto Lingüistico de Verano. G irard, V. 1971. Proto-Takanan phonology. University o f California Publications in Linguistics, 70. Berkeley and Los Angeles: University o f California Press. Grasserie, R. de la. 1890. ‘De la famille linguistique Pano’, pp. 438-50 o f Actas del VII Congreso Mernacional de Amerkamstas, held at Berlin in 1888, Kensinger, K. M. 1963. ‘The phonological hierarchy o f Cashinahua (Pano)’, pp. 207-17 o f Studies in Peruvian Indian languages I, SIL Publications in Linguistics, 9. N orm an: SIL and University of Oklahoma. 1985. ‘Panoan linguistic, folkloristic and ethnographic research: retrospect and prospect’, pp. 224-85 o f South American Indian languages: retrospect and prospect, ed. H. E. Manelis Klein and L. R. Stark. Austin: University o f Texas Press Key, M. R. 1968. Comparative Tacanan phonology. With Cavineña phonology and notes on Pano-Tacanan relationship, Janua Linguarum Series Prâctica, 50. The Hague: M outon, Kneeland, H. 1973. ‘La frase nominal relativa en mayoruna y la ambigüedad’, pp. 53-105 of Estudios panos II, Serie Lingüística Peruana, 11. Peru: Instituto Lingüistico de Verano. Loos, B. 1980. Textos capanahua, vols. la n d I I, Com unidades y culturas peruanas, 17. Peru: Instituto Lingüistico de Verano. Loos, B, and Loos. E. E. 1973. ‘La estructura semántica y fonolögica de los prefijos verbales en capanahua’, pp. 63-132 o f Estudios panos /, Serie Lingüistica Peruana, 10. Peru: Instituto Lingüistico de Verano. Loos, È. E. 1963. ‘Capanahua narration structure’, pp. 69R-742 o f University o f Texas Studies in Literature and Language, 4, supplement. 1969. The phonology o f Capanahua and its grammatical ha.m. Mexico: SIL. 1973a. 'Algunas implicaciones de la reconstrucoion de un fragmento de la gram ática del p roto-pano’. pp. 263-82 o f Estudios panos II, Serie Lingüistica Peruana, 11. Peru; Instituto Lingüistico de Verano. 1973b. ‘La construcciön del reflexlvo en los idioraas panos’, pp. 161-261 o f Estudios panos II, Serie Lingüistica Peruana, 11. Peru; Instituto Lingüistico de Verano. 1973c. ‘La señal de transitivldad del sustantivo en los idiomas panos’, pp. 133-84 o f Estudios panos I, Serie Lingüistica Peruana. 10. Peru: Instituto Lingüistico de Verano. 1976. Verbosperformativos: p artim las que tienen significadoperformativo o significado relacionado a los performativos en idiomas panos = Estudios panos V, Serie Lingüistica Peruana, 14. Peru: Instituto Lingüistico de Verano. 250 Eugene £ Loos 1986. ‘Capanahua syllable restructuring in relation to comparative studies’, pp. 291-301 o f Language in global perspective: papers in honor o f the 50th anniversary o f the SIL I935-I9S5,ed. B. Elson. Texas: SIL. Loriot, 1, Lauriault, E. and Day, D. 1993. Dicdonario Shipibo-Castellano, Shipibo, Serie Lingüistica Peruana, 31. Peru; Instituto Lingiiistico de Verano. M on tag, R. 1973. ‘La estructura semántica de las relaciones entre frases verbales en cashinahua’, pp. 107-57 o f Esiudios panos II, Serie Lingüistica Peruana, 11. Peru; Instituto LingiMstico de Verano, M ontag, S. 1981. Dicdonario cashinahua, voL I, Serie Lingüistica Peruana, 9. Peru: Instituto LingUistico de Verano. Navarro, R. P. E M. 1903. Vocabitlario caslellano-quectma-pano, con sus respectivas gramáticas quechua y pana. Lima: Imprenta del Estado. (Reprinted in d’Ans 1970.) Parker, S. G. 1992, Dalos del idioma hiiariapano. Documento de trabajo, 24. Peru: Instituto Lingiiistico de Verano, 1994. 'Coda epenthesis in H uariapano’, UAL, 60,95-119, Prost. G, and Prost, M. 1962. ‘Signalling of transitive and intransitive in Cliacobo (Pane)’, UAL, 28,108-18. Russell, R. 1975. Una gram átka transformaeioiml del amahuaca —Estiidios panos IV, Serie Lingüistica Peruana, 13. Peru: Instituto Lingttistico de Verano. Scott, E. and Frantz, D. G. 1973. *La pregunta en sharanahua y constreñimientos propuestos sobre la perm utacion de la pregunta’, pp. 185-209 o f Esuidiospanos /, Serie Lingüistica Peruana, 10. Peru: Instituto Lingüistico de Verano. Shell. O. A. 1957, ‘Cashibo II: grammemic analysis o f transitive and intransitive verb pat­ terns’, /J/IZ., 28.179-218. 1965. ‘Pano reconstruction’, Ph.D. dissertation. University of Pennsylvania. Published in 1975 as Las lengtias pano y su reconstrucaön =Estudios panos III, Serie Lingüistica Peruana, 12, Peru: Instituto Lingiiistico de Verano, 1973. ‘Los m odos del cashibo y el análisis del performativo', pp. 23-62 o f Estudiospanos 1, Serie Lingüistica Peruana, 10. Peru: Instituto Lingiiistico de Verano. 1987. Vocabitlario cashibo-cacataibo, Serie Lingüistica Peruana, 23, Peru: Instituto Lingüistico de Verano, Young de Hyde, S. 1973. ‘El verbo reHexivodel am ahuaca', pp. 9-51 o f Estudios panos II,Serie Lingüistica Peruana, 11. Peru: Instituto Lingüistico de Verano. 1980. Dicdonario amahuaca, Serie Lingüistica Peruana, 7. Peru: Instituto Lingüistico de Verano. Makû S IL V A N A A N D V A L T E IR M A R T I N S in tro d u c tio n ' 1 The M akù family consists o f four languages belonging to seven tribes - see table 9.1. O ne is spoken around the M iddle Rio Negro in Brazil, w ith the others ranging from the U pper Rio Negro to the region o f the Vaupés and Japurá spanning Brazil an d Colom bia (see map 8).^ The two groups belong to different linguistic areas and greatly difier in their gram m atical structure. T he M aku^are nom adic hunters and gatherers. They have a low socio-economic status am ong other indigenous peoples o f the region - speakers o f A raw ak and Tucano languages, who specialize in slash-and-burn agriculture. The M akû live in small groups along little streams, avoiding big rivers, and providing their neigh­ bours with fish-poison (dm bo) and game, receiving m anioc in exchange. As jungle dwellers, they are believed to possess supranatural magical power. Like the Pygmies in A frica, they are despised and feared at the sam e time. ' Wc are grateful to Elias Coelho for providing information on Hupda. ^ Puinave-spoken in the region o f the Inirida river in Colom bia (see chapter 13 b e lo w )-h a s sometimes been linked to M akù, as a Makii-Piiinave family. In fact, no genetic relationship between M akù languages and Puinave has as yet been proven. ^ The term ‘M aku’ is used as a pejorative term fur Indians in the regional varieties o f Portuguese spoken in the Upper Rio Negro. This term m ay be of Arawak origin; accord­ ing to Koch-Griinbcrg (1906b; 877) the terra was first used by Arawak peoples with refer­ ence to the original nomadic population, cf Baniwa-lCurripako ma-aku (negation-speak) ‘those who cannot speak’ (also see Weir 1984; 1,5). M akii languages considered here should not be confused with the following other groups to which this name is sometimes applied; (i) (ii) (iii) Makii, or M áku, an endangered language spoken by a few old people on the river Auari, tributary o f the Uraricoera which flows into the Rio Branco, Roraima, Brazil (see Migliazza 1985) - this is briefly discussed in §1.6 o f chapter 1.1 below; Mako, also known as Cofán-M akü, spoken around the lake o f Cuyabeno, in Cokimbia and Ecuador (this falls outside Amazonia); M akù, also known as Sáliba-Maco, or M aco-Piaroa, which is a group of the Sáliba-Piaroa family (see §3 of chapter 13 below), spoken in the savannas between the lower Ventuari and Orinoco in Venezuela. 252 Silvana and Valteir Martins They have traditional econom ic ties with A rawak- and Tucano-speaking peoples. Arawak- or Tucano-speaking subclans usually enter into a ‘m aster-underling’ rela­ tionship with a particular group o f M akù (hence denom inations o f the type: M aku o f Kubeo or M aku o f Tugano, in table 9.1). The M akü are m ainly endogamous, and they do n ot form pa rt o f the m arriage network in the Vaupés region, being con­ sidered inferior and ‘unmarriageable’ (Jackson 1976; and chapter 14 below). They lack the ‘obligatory’ multilingualism associated with ‘linguistic’ exogamy, so strik­ ing am ong the Tucano- and Tariana-speaking peoples o f the Vaupés. ' M akù-speaking men usually have some knowledge o f the Tnpinanibá-based creole Lingua G eral (in the region o f M iddle and U pper Rio N egro), o f Tucano (in the Vaupés), o r o f Portuguese and/or Spanish. Women and children tend to be monolingual. The M aku are believed to be the original population o f the U pper R io Negro (N im uendaju 1982: 169; G a lv io 1979: 147), subsequently conquered by Arawak-, and then by Tucano-speaking peoples. A ccording to the oral tradition o f the D esana, Tuyuca, Kubeo, Tariana and Baniwa peoples, a few M aku groups actually lost their own languages and started using languages o f their ‘conquerors’; these Torm er M ak ù ’ usually have a lower status in the internal hierarchy o f their tribes (see K och-G riinberg 1906b: 878; Janet Barnes p.c.). 9 M akû 253 Table 9.1 The M akù languages in Brazil and Colombia Middle Rio Negro ( Brazil) 1 Nadëb-Kuyawi (Nd-Ku) N adëb (Nd) is also known as N adöbö, Anodöb. M akunadöbo, G uariba, G uaribatapuyo, Kabori, or Cabori, and Xiriwai. There are around 400 speakers mainly o n the Uneiuxi river - a tributary o f the Middle Rio Negro - and along the Japurá river. Kuyawi (Ku) has around 20 speakers in the village Bom Jardim, oti the south shore o f the M iddle Rio Negro. The Upper Rio Negro and Vmipés ( Brazil and Colombia) 2 Daw (D) (regional pejorative name Kami) There are around 83 speakers o f Daw, currently living in one community near Sao Gabriel da Cachoeira, Brazil. 3 Hupda-Yuhup (Hu-Yu) H upda (Hu) is known as M akù o f Tucano and is spoken by around 1,900 people, occu­ pying a considerable area between the rivers Papuri and Tiquié in Brazil and Colombia. Yuhup (Yu) is spoken by around 4(K) speakers in 10 small communities between the U pper Rio Negro, the Tiquié and their tributaries, and in the region o f the U pper Japurá, Apaporis and Traira, and C aqueta, in Brazil and Colombia. 4 K akua-N ukak(K k-N k) K akua (Kk) is also known as Bara, or M akù o f fCubeo, or M akù o f G uanano. There are approximately 220 speakers, along the lower Vaupés an d the Papuri, in Colom bia near the Brazilian border. N ukak (Nk) is spoken by around 300-600 speakers scattered in jungle regions between the Guaviare and Inirida rivers, in Colombia. M ost o f the M akù o f the M iddle and U pper Rio N egro were contacted by white people in the early tw entieth century. Some groups, e.g. the N ukak, were com pletely unknow n to the outside w orld until the 1980s. 2 GENETIC CLASSIFICATION The M akii languages are listed in table 9.1. A bbreviations used in this chapter are given after each name. T h e genetic unity of M akù languages was suggested by K och-Griinberg (1906b), Rivet an d Tastevin (1920), and then by N im uendaju (1950/5, 1982), Recent com ­ parative studies show affinities in pronom inal paradigm s (but not in o th er gram ­ matical markers); about 300 cognates have been established across the family (V. M artins forthcoming). Examples o f lexical cognates are:'* N d, Ku ?ib, Yu, H u, Kk ?ip, D, Nk ?i:p ‘father’; N d, Ku tib, Yu, H u, Kk tip, D tip ‘egg’; * Raised figures in Kakua examples indicate tones. 254 Silvana and Valieir Martins Table 9.2 Cognate percentages between the M aku languages Nd-Ku 50 50 35? D 75 35? Hu-Yii 35? N d mi, H u, Yu tm h 'river*, K k mah ‘w ater’, D m l? Inside a liquid or fire’; Ku, H u, Yu tag", D tag ‘to o th ’; Hu môj, N k m i ‘house’, D m.3j ‘tem porary h u t’. N adëb and Kuyawi, H upda and Yuhup, K akua and N ukak are mutually intelli­ gible; these pairs have around 90 per cent o f their lexicon in com m on. D aw is not mutually intelligible with any o f these. Percentages o f cognates on a 200-item list are shown in table 9.2. The lexical data on K k-N k are scanty and these percentage figures are provisional The language o f the Middle Rio Negro group, N d-K u, is polysynthetic and highly prefixing. Nd has up to nine prefix positions (which is very unusual for an A m azonian language) and extensive incorporation o f one o r m ore nouns, postpo­ sitions and adverbs. It is predom inantly head-m arking, w ith elements o f m orpho­ logical and syntactic ergativity. Roots are polysyllabic, and there is a strong tendency towards verb-initial order. In contrast, Hu-Yu and D have few suffixes and alm ost no productive prefixes; K k-N k does have one prefix position. D and H u-Yu appear to have developed new suffixes out o f recently gram maticalized verb roots. They com bine elements o f head marking with dependent m arking, and are consistently nominative-accusative. C onstituent order .is verb-final. In D and Hu-Yu words tend to be mostly monosyllabic, unlike N d-K u and KkN k. This tendency can be illustrated with borrowings from Portuguese; only the stressed syllable o f the source noun remains in D: Dâw Portuguese y ë l? dinheiro ‘m oney’ pew chapéu ‘h at’ lê j k ite ‘m ilk’ D, Hu-Yu and K k-N k are m arginal members o f the Içana-Vaupés linguistic area due to long-term contact with Tucano speakers (see chapter 14); area! diffusion 255 9 M akti from T ucano in to the M akù o f the U pper Rio Negro may be the reason why these languages are so different from N d and Ku. T here are enough differences between N d-Ku and the others to establish them as a distinct group within the family. There are two possible family trees, given in dia­ gram s 9,1 an d 9,2; detailed com parative work will be needed to decide between these. proto-Maku Nd-Ku D Hu-Yu Kk-Nk Diagram 9. i The Maktt family tree: version / proto-Maku Nd-Ku D Hu-Yu Kk-Nk Diagram 9.2 The Makù family tree: version 2 The inform ation available on M akù languages varies in both quality and reliabil­ ity. Weir (1984, 1990,1994) has provided an excellent partial description o f N adëb. The only d ata on Kuyawi consist o f around 600 lexical items and phrases collected by Valteir M artins in January 1995. Silvana M artins (1994) has w ritten a prelimi­ nary gram m ar o f Dâw, and Valteir M artins (1994) has produced a detailed descrip­ tion o f the phonology and phonetics o f this language. Limited data on H u-Yu and on K k are found in Briizzi (1961), Del Vigna (1991), Brandao Lopes (1995), G iacone ( 1955) and M oore and F ranklin (1980). Studies on K k are C ath cart (1979), and C ath cart and Levinsohn (1976, 1977). There is an excellent anthropological study o f K ak u a by Si!verwood-Cope (1990). Only a few word lists have been recorded for N ukak (H uber an d Reed 1992). Valteir M artins is com m encing a com ­ parative study with the eventual aim o f reconstructing aspects o f proto-M aku. 3 PH ONO LO GY C onsonantal systems vary in their complexity, from 17 phonemes in D and 18 in N d, to 12 in Hu-Yu and 12 in K k. See table 9.3. N asalization is a syllable prosody in Hu-Yu and in K k-N k, so that Wm, din and g //7 are allophones. D and H u-Yu have prenasalized stops as word-final allophones o f simple stops. 256 Sihana and Valteir Martins Table 9.3 Consonants in M akit languages stop fricative affricate nasal liquid glide In all languages D also has: p, b, t, d, k, g, 7 h tj Î / d3 m, n I Hu-Yu also has: Nd also has: r W .m m, n, ji. r Kk also has: J .3 5 r w, y Table 9.4 Vowels in M akù languages In all languages (e.\cept Kk) D also has: Hu-Yu also has: Nd also h i,î,u ,ü e, o i,» ë, Ô t,î tu Ÿ 3 a,E,È s t , c , A, 3 æ æ a, à Table 9.5 Tones and stress in M akù languages Nd-Ku D Kk-Nk Hu-Yu no tones, non-eontrastive stress on the final syllable two tone œ ntours; almost predictable four tones not known N d, D and Hu-Yu have very complicatetl vowel systems, quite atypical for northwest Amazonia; see table 9.4. In contrast, K k has a ‘typical A m azonian’ flvervowel system (i, e, a, o, plus high central i), just like many Tucano languages. N asal vowels are phonemic only in D and Nd. Syllable structure is CV(C); D and Hu-Yu have vowel assimilation at morpheme boundaries. Velar stops tend not to occur in word-initial position. D has a low tone co ntour in 70 per cent o f C,VCj syllables with a voiced coda; all the syllables with a voiceless coda have a high tone (the vast m ajority o f words are monosyllabic). Tonal alternations may indicate a change in word class, e.g. w k:d (low) ‘eat’, H'£'£/(high) ‘food’. Proto-M akii probably did not have any tone or pitch accent; these could have developed under the areal influence of Tucano languages, o r as an independent innovation (V. M artins forthcomitig). 257 9 M akù 4 W ORD STRUCTURE The M akù languages o f the U pper Rio Negro and Colom bia are agglutinating and predom inantly suffixing with a few elements o f fusion across morphem e bou n d ar­ ies. D and Yu have vestigial prefixes;* in Kk just the subject (A/S) is cross-referenced on the verb with a prefix. In contrast, N d has up to nine prefix positions. N d prefixes are them atic, aspect, valency-changing, locational and subordinatirig. Weir (1984: 50) reports that their ordering appears to be determ ined by their phonological form. Some o f these - locational and subordinating, possibly some derivational and thematic - developed from incorporated postpositions and adverbials. There is also a set o f cross-referencing proclitics. (1) shows a typical verb in N d (Weir 1984; 51). (1) ga-na-ni-wan THEME-THEME-NEG + ASPECT-Spill ‘It does not s p ill’ A nother typologically unusual property o f N d is the position o f incorporated adverbials, nom inals and adpositions before the root (see (11)). M any other languages o f the A mazon tend to postpose incorporated nom inals/adverbials (see (n) in §2 o f chapter 1 above). Fossilized prefixes and traces o f preverbal incorporated adverbs are found in D, e.g. kcd-pij (inside-be.full) ‘overcom e’; fa f - fin (inside-count) ‘think’, lit. ‘count inside the head’; c f H u (M oore and Franklin 1980: 3): (undem eath-pass) ‘pass underneath’. A reflex o f a prefixed noun m arker in- in N adëb is found in Daw m ar ‘inherent possession’, and in negative predicates m ay and mëh. Tliis may indicate that proto-M akù had m ore prefixing tendencies than the m odem languages o f the U pper Rio Negro; it is quite likely that these languages acquired a predom inantly suffixing profile under the areal pressure from. East Tucano. 5 W ORD CLASSES The only open word classes are verbs and nouns. In D and Hu-Yu adjectival con­ cepts are expressed with stative verbs, while in N d ‘adjectives’ are a subclass o f obligatorily possessed nouns (Weir 1984; 84). There is a closed class o f positive and negative copula verbs used in equational, attributive and locative-existential clauses. * For instance, a prefix ?ih- is used to form names, e.g. wsèb ‘pigeon’, 71h-wæh ‘pigeon’ (proper .name). ^ " 258 Silvana and Valteir Martins 6 NOM INAL c a t e g o r i e s : POSSESSION, CLASSIFIERS, NUMBER Alienable and inalienable possession is distinguished in N d, and D and Hu-Yu, but it is marked in different ways. Nd has three classes o f nouns (Weir 1984: 83-7): (a) nouns which can only occur with a preposed possessor, e.g. Subih mooh ‘S ubih’s h and’; (b) those which can optionally occur with a possessor, e.g. Subih id ) ‘Siibih’s house’; (c) those which require a classifier when possessed, such as animals, plants, som e kinship terms, etc., e.g. Subih waa m asccl (Subih POSSV.CL:FOOD banana) ‘S ubih’s banana’ (one can n o t say *Subih masccl). In D and Hu-Yu both alienable and inalienable possession is m arked by juxtapo­ sition (Possessor-Possessed), e.g. D yam dum (dog tail) ‘dog’s tail’. D is unusual in employing ‘possessor classifiers’ - suffixes which attach to posses­ sors in alienable possession constructions; which is used depends on whether the possessor is anim ate o r inanimate: -de: ? ‘inanim ate possessor’, -éj ‘anim ate pos­ sessor’: see (2) and (3). (2) yud daw-tog-ej c lo th in g human-female.child-CL:ANiM,POSSR ‘The c lo th in g is a g irl’s’, o r ‘g irl’s c lo th in g ’ (3) yak kaw-wg?-de:? m a n io c garden-up-CL:iNAN.POSSR ‘m a n io c fro m a g a rd e n ’ D and Hu-Yu also have generic noun classifiers in noun phrases, e.g. D daw tog (C L :H U M A N girl) ‘a girl’. These are only rarely found in A m azonian languages. D an d Hu-Yu (but n o t N d) have very productive nom inal com pounding; nominal com pounds are sometimes hard to distinguish from classifier NPs, e.g. D dâw-Jobk e d (person-hand-inside) ‘the inside part o f a h an d ’; dâw -tim -bsk-tjën (person-eyeskin-hair) ‘eyelash’. These examples have ju st one stress, on the last syllable; this shows that they form one phonological word (unlike classifler-noun constructions). K k is the only M akù language to have a masculine/feminine gender distinction in 3rd person singular pronom inal prefixes, independent pronouns and dem onstra­ tives (C athcart and Levinson 1976: 26); this could be the result o f Tucano influence. D has five locative postpositions. Their choice depends on the physical properties o f the referent o f the head:** fca/*inside a bounded object’, m t? ‘inside liquid, or fire’, bit ‘underneath an object with an u pper boundary’, w a ? ‘above unbounded object’, * Similar constructions, termed ‘locative classifiers’, are found in C arib languages, and in Palikur and Lokono from the N orth Arawak family; see-§4.2 o f chapter 3 above. 9 M akû 259 / « / ‘inside a m ixture’. In (4), ke d is used with the noun ‘canoe’. In (5a), m i? is used with a noun ‘river’. (5b) is ungram m atical. (4) Jo:-ked canoe-iN : BOUNDED ‘ill a c a n o e ’ (5a) n3:J-piJ-mi? river-sm all-iN:LiQUiD ‘in a sm all river’ (5b) *x3-mi? canoe-iN xiQ U iD D and Hu have optional plural markers used with hum ans only (Daw -dâr; H upda -de - as in hup ‘person’, de ‘plural’). N d has a num ber of irregular singular/plural pairs, e.g. txaah/taah ‘soniaons’, f y n g /f y y ‘woman/women’ (Weir 1984: 83). Personal pronouns distinguish three persons and two numbers; there is also an inclusive and exclusive distinction in 1pi. D also has three forms for 3rd person sin­ gular when referring to hum ans (definite, definite focused and indefinite), and two for 3rd person plural (definite and indefinite); the collective p a y e e is often used as an im personal pronoun. 7 VERB STRUCTURE Verb structure is dram atically diflerent in N d, and in D and Hu-Yu. N d divides verbs into transitive and intransitive, with a few S = A ambitransitives. In D every verb seems to be am bitransitive, m ost o f them of type S = A w ith just a few S = 0 . N d has an extremely com plicated verbal structure with up to nine prefix posi­ tions. Prefixes are classified into six types (Weir 1994: 293-4): (i) the formative prefix a-, attached to the verb root in the absence o f other (ii) aspect prefix i- (obligatory w ith some verbs); prefixes; (iii) asp ec tu a l a n d m o d a l prefixes; (iv) them atic prefixes; (v) (vi) valency-changing prefixes; subordination prefixes. D and Hu-Yu have a m uch less complicated verbal morphology. There is a great deal o f verb com pounding (root serialization) used for encoding aspectual meanings, 260 Silvana and Valteir Martins as in (6), from D, where the verb ‘be straight’ shows the immediate character of the action. (U p to five verbs can be com pounded.) (6) yô:h bs:-him -yaw medicine spili-go-be.straight ‘The medicine spilt straight away.’ D and H u distinguish three tenses: present (unmarked), p ast (D -e?) and future (D ~ey). In N adëb every verbal root and suffix distinguishes indicative from non-indicative. Noo-indicative forms are used in imperatives, nom inalizations and interroga­ tives; they are m arked with vowel change and/or the voicing o f a final consonant (Weir 1984: 39). Some verbs also m ark w hether the S/A argum ent is singular or plural. Plural verb forms can be suppletive; o r derived with a prefix o r vowel and/or consonant alternations. There are a num ber o f aspectual prefixes (habitual, continuative, directional, etc.). There are only a few suffixes (diminutive, ingressive and completive; Weir 1984; 45). While H u and K k have only one im perative (H u -kern, K k reduplication o f the final vowel), N d distinguishes several imperatives (simple, indirect, permissive). D distinguishes im perative (oh) and prohibitive (îli). 8 VALEN CY -CHAN GING DERIVATIONS D, Hu-Yu and K k have no valency-reducing derivations. In contrast, N d has a prefix ka- ‘reflexive, reciprocal’ (Weir 1984:47).’ It is also used as an intransitivizer, and to m ark non-agentive passive (1984: 107). N d also has a m orphological causative, marked with da-. T he only example given by Weir ( 1984:47) is with an intransitive verb. A nother technique for marking caus­ atives is verb com pounding as illustrated in (7) (1984: 44). (7) taoh Subih na-boh-yat p ig Subih THEM E-shoot-lie d o w n ‘Subih killed a pig.’ (lit. ‘he shot; it is lying down’) In Kakua, morphological causatives (suffix - C athcart 1979:41 ) can be formed on intransitive - and probably also on transitive - verbs, while D and H u use peri­ phrastic constructions, as in (8) and (9), both from D: ’ It was noted in chapter 3 that many North Arawak languages use the same marking for reflexives, reciprocals and agentless passives; a widespread marker is (-)ka(-), so one can not exclude the possibility of borrowing or areal diffusion. The causative prefix da- could also well be of Arawak origin. 261 9 M akù (8) tih ?ip we:d wa;y tih-ï:y? he father eat order he-o.TOPiCAL ‘His father m ad e him eat.’ (9) äh d o? haw na;J-î:y? Isg causative b oil water-o. topical 'I m ade the water b o il,’ D has a reported/non-reported evidentiality contrast. Evidentiality is lacking from the oth er languages. 9 INCORPORATION A striking property o f N d (n o t found elsewhere in the family) is noun incorpora­ tion o f two kinds. Similarly to other A m azonian languages (e.g. G uahibo) S/O can be incorporated only if obligatorily possessed, and the incorporated noun is placed before the verb. The original possessor becomes the object (Weir 1990: 323ff.); anim ate nouns rarely incorporate. T he effect o f incorporation is 'discourse' prom i­ nence o f the new object. C om pare (10) and (11): (10) Subih tob iih ta-m a Subih h ou se Isg THEME-make T am m aking S ubih’s h ou se.’ (11) Subih Subih îih tob-ta-m a Isg house-THEME-make ‘I am m aking a house for Subih.’ (lit. T am house-m aking S ubih’; the benefit for Subih is em phasized) N d allows m ore than one noun to be incorporated (similar to Cuiba, from the G u ah ib o family - see chapter 13 - but unlike the majority o f A m azonian lan­ guages). C om pare (12), lacking incorporation, with the incorporation in (13). It is also possible to incorporate two or even three nouns as in (14) and (15), but this is not com m on. The incorporated nouns are underlined. These examples are from Weir (1990). (12) a hooñ 2sg + P0SSR grandm other tab h o u se nooh kad m ou th uncle ga-jaa THEME-close dak be.suspended = , ‘Uncle closed’the d o o r o f your grandm other’s house.’ ■ 262 Silvana and Valteir M artins (13) a hooñ 2sg + P0SSR grandm other tab kad no o h -ga-iaa house uncle m o u th -THEME-close dak be.suspended ‘U n d e closed the door o f your g randm other’s house.’ (lit. ‘U ncle mouth-closed your grandm other’s house’; the effect on the door is em phasized) (14) a hooñ kad tob-nooh-ea-iaa 2sg + possR grandm other uncle h ou se-m ou th -THEME-close dak be.suspended ‘U ncle closed the do o r o f your grandm other’s house.’ (lit. ‘Uncle house-m outh-closed your grandm other’; the eflect on the house is em phasized) (15) om kad h o o ñ-tjb-nooh-ga-jaa dak 2sg uncle grandm other-house-m outh-THEME-close be.suspended ‘Uncle closed the door o f your grandm other’s ho u se.’ (lit. ‘Uncle grandm oth er-house-m outh-closed y o u ’; the effect on grandm other is em phasized) Adverbs an d adpositions can also be incorporated. Unlike some other A m azonian languages, these are preposed to the verbal root. Incorporation o f adpositions results in applicative-like derivations: if the verb is intransitive, the argum ent o f the postposition becomes O, and the original S becomes A. If the verb is transitive, the argum ent o f the postposition is prom oted to O, and the old O is dem oted to the periphery (Weir 1990: 326ff.). Consider (16) and (17). The incorpo­ rated postposition in (17) is underlined. (16) (17) CEj hxDoh go father FORMATivE-go.downriver canoe ‘F ather g o es downriver in a ca n o e.’ a-h in g in hxaahjj ee^ la -h in g canoe father in-go.downriver ‘F ath er goes downriver in a canoe.’ (lit. ‘Father goes downriver-in a c anoe’) This affects the discourse properties o f ‘canoe’. Incorporation o f a postposition can be a strategy for topicalization (Weir 1984: 136); the transitivity o f the verb is then not affected. 263 9 M akû Table 9.6 Ergative cross-referencing in Nd A/S = 0 : ‘ergative’ 2sg. 3sg, 3pl A = S = 0 : ‘neutral’ Isg, Unci, lexcl, 2pl T able 9.7 Cross-referencing proclitics in N d S/O A 2sg ôm 3sg 0 m a- ~ Sm ta­ 2pl 3pl (18) baah la- 0 ta-koliay 3-claw ta -ta 3-food ta-m i-so o 3- w it h + ASPECr-take ta-kolàay 3-claw me w ith ‘As for his claws, it is with his daw s that he takes his food.’ 10 GRAMMATICAL RELATIONS Nd is one o f the most ergative languages in the A m azon region, while D and HuYu are consistently nominative-accusative.* In N d, an overt reference to A outside the predicate is obligatory, but the S/O argum ent may be omitted. This language also uses constituent order SV o r VS and OAV o r AVO, and it has a syntactic pivot S/O in coordination (Weir 1984; 89-91). There is also a split-ergative pattern for personal pronom inal proclitics dependent on person and num ber - see tables 9.6 and 9.7. D and Hu-Yu employ cases for m arking core gram m atical relations.* D has an unusual pattern o f core cases dependent on topicality o f O and o f A/S. This pattern may be the result o f areal diffusion from Tucano languages. O is unm arked when it precedes the predicate as shown in (19). (19) laka:h^ mëtQ Jooh-ê? hen jacam im peck-PAST Ä hen pecked a jacam im bird.’ ® Kk seems to use distinct cross-referencing markers for A. S and O; this needs further check­ ing. ’ Kk seems to have a suffix (or enclitic?) which m arks both O and addressee, -dih* (Cathcart and Levinsohn 1976: 26-8). 264 Silvana and Valteir Martins When it is topical and follows the predicate, O is m arked with (20) laka:h^ Jooh-e? hen as in (20). m êî-î:y? peck-PAST jacamim-o.TOPiCAL ÄS for the jacam im bird, a hen pecked it.’ The O suffix can be om itted when pragm atic context is likely to resolve any pos­ sible ambiguity. In a clause like (21), it is highly unlikely that a deer could kill a jaguar, so the object m arker is usually omitted. (21) ?y am ji?^ yflt to:(4:y?) ja g u ar kill deer(-o.TOPiCAL) ‘A Jaguar killed the deer.’ A/S can be marked w ith an enclitic -teh i f it is in contrastive focu s and is highly agentive. In (22), the focus is on II, a mythological character.*® (22) ?aaten ?il-teh warn hew m ij then 11-ag entiv e .a / s sm oke much jacam im ‘Then it was II who smoked a lot o f jacam im birds.’ Oblique constituents are m arked with clitic postpositions. D is unusual in having an instrum entai and three comitatives: -red ‘instrum ental’, -did ‘w ith’ (equal rela­ tions between participants); -rûy ‘with’ (one participant is superior to the others, e.g. chief, elder brother, etc.); - / « / ‘with, am ong’ (with inanimates: a mixture, e.g. coffee with milk). D, Hu-Yu and K k have just one locative marker (D -hid, K k -p ‘to, from, in’), like Tucano languages. In contrast, N d marks only ablative (bf) and locative (liën). It is possible th a t proto-M aku could have been ergative, and th a t the languages o f tlie U pper Rio N egro lost the ergative marking under the areal influence o f Tucano languages. The main argum ent for this is the loss o f ergative-type marking in constituent order in the languages o f the U pper Rio Negro, accom panied by the loss o f proclitics and prefixes. 11 NEGATION All the M akii languages have com plicated mechanisms for m arking negation and have a special prohibitive marker. Nd has three negative m orphem es: dooh, used to A similar pattern is found in Tariana (North Arawak) and Siona (West T u o n o ). See chapter 14 below. 265 9 M akû negate non-im perative m ain clauses (and as a negative answer); na, used to negate dependent clauses; and m anih, used to negate im perative clauses (Weir 1984: 148-263; 1994), Besides simple negation, D has em phatic negation (contra-expectative) -ta?, existential and possessive negation -ëh, and prohibitive -Th. 12 SYNTAX D has a num ber o f coordinating and subordinating verbal enclitics, while N d marks subordination and relativization with verbal prefixes. H u seems to have a relativizing suffix (-?ep) which goes on verbs (M oore and Franklin 1980: 15). In N d, only core argum ents can be relativized. To relativize other argum ents, they must be p u t into O position (via incorporation o f a postposition). The relativizing m orphem e appears within the relative clause, and the com mon argum ent is om itted from the m ain clause. (23) [kapehg c o f fe e je m yesterday íih ^ Isg i- jë e A S P E c r-b u y + in d ic doo] Subih REL Subih i-eek ASPECT-drink + in dic ‘Subih is drinking the coffee which I bought yesterday.’ N o such restrictions exist in other M akù languages - any argum ent in the main clause can be relativized. C om plem ent clauses in N d are marked w ith subordinating verba! prefixes, while D uses simple juxtaposition o f clauses. Only Kk has switch-reference, which has possibly developed through diffusion from Tucano languages. 13 LFsXICON D an d o th er M aku languages o f the U pper R io N egro are highly ‘verbal’; ab o u t 75 per cent o f the words in texts are verbs. D has numerous verbs for ‘carrying’, ‘cutting’, ‘falling’ (depending on w hether it is a person, o r a fruit) and ‘disappear­ ing’ (w hether it is in the jungle o r elsewhere). Similarly to East Tucano languages but unlike N d - D and H u-Yu have ju st one word for ‘sun’ and ‘m oon’. D has a fascinating system o f counting. T here is a word iV«ë?*one, alone’, and ik m ‘p air’ (used for two), and mutu-wap ‘three’ (also used for ‘m any’). To say ‘four’, speakers show four fingers in twos saying ?m e? maab ‘one brother’ (each o f the fingers has a ‘brother’); ‘five’ is ? m ë? maab m eh ‘one does not have a brother’ (this goes on up to 10). 266 Silvana and Valteir M artins BIBLIOGRAPHY Brandào Lopes, A. 1995. ‘Fonologia da lingua Yuhup: uma abordagem nio-linear*, MA thesis, Universidade Federal de Santa Catarina, Florianépolis, Brazil. Brözzi, A. A. da S. 1961. Discoleca etno-lingtilsHco-muskaldas tribos dos rios Vaupés, Içana e Cmmbiiii. Sao Paulo: Inspectoria Salesiana. Cathcart, M. 1979. ‘Fonologia del cacua’, Sislemas fonotégicos de idiomas colombianos, 4.9-45. Cathcart, M. and Levinsohn, S. 1976. 'Codificacion de la progresion cronolôgica del discurso narrativo en e! idioma cacua’, pp. 1-31 o f Serie sintcictica 3: Estudios en cacua, Jupda y saliba. 1977. ‘The encoding of chronological progression in Cacua narratives’, pp. 69-94 o f Discourse grammar, studies in indigenous languages o f Colombia, vol. II, ed. R, Longacre and F. Woods. Dallas: SIL and University o f Texas at Arlington. Cathcart, M. and Lowers, L. 1976. ‘La pelea entre ,4qui y Victoria, texto del cacua’, pp. 37-82 of Serie sinláctica 3: Estudios en cacua, Jupda y sáliba. Del Vigna, D. 1991. ‘Segmentes complexos da lingua Yuhup’, MA thesis, Universidade Naeiona! de Brasilia, Brasilia, Brazil. Galviio, E. 1979. ‘A culturaçâo indigena no rio Negro’, pp. 135-92 o f his Enconlro de sociedades. Rio de Janeiro: Paz e terra. Giacone, A, 1955. Pequena granmtica e dicionário pnrtuguês ubde-nehern ou macii, Recife; Missao indigena salesiana. Huber, R. Q. and Reed. R. B. 1992. Comparative vocabulary: selected words in indigenous lan­ guages o f Colombia. Santafé de Bogota: Asociaciôn Institute Lingüistico de Verano. Jackson, J. 1976. ‘Vaupés marriage: a network system in the Northwest Amazon’, pp. 65-93 of Regional analysis, vol. II: Social systems, ed. C, A. Smoth. New York: Academic Press. Koch-Grünberg,T. 1906a. ‘Die Indianer-Stämme am oberen Rio Negro und Yapurá und ihre sprachliche Zugehörigkeit', Zeilschrifl fur Ethnologie, 38.167-205. 1906b. ‘Die Sprache der M aku-Indianer’, Anthropos, 1.877-906. 1909/10. Zwei Jahre iinler den Indianern. 2 vols. Stuttgart. 1913. ‘Abschluss meiner Reise durch Nordbrastlien zum Orinoco, mil besonderer Beriioksichtigung der von mir besuchten Indianerstämm e’, Zeitschrift fiir Ethnologie, 45.448-74. 1928. Vbm Roroima ztim Orinoco: Ergebnisse einer Reise in Nordbrasilien und Venezuela in den Jahren 1911-1913, vol. IV: Spraclien. Stuttgart. Martins, S. A. 1994. ‘A nálise da morfosintaxe da lingua D a w {M akit-Kam â)esuaclassificaçào tipologica’, M A thesis, Universidade Federal de Santa Catarina, Florianépolis, Brazil, Martins. V. 1994. ‘Fonologia da lingua Daw’, MA thesis, Universidade Federal de Santa Catarina, Florianépolis, Brazil. forthcoming. ‘Estudo historico-comparativo da familia iingiiistica M akii’. Migliazza, E. C. 1985. ‘Languages of the Orinoco-A m azon region: current status’, pp. 17-139 of South American Indian languages: retrospect and prospect, ed. H. M. Klein and L. Stark. Austin; University o f Texas Press. Moore, B, J. 1977. ‘Some discourse features o f Hupda M acu’, pp. 3 5 ^ 2 of Discourse grammar studies in indigenous languages o f Colombia, vol. II, ed. R. Longacre and F. Woods. Dallas; SIL and University o f Texas at Arlington. Moore, B. J. and Franklin. G. 1980. Breves notkias da lingua Maku-Hupda. Brasilia; SIL. Nimuendaju, C. 1950/5. ‘Reconheciinento dos Rios Içana, Ayari e Vaupés’, Journal de la Société des Américanistes, n.s., 39.125-82; 44.149-78. 1982. Textfis indigenistas. Sao Paulo: EdiçSes Loyola. 9 M akíi 267 Rivet. P., fCok, P. and Tastevin, C, 1924-5. ‘Nouvelle contribution à l’étude de la langue m akù’, UAL, 3.133-92. Rivet, P. and Tastevin, C. 1920. ‘Affinités du m akû et du puinave’, Journal de la Société des Am éricm istes,n.i., 12.60-82. Silverwood-Cope, P. L. 1990. Os Makû: p o w caçador do Noroeste da Amazônia. Brasilia; Editora da Universidade N acional de Brasilia. Weir, E. M. H. 1984, ‘A negaçâo e outras tôpicos da gram àtica Nadëb’, M A thesis, Universidade Estadual de Campinas, Brazil. 1990. ‘Incorporation in N adëb’, pp. 321-66 o f Amazonian linguistics: studies in Lowland South American Indian languages, ed. Doris L. Payne. Austin: University o f Texas Press. 1994. ‘N adëb’, pp. 291-323 o f Typological studies in negation, ed. P. Kahrel and R. van den Berg, Amsterdam: John Benjamins. 10 Nambiquara IV A N L O W E I in t r o d u c t io n ’ The Nambiquara^ family consists o f three languages: {I ) Southern N am biquara dialect complex (2) M am ainde/N akarothe (3) Sabanes The Southern N am biquara dialect com plex com prises o f the following twelve mutually intelligible dialects (superscript figures indicate tones - see below): (a) (b) (c) (d) (e) (0 Ne^su^ (g) KiHâuJlhti* (h) Ha^hài?He^su^ Ha^lot^te^ii^ (i) (i) w (I) A ’l?uh ' n?e%aHa-’ki^te^su^ Wa-’ka^li-’te^su^ Wa%uh’s?u^ A-’län'te^su^ Waik^tiHe^su^ Ka^lii%hâi^ko-’te% tf Tu-’ka^l ?u^l a%o^ te^u* Wai-’su^ The author is grateful to his SIL colleagues M enno and Barbara Kroeker and Peter Kingston for the map o f the Nam biquara area and for information on Mamainde, And also to the residents of the villages of Serra Azul, C am araré and Campos Novos for their patience in teaching him the language. The origin of the name N am biquara (alternatively spelt Nam bicuara or Nam bikwara) is interesting. Lévi-Strauss (1948: 361) states: The Nambicuara {Nambikuara, Mambyuara, Mahibarez) have been identified, only recently. Nambicuara, meaning ‘long eared’, was originally a Tupi nick­ name used since the 18th century for the little known tribes o f the western and northern parts o f the Serra dos Parecis. These tribes had large ear and lip plugs, like those of the Suya and Bolocudo, and were called Beiços de Pau, ‘Wooden M ouths’, by the rubber collectors and gold miners. About 1830, they began to make hostile sorties from the region o f the upper Sangue River, W hen, in 1907, General Candido M ariano d a Silva Ronddn discovered im portant tribes in the Serra do Norte, he identified them with the Nambicuara o f the old literature. Thus, Nambicuara designates a tribe other than the 'Long Ears’, or ‘Wooden M ouths’, to whom it was originally applied. 270 Ivan Lowe T he N orthern N am biquara subgroup consists o f two languages: the mutually intel­ ligible dialects (m) M am ainde and (n) N akarothe, and the single-dialect language (o) Sabaoes. M ap 9 shows the approximate locations o f the dialects o f N am biquara. These are numbered 1 to 12 and are given below. A fter each location nam e is the approx­ im ate num ber o f residents there, followed by the dialects spoken there according to the labelling (a) to (o) used above. All the locations are within the State o f M ato G rosso do N orte, Brazil, and are bounded by the Rio Papagaio to the east and Rio G uaporé to the west, both of which flow northw ard and are eventual tributaries o f the Amazon. (Tlie tributary sequences are Papagaio-Juruena-Tapajés-A m azon and G uaporé-M adeira-A m azon.) 1 C am pos Novos, 48 (a, b) 2 Cam araré, 73 (b) 3 Posto N am biquara, 182 (c) 4 U tiariti, 62 (c, d) 5 G alera, 64 (e) 6 Fazenda Estrela, 82 (f, g) 7 Fazenda Zilo, 67 (h) 8 9 Q uatro Pontos, 44 (e, h) Sararé, 51 (i, j, k, 1) 10 M am ainde, 103 (m) 11 N akarothe, 25 (n) 12 Sabanes, unknow n (probably less than 50) (o) The description in this chapter is o f (b) K i’täu^lhu^, a dialect o f Southern N am biquara, and is based on some 2,000 pages o f oral te.\t and some 8 years of res­ idence in N am biquara villages. T he dialect o f N orthern N am biquara that has been extensively studied is M am ainde (e.g. E berhard 1995). Its lower-level phonology and m orphosyntax is no t unlike that o f thé southern dialect which is to be the main concern o f this chapter. Brief notes will be m ade from time to time in the descrip­ tion below when there are significant differences between the structures o f the two dialects. The first large-scale contacts o f the N am biquara with the outside world were with the Rondon expedition o f 1911. Estimates o f their population at that time run from 20,000 to 50,000. Sadly, epidemics immediately after the Second World W ar reduced their num bers to less than 500 by the early 1950s. Happily, their numbers have increased significantly in the last 15 years; in some villages the population has m ore than doubled. M ost o f the population is young. Until about 20 years ago, the N am biquara lived a traditional hunting and gathering life, supplem ented by some 271 10 Nainbiquara Table 10.1 Nambiquara consonants stops im plosives fricatives nasals liquids sem ivowels labial co ro n al velar glo ttal P t d s n 1 k ? (0 (m ) w y h Table 10.2 Nambiquara vowels slash-and-burn agriculture. T here was little contact with the surrounding Brazilian culture, and only a few o f the men spoke even rudim entary caboclo Portuguese. Today m any o f the younger people speak good Portuguese. T he children, however, still learn the vernacular and all N am biquara use it. 2 PHONOLOGY The phonem es o f N am biquara are given in tables 10.1 and 10.2. T he (f) and (ra) within parentheses can occur only in loan words. However, the nasal phone [m] and its prestopped counterpart [bm] can occur as variants o f the phonem e /n/. (The phonem ic system o f M am ainde has no contrastive implosive /d/, and /m / is quite rare.) T here are three contrastive contour tones m arked as 1 (down-gliding), 2 (upgliding) and 3 (level). (M am ainde has a fourth tone which is at a lower level and which occurs quite rarely.) N asalization (shown by ” over a vowel) is contrastive on vowels. Laryngealization (shown by . under a vowel) is also contrastive (not merely conditioned by the presence o f a glottal stop), on both oral and nasal vowels. A spiration can occur contrastively on the voiceless stops /p/, /t/, /k/, on the nasal /n/, an d the liquid /I/. Clearly the aspirated versions o f these sounds could have been interpreted as five extra phonem es. But, in the interests o f economy, they have been 272 Ivan Lowe interpreted as clusters with /h/. This interpretation also leads both to more symmet­ rical distributional statements, and to simpler m orphophonem ics. 2.1 The variants o f phonemes The stops /p/, /t/, /k/ have variants that freely fluctuate from voiced to voiceless in all environments: [p}~[b], {t]~[d], [k]~[g]. The nasal /n/, when occurring in the coda o f a non-word-flnal syllable, has prestopped variants [bm]~[dn]-|gr|] when the syl­ lable nucleus is an oral vowel, and the whole stop-nasal sequence assimilates to the place o f articulation o f the initial consonant o f the following syllable. When the nucleus is a nasal vowel, however, the variant is simply a nasal consonant [m]~[n]~[ij] that again assimilates to the place o f articulation o f the initial consonant o f the fol­ lowing syllable. T he liquid /1/ has variants [I] and [r], the lateral preceding the back vowels /u/ and lot, and the flap preceding the front and central vowels /i/, /e/ and /a/. As for the vowels, the high back /u/ has a variant range over [u]-[ij]-[iHa] in unstressed syllables. The low central /a/ is realized as [3] io unstressed syllables. 2.2 Syllable structure The syllable structure is (C(CCC))V(C). By far the m ost com m on syllable type is CV. The optional final C slot can be filled only by the voicelessstops /t/, /k/, /?/, the nasal Ini and the semivowel /h/. As for the onsets, the first slot can be filled by any consonant in the inventory, but subsequent onset slots can be filled only by the glides /w/, /h/, /?/. The initial syllable o f the following word is an example o f a syl­ lable with a maximum onset: (I) [kw h?a?-'ka?-’li^su^ ‘kind o f deer’ 2.3 Stress Stress is realized as increased intensity, with an increased pitch range on the falling and rising glides I and 2, and with a lowered-level pitch on the level tone 3. It comes on the last syllable o f a morphem e cluster. 2.4 Morphophonemics The surface forms o f b oth nouns and verbs follow a few m orphophonem ic rules. For verbs there are various different surface forms o f the Isg and 2sg subject suffixes and the 3sg negative subject suffix. For nouns there are various different surface forms for the different nominal endings. The m ost econom ic way to describe 273 10 Nambiquara these is by setting up underlying forms for the verb or noun roots (or stems) from which the various surface forms can be easily derived. Specifically, we take the verb roots (or stems) to have underlying forms which can end in a vowel o r in one o f the consonants n, t, h. I, and the underlying form s o f the Isg an d 2sg subject suffixes to be -a'- and -in' respectively (as shown in table 10.7). T hen, for example, the verb ro o t with underlying form suP- ‘hit’ (ending in I) would derive the surface forms o f its suffixes typically as follows: U N D E R L Y IN G F O R M stiF-a'-ra^ SURFA CE F O R M sxi^la'ta^ hit-lsg-PERFV ‘I hit’ (recent past) w here the crucial step in the derivation is resyllabification whereby the final closing co n so n an t o f the underlying form o f the verb root becomes the initial consonant in the surface form o f the Isg (or 2sg) subject suffix. H ence the Isg subject suffix will have surface forms -a'-, -na'-, -ha'-, -la'- depending on which consonant, if any, closes the final syllable o f the underlying form o f the verb root. Similarly noun roots have underlying forms which can end in a vowel o r in one o f th e consonants n, t, h, /, s, k (see §3.3). The definite noun suffix in its underlying form is -a^, and again we get the correct surface form s o f the definite nouns by a derivation where the crucial step is a resyllabification. Here are some examples of definite nouns in their underlying and surface forms. U N D E R L Y IN G F O R M s? ih^-a* SU R FA C E F O R M s? i^-ha^ house-DEF ‘the h o u se’ • wäF-a^ wiMa^ cloth-DEF ‘the clo th ’ kat^-a^ kat^-ta^ stick-DEF ‘the stick’ N ote th a t when the underlying form o f the noun ro o t ends in a voiceless plosive, that voiceless plosive is doubled in the surface form. 3 M ORPHO LOG Y The m orphology is largely agglutinative, but there are also some com plex m or­ phem e fusions in some o f the verb forms. The language is head-m arking and pre­ dom inantly suffixing, but there are a few prefixes. 274 Im n Lowe There are three open word classes in N am biquara, verbs (§3.1), adjectives (§3.2) and nouns (§3.3), and four closed word classes, pronouns (§3.4), adverbs (§3.5), interjections (§3.6) and ideophones (§3.7). 3.1 Verbs 3.1.1 Main verbs M ain verbs are suffixed for m ood, person-num ber and aspect. M ood can be either indicative or imperative, and indicative verbs are Turther suffixed for tense-evidenîiality. ( i) In d ica tive verbs Indicative verbs can be either declarative (asserting inform ation w ithout any degree o f doubt) o r dubitative (giving inform ation but qualified by varying degrees o f doubt). Indicative main verbs are inflected for the following obligatory categories: subject person: first, second, third subject number: singular, dual, plural speaker number; singular, plural aspect; perfective, imperfective tense; future, present, recent past, mid past, rem ote past (M am ainde has two futures, one o f which is more definite than the other. It also distinguishes recent past, mid past and d istant past.) The non-future N am biquara forms are further obligatorily inflected for: evidentiality: observational, inferential, quotative, internal support newness: given, new (with the usual meanings) Evidentiality has to do with the kind o f evidence that the speaker adduces to support their statem ent. The four main subcategories m entioned above have thé fol­ lowing meanings: with observational support the speaker claims to have seen the action they are reporting; w ith inferential support, the speaker’s claim can be based either on seeing an associated sim ultaneous action and m aking an interpreta­ tion therefrom , or on seeing a set o f circumstances which must have resulted from a previous action and m aking an inference; different suffixes m ark these two options; 275 10 Nambiquara Table 10.3 Nam ibiquam new-information verbal suffixes -wa^ -ra^ -n(i- ~-nïn-ta-^ -na^ -ta' -he’ -nha^ -hê^ -he' -na’ impcrfective aspect perfective aspect inferential observed circumstances observed action quotative mid past tense external present tense internaf recent past tense internal mid past tense internal action currently observed by both speaker and hearer with quotative support the speaker is simply passing on inform ation they have h eard from another speaker; with internal su p p o rt the speaker reports their ‘gut feeling’ that that which they assert m ust be so. (M am ainde has the distinctions o f observational versus inferential versus quotative support, b u t there does n ot seem to be any distinction between given and new infor­ m ation expressed in the verb m orphology.) Table 10.3 gives the surface fprms o f some new -information form s with some forms showing typical category com binations. H ere are some typical exam ples o f new -inform ation verb forms: ( 2) wa'kon^-na-’-ra^ Work-ACTION.CURRENTIY.OBSERVED.BY.BOTH.S.&H.-PERFV ‘He is w orking.’ (3) w a’kon’-na^-ra^ WOrk-OBSERVED.ACriON-PERFV ‘He worked.’ (1 observed, recently) (4) w a% on^-ta'-he'-ra^ WOrk-QUOTATIVE-MID.PAST.INTERNAL-PERFV 'H e worked.’ (I was told, in the past) (5) w a’kon’-nTn*-ta^-wa^ WOrk-lNFEREOTtAL-OBSERVED.CIRCUMSTANCES-IMPERFV ‘He worked,’ (I inferred from observed circumstances recently) 276 hart Lowe (6) wa^kon^-nm^-na^-wa^ WOrk-INFERENTIAL-OBSERVED.ACTION-IMPERFV ‘He is w orking.’ (I inferred from observed associated action from a distance recently) Given-inform ation form s have many fused morphemes, so I merely give a few rep­ resentative forms here, w ithout attem pting any m orphem e segmentation. (7) wa-*kon-’taitki^tu-’wa^ (8) w a^onH e'ni^tu’wa^ (9) wa-’kon^te^nü'ti^tu^wa^ ‘He w orked.’ (mid past, we b o th saw him do it) ‘He w orked.’ (recent past, we b oth saw him do it) ‘He w orked.’ (recent past, we both inferred from circumstances) ( 10) w a% onH a'tet'ti^tu% a^ ‘He w orked.’ (recent past, we were both told) ( ii) Im p era tiv e verbs Imperative m ain verbs are suffixed for: speaker number: singular, plural subject person-nuinber: 2.singular, Z.diial, 2.plural aspect: perfective, imperfective There are three diflerent forms o f positive im perative verbs, one expressing an action to be done in the immediate future, and two others expressing an action to be done in the more distant future. However, there is only one form of negative imperative verbs. (M am ainde has only two positive imperative forms, one strong, the other more gentle, almost a permissive. There is also one negative imperative form.) In reported speech and reported th o u g h t quotes, the main verb takes on a different set o f aspect markers from those used when the m ain verb is in an ordi­ nary statement. T hus contrast: ( 11) wa%o’-na'-tu'-wa* w ork-1Sg-FUT-IMPERFV.STATEMENT ‘I will w ork.’ ( 12) wa% o’-n a'-tu '-w i' w o r k - 1Sg-FUT-IMPER FV.QUOTE ‘“ I will w o rk ” , he said.’ 10 Nambiquara 111 Table 10.4 Nambiquara suhordimte verb inflections Verb-like inflections: 1 temporal sequence 2 conditional -k?ai^na%tu^ (highly probable), -k?e^la^te%?ai’ (less probable) 3 adversative -ta' 4 high-level switch -ín a % a't? e' Nominal-like inflections: 5 spatio-temporal-circumstantial 6 (13) -nü^Ia^,-te % a ',-k a ’tu \-te ^ k a % a ‘ causal -tàu^à' -ha^k?ai’, -jut’stf, -jau’s?u^ -k?e’su^ wa%o^-na'-tü*-?à' w o rk -1Sg-FUT-IMPERFV.THOUGHT ‘“ I will w ork”, I think.’ (or ‘I intend to w ork’) 3.1.2 Subordinate verbs Subordinate verbs fortn the nucleus o f subordinate clauses. They can only take indicative m ood, and they are suffixed for subject person and number, a n d speaker number, exactly like indicative m ain verbs. But the rem ainder o f the subordinate suffix strings are different. Some subordinate verbs end with nom inal-like inflec­ tions, b u t others do not. We thus divide the categories o f subordinate inflection into two subclasses. These inflections are given in table 10.4. In the first subclass, th a t o f the verb-like inflections, the subordinate verb consists o f a verb root followed by person-num ber sufiixes and finally a subordinate inflec­ tion. T here is no possibility o f any further inflection following the subordinate inflections. Specifically, none o f the tense-aspect-evidentiality inflections found on main verbs, nor any o f the definiteness-dem onstrative-evidentiality-causality suffixes found on nominalizations, are possible on any o f the subordinate verb forms o f this first subclass. N o r do the subordinate verbs o f this first subclass fill syntactic slots that are norm ally filled by nouns. In the second subclass, th a t o f the nominal-like inflections, the forms are quite different. There is firstly the verb root, followed by the person-num ber suffixes, and then by the nominal-like subordinate inflection. Specifically, o f the four causal suffixes, th e last three ja i^s? i^, k ? é s iê all end in -si^, the indefinite ending for nouns. T he last m orphem e -k?al^ on the first causal ending -hcêkîai^ is also fre­ quently found on nouns - for example. 278 Ivan Lowe (14) wês^-a^-k?ai^ child-DEF-CAUSAL ‘The child therefore [I will go back, pick her up and return].’ In the text from which this example was taken, the speaker had been carrying his small child in his arm s on a journey but both had been caught in a violent rainstorm. The father left his child at a shelter in the care of others and carried on to a village to fulfil an errand. Now, however, with his errand accomplished a nd the rain having also abated, the father feels he can attend to his child. This is when he says (14) ‘The child THEREFORE I will go back, pick her up and return.’ In the sequence of things that he had to do, the time had logically come for him to deal with the child. It is difficult to distinguish between the meanings o f the tem poral sequence suffixes. If the spatio-tem poral-circum stantial option is chosen, there are further options o f evidentiality possible. The meanings o f the four causals have been dis­ cussed at length in Lowe ( 1990). They relate to mental motivation (e.g. I saw the rain coming, I thought that I didn’t w ant to get wet, so I went hom e) and situational motivation (e.g. I feel hungry, so I will go hunting), am ong other factors. Here is one example to illustrate the usage o f each o f the verb-like subordinate clauses. 1 Temporal sequence. Both the subordinate verb and the main verb describe actual events in time sequence in the story. (15) í-’ha^no'-na'-ka^tu^ a ’-na'-hé^-ra^ approach-1Sg-SEQUENCE sh o o t-1Sg-MID.PAST. EXTERNAL-PERFV ‘Coming up to it [the pig], I shot [it].’ 2 C onditional. The conditional clau.se describes a hypothetical event or state o f affairs. The main clause which almost always follows the con­ ditional clause, describes a situation contingent upon the truth o f the conditional. (16) hi^ye-’-k a ’lo^-a^ ?wa^-a^-k7ai^na’ntu^ plane-CL:FLAT.SHEET.LiKE-DEF come-descend-CO ND i^sa^?yau-’-sa^-lhun'-nha^-w a^ rise-lsgO-want-PRES.lNTERNAL-lMPERFV ‘If th e a e ro p la n e la n d s I w a n t to fly [in it].’ N ote that the aeroplane is classified as flat-sheet-like because of its wings. (See §3.3 for more inform ation on the use o f classifiers.) 3 Adversative. In the following example, the subordinate or adversative 279 10 Nambiquara clause comes first and describes a situation. The main clause follows and describes a situation contrary to the expectation o f the first. (17) a^lü'-ai^li^ í ’h i t '- t a '- t a ' ta p ir-th is chase-Isg-ADVERSATIVE á^n ü '-s?á^ d iv e-a n d a-’?yo^ha^7ti'ha^?ti^sai' k?i^ ?ai^-lh?a^ swim.across go-and a u n ’-sa^-he^-ra^ escape-1sgO-REC. past , in t e r n a l - p e r fv ‘I chased this tapir, but it dived (into the river) and swam across, and escaped from me.’ 4 High-level switch. The subordinate ending is -?na^ha't?e'. It is only used to signal a m ajor switch in topic or in participant attention within a narrative discourse. (M am ainde has a form with essentially the same discourse function.) (18) ?ï^-ya%in'-tâu^la^ y a ’hin'-?na^ha't?e' go.aw ay-2dl-w hen 2dl-HlGH.LEVEL.SWiTCH dai^na^-sa^-k?ai-’lhu^ h â i’-s?à^ Isg.D E F .cuR R -m o re.to .co m e-th erefo re o n ly -a n d s?a-’- n a '- tü ' stay-1 sg -intend ‘W hen y o u tw o g o aw ay I, o n th e o th e r h a n d , in te n d ju s t to sta y (h ere ).’ In the preceding textual context o f this example, the speaker has been talking about the plans o f his two hearers. Now-he switches over to talking about him self and what he plans to do. This latter topic takes up much o f the discourse that follows. 3.2 Adjectives Adjectives are inflected for the categories o f tense, aspect and evidentiality just like third-person verbs. In m ood they are limited to the indicative. (19) wain-’-na-’-ra^ Straight-ACTION.CURRENTLY.OBSERVED.BY.BOTH S.&.H-PERFV ‘It is S tra ig h t.’ ( o r ‘c o rre c t’) (20) w i'-na^ -ra^ gOOd-ACTION.CURRENTLY.OBSERVED.BY.BOTH.S.&.H-PERFV ‘It is g o o d .’ 280 Ivan Lowe Table 10.5 N am biqm ra classifiers kat’ SU -’ ki-’ en‘ yau^ n?àn^ ka»lo’ nün^ thín^ ko?’ jah-Mo^ ka?’lu^ yut-’ jul^ jau’ k?e’ te2 stick-like bone-like round hole-like or stomach-like liquid leaf-Iike flat-sheet-like powder-Iike house-like country man woman work situation word or thought patient agent or attribuand Adjectives could be regarded as a subclass o f verbs with certain limitations on their pronom inal and m ood inflections. To convey the inform ation that a first- (or second-) person referent is the bearer o f an attribute, the following kind o f con­ struction is used; (21) nün’kün^-te^-sa'-wa^ Strong-ATTRIBUAND-IsgC-IMfERFV ‘I a m s tr o n g .’ Fn (21) the IsgC is a first singular copula form (see table 10.7). 3.3 Nouns Noun roots, like verb roots, may end in a vowel, or in one o f the consonants n, t, h, /, s, k. N oun stems often involve a classifier suffixed to the root. T he classifier often describes th e shape o f the n o u n referent but there are a few m ore abstract classifi­ ers as well. A list o f classifiers is given in table 10.5. The last four classifiers listed in table 10.5 can only occur in nominalizations. All o f the others can occur in both noun stems and in nominalizations from verbs. (M amainde has classifiers covering essentially the same semantic categories but some o f the M amainde forms are very different.) Below are some examples of classifiers on noun stems. In each case the classifier comes immediately after the noun root. 281 JO Nambiquara (22) wa-’lin^-su-’-siimani0C-CL;80NE.UKE-INDEF ‘m anioc root’ (indefinite form ) (23) h u tí-én'-su^ shooter-CL;HOLEXIKE-INDEF ‘sh otgun’ (the gun barrel is the h ole) (24) huk^-ki-’-su* ShOOter-CL:ROUND-INDRF ‘b ow ’ (th e arc o f th e bow is ro u n d ) Classifiers can be used as deverbal nominalizers: for example, (25) s?i^ha^ îyau^-ain'-thî^-na^ h o u se live-3non.sg-CL:HOUSE.nKE-DEF ‘The house they live in.’ (26) wan^ta^ e^e-’-k?i^-sain'-jau7^-ai^na^ word speak-to-they.to,m e-CL:woRD-this.definite ‘This word th a t they spoke to me.’ Classifiers also occur on adjectives used as modifiers in an NP: for example, (27) wa*la* c lo th wi^win^-ka^lo’-a^ b liie -c i- ; f l a t . s h e e t . u k e - d e f ‘the blue cloth’ They are no t used with an adjective when it is em ployed predicatively: for example, (28) wä^la^ cloth wiVin^-na^-ra^ blue-ACriON.CURRENTLY.OBSERVED.BY.BOTH S.&.H-PËRFV ‘The cloth is blue.’ Classifiers are also used w ith num erals: for example, (29) a’l ä V so'l?i^ k i’-h a'li' hut^tit’-ta'-hi^-ra^ parrot only CL:R0UND-tWO shoot-lsg-MiD.PAST.EXTERNALPERFV ‘I shot only tw o parrots.’ (parrots are viewed as round objects) Classifiers also figure frequently in the syntax and will be referred to again in §4.5. N ouns can be suffixed for definiteness and causality. D efiniteness has three values - indefinite o r definite {-cf) a t conditional (-la^). Indefinite nouns can be 282 Ivan Lowe Table 10.6 Nambiquara nominal endings -a^ -ai^na^ -in-’ti^ -ait-’ta’li^ -ait-tà^ -nü'tà^ -nO'tai^na^ -au-’të?'!à^ definite, unmarked definite, current observational, recent past, give: observational, mid past, given observational, mid past, new inferential, definite, unmariced inferential, current quotative, mid past, given further inflected for causality but for no other categories. Definite nouns, however, can be further inflected for demonstrativeness, spatio-tem porality, evidentiality and causality. However, only a limited num ber o f com binations o f time and cvidentiality have been observed even in a very large corpus o f nouns. F or evidentiality, three categories, observational, inferential and quotative, have been observed on nouns, but for newness, only given-inforniation inflections occur. O bservational evidential­ ity seems to exclude ‘current’ (where current means ‘at the time and place reached in the discourse’). A nd curiously, in folktales it is the inferential endings which are the ones m ost frequently found on nouns (where one m ight expect that a quotative evidentiality was m ot« appropriate). The endings th a t have been found on N am biquara definite nouns are given in table 10.6. Since the glosses on these sufTix sequences may appear rather abstract, here are a few exatnples o f their use on actual nouns: (30) wa-’lin^-su^’-a2 mani0C-CL:B0NE.tIKE-DEF ‘the m an ioc root’ (31) wa^’lin-’-su^-ai-na^ mani0C-CL:B0NE.L!KE-DEF.CURRENT ‘This m anioc root which we b oth see before us now.’ (32) wa'linJ-su^-n’ti^ mani0C-CL:B0HE.LIKE-OBSERVATIONAL.RECENT.PAST.GlVËN ‘The m anioc root that both you and I saw recently.’ (33) wa-'lin^-su^-ait^ta^’ii^ manioC-CL:BONE.LIKE-OBSERVAT10NAL.MlD.PAST.OtVRN ‘The m an ioc root that both you and I saw som e tim e p a st.’ 283 10 Namhiquara Table 10.7 Namhiquara pronouns Isg liion.sg.inc! Inon.sg.excl 2 sg 2 dl 2 pl 3sg.masc 3sg.fem jnon.sg (34) Subject Object Copula -a' -kj* -sî'n a ’ -in' -ya’h in '. -yar’lhin'- -saJ~-sa> -nin* -sa^sin' -?na^ -îna^li^ -îna^ii-’ -sa' dai^na^ da*wâ'ki%a'ai^na^ -sin' 0 0 0 0 -ain' -ain' wîâi^n wTä^nQn’ka^tai^na^ w?à 2nün’ka 2tai% a 2 jah 'la i^ a ^ ta‘?ka’?lai-na^ tûn'ka^tai^na^ -la' -la' wa-’lin-'-su-’-aitñä^ manioc-CI-;BONE.LIKE-OBSERVATIONAL.MtD.PAST.NEW ‘The m anioc root that 1 saw (but you didn’t) some time past at some distant p la ce/ (35) wa-’lin^-su^-nO'ta^ mani0C-CL:B0NE.UKE-lNFERENTIAL,DEF.UNMARKED ‘The m anioc root th a t must have been a t some time past, as inferred by me (but n o t by you).’ 3.4 Pronouns The suflRx pronouns and the free pronouns are given in table 10.7. Note that second person distinguishes singular, dual and plural, while first and third persons have ju st two numbers, singular an d non-singular (covering dual and plural). T h e reflexive p ronoun suffix is -nha- and the reciprocal is -n h yé, both valid for ail persons- 3.5 Adverbs T here is really no well-defined class o f adverb words. However, adverbial func­ tions can be realized by adjectival roots prefixed to a verb stem. Such adjectival ro o ts are true prefixes; they obey w ord-internal m orphophonem ic rules. For example; 284 Ivan Lowe (36) w i'-lâ u ^ ? F -s a ^ -h ë '-ra ^ gOOd'Sleep -IsgO-MID.PAST.INTERNAL-PERFV ‘I slept well.’ C om pare this with: (37) âu-’?i^-sa-’-h ë'-ra^ sieep-lsgO-MlD.PAST.INTERNAL-PERFV ‘I s le p t.’ 3.6 Interjections Some examples are hai^yo^'' ‘Right!, O K !’, la^wi' ‘N o way!’ A s expected, these forms do not take affixes, and do n o t conform to the regular phonological template o f th e language. 3.7 Ideophones Ideophones are words whose sounds depict the action they refer to. Examples are (ka^Iagn-’] ‘sound o f a heavy fall’ and [taubm ’] ‘sound o f footsteps’. The unm arked constituent order for both main and subordinate clauses is AOV, SV. The gram matical relations A, S, O are m arked by cross-referencing pronominal suffixes on the verb; A and S being m arked by subject pronouns, O by object pro­ nouns. Topicality considerations change the unmarked word o rd er so that the topic comes initially. 4.1 Main clauses M ain clauses can be intransitive m otion, intransitive non-m otion, stative intransi­ tive, descriptive, existential, transitive. Intransitive m otion clauses describe the m otion o f the subject as in: (38) ka?-’na-*haW na' â^ni^-ha'-ra^ this.m orning run-lsg-PERFV ‘This m orning, I ran.’ Intransitive non-m otion clauses describe the intransitive non-m otion action o f an agentive subject as in: 285 10 Nambiquara (39) âu^?P-na'-tu'-w a^ s l e e p - 1Sg-FUT-IMPERFV ‘I am going to sleep.’ Stative intransitive clauses describe the state experienced by a patient subject as in: (40) dai^ri^ kat-’jan^jah'lai^na^ yuh^li^-sa^-nha^-wa^ Isg.VAGUE.cuRR white.m an fear-lsgO-PRES.iNTERNALIMPERFV 'I am afraid o f the white m an.’ Descriptive clauses describe an attribute o f an individual (or attribuand) as in; (41) sa?% e’na^ forest wi'-na-’ra^ good-is ‘The forest is good.’ Existential clauses assert the existence o f som ething as in: (42) a^lén-’su^ arm adillo.hole na^hê^ra* vv'as ‘There was an arm adillo hole.’ C opula clauses do n o t have an overt surface copula verb as such, but the copula relationship is expressed by suffixing the noun ro o t predicate w ith the copula pro­ nouns (see the third colum n in table 10.7). For example: (43) daigna1 a^hoh^li-’-sa'-wa^ S g .D Ë F .C U R R e x p e r t- ls g C O P U L Ä - lM P E R F V ‘I am an expert.’ (a^hoh-’li^su^ is the underlying noun ‘an expert’) Transitive clauses have the unm arked constituent o rder AOV as, for example, in; (44) dai^ri^ d?uh'-ai^na^ të^-a'-tu'-wa^ ISg.VAÜUE.CURR WOman-DEF.CURR take-lsg-FUT-IMPERFV ‘I will take (m arry) this woman,’ However, discourse considerations often front the object N P and give an OAV order. T here are, in fact, n o case m arkings o f any sort on the core syntactic N Ps to distinguish functions such as A, S and O. However, in a few instances where the subject also has high topicality, the “highly topical subject’ suffix, -h îfsen ^stê, is used, as in: (45) w ê ’-sa^ k w |-’lha^-kaun-’-jah'-lait'tâ^-h?i^sen^su^ child-DEF name-young-male-MiD.PAST-HiGHLY.TOPîCAL.suBJECT 286 Ivan Lowe su^lhà^-k?i^-sain' na^-ri' m urder-to-he.to.m e act io n .c urrently .observed . by.both .s.& h .- PERFV Free transiation: ‘It was M r K wÿlha^ the younger who m urdered my child.’ Literal translation: ‘The child, it was M r Kwä^lha^ the younger who m urdered him of me.’ {Note incidentally that the present tense and the ‘observed by both speaker and hearer’ evidentiality is used by the speaker to report an action that neither the speaker no r the hearer ever saw! This is a rhetorical device to m ake the report m ore vivid.) There are also many other cases in which the preferred strategy is not to use a single clause with two third person N P constituents, but rather two successive clauses with a repeated verb, introducing first the A o f the action and then the O in the two clauses. For example, (46) ä^nO^-a^ ka^la'tVi^ anW -so'?i^-hyu'-nQ ^la^ people-DEF m any by.hand-gather-they.to.him-SEQUENCB ‘M any people gathered up (the corpse) by hand, ?yo^wen^-kah^-la^-ta^-ha^ su^lhà-''hyu'-k?e-’ name-male-DEF-son-DEF m urder-they.to.him -0 -nû'ta^-k?ai-’lu^ -INFERENTlAL.DEF.UNMARKED-therefore so'?iMiyu'-nQ^lagather-they.to.him-SEQUENCE they gathered up M r ?yo^wen^’s son w ho had been m urdered, and sa^nî-^-kîi^-nhyain'-ta'-hê'-ra^ bury-tO-they.tO.him-QUOTATIVE-MID.PAST.lNTERNAL-PERFV they buried him (it was reported).’ D itransitive clauses: there are certainly verbs like ‘give’ and ‘speak’ that can take three non-oblique argum ents, but alm ost never in text do we find a single clause with three overt N P constituents. Rather, we find the inform ation spread over two or m ore clauses as, for example, in: (47) ?yo2wenî-kah-'la24a2-ha2-sa%?ai^lunam e-m ale-son-DEF-therefore M r ?yo^wen^’s son 287 10 Namhiquara kw|^Iha^-kaun-'-jah‘-lai%a^-sa^k?ai’iu^ nam e-young-m ale-this-therefore M r Kw|-’lha^ the younger ï-*ye’-k?i^-nhyain'-ta'-hë'-ra^ say-tO-he-.tO.him-QUOTATfVE-MlD.PAST.lNTERNAL-PERFV he spoke to him íy o ’wen^-kah’la^-ta^-hai^na^ wân^t?i^-k?ai’lu^ sa%o'?i^-n 0^1a^ nam e-m ale-son-this word-therefore take-SBQUENCE î-V '’-k?i^-nhyain ' -te^na'lh?i' say-to-he. to. h im - Q u o T E .o P E N E R This M r ?yo^wen^’s son’s words then which he took and said to him were [‘content o f q uote’]. ‘M r ?yo^wen* says to M r kw |*lha-, M r ?yo^wen^’s words which he takes and says to him were [“content o f quote”].’ In this example, the sim plest translation into English would be in the form o f a single ditransitive clause ‘M r ?yo% en^ says to M r kw ÿlha^ “content o f q uote” ,’ There is the verb ‘say’, the subject (M r ?yo V en^), the indirect object (M r kw ^’lha^), and the direct object (‘these w ords’ plus the content o f the quote). However, in N am h iq u ara, the same inform ation is invariably spread over two clauses as above. T he first clause introduces the speaker and addressee, and has the verb ‘say’ but it is am biguous as to who the speaker is. T he second clause has a verb phrase ‘he take and he say to him ’, and the subject o f this verb phrase is M r Y, and the object is ‘w ords’. It is now finally clear who the speaker is. 4.2 Subordinate clauses Subordinate clauses have the same constituent order as main clauses, and the verb inflections are as already set out under subordinate verbs. Relative clauses consist of a clause whose nucleus is a nominalization. Either the subject or the object o f the em bedded clause can be lïlativized. There is no surface relative pronoun form as such. The relative clause usually modifies a noun head, as shown in the following examples: (48) i^nO^-a^-nu'ta^ persOn-DEF-INFERENTIAL.DEF.UNMARKED o h ’n au ^ ^ above y?au-'-k?i2-jah‘lo2-nû'tâ2 Stay-tO-male-INFERENTIAL.DEF.UNMARKED ‘the person who stayed up high’ (subject relativization) 288 Im n Lowe (49) ä^nO^-a^-nü'tá^ wa-’tïn^ka’lh ? a’lha^ person-DEF-iNFERENTiAL.DEF.UNMARKED anteatef sa^w?e^-k?i^-jah'lo^-nü'tä^ put-tO-male-INFERENTIAL.DEF.UNMARKED ‘the person w h o was put up there by the anteater’ (object relativization ) (50) a^hü^lyau^?a^ water ?u^-k?e^-su^ far-O-iNDEF ‘the water which was far away’ H eadless relatives also exist, as in (51): (51 ) à^ïwân^-kîe-’-su^ left.over-O-iNDEF ‘what was left over’ 4.3 Clause coordination C lause coordination is effected by using either one o f the coordinate endings, - f or -ha'k?aP, in place o f the tense-aspect endings on the verbs in each o f the coordi­ nate clauses. T hen, in m ain clauses, the tense-aspect-evidentiality for the whole coordinate construction is expressed by one o r more suffix strings at the very end of the coordination. Thus, for example: (52) wä^la^ wa^hi-’l-i^ wâ-ho^?-i^-na'-tû'-?â' cloth es wash-COORDINATE bathe-COORDlNATE-lsg-FUTIMPERFV.THOUGfrr na'-hè^-ra* Isg-REC.INTERNAL-PERFV ‘I intend to wash my clothes and to take a b ath .’ (H ere the suffix strings na'íti‘?ä‘ n a 'hë^ré express the tense-aspect-m ode-evidentiality for the w hole coordination.) (53) nfin^t-ai^li^ stf-la'-ha% ?ai^ sa-’nai^-ai^li^ lizard-DEF.REM kill-lsg-COOROINATE armadilio-DEF.REM su^-la'-ha^k?ai^ n a ’-hê^-ra^ kill-1Sg-COORDINATE 1Sg-MID.PASXINTERNAL-PERFV ‘I killed a lizard and an arm adillo.’ It is difficult to get any lower-ievel distinction in the meanings o f these two coordi­ nation endings. Their discourse functions are rather different in that the -h é k îa fi 289 10 Nambiquara ending implies th at there is som e logical connection with w hat has been mentioned in the preceding context. T he -r’ ending by contrast seems unmarked. 4.4 Noun phrases Noun phrases can be descriptive, possessive (alienable and inalienable), appositional or quantifying. Descriptive (54) yu^la- käin^-te^-a^ k n ife b ig -th is -D E F ‘t h e m a c h e t e ’ Possessive (alienable. i.e. optionally possessed) (55) ja h ’la^ huk% ?a^ he bow ‘his bow’ Possessive (inalienable, ie . obligatorily possessed) (56) da^w i' ne’ki^su- our head ‘hum an head’ An inalienably possessed noun m ust take a possessive pronoun, the citation form involving dcPwä' ‘o u r (inclusive)’. Appositional (57) k?a^yuh^?a’ tu ’ha^ game.food honey drink-Isg-FUT-iMPERFV î-’nà^-na'-tu'-w a^ ‘I will drink honey.’ (Here the N P object o f the clause consists o f a generic noun, ‘game food’, and a specific noun, ‘honey’, in apposition.) Quantifying (58) ?yo^ha-’ka' people ‘all (59) th e all people’ ya^na'la^ h a 'li' ja g u ar two ‘two jaguars’ 290 Ivan Lowe 4 .5 N o m in a liz atio n s All classifiers may, as m entioned in §3.3, also function as nominalizers: (60) s ? a ^ - a in '- th t’-nastay-3non.sg-CL:HOUSE.LiKE-DEF ‘th e h o u s e (o r village) w here th ey s ta y ’ As m entioned before, there are four classifiers which do not serve to derive noun stems from n oun roots, but which only function as nominalizers, for example, (61) a2yen^-k?a^ e % ? i'-sa n '-ju t^ -ta ^ things-DEF T --a'-tu '-w a^ speak-you.to.me-siTUATiON-DEF see-lsg-FU T-iM PER Fv ‘The situ a tio n o f the things y o u sp o k e to me a b o u t. I'll tak e a lo o k a t.’ Classifiers figure quite prominently in the syntax. They can function like ‘a n a­ phoric pronouns’ to refer back to an item mentioned previously in the text, as, for example, in; (62) hï'na% u^ wa^lin^-su-’-nti^ today i ’-a'-ra^ m an io c-C U B O N E .L IK E -O B S E R V A T lO N A L .R E C .G IV E N plant-1 s g -P E R F V ‘Today I planted the m anioc roots that we both saw earlier in the day. ? y ä n 'ta ' but su-’-ha'li' CL;BONE.LiKE-few ot's?â-’ â^-sï'na'-ra^ e x tra lea v e -1non.sg.excI-PERFV But we left behind a few extra roots. na'-su-’-ai*na^a^-k?ai*Ihu^ lsg.POSSV-CL:BONR.LiKE-this-therefore These my roots therefore, kîa-’na^haHai^nà^nta^ te^-yà'-îne^ to m o rro w this-again-MANNER p - a '-tû '-? à ' na'-he^-ra^ p la n t-1Sg-FUT-IMPERFV.THOUGHT Isg.POSSV-REC.INTERNAL-PERFV I intend tom orrow to plant as before [i.e. as I ’ve done today].’ Note, in the vernacular of this example, the classifier sié ‘bone.like’ occurs three times (once on each line o f the surface forms, where its gloss has been capitalized to facilitate reader recognition). It is easily seen that the second and third occur­ rences o f Í 1Í-* refer anaphorically to the first occurrence. 10 Nam biquam 291 However, I do n o t analyse classifiers as pronouns as they can never occur as pro­ nom inal affixes on m ain verbs. M oreover classifiers can be possessed, but unlike pronouns they can never be possessors. Thus, from the above example, we have: na'-su^-ai^na^sa^-k?ai'lhu^ ls g .P O S S V - C L :B O N E .L iK [ i- th is - lh e r e f o r e where the classifier j i / ‘bone-like’ refers anaphorically to w i^lii^stJn lf ‘the m anioc roots’, first introduced in the first line o f the text in the example, and the na' ‘Isg.POSSV’ is a possessive pronoun. T he whole expression is therefore a possessed nom inal, and the classifier -su^ is possessed. B IB L IO G R A P H Y Eberhard, D. 1995. Mamaindé stress: the needfor strata. Dallas: SIL and University o f Te.xas at Arlington. Kroeker, B. J. 1972. ‘M orphophonem ics o f Nam biquara’, Anthropological Linguistics, 14.19-22, Kroeker, M. H. 1975. ‘Them atic linkage in N am biquara discourse’, pp. 361-8 o f The thread o f discourse, ed. J. E. Grimes. The Hague: M outon. Lévi-Strauss. C. 1948. ‘The N am bicuara’, pp. 361-9 o f Handbook o f South American Indians, vol. 111. ed. J. H. Steward. Bureau o f American Ethnography, Bulletin 143. W ashington, D.C.: Smithsonian Institution. Lowe, I. 1972. ‘On the relation o f formal and sememic matrices with illustrations from N am biquara’, Foundations o f Language, 8.360-90. 1986a. ‘Topicalisation in N am biquara’, pp. 131-47 o f Sentence initial devices, ed. J. E. Grimes. Dallas: SIL and University of Texas at Arlington. 1986b. ‘Participants in N am biquara myths and folktales’, pp. 187-98 of Sentence inilial devices, ed. J. E. Gritnos, Dallas: SIL and University o f Texas at Arlington. 1990. ‘Cause and reason in N am biquara’, pp. 543-73 o f Amazonian linguistics; studies in lowland South American languages, ed. D oris L. Payne. Austin: University o f Texas Press. Price. P. D. 1976. ‘Southern N am biquara phonology’, UA L, 42 1978. ‘The N am biquara linguistic family’, Anthropological Linguistics, 20.14-37. 11 Arawá R . M . W. D I X O N 1 INTRODUCTION* The A raw á language family was first posited by Brinton (1891:293) on the basis of vocabularies o f A raw á and Paum ari. It consists o f six languages, shown in table 11.1. They are located around and between the Purus and Juruá rivers, m ajor south­ ern tributaries o f the A m azon. All are in the southern p art o f the state of A m azonas, Brazil, except for the language w ith the m ost speakers, K ulina, which extends to the neighbouring state o f Acre and over into Peru. The accom panying m ap shows the approxim ate locations o f the A raw á languages and dialects. For A raw á, after which the family is nam ed, all we have is about fifty words taken down by the English explorer Chandless in 1867 (these are rather well transcribed); the tribe was wiped o u t shortly afterwards by an epidemic o f measles. The tribes speak­ ing M adi, Deni and Kulina are semi-acciilturated - wearing clothes and using guns, salt, sugar, etc. - but still live a fairly traditional life within the jungle, supporting themselves by slash-and-bum agriculture, fishing and hunting. M ost m em bers o f these tribes speak a little Portuguese (or Spanish, in Peru), as a second language. The Sorow ahá were only contacted in 1980; they are m onolingual and live a fully traditional life (w ithout clothes or guns). The Paum ari have h ad m ost intensive contact w ith non-Indians. Only the older people speak Paum ari, the younger ones using ju st Portuguese o r a m ixture o f Portuguese and Paum ari. M aterials available are o f uneven quality. Dixon and Vogel have an extensive gram m ar o f Jaraw ara in an advanced stage o f preparation. SIL teams have done som e linguistic work on the Jam am adi and Banawá dialects o f M adi, on D eni and on K ulina (from the Peru side), producing m anuscript sketch gram m ars for Deni and K ulina and draft dictionaries for all four varieties. For Paum ari there is a ' Thanks are due to the following missionary linguists for making available draft dictionaries and for discussing grammatical topics: A lan Vogel (Jarawara), Shirley Chapm an (Paumari), Robert and Barbara Campbell (Jamamadi), Ernest and Barbara Buller (Banawá), Edson M assamiti Suzuki and M árcia dos Santos Suzuki (Sorowahá), and Fred and Paula Boley (Deni). 294 R. M. W. Dixon Table 11.1 The /Ira tra language fam ily 1 Paumari (c. 600, only c. 200 speak the language) 2 M adi, spoken by three tribes, each with its own dialect: Jarawara (c. 150) Jam atnadi (i. 190) Banawá (c. 80) 3 Sorowahá (c. 100) Deni-Kiilina subgroup 4 Deni (c. 1,000) 5 Kulina (or M adiha or M adija) (c. 2,500) 6 Arawá (extinct since about 1880) Table 11.2 Cognate percentages between the Arawá languages Paumari 50 34 37 43 40 54 Madi Sorowahá 61 60 79 34 31 53 Deni 76 72 Kulina 67 lengthy gram m atical sketch, o f good quality, by Chapm an and Derbyshire (1991); C hapm an is currently putting the final touches to a draft dictionary. M issionaries from Jovems C om U m a Missao (JO C U M ) have done some preliminary linguistic work on Sorowahá, with a short dictionary in draft form and a little grammatical inform ation (we have Sorowahá data relating to only some o f the points dealt with below). M onserrat and Silva (1986) and Silva and M onserrat (1984) have publi.shed a short gram m ar and dictionary o f a dialect o f Kulina spoken in Acre. Percentages o f cognates on a 115-item lexical list (80 nouns, 4 adjectives, 31 verbs) are shown in table 1 i. 2. N ote that the A raw á figures are based on Chandless’s word list and the denom i­ nators vary between 19 and 25; these figures have a lower reliability than the others. The three M adi dialects - Jaraw ara, Jam am adi and Banawá - share about 95 per cent vocabulary, have very similar gram m ars, and are fully m utually intelligible. T here are enough gram m atical ditferences between D eni a n d K uhna to estab­ lish them as distinct languages, but they plainly constitute one subgroup. N o other subgrouping is justified on the inform ation currently available. M adi has strong lexical an d gram m atical similarities w ith D eni and K ulina whereas Paum ari is m ore divergent. As m entioned in chapter 1, there m ust have been a great deal o f 295 I l Arawá Table 11.3 Likely consonant system fo r proto-Arawâ bilabial voiced imploded stop voiced stop voiceless stop aspirated voiceless stop voiced affricate voiced affricate/fricative aspirated voiceless affricate nasal liquid semi-vowel 'b b p p'' m apico-dental/ alveolar 'd d t t'' dz ts ts'' n r apico-palatal dorso-velar glottal g k k'' ? t{ (w) h w trib al merging d uring the past few hundred years. T h e Paum ari and Sorow ahá tribes m ay be the result o f such mergers, one o r m ore tribes speaking an A raw á language com ing together w ith a tribe o r tribes speaking languages from other families; this would have resulted in a significant substratum vocabulary o f nonA raw á origin. 2 PH ONO LO GY There is a straightforw ard (C)V syllable pattern in all languages. For proto-A raw á, 'I have provisionally reconstructed the consonant system set out in table 11.3 (based on D ixon, forthcoming-b). Chapm an and D erbyshire (1991: 347) m ention that in Paum ari there is a contrast between a flap vibrant [r] and a retroflexed grooved reverse flap vibrant [f] (although this has a very low functional load). These two sounds may have been present in proto-A raw á. All o th er languages have a single liquid h i, usually with both rhotic and lateral allophones. P aum ari has the m ost conservative system. Both *ts and *ts'‘ have become s; *t has becom e a glottal stop medially but is retained initially; */)''has been replaced by f , and *dz has become d?- The Paum ari vocabulary includes a substratum (which may com e from another A raw á language) where *p> ?, *dz> s, *pf'>h and *s> h (see D ixon 1995; 291-3). D eni, K ulina and M adi have lost the im ploded voiced stops, w ith initially a n d > /j medially, and *’d > t in all positions. They have also lost g (generally replaced by w) and m ost o r all occurrences o f the glottal stop. The only other change in Deni and Kulina is M adi has undergone further changes. First, aspiration was lost, w ith *p''>p. 296 R. M . K Dixon Table 11.4 The consonant system o f the Jamamadi and Banawà dialects o f M odi apicodental bilabial voiced stop voiceless stop voiceless fricative nasal liquid semi-vowel apicoalveolar d t b laminopalatal dorsovelar k h s <!> m glottal i n r (w) w Table 11.5 ProtoArawá vowels front back high mid low */'■>?, * !é> k. Then p became a bilabial fricative 4>, all o f * t^ and * t f fell together as s] and *dz became a lam ino-palatal stop, ^ (with semi-vowel y as an allophone). The co n sonant system o f the Jam am adi and Banawá dialects is set out in table 11.4. In the Jaraw ara dialect there has been a further change, with d and t falling together as t, giving ju st 11 consonants, as com pared with 2 1 in Paum ari and 17 in Deni and Kulina. Sorowahá has undergone similar changes to M adi, w ith the addition o f <p>/i. However, it retains *g, and probably the contrasts (as s /f) . The Arawá word list also includes som e instances o f g; for this language we can infer *p> f, * p * > /o r Ii; * d s> s and *’d > t. Proto-A raw á h ad a system o f four vowels, given in table 11.5. These are m ain­ tained in Deni, K ulina and M adi. In Paum ari and Sorowahá, e has been lost, gen­ erally being replaced by i in the third syllable of a root and by a elsewhere. It is unusual in A m azonia to find a vowel system that lacks a high central vowel HI. In fact, younger speakers o f Sorowahá have innovated [i] as an allophone of any of the three vowels - HI, /a/ and loi - in unstressed position. In the three languages in which e is retained, it plays a pervasive role in engender­ ing assimilation o f a preceding a, e.g. *ama-ne ‘blood-M A SC ’ has become eme-ne in Deni, Kulina and M adi (it is ama-na in Paumari). I I Arawà 297 Table 11.6 L exem es fo r ‘m oon, 'sun and ‘thunder’ in Arawá languages sun Paumari Jamamadi Jarawara Banawá Sorowahá Deni ICulina Arawá masiko abariko abariko abariko masiki abadziko abadziko masiko safini mahi bahi mahi masiki mahi mahi m ahi ‘thunder’ bahi bahi bahi mahi bai bahi s'*omaroro (not known) T here is an interesting link between semivowel w and vowel o (this was first pointed o u t in A datns Lichlan and M arlett 1990), T here is no contrast between oand wo- at the beginning o f a word, or between -owV- and -oV- or -Fwo- and -Vo(where K is a vowel other than o) medially, and there is no example o f -m w-. However, we d o get a contrast between V- and wV- initially (e.g., *adami ‘hill’ and *wadaml ‘dream ’) and between and medially. It is possible to treat w and o as allophones o f a single phonem e (which we can call lOf) b u t the cost o f this ‘econom y’ is that a syllable boundary must always be marked. T hus /O ina/ could be either oinahwina o r wina, which are different words. The first w ould have to be show n as /O .ina/ and the second as /O ina/, where •-dary. indicates a syllable boun- O ne characteristic o f A raw á languages is a profusion o f lexical homonymy, in which speakers appear to delight. The neutralizations that have taken place (e.g. *'d, *d, *i an d all falling together as t in Jaraw ara) have added to this number. For instance, in Jaraw ara there are three nouns }i4>o: (a) ‘ham m ock’,(masc.); (b) ‘fire, firewood’ (fem.); and (c) ‘m uriti palm (M auritia vinifera)' (masc.). A nd there are four verbs saa: (a) ‘shoot fish with arrow ’; (b) ‘release, let go’; (c) ‘peel, strip’; and (d) ‘vom it’. A raw á languages are also prone to referential merging, as illustrated in table 11.6 for the words for ‘m oon’, ‘sun’ and ‘thunder’. It will be seen that the Jam am adi dialect o f M adi m aintains the original m ahi ‘sun’ and bahi ‘thunder’. In Jaraw ara, bahi ‘th u n d e r’ has been extended also to cover ‘sun’ while in Banawá m ahi ‘sun’ has been extended also to cover ‘thunder’. In Sorowahá, m asiki ‘m oon’ has been extended also to cover ‘sun’. (Paum ari appears to have innovated safini for ‘sun’, but it does have a lexeme mahi m eaning ‘day’ o r ‘tim e’ or, as a possessed noun, ‘life’, as in ‘my life’.) 298 K M .W . Dixon W ORD CLASSES Arawá languages have two open lexical classes - nouns and verbs. They typically have a small closed class of adjectives, which can only m odify a noun within an NR A class o f about fourteen adjectives is given for Deni and one o f similar size for Jarawara. These include reflexes of *'bo’de ‘old’ and *dsa'di ‘new ’ and also terms for ‘big’, ‘sm all’, ‘o th er’. There are closed classes o f pronouns, demonstratives, post­ positions, clause linkers and interjections. These divide Into two subclasses - free and possessed. Each free noun has an inher­ ent gender, m arked not in the form o f the noun but in the form o f modifiers within an N P and suffixes to the verb. There are two genders - feminine, w hich is the func­ tionally unm arked term (for instance, all pronouns are cross-referenced as femi­ nine), and masculine. A possessed noun follows a free noun o r pronoun and agrees with it in gender. There can be well over 100 possessed nouns, including body parts, parts o f trees, and such things as ‘nam e’, ‘hom e’, ‘dream ’, ‘food’, ‘path’ and ‘container’. There are also possessed nouns referring to orientation, e.g. ‘in front o f ’, ‘on top o f ’, ‘inside’, ‘by the side o f ’; these correspond semantically to prepositions o r postpositions in o th er languages. In proto-A raw á, gender was marked on possessed nouns by the suffix -ni for fem­ inine and -ne for masculine. All languages retain feminine -hi, except M adi where it has been lost, often leaving a phonological trace, e.g. *noko-ni> noki ‘eye + F E M ’, *m ata-ni> m ate ‘buttock, tree stum p-i-FE M ’. In all languages masculine *-ne has been lost from some, but not all, possessed nouns. In D eni, K ulina and M adi it has often engendered vowel assimilation before being lost, e.g. *mata-ne>m ete-ne> m ete ‘buttock, tree stum p +-MASC’. A s a result o f these changes, in M adi about half the possessed nouns have identical feminine and.m asculine forms, e.g. *’dd'dinil*'da'di-ne> tatiltati ‘head’. In all other languages the gender distinction is m ain­ tained, through the retention o f suffix -ni. (There is a full discussion of possessed nouns in Dixon 1995.) Just in Paum ari, D eni and Kulina there is a further division o f nouns, into those that require a cross-referencing prefix ka- on the verb and on some nom inal m od­ ifiers (when the n oun is in pivot function in the clause) and those that do not. A rel­ atively small num ber o f nouns belong to the ka- class; they include some possessed nouns, and free n o uns o f a num ber o f semantic types. There is a partial semantic basis - containers and things in containers (e.g. ‘pan’, ‘canoe’) and things m ade up J I Arayvá 299 o f parts (e.g. ‘house’, ‘plane’) or o f particles (‘sugar’, ‘banana m ash’) are generally ka- class. Some nouns can be ka- o r noa-ka-, with a meaning difference, e.g. in Paum ari sa?ai is ‘hand’ as ka~ class a nd ‘finger’ as non-ka-, while siho is ‘fire’ as kaand ‘firewood’ as non-ka- class. 5 NOUN PHRASE STRUCTURE The general structure o f a noun phrase is: (a) alienable possessor. This can be a full noun phrase, and is followed by (b) noun or pronoun. This supplies the gender - and the ka- specification (c) adjective(s) - one or m ore members o f the closed class (some show dis­ possessive m arker *kha. - for the whole noun phrase. tinct gender forms). (d) possessed noun(s), agreeing in gender w ith the head noun. T here may also be a plural m arker, e.g. deni in both Deni and Kulina. This structure can be illustrated by a sample noun phrase in Jaraw ara. N ote that the n oun jom ee ‘dog’, is both alienable possessed and inalienable possessor: ( 1) (b) (c) (d) [fana (a) kaa ] jom ee bote teme woman(fem) possv dog(masc) old foot + masc ‘the w oman’s old dog’s foot’ W ithin an N P showing alienable possession it is the possessed noun which is head. However, the question o f w hat is the head o f an N P which consists o f a pronom i­ nal possessor and a possessed noun is not a simple m atter. An N P such as o-mano ‘my lower a rm ’ triggers fem inine agreement on the verb, relating to the Isg o- (all pronouns are cross-referenced with the unm arked gender, feminine). However, the N P counts as 3rd person on other criteria (e.g. in an O -construction where the verb will show prefix hi- if both A and O argum ents are 3rd person - see §10 below). T hat is, from the point o f view o f gender agreement the pronom inal possessor appears to be the head, b u t from the point o f view o f person agreement it is the possessed noun that appears to be the head. D em onstratives are typically apposed to noun phrases (rather than acting as direct modifier o f a head noun). T here is a limited set o f enclitics or postpositions th a t may follow a n oun phrase, indicating its function in the clause. These include: 300 R. M. W. Dixon (i) accusative -ra can follow a noun phrase in O function in Paum ari, in Sorowahá and in the Jam am adi dialect o f M adi (in Jam am adi it may only attach to an O N P if the A argum ent is third person). In the Jaraw ara dialect it is now restricted to occurrence on pronouns. (il) ju s t Paum ari has enclitic -a with a locative and instrum ental meaning. It is also used to m ark a noun phrase in A function in an 0-co n stru c­ tion (see below). (iii) in P aum ari, has directional m eaning ‘to ’. In D eni, Kulina and M adi it has a much wider function, covering some o r all o f locative (‘a t’), allative (‘to ’), ablative (‘from ’), com itative and instrum ental (‘w ith’). It is likely th a t in p roto-Arawá m ost (or all) verbs could function in either o f two ways (with a m eaning difference) - either accepting prefixes and suffixes themselves, or having prefixes and suffixes added to a following auxiliary, -na- or -ha- (the forms are -ni- and -hi- in Paum ari). This is retained to some extent in Paum ari, e.g. we get -mitha- (with affixes on the verb) ‘hear’ and mitha -ni- (with affixes on the auxiliary) ‘listen’. In some languages verbs have separated out into two subclasses - iniecting (taking affixes themselves) and non-inflecting (requiring an auxiliary to carry the affixes). There is a partial semantic basis - for instance, m ost stative verbs (corre­ sponding to adjectives in other languages) are inflecting, e.g. ‘be hard’, ‘be cold’, ‘be black’, ‘be sweet’, ‘be angry’. There is also a division of verbs into intransitive (e.g. ‘sit’), transitive (e.g. ‘hit’), ambitransitive where S o f the intransitive corresponds to O o f the transitive (e.g. ‘break’), and am bitransitive where S corresponds to A (e.g. ‘paddle (a caooe)’). There is no correlation between transitivity and the inflecting/non-inflecting division. In Madi and K ulina (at least) there are verbs with suppletive forms depending on whether the S o r O argum ent is singular, dual or plural; these include ‘sit’, ‘lie’, ‘pu t’, ‘fair, ‘h it’ and ‘kill’. 7 P R E D IC A T E S T R U C T U R E The only obligatory constituent o f a clause is the predicate. Its structure is typically; (a) object pronoun (a separate word); obligatory in a transitive clause. (b) subject pronoun (may be a prefix o r a separate word); obligatory in all clauses. I l Arawá (c) 301 prefix ka-; this can have a num ber o f functions: (i) cross-referencing a /ca-class noun in pivot argum ent function; (ii) m arking an applicative derivation, in which a transitive stem is derived from an intransitive root, with underlying S becom ing A an d a peripheral argum ent being prom oted to be O, e.g. 'laugh + a t’; (iii) a variety o f other semantic functions, e.g. dual subject, ‘accom panied by’. Inside’. (d) causative prefix na- o r niha-; in all languages this derives a transitive stem from an intransitive root, underlying S becoming surface O. In some it can also be used with transitive roots, the underlying A becom ­ ing surface O and the underlying O going into a peripheral slot. In P aum ari a transitive verb must first be delransitivized (by suffix -a) before taking causative prefix na-. (e) verb root, inflecting or non-inflecting. (f) auxiliary verb -na/i- or -Imli- (there are other, minor, possibilities in Deni and Kulina); obligatory with a non-inflecting verb root, taking prefixes (b)-(d) and all suffixes. Arawá languages have rich suffixal possibilities - several score forms (organized into fifteen o r more slots), b u t all are optional. T he m ajor types are: (g) a large class referring to location (‘on the ground’, ‘in the w ater’), direc­ tion (‘up’, ‘alongside’), duration (‘all night long’, ‘continuously’), and ‘following’, ‘alm ost’, ‘again’, etc; th e same sorts o f m eaning are expressed by verbal suffixes across languages o f the family b u t the actual forms tend to differ. Just a few form s recur, e.g. *-ihima ‘upriver’. (h) a negative suffix; the A raw á word list shows a negative suffix -raha. K ulina has two negators: stative verbs take -ra im mediately after the root or auxiliary while non-statives have -/iera(fem)/-/iara(masc) after tense-aspect. D eni ju st has -heral-hara. Jaraw ara has -ra which comes either before tense-aspect or after m ood (when there is no tense-aspect suffix). Paum ari and the other dialects o f M adi have -ra, before tense-aspect. (In Sorowahá there is a negative suffix with quite difterent form, -nasio.) In M adi (and probably in some o ther lan­ guages) the only way o f expressing negation is through a verb plus neg­ ative suffix; there is no interjection ‘n o ’. (i) tense-aspect suffixes; all languages have a set o f suffixes for indicating reference to time, b ut there is considerable semantic variation. Chapm an and Derbyshire (1991:275ff.) state that there is no specification o f tense as such in Paum ari; but there are aspectual suffixes ‘completive’, ‘con­ tinuous’, ‘iterative’, ‘durative’, etc. A t the opposite extreme, Jarawara 302 R. M .W . Dixon has three past tenses - immediate past (a few seconds to a few weeks ago), recent past (a few weeks to a few years) and far past (many years). We also typically find suffixes indicating intention, predication and general irrealis. In Paumari there is a post-verbal particle with future ref­ erence, -bam', Madi has future suffix -bana and Sorowahá has -ba. in Jaraw ara each o f the three past tenses is m arked for w hether or n o t the event described was actually seen by the speaker. It is reported th at Deni has verbal suffixes which include ‘speaker did not know this at the time referred to ’ and ‘speaker knows this but other people referred to do n o t’. This param eter o f evidentiality has not been studied in the other languages. (j) m ood: there are verb-final suffixes indicating declarative, interrogative an d imperative. Polar interrogative -Ä:í(fem)/-A:o(masc) is found in D eni, Kulina and Sorowahà. Simple imperative is m arked by -hi{(em)l -/w(masc) in M adi and K ulina, by -ho in Deni, and by zero in Paum ari an d Sorowahá. There can be more complex m ood m arking; Jaraw ara has distinct suffixes for im mediate {‘do now’) and distant ('do later o n ’) imperatives, in both positive and negative polarity. M ost tense-aspect and mood suffixes have distinct feminine and masculine forms, agreeing in gender with the pivot argum ent. All languages except Paum ari include a final suffix *-(ha)ro(fem )l*-(ha)ri(m asc). In Sorowahá this is added to a verb ju st as a m arker o f gender, when no other suffix has been chosen. In Deni it is glossed as ‘neutral aspect’ and in K ulina as ‘completive’. In M adi it has become the recent past eyewitness form, and is also included at the end o f various deictics and subordinate clause m arkers Individual languages have other kinds o f affixes. In Kulina, for instance, there are separate suffixes for plural object and for plural subject. Jaraw ara has a suffix, -ranw, marking that a core argum ent (in any function) has a feminine noun as head and plural reference. Arawá languages exploit verbal reduplication throughout their gram mars, often using all o f initial CV-, initial CVCV- and final -CV. This can be used for semantic effect (‘do a b it’, ‘do with force’) and to derive a nom inal, e.g. koro ‘to fish’, kokoro ‘fishing line’ in Kulina. (Dixon and Vogel 1996 provides a full discussion o f redupli­ cation in Jaraw ara.) 8 PRONOUNS The main pronom inal forms are shown in table 11.7. Those in row (i) function as subject o f a verb (in slot (b) o f predicate structure) and as possessor o f a possessed 303 I l Amwá Table 11.7 M ain pronominal fo rm s in Arawá languages isg ' 2sg 3sg Inon-sg 2 non-sg 3non~sg 0 a(ri)- awa- wa- 0 iari Ei- + redupl. ivd deni 0 iia tiwa deni poni deni (fem) powa deni (masc) ai at dee dee mai, mee mai, mee 0 hine (fem) hiwa (masc) ee ee tee tee mee mee, mati 0 ari ari (not known) tima (not known) madi 0 Deni o(H) owa tilia o(ii) owa titiwa 0 Jamamadi (i) o(ii) owa titiwa 0 (i) 0 m owa tittwa ) 0- i- Paumari (0 Kulina (0 Jarawara Sorowahá (i) (ii) aro itaa poni (fem) poa (masc) poni (fem) powa (masc) hini ini, iri poni deni (fem) poa deni (masc) 0 noun. Those in row (ii) are the free forms, also generally functioning as object (in slot (a) o f the predicate), sometimes with the addition o f accusative suffix -ra. Paum ari has nothing corresponding to the set (ii) o f other languages; here a further set o f prefixes (Isg ho-, 2sg Inon-sg hari-, 2non-sg a?oni-) can be added to one o f a small num ber o f forms (e.g. -ra, accusative, -wani, contrastive). Alienable possession is generally shown by a prefix o f set (i) added to possessive marker *-kha. Again, Paum ari is different, with forms such as Isg kodi (possibly < o-kadi), \nox\-i%a-kadi. Only in the M adi dialects is there a distinction between inclusive and exclusive in 1non-sg. The inclusive form is ee, corresponding to general Inon-sg in other lan­ guages, and odaa (otaa in Jaraw ara) has been innovated as the exclusive pronoun. 3sg is always zero in set (i). T he 3sg set (ii) forms in Dent and Kulina (p o n i and p o (w )d ) are used for emphasis. T he 3sg set (ii) form s in the M adi dialects {hint in Jam am adi and hineihiwa in Jaraw ara) occur only in reflexives. The M adi dialects (and probably also Sorowahá) have developed a 3non-sg pronoun - in sets (i) and (ii) - from the original noun *madi 'person’; we can observe changes *m adi> m ai> mee. Just Paum ari has a 3non-sg prefix, n'a-. C om m enting now on the I si and 2nd person forms: (a) Isg shows o- for verbal and possessive prefix, in all languages. (The 1869 Arawá word list includes ‘usafá, h an d ’ and ‘otam á, foot’, which plainly include prefix 0- ‘my’.) The set (ii) form is owa in Deni, Kulina and M adi and aro in Sorowahá. 304 K M . W. Dixon (b) 2sg set (i) was plainly prefix *«'- in proto-A raw á, w ith the t being lost in Paum ari and Sorowahâ. The set (ii) form is tiwa io M adi and Kulina, tia in Deni and taa in Sorowahá. (Paum ari has a couple o f traces o f an archaic ti, e.g. in the greeting m ’a m (ti) which involves the 2sg free pronoun iwani plus an optional -ti) In Sorowahâ Isg o- and 2sg i- are prefixes to possessed nouns but infixes on a verb beginning with a consonant, following this consonant. Thus, from gcmia- ‘see’, we get g-o-ania T see’, and from sawa- ‘w ash’, there is s-i-awa ‘you w ash’. In fact the infix is phonetically an off-glide, i.e. [g^ania] and [s>'awa] respectively. (c) For 1non-sg, ari occurs as the set (ii) form in Deni and Sorowahâ and as set (i) prefix in Paum ari. Just Deni and Kulina have a set (i) prefix it is not clear whether this should be reconstructed for proto-A raw á o r w hether it is an innovation in DeniKulina (perhaps as the reduced form o f an earlier prefix ari-}. In M adi and Sorowahâ the same free form is used in both sets. It seems likely that *ari reduced to at in Jam am adi and then to ee in Jaraw ara. The Kulina form ia may also relate to *ari (or there may be some other explanation for this). (d) 2non-sg shows the greatest variation. T he M adi dialects have free form dee (tee in Jaraw ara) in both set (i) a nd set (ii) while Sorowahâ has lima in set (ii). Paumari shows a prefix oina-. Kulina has the sam e prefix, ti-, as 2sg and adds a plural suffix -deni. Deni also uses prefix ti- but here plural is shown by reduplication; roughly, if a verb (om itting the final m ood suffix) ends in Cl giving-F, then Is repeated before F ,# , e.g. ti-kha-thima-aro (‘2-be in m otion-U PSTR E A M -D EC L + F E M ’) ‘you (sg) go upstream ’ and ti-kha-thi-a-ma-aro ‘you (pi) go upstream ’. 9 DEMONSTRATIVES AND INTERROGATIVES There is typically a rich set o f demonstratives, but the semantic details and forms vary from language to language. M ost Arawá languages have a single root underlying m ost or all o f ‘who’, ‘w hat’, ‘where’, etc. This is naliina (with variants niha and hana) in Paum ari, nelteko in Kulina and ako in Deni; in Deni and K ulina fem inine- ( ha)ro and m asculine-(h a jri are added to the basic root to form ‘w ho’. In Jam am adi ‘w ho’ is ebenike and in Jaraw ara it is hike (fem) and hika (masc) which may be further cognates. 10 CONSTRUCTION TYPES There are two transitive construction types, the choice between them depending on w hether the A o r the O argum ent is pivot (topic) within the discourse in which the clause appears. See examples (2-3) below. The main characteristics o f the two construction types are: 305 I I Arawá A -C O N STR U C TIO N O -C O N STR U C TIO N The A argum ent is discourse T he O argum ent is discourse pivot pivot The A N P is generally not included T he O NP is generally not included The O NP norm ally s ta t« i T he A N P normally stated Verbal suffixes agree w ith A Verbal suffixes agree w ith O O m arked by accusative -ra A m arked by ergative -a (only (in P aum an and Jam am adi) No v e r b a l p r e fix if b o th A and O are 3rd person in Paum ari) I f both A and O are 3rd person, verb shows p r e f ix bi- in Paum ari, hi- in M adi, and i- in D eni and K ulina The two construction types have different preferences for constituent order. In the few textual instances where both A and O NPs are stated, M adi prefers AOV for an A -construction and OAV for an O -construction while Paum ari prefers OVA for an A -construction arid AVO for an O -construction. (T hat is, the pivot N P is preferred in clause-initial position in M adi and immediately after the predicate in Paum ari.) The A -construction appears to be the unm arked construction type, and can be used for all possibilities at A and O. The O -construction is more restricted - in Paumari it may only be used if the O is 3rd person, in Deni and K ulina only if both A and O are 3rd person, and in Jaraw ara only if either A or O (or both) is 3rd person. Sentence (2) gives an exam ple o f an A -construction a nd (3) of an O -constniction from Jaraw ara. Each is preceded by an intransitive clause to provide discourse context. (2) fan aj to-ka-hara-ke w o m a n (fe m ) A W A Y - b e .in .m o t io n -iM M .P A S T .e y e w itn e s s + fe m -D E C i. + fern awijj tapir(m asc) 0 -m ita-hara-ke 3 s g - h e a r - iM M .P A S T .e y e w it n e s s + fe n i-D E C L + f e m ‘The w oman went o u t [into the forest] and heard a tapir.' (3) aw ij ka-ke-hare-ka t a p ir (m a s c ) b e .i n .m o t i o n - c o M iN G - iM M .P A S T .e y e w it n e s s + m a s c -D E C L + m a s c fa n a ^ w o m a n (fe m ) h i- in it a - r e - k a O .c o n s t r u c t i o n - h e a r - iM M .P A S T .e y e w it n e s s + m a s c -D E C L + m a s c ‘A t a p ir c a m e [ i n t o t h e v i l l a g e ] a n d t h e w o m a n h e a r d h i m . ’ 306 R M. W Dixon In {2), fa n a ‘w om an’ is the pivot running through the two clauses. The transitive clause is then an A -construction in which fa n a is the understood A and awi ‘tapir’ the stated O. It is recognizable as an A -construction by the absence o f a verbal prefix In- and by the feminine forms of tense and m ood suffixes, agreeing with the A argu­ m ent ‘w oman’. In (3), awi ‘tapir’ is the pivot linking the two clauses; it is in S func­ tion in the first and in O function in the second. The latter is an 0-construction, m arked by verbal prefix hi- and the m asculine forms o f tense and mood suffixes, agreeing with the O argum ent. (A fuller discussion o f A -constructions and O -constructions in Jaraw ara is in Dixon forthcoming-a.) Arawá languages also have com plem ent clauses (filling a core slot in the main clause) and a rich array o f other types o f subordinate clause constructions. Just Paum ari also has a passive derivation (in addition to A -constructions and 0 -c o n ­ structions). BIBLIOGRAPHY Adams, P. 1987. ‘Madija grammar sketch’ (87p typescript, SIL Peru). Adams Lichlan, P. and M arlett, S. 1990. ‘Madija noun morphoiogy’, U A L , 56,102-20. Brinton, D. O. 1891. The American race. New York: Hodges Publisher. Chandless, W. 1869. ‘Notes on a journey up the River Juruá’, Journal o f the R oyal Geographical S ociety, 39,296-311. Chapman, S. and Derbyshire, D. C. 1991. ‘Paum arí', pp. 161-352 of H andbook o f Amazonian languages, vol. Ill, ed. D. C. Derbyshire and G. K, Pullum. Berlin: M outon de G ruyter Dixon, R, M, W. 1995. 'Fusional developmetjt o f gender marking in Jarawara possessed nouns', U A L , 61.263-94. forthcoming-a. ‘A -constructions and O-constructioiis in Jaraw ara’. forthcoming-b. ‘Proto-Arawá phonology’, Dixon, R. M. W. and Vogel, A, R, 1996. ‘Reduplication in Jaraw ara’, Languages o f the World, 10.24-31. forthcoming. The Jarawara language o f southern Amazonia. Koop, G. and Koop, L. 1985. Dicionário Denl-Portugués. Porto Velho: SIL. (Experimental edition of ten copies; includes grammatical sketch, building on M oran and M oran 1977.) Monserrat, R. M, F. and Silva, A. O. (Kanau). 1986. G ram ática d a lingua Kiihna. Acre: Conselho Indigenista Missionário. M oran, P. and M oran, D. 1977. ‘Notas sobre morfologia verbal Deni’, pp. 29-71 o f Série Lingiiislica 7, ed, L, Bridgeman. Brasilia: SIL. Silva, O. (K anau) and Monserrat, R. M , F 1984. D icionário Kulina-Português e Porluguês-Kuiina ( dialeto do Igarapé do A njo). Acre; Conselho Ittdigenista Missionário. 12 Small language families and isolates in Peru M A R Y R U T H W IS E 1 in t r o d u c t io n ; t h e l a n g u a g e s a n d f a m il ie s The m em bers o f five small language families - C ahuapana, Jivaro, Zaparo, PebaYagua and W itoto - are located in northern Peru and across the borders in Ecuador, C olom bia, an d Brazil. T he speakers o f these languages, with the exception o f Jivaro and possibly C ahuapana, were am ong the ethnic groups whose populations were decimated during the ‘rubber boom ’ at the tu rn o f the twentieth century. T he indig­ enous population o f the Putum ayo River region dropped from 50,000 to 7,000-10,000 during the first decade o f the tw entieth century. The Z aparo lan­ guages have lost ground continually to Quechua; they and all o f the families except Jivaro are also losing ground to Spanish. O ne o th er small language family might be added to the five discussed in this chapter, namely H arakm bet (or H ate) o f the southern Peruvian jungle. However, the cu rren t consensus is that it should be considered a single language isolate. Two isolates are alm ost extinct. Language loss appears to have occurred in M unichi in the form o f morphological simplification (G ibson 1996: 26). It was impossible to record any folklore am ong the Taushiro since all o f the old story tellers died in an epidemic in about 1964 (N. Alicea p.c.). Table 12.1 Jists the languages w ithin families o r as isolates, as well as other lan­ guages o f northern Peruvian A m azonia; alternative nam es within parentheses are followed by location, population, and com m ents on literacy, bilingualism and dialects.’ The accom panying m ap shows the approxim ate locations o f these lan­ guages and isolates. For m ore than thirty years, there have been governm ent-spon­ sored bilingual schools am ong the C hayahuita, A guaruna, Achuar, W ambisa, Shuar, A rabela, Yagua, Bora, O caina, M urui W itoto, C andoshi and T icuna. The first A g uaruna-S panish school began in 1953; am ong the A m arakaeri and Iquito, ' Unless spectfii ;d otherwise, references to bilingualism ittdicate proficiency In the vernacular and Spanish. 308 M a ry Ruth Wise M ap 11 Small language families and isolates of Peru with approximate locations (note that national boundaries are only approximate) bilingual schools operated for a few years only. Larson (1981) gives a summary history o f the first twenty-five years o f bilingual education am ong the A guaruna and attribiites the on-going success o f the program m e - approw mately 80 per cent are literate - to the training of indigenous leaders who, in turn, trained the next gen­ eration o f indigenous leaders, Trudel! (1993) summarizes various factors affecting literacy acquisition in the vernacular and in Spanish am ong several groups, includ­ ing the A guaruna and Yagua. The practical orthographies are all basically pho­ nemic and are adapted to Spanish insofar as possible. The languages o f the C ahuapana family (C hayahuita and Jebero), while clearly £> CL n g .5 ä Sî £ J '2 O. a eî go 3 a & « 1 s 2 H .i D. .ti D I S s s I _ a g i 8 s a 8D I S 1 2 S •3 s-i § ^ I® 3 5 | £ § :s s 5. 3 ? < 3 -g is < I 3 -1 1 â i i § S 'C x : -a c a,-a O ÎÔ D. iC | ‘l £ I tj ÏÔ i« g I cl o «s -S X cu to II 11 < iî 3 Is s fa­ 0 .5 1° ■s ö — (A B .2 o ro if Z .S •:= il eu U s'J Ö TJ —•■ ■ §1 I CJ I *1 e § gI S.S: “r s l 2 Î lit -3 U I .£ ä 3 op ?3 -J 'i > i 2 â Ñ o. (3 § O §* £ ÍD «I J I^ = i è ^ á; 4 â Æ •Çf I I íl a s - l 1 |._-32 I c 3 S 5 | z s s T3 ÎJ uj «n ■5S o 2; B S < -c g H a. g _"O "C I 2 1 > .0 I g E< a-Si a , ■= a 2 h «i 1) 2î t/5 -J= H i S i l l g 2 u 'B 5I ^ § 1 1s 1C il■ o. S £ S ^5 S à 3 S XI a > < â ÍC C ^ E5 o .t; ’C îs g U g- > p c S « “O ï. I .S I s I §D :2 o > O IS £ i l l ^ o" S) j a 'l l C S .S- g 2 e -S a f 5 B ° S g ,S JD IS « i, « Í .1 I I I I araI ÖI s £ 5 f2 p 312 M ary Ruth Wise related, are mutually unintelligible. No reconstruction has been proposed, no r has membership in a larger family o r stock been confirmed by com parative work. The Jivaro (Shuar) family includes A guaruna (the m ost diverse), Wambisa, A chuar and Shuar. C andoshi has sometimes been classified as a member o f a S huar-C andoshi family b u t David Payne (1989) considers his earlier (1981) com ­ parative reconstruction o f proto-Shuar-C andoshi to be unjustified. It has also been erroneously classified as Z aparo by some authors.^ Little data are available on Peba and Yameo and no com parative work has been done on the Peba-Yagua family.^ Languages o f the Z aparo family are spoken in the eastern jungle o f northern Peru and southeast Ecuador. The whole family is nearly extinct with fewer than 200 speakers in total. Only A rabela and Iquito are spoken by m ore than 10 people, and only Arabela has even a slim chance o f survival beyond the present generation o f older adults. (See Stark 1990 for a description o f the sociolinguistic situation.) Taushiro is possibly a Z aparo language, b u t its classification has not been con­ firmed. The W itoto family includes two m ain branches: B ora-M uinane and W itoto-O caina. The two B ora-M uinane languages are: Bora, including the Mirafla dialect which is about 94% intelligible with Bora, and M uinane, n o t to be confused with M uinane W itoto. Resigaro has sometimes been included in the family but David Payne (1985) dem onstrates that the apparent relationship is the result of heavy borrowing, with consequent phonological change and some grammatical borrowing; Resigaro is clearly an A raw ak language (see also A schm ann 1993: 3). H arakm bet has sometimes been erroneously classified as Pre-andine Arawak. A pparent cognates are probably due to borrow ing and to the presence of some widespread gram m atical forms (cf David Payne 1990b). There has been some con­ vergence o f dialects, or rather extension of A m arakaeri, at least in the D ominican M ission at Shintuya (Van den Eynde 1972:4; see also Lyon 1975). 2 PH ONO LO GY M ost of the languages in the five families differ from areal patterns in one or more traits. In the Jivaro languages there are few restrictions on syllable-final consonants so that consonant clusters are very frequent. Syllables are usually open in the other ^ David Payne (1989) suggests a possible deep genetic relationship between Candoshi and Arawak; he presents thirty-two basic vocabulary forms with CVC resemblances and several apparent grammatical cognatcs. Gnerre ( 1988) suggests a possible deep genetic relationship between the Jivaro and Arawak families. N either suggests that Jivaro languages and Candoshi are more closely related to each other than to Arawak. Doris Payne (1984) hypothesizes a possible Zaparo-Y agua connection. 313 12 Sm all language fam ilies and isolates in Peru Table 12.2 Jebero (Cahiiapana) consonants bilabial dental alveolar post-alveolar palatal velar labio-velar glottal plosive P ejective m nasal fricative flap ejective flap lateral approximant t n s r r’ c k k’ Ji 0 Ï / 1 A J j w Table 12.3 Jebero (Cahuapana) vowels front dose near-close open central i back u i a languages. A guariina (Jivaro) has no liquids while Jebero (C ahuapana) has three. Both fricatives and affricates occur in some but others have only one o r the other. Pitch accent and high-low tone contrasts are more frequent than contrastive stress. A four-vowfel system is the m ost frequent b u t several languages have six. T h e consonant and vowel inventories o f Jebero (C ahuapana) are shown in tables 12.2 an d 12.3 respectively (adapted from Bendor-Sam uel 1961). C hayahuita lacks c and the glottalized consonants, has only one liquid (r) and two nasals (bilabial and alveolar). In Jebero, occlusives are voiced after nasals. The phonem e lÁl is ‘produced by the blade of the tongue in the palatal region w ith the tongue tip down behind the bottom teeth’ (Bendor-Sam uel 1961: 13); j i s a frictionless continuant. Syllables can be open o r closed. T he central near-close vowel is lower in Jebero than in C hayahuita. ?, r-quality and nasalization are analysed as syllable prosodies in Jebero (1961: 20). Vowel length, aspiration, giottalization [vowel + ?] and nasalization are considered to be p art o f the syllable nucleus in C hayahuita (H art and H art 1976) so that syllables are open. Stress is not contrastive in either language. In Jebero the first syllable o f a disyllabic word is accented and the second o f words w ith three o r more syllables; there is some m odification with certain suffixes. 314 M ary Ruth Wise Table 12.4 Aguanm a consonants plosive nasal affricate fricative approximant bilabial alveolar P m t n ts s post-alveolar palatal velar labio-veiar k glottal Î 0 tj Í h w j Table 12,5 Aguanm a vowels front central back close open The Jivaro languages are characterized by many consonant clusters wordmedially. The consonants and vowels o f A guaruna are shown in tables 12.4 and 12.5 (adapted from David Payne 1981). Nasal and oral vowels contrast. As in m any A mazonian languages, nasalization can extend over a whole series o f vowels and semivowels; nasal consonants are pro­ nounced as prenasalized voiced plosives in the environm ent o f oral vowels. (See David Payne 1976 for a detailed analysis o f nasality.) David Payne ( 1990a) proposes that A guaruna has a type o f pitch accent in which stress (intensity) does not neces­ sarily coincide with high pitch (accent) as shown in (1).“* This analysis is probably applicable to the other Jivaro languages. The reflex o f /*o/ is /r/ in syllable-initial position (often syllable-final after vowel deletion) in all o f the languages except A guaruna which retains /rj/; consequently there are no liquids in A guaruna. Voiceless word-final vowels are contrastive but are no longer pronounced by many speakers. C ontrastive vowel length is interpreted as a vowel sequence. Syllables may be open or closed (2a). Vowel deletion and/or metathesis result in many consonant clusters ( 2b). (1) (David Payne 1990a: 165) [ÏÏjÏÏi] /tjü u / (David Payne 1981: 336) [ k ä ^ ] /káap/ ^ ‘m onkey (nom )’ v s [’tju in j /tju á n / ‘monkey (a c c ) ’ ___^ ‘vine’ vs [kaap] /kaáp/ ‘gnat’ '* Throughout this chapter the acute accent ( ') is used to indicate high pitch or tone. Low tone is left unm arked in those languages where tone is contrastive. 315 12 Sm all language fam ilies and isolates in Peru Table 12.6 Yagua consonants bilabial plosive P [P". P] m [m*, m, m*’] nasal affricate fricative flap ttpproxiinant alveolar post-alveolar palatal velar labio-velar glottal t[t] n (n , ji, n**] ts [ ts ,tJ ,J .s ] tj[ tj] k[k] h[h] I-W j [)] w (w, P] Table 12,7 Yagua vowels front close close-mid open-mid open (2a) (2b) central i [i. I] e [e] back u [u, u , i] o (3 | a [a, * , e] (David Payne 1981) /namar)/ ‘m eat’ /su w a tj/ ‘lungs’ /piokir)/ ‘good’ (W ipio c ía /. 1996) /saiuiák/ ‘kind o f leaf’ vs /saunkágtin/ ‘stream where plants w ith these leaves ab o u n d ’ /junüm at/ ‘to draw near’ vs /juniím tut/ ‘to draw near to som e­ thing or som eone’ The phonem es of Yagua and their principal allophones except for m ost palatal­ ized consonants are shown in tables 12.6 and 12.7 (adapted from D oris Payne 1985b; 16-17). Powlison (1995; 3()-3) sum m arizes the effects o f /j7 on vowels: for example, /wja/ [Pja], /w jaa/ [Pee], /wjaj/ [(Je], /w jaaj/ [PjeeJ. T hus a phrase like r a j-tä r jü j-rü j tsaim u tsij-ra is pronounced [ r a tjê rjû r jü j si:m usirja]. W henever a m orphem e ending in /j/ immediately precedes a m orphem e beginning w ith any consonant o th er than an alveolar o r palatal obstruent, there is m etathesis o f /j/ and the consonant. M orphem es d o not end in consonants other than /j/ (D oris Payne 1985b: 17). As in the Jivaro languages, nasal consonants are pronounced as prenasalized plosives, 316 M ary R uth Wise Table 12.8 Bora consonants bilabial alveolar post-alveolar palatal plosive nasal affricate velar labio-velar glottal k k'’ kP[ki>, k“^ 7 tj If fricative X[x, c] flap appro Jsimant that is with an oral release, in the environm ent o f oral vowels. There are two con­ trastive tone levels; som e syllables have inherent high tone; others inherent low tone; others have no inherent tone (1985b: 17-18). The phonological systems o f the W itoto languages vary considerably. Bora and M uinane have two contrastive levels o f lexical tone. Tone is also very im portant in the m orphosyntax o f Bora, as shown in (3). Note, for example, that high tone occurs on the penultim ate vowel in the citation form o f ‘chief’, low tone when it is subject, and high to n e on both the penultim ate and antepenultim ate when it is the possessor. Similarly the nom inal form o f m a h tfá h m high tone on the first syllable but the verbal forms do not, except in the future tense. (3) (Tliiesen 1996: 15) áP'éhuüpe ‘chief’ ápiéhuupe m a h tjö ‘the chief eats’ áP'éhuupée m a h tjá ‘the chief ate some time ago’ á|Jiéhuupe m á h tjo ö ‘the chief is going to eat’ áp 'éh áápe m á h tjo ‘the chief’s food’ Each syllable has either a high o r low tone and only one mora; the tones are the basis for sending messages a distance o f several kilometres using a pair o f hardwood signal drum s ab o u t five feet each in length (Thiesen 1969). M uinane also has con­ trastive tone while O caina has contrastive pitch accent (A schm ann 1993:3). W itoto, except for M urui, has contrastive accent (stress). The phonemes o f Bora are listed in tables 12.8 and 12.9. All o f the consonants except /r/ and /k**/ have palatalized counterparts; /t j/ and /tj''/are the palatalized counterparts o f /ts / and /ts ¥ . The phonem e /j/ functions m orphophonem ically as the palatalized counterpart o f Irl following HI. Bora is unusual in that it h as three close o r near-close unrounded vowels. 317 12 Sm all language fam ilies and isolates in Peru Table 12.9 Bora vowels front close near-close close-mid open-mid open central back i (i, i] lu i [}, i] ç[ç, e (e , æ, ej a [a, a] Table 12.10 Arahela consonants bilabial alveolar plosive p nasal m fricative flap approximant post-alveolar palatal velar t[t, t:J n[n, n:, n:''] s J [J. J:] r [r, r, n , r:] labio-velar glottal k [ k , X, g , ¥ , k :] h j []• j) w [w, w] Table 12.11 Arabela vowels close close-mid open front central back i e[e, i] } u [u, u] a [a, se, e, o, a] O eaina also has palatalized consonants but not a full series. It is unique in that fortis an d lenis nasals contrast. T he fortis feature may be actualized either as length or as tense articulation o r as both {Agnevv and Pike 1957: 24). The lenis nasals are pronounced w ithout com plete closure (Leach 1969: 164). Voiced and voiceless plo­ sives contrast. There are twenty-six consonants, five oral vowels, five nasal vowels and contrastive pitch accent. W itoto is unique in th a t there is no voiceless bilabial plosive; N ipode does have [p] as an allophone of the voiceless bilabial fricative; like Bora, there is no [s]. M urui W itoto has a dental fricative /0/. N ipode has voiced implosives / 6/ and iá l wordmedially, an d M uinane has gem inate plosives word-medially. Sequences o f three vowels are com mon in W itoto. The co n sonant and vowel inventories o f A rabela from the Z aparo fam ily are shown in tables 12.10 an d 12.11 (adapted from F. Rich 1963). 3 18 M ary Ruth Wise Zaparo differs from A rabela in having alveolar and postalveolar afTricates and a glottal plosive. Iq uito and Zaparo each has only four vowels; close front and central, open central, and one back vowel varying between close and mid. In all three lan­ guages long and short vowels contrast. In Arabela, at least, there is palatalization an d labialization o f consonants (w ritten as Ci and Cu in the examples) in syllables following i and u, respectively. In A rabela /h/ has a nasal quality and vowels are nasalized following /h/ and nasal consonants (F. Rich 1963; 197). With the excep­ tion o f a phrase-final fortis /h/ and [?] for em phasis in A rabela, syllables are open. A tw o-tone system is reported for Iquito (Eastm an and Eastm an 1963; 146).’ The isolates present several phonological features which differ from areal patterns: Ticuna has a very complex tonal system as well as laryngealized vowels (L. Anderson 1959a,b, and M ontes Rodriguez 1995).* C andoshi has a voiceless retroflexed postalveolar affricate, and the maximum syllable is CCCVC. Taushiro is the only Peruvian language in which there are no bilabial or labio-dental consonants. 3 M ORPHOLOGY Word classes in m ost o f the languages include open classes o f noiuis, verbs and adjectives. Closed classes usually include adverbs as well as pronouns and conjunc­ tions. In most o f the languages, adjectives are more likely to occur as predicate adjec­ tives than in the noun phrase but the adjective class as such is not necessarily closed. All o f the languages o f the five families are agglutinative, although strings o f affixes tend to be shorter in the Jivaro languages than in the others. Suffixes predom ­ inate but there are causative prefixes in Jivaro, C ahuapana and H arakm bet. The C ahuapana languages also have about two dozen stem -form ing verbal prefixes. In Zaparo, Bora and Yagua, person m arkers are prefixed to possessed nouns. In Yagua, Iquito and H arakm bet, and sometimes in A rabela and Zaparo, subject person markers are prefixes also. C ahuapana, Jivaro and H arakm bet verbs crossreference both subject/agent and object. The cross-referencing m orphemes are pre­ fixes in H arakm bet; declarative, dubitative and im perative sets differ in some subject/agent-object forms. The W itoto languages and Yagua have dual, as well as singular and plural, numbers (a dual num ber is unusual in A mazonia). C ahuapana languages and H arakm bet have dual num ber in the first person only. ’ Further analysis might show that Iquito has a pitch-accent system. ‘ Anderson (1959a,b) and Soares (1995) describe Ticuna as having five tones as well as glides. In a recent auto-segmental analysis, M ontes Rodriguez (1995) agrees with Anderson’s pho­ netic transcription but considers that high and low tones comprise the basic opposition and that a third mid tone is relatively weak and might best be considered as an underspecifled element. 319 12 Small language families and isolates in Peru 3.1 Nominal morphology 3 . 1.1 C lassifiers In Yagua, classifiers are infixed into the num ber ‘one’ and suffixed to other num bers; classifiers occur w ith nouns, dem onstratives, adjectives and verbs. They are especially im p ortant In Bora where m ore than 350 have been identified. In W itoto an d Peba-Yagua, the only families with gender distinction, masculine or feminine gender suffixes function as classifiers in w ords referring to anim ate nouns. In Bora classifiers occur in pronouns also. Examples from B ora w ith -?áám i ‘leaf’ follow: (4) (Thiesen 1996: lOfr-7) ki’aahák'’iu ‘know ledge’ kPaahák’’iu?áám i ‘b o o k ’ m it''ja ‘big’ mítW a?áámi ‘big leaf, book, etc.’ ts*'a- ‘o n e’ ts''á ?aám i ‘one leaf, book, etc.’ ts*"!- ‘an o th er’ ts*’i?% ám i ‘another leaf, book, etc.’ ‘conjunction, á?aám i ‘that leaf, etc., anaphoric m entioned in preceding sentence’ pronoun’ p*'an<ct''iue- ‘im p o rtan t’ p^án-'ét'‘itie?áám i ‘the m ost im portant (leaf, book, etc.)’ In the W itoto family and Yagua, num erals agree w ith their head noun as in (5); in Bora and Yagua, dem onstratives do also.^ T he m odifier-in descriptive noun phrases may optionally agree, as in ( 6). (5) (Yagua; D oris Payne 1985b: 114) tá-nu-kii n íín u o n e - C L :th ie k .p o le - o n e p o le ‘one pole’ o r ‘one tree tru n k ’ ( 6) (Yagua: D oris Payne 1985b: 117) rá-b ii rd n aj(-b ii) (rá,ia{-bii)) its-C L :flo w e r re d (-C L :flo w e r) ‘its red Bower’ The data í o r not. t to determine if this is true o f all W itoto languages 320 M a ry Ruth Wise In C ahuapana and Zaparo, one o f the frequent noun derivational suffixes is an attributive (‘ow ner o f ’), as in so?ja-wan (wife-ATTRIB) ‘one who has a wife’, ma?wan (things-A TTRIB) ‘a rich person’ (Chayahuita: H art 1988: 261). 3.1.2 Case N ominative-accusative type case markers occur in all languages o f the Cahuapana, Jivaro and W itoto families, and in Candoshi and H arakm bet. In Bora and the Jivaro languages, the nom inative is zero. W itoto itself optionally m arks both nominative and accusative (the object is obligatorily marked when the subject is third person). In C ahuapana both may be marked or both may be unmarked; the subject is usually marked only to avoid am biguity or for emphasis and the object only for emphasis. In Arabela, subject and object functions are usually marked by constituent order only but -ri may m ark the subject when it precedes the verb. The -ri subject marker occurs prim arily in transitive constructions and may be indicative that Zaparo lan­ guages are partially ergative. A nother ergative-like feature in all o f the Zaparo family is the fact that there are two sets of pronouns, the second o f which could be consid­ ered absolutive since it Is used for the object and for the subject o f stative clauses. The useof one set or the other, however, depends primarily on constituent order; the first set isused for agents/subjects and also for objects when they precede the verb. lo Yagua, definite object enclitics attach to any word which follows the verb and directly precedes the object; or they attach to the verb when the object is not a fully specified noun phrase, as in (7d). Com pare the agent and object references in (7). (7a) (T hom as Payne 1983: 176-7) tsa-h u u j-m áa A nita [tsahOQ m jáá] 3 sg .A -fall-P E R F V A n ita ‘A n i t a f e ll.’ (7b) tsa-puO tJi Pauro-nii A nita 3sg.A-carry Paul-3sg.O A nita ‘Paul carries A nita.’ (7c) tsa-püOtJ'i-níf 3sg.A-carry-3sg.O A nita A nita ‘H e carries A nita.’ (7d) tsa-pQOtXi-ni! 3sg.A-carry-3sg.O ‘H e carries him/her.’* • Thom as Payne glosses this example Powlison all gloss -nil as ‘3sg’. . . him/her/them’ but he, D oris Payne and Paul 12 Sm all language fam ilies and isolates in Peru 321 Thom as Payne (1983; 180) argues that this kind o f system can lead to an ergative reanalysis and places Yagua typologically with those languages which utilize erga­ tive construction types as the unm arked transitive construction in discourse, where the direct object is not being introduced into the discourse for the first tim a The cases in M urui W itoto are: -d i ‘nom inative’, - m ‘accusative/oblique’, -da ‘instrum ental’, - n ‘dative’, -m o ‘allative’, -m ona ‘ablative’, -ira/ii‘locative’. In Jivaro, accusative an d dative cases are n o t distinguished; the m arker in A chuar is -n l-nä-an/-un. O ther A chuar cases are -haT ‘instrum ental, com itative’, -kl-ak ‘means’, -num l-nam ‘locative’, - V ‘genitive’, - V ‘vocative’. In Z aparo an d Yagua there are not only basic locative suffixes or postpositions but m ore than twenty-five postpositions indicating m ore exact location in time or space; m any o f them are com pound, e.g. A rabela -ka ko ‘on, above’, -h i ‘from ’, -ka ko -h i'fro m above’; -ko m a ‘below’, -ra ‘for’, -kom a-ra ‘downwards’. The Yagua applicative verbal suffix and the ‘instrum ental, com itative’ case m arker have the same shape a nd are no do ubt etymologically related (D oris Payne 1985b: 178). T his is also true for the Arabela suffix -tal-tia ‘applicative or instrum ental/comitative’. 3 .2 P ronoun s C om m on to all five language families are personal, dem onstrative, possessive and interrogative pronouns. Singular, dual and plural are distinguished for all persons in the W itoto and PebaYagua families. In C ahuapana languages, dual is distinguished only for first person inclusive (and first dual inclusive also has an im personal sense).* F irst person inclu­ sive and exclusive are distinguished except in Jivaro - where first person singular is used for exclusive and first person plural implies inclusive - and for the W itotoOcaina branch o f W itoto. D em onstrative pronouns distinguish three degrees of distance and are often partially identical with third person pronom inal forms, as can be seen in the case o f M urui W itoto in table 12.12 (from S. A, Burtch 1983, vol. II: 149-50). (Encliticized subject person markers are listed only if they ditTer from the free forms.)'® Jivaro does n o t distinguish singular and plural in third person, and the dem on­ strative for second degree o f deixis is the same as one o f the third person pronouns, as shown in table 12.13 for A chuar (from Fast, Fast an d F ast 19%: 31,33). Possessive pronouns in Jivaro and C ahuapana are form ed by the addition o f a case m arker to the personal pronoun, e.g. t a ‘1’, k a k in ‘mine’ (Chayahuita: H . H art 1988: 262-3). ’ ‘Impersonal’ is used in the sense of ‘one (does)’, i.e. an unspecified subject or agent. The grave accent marks the final vowel of a diphthong in Witoto. 322 M ary Ruth Wise Table 12.12 Muriii Witoto pronouns Dual Singular raasc 1 Plural kokD kaiîi kuè fern kaljiadi DmuikD, -DmiukaAi masc 2 3 fem 3 amaûi omujjiailj masc imuie iadiraaiaûi imakui fern ijiaijia iaûijiuaûi ijialjiuadi -maki neut iel-e Near ( hi-) masc anim bimuie bimakui fern anim bijiaijio bijiaijiuaiii Within view ( ba-) masc anim baimuje baimakui fem anim baijiaiji 3 baijiaijiuaili Far (out of sight) ( m -) masc anim naimuie naimakui fem anim naijiaijiD naijiaijiuaib neut bie neut baie neut naie D em onstrative or interrogative pronouns usually substitute for relative pro­ nouns, as in A rabela. ( 8a) (R. Rich 1999; 46) tejanoteni? 'W liieh is it?’ ( 8b) (R. Rich 1999: 426) kia pa-re tejaiio kia pani-ja-ni 2sg to.take-iMPER which 2sg to.like-coNT-1 R ‘Take whichever one you like.’ As shown in (8 b), indefinite pronouns are sometimes based on the interrogatives. !n C hayahuita they are separate: a?nakin ‘some, som eone’, Insosona ‘w hoever’, m a?sona ‘w hatever’, k o insonta? ‘no one’, k o m anta ‘nothing’ (H. H art 1988: 265). 323 12 Sm all language fam ilies and isolates in Peru Table 12.13 Achuarpronouns Singular wi ám í 1 2 3 Plural ii átum present or in sight absent áu ni Demonstrative hü áu nu near present present (in sight) absent (out of sight) 3.3 Adjectives In some o f the languages adjectives are not clearly distinguished from nouns in that they may occur with m ost o f tlie nom inal affixes. However, when they function as nouns they often have a slightly different meaning, e.g. A chuar ápu ‘fat’ (as adjec­ tive) but ‘the chief’ (as noun); p isu ‘hard’ but p/sti-r/hard-3sg ‘its hardness o r firm­ ness’ (Fast, Fast and Fast 1996: 35, 236). 3.3.1 Comparative and superlative grades Com parative and superlative grades are not always distinguished. In C hayahuita the com parative/superlative is form ed by reduplication o f the adjective stem , e.g. noja ‘g o o d ’, noja noja ‘b etter/best’. The standard precedes the com pared form , as in (9). (9) (H . H art 1988: 267) ipi k ira n w a?w i w a ? w -ijin ag o u ti fro m sm a ll small-ADJ in a -s o ? that-BMPH ‘T h a t [anim al] is sm a lle r th a n th e a g o u ti.’ In Yagua the standard, which is a postpositional phrase, follows the com parative form, as in ( 10). (10) (D oris Payne and T hom as Payne 1990; 288) h á á rjij tsám ij A nita raj-jan ü h ö very good A nita Isg-m ore.than ‘A nita is prettier than I.’ Com parison can also be m ade by juxtaposition o f clauses, as in (11). 324 M ary Ruth Wise (11) (D oris Payne and T hom as Payne 1990: 290) h á á rjij rá-m u ritjééw a h a ts ü k a ru néé h á á rjij m uch iN A N -c o s t su g ar neg m uch rá-m u ritjééw a h arü tsij iN A N -c o st ric e ‘Sugar costs more than rice.’ (lit. ‘Sugar costs a lot, rice costs not much.’) 3.4 Verbal morphology Verbal m orphology in all o f the families is considerably m ore complex than that o f nouns and oth er word classes; so much so that in narrative text many clauses consist ju st of a verb. In Jivaro an d C ahuapana, suffixes referring to both the agent and the object occur; in object-agent order in Jivaro and in the opposite order in Cahuapana. Wlien there is no suffix agreeing with the object, it is third person, as in (12). (12) Chayahmta (H. H art 1988: 273) aw ir-in-ko ‘He hits me.’ aw ir-in-kin ‘H e hits you.’ aw ir-in ‘He hits [him /her/them].’ In Iquito and Yagua a prefix agreeing with the subject/agent occurs, and in W itoto a suffix. In Bora the classiiers can occur as verbal suffixes agreeing with a third person subject/agent as in (13b). F irst and second person subjects are inde­ pendent pronouns preceding the verb, as in (13a). (13a) (Thiesen 1996: 58) 6 ts iin é (13b) ‘I r a n . ’ (Thiesen 1996: 108) t''áápo-dpe treat-masc.sg ‘he treats (the sick)’ t''ááp o -tse treat-fem .sg ‘she treats (the sick)’ Incorporation o f the object is described for C ahuapana and H arakm bet. The object thus incorporated may be a noun stem o r a classifier. (14) Chayahuita (H. H art 1988: 274) a ? p i-n a ja -ti-r-a w i t o . I i g h t - e y e - A P P L i c - i N D i c - 1s g . A ‘I s h o n e t h e lig h t in ( h is ) e y e s .’ 12 Small language families and isolates in Peru 325 C om pound verbs are described for Bora (13), Chayahuita (16) and H arakm bet (17). ( 15) (Thiesen 1996; 60) i?h M ^á-tujhk'’énui-ujpe / tm hk''énui-m pe to.speak-to.begin-3sg.masc / to.begin-3sg.masc ií?h'áipá-ne to.speak-action.in.general ‘H e began to speak.’ (16) (H . H art 1988:271) n ito-ta?a-r-in to.know -to.run-iN D ic-3sg ‘H e know s h o w to run.’ (17) (Tripp 1995:205) e ? -tô ë ? + i N F - g u a r d /p o s s e s s e ? -b ie i? > e ? b ie jtö é ? iN F -to .d ie ‘t o g u a r d a d e a d a n i m a l ’ In C andoshi repeated action is expressed by repetition or reduplication o f the verb root- In a series o f individual actions the repeated root is followed by an inflected pro-verb as in (18); in continuous repetition the reduplicated verb root occurs in a finite verb, as in (19). (18) (S. Tuggy 1982:41) T p o ts kos kos kos people to.arrive to.arrive to.arrive to.do.thus-CURR.PAST-3pl ajira-g-ana ‘A group of people arrived, then another group, then another group.’ (19) (S. Tuggy 1982:41) T p o ts k a ro -w a ro -ta r-ta m ta - 0-ja people to.descend-to.descend-HABiTUAL-also-3-EMPH ‘A m ultitude o f people descended also.’ C hayahuita distinguishes aspect but not tense in the verb. Jebero and languages of the other families have b o th tense and aspect suffixes. In Yagua, Jivaro and C ahuapana the unm arked tense is the present. W itoto itself distinguishes non­ future and future tenses. Jivaro distinguishes five past tenses: im mediate past, recent past, rem ote past, habitual past (over a long period o f time) and reportative past. Jivaro also distinguishes im mediate, definite and indefinite futures. Exam ple (20) gives A chuar phrases with the irregular verb titin ‘to say’, and (21) gives further examples o f contrasts in tense. 326 (20) M ary Ruth Wise (Fast, Fast and Fast 1996: 75-7) táw ai (21) ‘he says’ tu rd tu i ‘he says to m e’ tiji ‘he said (im mediate)’ tim iaji ‘he said (recent or rem ote)’ tin u ja ji ‘he used to say’ tá u ‘he said (reported)’ tita tu i ‘he will say (im m ediate/definite)’ titin u ilai ‘he will say after some tim e’ (Fast, F ast and Fast 1996: 4 5 -6 ) jà à û tak ák m a-s-m a-m í yesterday to. work-PER f v - r e c . PAST-2sg ‘You worked yesterday (recent).’ am t-Ja tak ák m a-s-m ia-m í 2sg-also tO.WOrk-PERFV-REM.PAST-2sg ‘You worked som e tim e ago (rem ote).’ jam áikia tak ák m a-s-ta-h ai now to. work-PER FV-IMM. PAST-1 sg ‘I am g o in g to k a jin wi tom orrow ‘I am g o in g work now (im m ediate).’ Isg tak ák m a-s-ta t-h a i to .w o r k - P E R F V - F U T .D E F - ls g to w o r k tom orrow ( d e f i n i t e ) . ’ Aspects distinguished include perfective (or com pletive o r terminative), d u ra tiv e ( o r p ro g re s s iv e , in c o m p le tiv e ) a n d in c e p tiv e . D irectional suffixes occur in all o f the language families except Jivaro. Iquito has a representative set: - w it ‘up, upriver’, -kuaa ‘down, dow nriver’, -k u w ii ‘returning’, -sa w ii ‘arriving’, ~maa ‘indefinite in tim e/direction’ (E astm an and Eastman 1963: 180), In Yagua, categories having to do with m ovem ent are ‘bounded* (‘upon arrival’, ‘upon departure’, ‘en route’, ‘stationary’) and ‘unbounded’ (e.g. ‘while wandering aro u n d ’). In Jivaro, C ahuapana and W itoto desiderative affixesoccur, especially if the subject o f the com plem ent and ‘to w ant’ are identical. The desiderative C ahuapana is a prefix (ja- in Chayahuita). (22) M u n ii Witoto (Burtch and Wise 1968: 23) kuio-aka-d-e see-want-THEMATic-3sg ‘He wants to see.’ in 12 Small language families and isolates in Peru 327 All o f the languages have m orphological causatives. In C ahuapana and Jivaro languages and H arakm bet a causative prefix consisting o f a single vowel is iden­ tified by David Payne (1990b: 78) as a wide-spread gram matical form in South Am erican languages. These languages, as well as the others, also have causative suffixes. In m ost cases the causative affix can occur w ith both transitive and in tran ­ sitive verb roots. (23) Achuar (Fast, Fast and Fast 1996: 38-9) w aká-tin vs i-wiák-tin to .c lim b -iN F C A U S - tO .c lim b -I N F ‘t o c l i m b ’ ‘to lift’ takákm a-s-m ia-ji taká-m tik-sa-m ia-ji tO.WOrk-PERFV-REC.PAST-3sg tO.WOrk-CAUS-PERFV-REC.PAST-3sg ‘he has w orked’ ‘he has m ade him w ork’ In W itoto itself one o f the causative suffixes is -ia . A valency-changing suffix with the consonant i is wide-spread in A m azonian languages (Wise 1993). In Zaparo, Peba-Yagua and C ahuapana (and possibly Jivaro although it has not been analysed as such), a suffix with t functions as an applicative" so that norm ally oblique noun phrases are treated as direct objects. The applicative may also introduce another argum ent although it sometimes remains implicit. In C ahuapana the suffix -tii-ta may verbalize, transitivize, change im personal verbs to intransitive, detransitivize, and change transitive to ditransitive, as in (24) (- r ‘indicative’; -in ‘third singular agent or subject’; third singular object is zero). (24) Chayahuita (M. H art 1988: 269^70) ira ‘trail’ (noun) ira-ti-r-in am a-r-in ‘he/she bathes’ am a-ti-r-in taji-r-in ‘becomes night’ ta ji-ti-r-in (lit. ‘it nights’) ‘he/she walks’ ‘he/she bathes him /her’ ‘it becomes night in the place where he/she is’ (lit. ‘it nights on him /her’) ‘he/she obeys nati-tt-r-m ‘he/she obeys’ him /her’ a?pa-r-in 'he/she sends it’ a?p a-ti-r-in ‘he/she sends it to som eone’ In Zaparo languages, the meanings o f -tal~tia include: in a container, with contents, a group/plural, affection, perform the action carrying something (25), involuntary " See Greenberg (1987: 317-20) for discussion of a system of applicative classifiers in Almosan-Keresiouan languages. 328 M ary R uth Wise accompaniment (i.e. being carried), to do with a goal/reason, the subject is sick/old/wounded (26), and reflexivizer, i.e. intransitivizer. A n additional function, unique to Zaparo, is that it occurs in the main verb when the subject o f the infiniti­ val complement is different, as in (27). (25) (R . Rich 1999:431) kua m o re h a k a Isg m a n io c tiu r ii- tia - r e e - n ih ia stum ble-APPHC-cOM PL-lsg ‘I stum bled while carrying my load of m anioc.’ (26) (R. Rich 1999; 56) h a n ij a - r i n u -k o k o -h i kua Jikiorta-aJi Isg-S tra il-b y -o f Isg hurt-CL r o ji-jo - k o -h o - ta - rik io -w a -n i drag-CONT-COntact-MULTIPLE-APPLIC-IMPERFV-REPET-lR ‘W ith my wounded [leg] paining me at each step, I dragged myself along the trail and returned [home].’ (27) A m hela (R. Rich 1999: 91) h anija kia isg 2sg pani-tia-a want-APPLIC-CONT kia-nu-ni gO-INF-lR ‘I w ant you to go.’ Reflexive and reciprocal suffixes usually precede aspect and person suffixes, as in A chuar (28). (28) (F ast, Fast and Fast 1996: 38) tja rii-k -h a i vs tjard-m a-k-hai CUt-PERFV-lsg CUt-REFL-PERFV-lsg ‘I c u t i t ’ ‘I c u t m y s e lf’ In Bora there is one valency-reducing derivation which can be interpreted as reflexive o r passive according to the context. (29) (Thiesen 1996: 59) 6 k*’á?tá ? ín iíi-m eí Isg CUt-REFL ‘I cut m yself / 1 was cut.’ A suffix - j is reported for Yagua, which forms an intransitive stem from a transi­ tive root w ith O becom ing S.“ Doris Payne (1985b) uses the term ’anti-causative’ for this suffix. 329 12 Small language families and isolates in Peru (30a) (D oris L. Payne 1985b: 38) tsa-n 6 öta -m áá-rá 3sg-knock.down-PERFV-iNAN ‘She/he has knocked it do w n .’ (30b) tsa -n é ö la -j-m á á [sa n ö ö ta m já á ] 3sg-knock.down-DETRZR-PERFV ‘She has fallen d o w n .’ Evidentials occur in Bora and Zaparo. In A rabela there appear to be only reportative evidentials: - m is suffixed to the subject if it precedes the verb, as in (31); if there is no subject preceding, -kinia is suffixed to the verb, as in (32). (31) (R. Rich 1999: 80) kua k e-ja-na kua nik itio -ja kuno p u eja -n o -n i Isg father-pI-REP Isg that person-sg-lR give-coN T ‘They say my parents are giving m e to that m an .’ (32) (R. Rich 1999: 80) ku so-ja-k in ia sick-coNT-REP ‘They say he/she is sick.’ In Bora the reportative - îh a indicates that the speaker was not a witness o f that which he reports; the reportative -pa indicates that th e speaker is reporting what he heard. T h e two may occur together, as in (33). (33) (Thiesen 1996: 97) H o tsée-p á-?h á-p '’e lum tpá Joseph-REP-unseen-PAST k''uiuipé-?(56-ha-tuj escaped dark-room -house-from ‘Joseph escaped from jail a while back (th e o n e w ho told m e w as not a w itness).’ 4 SYNTAX 4.1 Constituent order The preferred constituent order is AOV, SV in the Jivaro, W itoto, C ahuapana fam ­ ilies and in A rabela and H arakm bet. T he preferred order in Iquito, Z aparo and Ticuna is AVO, SV; in Yagua and Taushiro it is VAO, VS. U rarina is the only Peruvian language in which it is OAV, SV. All o f the languages have postpositions. D emonstrative, quantitative and descriptive adjectives generally precede their 330 M ary Ruth Wise nouns, as in (34a); descriptive adjectives, however, can follow the noun, as in (34b), In com pound nouns the modifier is usually first, as in (34c) and (35). (34a) (34b) A m bela (R. Rich 1999: 94) k e ra a tia p u eree tu u k a sapitiaaha ‘m any large fish’ kua m ueruu kiari-nii Isg m achete new-CL ‘my new m achete’ (34c) k ahi-rikiako ‘axe-handle’ (35a) Chayahuita (H. H art 1988: 261) p im o-Japon ‘t'ragrant-soap’ (35b) /in iti-n iti to.be.thin/skinny-nose ‘n arrow nose’ ‘In Yagua the unmarked order o f constituents within the noun phrase is: D EM + N U M B ER + H E A D N + D ESC .A D J’ (D oris L. Payne 1986:447), as in (36). (36) (D oris L. Payne 1986: 447) h in ju u j ánuuhjuj these.two two kiw àâ fish hààm u-juj big-two ‘these two big fish’ In Chayahuita, descriptive adjectives occur much more frequently as the predi­ cate adjective o f equative clauses than in a noun phrase. (37) (H . H art 1988:292) to ? tir a tin tja n tjo p i claws long m a jo w i? very ‘Its claws (o f the sloth) are very long.’ Genitives usually precede their noun heads. In Yagua, however, if the possessor is expressed by an enclitic prefixed to the head noun and by a noun, the possessor noun phrase follows the head noun, e.g. sst-hûmün>u [sQOmun-'u] A lchico (3sgcanoe Alchico) ‘A lchico’s canoe’ (D oris L. Payne 1985b: 103). M ost kinship terms and body parts are inalienably possessed; however, the possessive construction does not differ, w hether the nouns are inalienable or alienable. 331 12 Sm all language fam ilies and isolates in Peru Interrogative words in content questions are usually the first element o r cliticized to the first element in the sentence, as in Zaparo (38); in that language, however, inter­ rogative particles in polar questions are always cliticized to the subject, as in (39). {Peeke et at. 1991: 14) (38) lia á jtjo r a k a what nam e ali t/ a 2sg be.iNTERROG ‘W hat is your nam e?’ (39) (Peeke eííí/. 1991: 13) m ah a-k á tja -ti á w iro -k a a ts á -k a -? raw -sg 2sg-lNTERR0G be e tle -sg ea t-C O N T -A N T IC lP A T O R Y ‘A re you eating a raw beetle?’ Negative affixes usually occur in the verb, as in M urui W itoto hikano-jje-d-e (ask-N EG -T H E M A TlC -3sg) ‘he did not ask’ (B. Burtch and Wise 1968: 23). N ega­ tive interjections are reported for all o f the languages, e.g. A chuar a tsá /tsá ‘n o ’. In Yagua there is no specifically passive construction; however, predicate nom inal constructions often convey a passive sense, as in (40). (40) (D oris L. Payne 1985b: 38) n á á jiá -tsí-n u m a a -rá j trampIe-O.NOMZR. ANiM.sg-now-1sg ‘I am now tram pled / 1 am now a tram pled one.’ In W itoto itself, as in (41), and Zaparo languages, as in (42), there are specific passive constructions; the agent may be explicit in W itoto. (41a) M urui Witoto (B, B urtch and Wise 1968: 25) bie em u rue o this earth you to.make-PASSV ‘The earth was m ade by you.’ (41b) hai (j>a-ka-m3 m arc G a ha b ie -m o n a a lre a d y to,prove-(»ASSV-when good th en th ere -fro m m a r e -h i- t- e g00d-FUT-INDJC-3sg ‘Now w hen it h a s b een proved g o o d , it will b e g o o d fro m th ere .’ (42) /lra ie /fl(R . Rich 1998) paa ta ri l.in c l now huura-ha-Ji-ja-ree-ni d i s a p p e a r - C A U S - P A S S V - c o N T - c o M P L - lR ‘We (our population) are being decim ated.’ 332 M ary Ruth Wise 4.2 Relativization Relative clauses génerally follow their head nouns if there is one. In C hayahuita the only m ark o f relativization is -so ? suffixed to the verb, e.g. iti aw ir-an-so? (agouti strike-2sg-REL) ‘the agouti which you struck/hit’ (H. H art 1988; 264). In Yagua, relative clauses - enclosed in square brackets below - consistently follow their head nouns; non-pronom inal relativizers are form ed with the dem on­ strative -h ij plus the neutral classifier - r a and the enclitic-tij, as in (43a), or a rela­ tive pronoun may be formed by adding the enclitic -iij to a pronoun such as rij, as in (43b). The verbs in both clauses are finite (43a) (D oris L. Payne 1985b: 69) r a m jitiw je rja h im jitja ra [h irjá tijl r a j-m u tiw je j-râ I s g -c o o k -iN A N tsa-tààrjO j hij-ra-tlj fo o d 3sg-buy d e m - c l :n e u t - r e l T o m ása-ra T o ta -JN A N [R E S U M P T IV E ] ‘I cooked the food that Tom bought.’ (43b) (D oris L. Payne 1985b: 71) néé tsâm ir-ja [rítjim já j tû w â â tju rij-tij-m iij NEC g o o d -C L :N E u r tsiim u] tsa-im u 3pl-REL-NEG listen 3sg-L0C ‘Those who don’t listen to him /her are not good.’ The com mon argum ent may have any function in the main clause and in the rela­ tive clause. in Jivaro languages there are no relative pronouns but third person pronouns may have this function. The more com m on construction is simply a nominalized clause, as shown in the A guaruna examples in (44). (44a) Aguaruna (Larson 1978: 296, 299) ta k ak tsu h i [suhum ainu-k] w e.do.not.carry.in.h a n d ‘We d o (44b) not w hat.m ight.be s o ld - T O P iC have anything to sell.’ Aguaruna (L&vson 1978:252,255) m ik itjik aints [m atjiriki-n one person m onkey-O ‘a person who tamed a m onkey’ tarjkumau] one.who.tamed 12 Sm all language fam ilies and isolates in Peru 333 In Bora an independent clause with the subject m arked by a pronom inal suffix may function as a subject, as in (45b), o r object, as in (45c); o r if there is a nom inal subject, the inanim ate m a rk e r-nel-n'e is suffixed to the otherwise independent verb in the relative clause, as in (45d). (45a) (Thiesen 1996:91-2) mit''->ane k'’ák'’ím eí-íp'e much work-3sg ‘He works h a rd .’ (45b) [mftWane k '’á k ’’ím ei-ípjé] ts ’'áhuik*'<Sö much w ork'3sg come > ‘The one who works hard is com ing.’ (45c) [m it^ane k ’’ák*'ímeí-ip' e-k*’e] m uch work-3sg-ACC 6 Isg a?t<5 pay ‘I paid the o n e w h o works a lo t.’ (45d) Ö Isg aht'''uim { [Jöáa saw John k>’ák''ímeí-nj e] w o r k - iN A N ‘I s a w John w orking.’ In Iquito, relative clauses are introduced by the dem onstrative pronoun iíná ‘this, that’. (46) (Eastm an and Eastm an 1963: 167) káá ki-nikii kiaa-m utiiiira [iíná NEG Isg-see 2sg-m otor that 2sg-used.to.have kiaa-m iijáákura] ‘I d o n ’t see the m o to r which you used to have.’ 4.3 Subordination and coreference/switcb-reference O ther noininalized clauses include com plem ents and purpose clauses. W hen the subject o f the verbal com plem ent is the same as th a t o f the main clause, the com ­ plement is norm ally nom inalized with an infinitive suffi.K as in A rabela (47). (C om pare (27) for use o f the applicative in the main verb if the subjects are different.) (47) (R. Rich 1999: 91) han ija k ia-ta kia-nu Isg 2sg-COMIT pani-ja-ni g O - lN F ‘1 w ant to g o w ith yo u .’ w a tlt-C O N T -lR 334 M ary Ruth Wise In Yagiia if the subjects are different a finite verb is used in the com plem ent with a non-coreferential person-m arking clitic, as in (48). (48) (D oris L. Payne and T. E. Payne ! 990: 337) [hiw jààta] [tsQûndâj] liij-w âàta tsa-hOnijáj T om áása 2sg-want 3sg-iook Tom ‘D o you w ant Tom to look/write?’ In Yagua, infinitival nom inalizations have adverbial functions also. Usually they are suffixed with a postposition and person-m arking clitics (equivalent to possessor reference), as in (49-50). If subject reference is om itted, it is understood that the subject o f the infinitive is the same as that o f the main verb. (49) (D oris L. Payne and T. E. Payne 1990: 337) tsatsii'm jaa him juutjààniihu-nü tsa-tsííj-m aa hij-him uütja-hanu-hO -níi 3sg-run-PF.RFV COREF-help-INP-ADLATlVE-3sg ‘H e has run to help him .’ (lit. ‘he, has run tow ards his, helping him.’) (50) (D oris L. Payne and T. E. Payne 1990: 337) [jaj [sââj] hij-â tsâ â j [siim b jcem ih û ju ra ] tsa-him jij-hanû-hQ -jù-râ 2sg-lRR give 3sg-eat-INF-ADLATIVE-COREF-INAN ‘Give it to him to eat.’ (lit. ‘give it to him, towards his, eating’) The Yagua corefereiice clitics indicate ‘coreference with a previously mentioned subject, possessor or object o f postposition within the same clause . , . Part o f the function o f the set II [suffixal] coreference clitic is to indicate reflexivity and reci­ procity’ (D oris L. Payne and T. E. Payne 1990: 361). C ahuapana a n d Jivaro, on the other h a n d , have subordinating verbal suffixes which indicate coreference or lack o f it, i.e. switch-reference, relative to the agent or subject o f the adverbial subordinate clause and that of the m atrix clause. In Chayahuita, switch-reference suffixes occur only in subordinate verbs with third person agents o r subjects, as shown in table 12.14 (from H. H art 1988: 489) with the sequential action, subordinate forms o f the intransitive verb s a k a tk in ‘to work’.’’ Switch-reference systems are highly developed in Jivaro languages. Long chains o f subordinate clauses occur. Larson (p.c.) reports a com plete A guaruna narrative consisting o f 136 subordinate clauses, including em bedded dialogue, marked for The morphemes could be broken down further but the whole does not always equal the sum o f the parts. 12 Sm all language fam ilies and isolates in Peru 335 Table 12.14 Chayahuita switch-reference suffixes Singular I 2 3 sakato-w ato SS DS excl dual incl sakato-w atan sakato-w aton sa k ato-w atjin sakato-w atoi sakato-w ati? sakato-vvatiwa? sakato-w atam a? sakato-w atona7 s a k ato -w a tjîn aî same o r different agent or subject; these are followed by one independent clause. A person m arker occurs in the verb, and the subordinating suffixes also m ark tem po­ ral and or logical relations. T he subordinate clauses are m arked by - k (a ) ‘topic m arker’ if they are part o f the m ain event line. B IB L IO G R A P H Y Agnew, A. and Pike, E. G. 1957. ‘Phonemes o f O caina (W itoto)’, UAL, 23.24 7. Alicea, N, 1975a. ‘Vocabularia Taushiro’, Dittos Eino-LingiHsticos 22 (microfiche). Lima: Institute Lingüístico de Verano. 1975b. Änálisis fonémioo preliminar del idiom a taushiro’, Datos Etno-IJngülsticos 23 (microfiche). Lima: Instituto Lingüísüco de Verano. 1975c. ‘Un texto taushiro’, D a to s E tn o-Lm giilslicos 23 (microfiche). Lima: Instituto Lingiiistico de Verano. I975d. ‘Análisis preliminar de la gramática del idioma Taushiro’, D a to s E tno-Lingilislicos 24 (microfiche). Lima: Instituto Lingiiistico de Verano. 1976. ‘Taushiro’, pp. 163-79 o f Loos 1976, vol. L Alicea, N. and Garcia, A. 1975. ‘Un texto historico de los taushiro’, D a to s E tno-Linguisticos 56 (microfiche). Lima: Instituto Lingiiistico de Verano. Anderson, D. 1962. Comersational Tiaina. Pucallpa; Instituto Lingiiistico de Verano. Anderson, Lambert. 1959a. ‘Ticuna vowels with special regard to the system o f five lonemes'. S érie L in gû istka Especial, 1.76-119. 1959b. ‘Ticuna vocabulary o f minimal tone words’, S érie Lingilistica Especial, 1.120-7. 1966. ‘The structure and distribution of Ticuna independent clauses'. Linguistics, 20.5-30. Anderson, Loretta and Wise, M. R. 1963. ‘Contrastive features of Candoshi clause types’, pp. 67-102 o f S tu dies in Peruvian Indian languages I, SIL Publications in Linguistics, 9. Norm an; SIL and University o f Oklahoma. Aschm ann, R. P. 1993. Proto ffftoioan, S lL an d University ofTexas at Arlington Publications in Linguistics, 114, Dallas: SIL and University of Texas at Arlington. Beasley, D. and Pike, K. L. 1957. ‘Notes on Huam bisa phonemics'. Lingua Posnaniensis, 6 . 1- 8 . Bendor-Samuel, J. 1961. The verbal p iec e in Jebero, Supplement to Word, ¥ol. 17, monograph no. 4. 336 M ary Ruth Wise Burtch, B. 1975. ‘Fonologia del idioma huitoto muriii', Datos Etno-Lingiiisticos 9 (micro­ fiche). Lima: Institute Lingiiisfico de Verano. Burtch, B. and Burtch, S. A. 1976. ‘Huitoto m urui', pp. 119-218 o f Loos 1976, vol. 0 . Burtch. B. and Wise, M. R. 1968, ‘Murui (W itotoan) clause structures’, Linguistics, 38.12-29. Burtch, S. A. 1983. D kchnario Huitoto Munit, Série Lingüistica Peruana, 20.2 vols. Pucallpa: Institute Lingüîstico de Verano. Costales, P. and Costales, A. 1975. ‘La familia elno-Hngiiistica záp ara’, Ethos, 1.3-33. Cox. D. 1957. ‘Candoshi verb inflection', UA L, 23.129-40. Eastm an, R. and Eastm an, E. 1961. (ms.) ‘Iquito gram tnar outline’, Informacion de Carapo, 108 (microfiche). Lima: Institute Lingüîstico de Verano. 1963. ‘Iquito syntax', pp. 145--92 o f Studies in Peruvian Indian languages 1 ,SIL Publications in Linguistics, 9. Norman: SIL and University o f O klahoma. 1976. ‘Fonologia del idioma iquito’, Datos Etna-Llngiiisticos 29 (microfiche). Lima: Instituto Lingüîstico de Verano. Espinosa, P. L. 1955. Contrihuiiones lingfiisticasy etnográficas sobre algmospueblos mdigenas de la Amazonia peruana. M adrid: Instituto Bernardino de Sahagùn. Eynde, E. van den. 1972. Léxicos y fonologia Arnaralcáiri-Wacipáiri (fíarákmbet o'Mashco'), Docum ento de Trabajo, 7. Lima: Centro de Investigadoti de Lingüistica Aplicada, Universidad Nacional Mayor de San Marcos. Fast, G. 1975a. ‘Estructura de la narracién en A chual’, Oatos Etno-Lingiiisticos 6 (microfiche). Lima: Instituto Lingüistico de Verano. 1975b. ‘Sistema fonolôgico del idioma achual’, Datos Etno-Lingüisticos 20 (microfiche). Lima; Instituto Lingüistico de Verano. 1981. Imroducdôn al idioma Ackuar. Docum ento de Trabajo, 20. Pucallpa: Instituto Lingüistico de Verano. Fast, G., Fast, R. and Fast, D. 1996. Dicclonario Ackmr-Shiwiar-Castellano, Serie Lingüistica Peruana, 36. Lima: Instituto Lingüistico de Verano. Fast, G. and Larson, M. L, 1974. Introducdôn al idioma Aguaruna, Docum ento de Trabajo, 3. Pucallpa: Instituto Lingüistico de Verano. Faust, N. I97L ‘Cocama clause types’, pp. 73-105 o f Tupi studies I, ed. D. Bendor-Samuel, SIL Publications in Linguistics and Related Fields, 29. N orm an: SIL and University of Oklahoma. 1972. Gramática Cocama: leccionespara el aprendizaje del idioma Cocama, Série Lingüistica Peruana, 6. Pucallpa; Instituto Lingüistico de Verano. Faust, N. and Pike, E. G. 1959. ‘The Cocam a sound system’. Série Lingíiística Especial, 1.10-55, Gibson, M . 1996. El Munichi: un idioma giie se extingiie. Série Linguistics Peruana, No, 42Pucullpa; Instituto Lingüistico de Verano. Gnetrc, M. 1988. ‘Evidence for a macro-Arawakan-Jivaroan connection’, paper presented at the symposium on Arawakan Linguistics during the 46th International Congress of Americanists, Amsterdam, July 1988. Greenberg, J. H. 1987. Language in the Americas. Stanford: Stanford University Press. Grimes, B. F. 1992 (ed.). Ethnologue.- languages o f the world. 12th edn. Dallas: SIL. H arl,G . and H art, H. 1976. 'L a fonologia del chayahuita’,i)oíoíir»io-I.m gÄií(c;os 28 (micro­ fiche). Lima: Instituto Lingüistico de Verano. 1981. ‘La cohesion en el texto narrativo del chayahuita’, pp. 69-151 o f Wise and Stewart 1981. H art, H. 1988. Diccionaria Chayaliiiita-Casteilano, Série Lingüistica Peruana, 29. Pucallpa: Instituto Lingüistico de Verano. 12 Sm all language fam ilies and isolates in Pern 337 H art, R. 1963. ‘Semantic com ponents o f shape in Araarakaeri gram m ar’, A nth ropologim l Lingm stics, 5.1-7. Heiberg Chávez, H. 1984. ‘Skizze eitier G ram m atik des A m arakaeri’, Ph.D. thesis (1982), University o f Tübingen. 1990. ‘Análisis funcional del verbo am arakaeri’, pp. 2 2 7 ^ 9 o f Temas de lingüfstica A m erindia, ed. R. Cerron-Palomino and G. Solis Fonseca. Lima: Concejo Nacional de Ciencia y Tecnologia and Oesellschaft fur Technische Zusammenarbeit, Jakway, M., Lorenzo, H. and Antonio, A. 1987. Vocabulario H m m b isa , Série Lingüistica Peruana, 24. Pucallpa: Instituto Lingiiistico de Verano. Larson, M. L. 1963. ‘Emic classes which manifest the obligatory tagmemesin m ajor indepen­ dent clause types o f Aguaruna (Jivaro)’, pp. 1-36 o f S tudies in Peruvian Indian languages I, SIL Publications in Linguistics, 9. N orm an; SIL and University o f Oklahoma. 1978. The functions o f reported speech in discourse, SIL Publications in Linguistics, 59. Dallas: SIL and University of Texas at Arlington. 1981. ‘Training to train: the key to an ongoing program ’, pp. 335-50 o f B ilingual£ducation: an experience in Peruvian A m azon ia, ed. M. L. Larson and P. M. Davis. Washington, D.C., and Dallas: Center for Applied Linguistics and SIL. Leach, L 1969. Vocabulario Ocaina, Série Lingüistica Peruana, 4. Pucallpa: Instituto Lingüístico de Verano. Loos, E. E. (ed.) 1976. M a teria lesp a ra estudiosfonologicos, Docuraento de Trabajo, 9 .2 vols. Pucallpa: Instituto Lingüistico dc Verano. Lyon, P. J. 1975, ‘Disiocacion tribal y clasificaciones üngüisticas en la zona del rio M adré de D ios’, X X X I X Congreso Internacional de Am ericanistas; J e ta s y M em orias, 5.189-207. Maniis, P. 1976, ‘Términos urarina’, pp. 5-50 o f Loos 1976, vol. H. 1977. ‘N otas sobre la fonologia del idioma urarina’, D a to s Em o-LingüisU cm 53 (micro­ fiche), Lima; Instituto Lingüistico de Verano. 1992. ‘Proposiciones de desarrollo y de apoyo y su expresiôn en las oraciones del urarina’, B efista Latlnoam ericana de E studios E tno-Lingülsticos, 7.15-45. Minor, E. E. 1956. ‘W itoto vowel clusters’, U A L , 2 2 .0 1 -7 . Minor, E. E. and Loos, E. E. 1963. ‘The structure and contexts o f W itoto predicates in nar­ rative speech’, pp. 37-66 o f Studies in Peruvian Indian languages 1, SIL Publications in Linguistics, 9. Norm an; SIL and University o f Oklahoma. Minor, E. E. and Minor, D. 1971. Vocabulario H uitoto M uinane, Série Lingüistica Peruatia, 5. Pucallpa: Instituto Lingüistico de Verano. 1971. R e s m m n d e la g ra m á lica [ W itoto J. Lomalinda: Instituto Lingüistico de Verano. 1976. ‘Fonologia del huitoto', S istem as F on dégicos d e Idiom as Colotnbianos, 3.59-67, 1987. Vocabulario bilingiie: H uitoto-E spaifol, E spañ ol-H u itoto (diaiecto M inica). Lomalinda: Editorial Townsend. Montes Rodriguez, M. E. 1995. Tonologia de la lengua T ia m a ( A m acayacti), Lenguas Aborigènes de Colombia, Etescripciones, 9. Bogota: Centro Colombiano de Estudios de Lenguas Aborigenes, Untversidad de los Andes. Payne, David L. 1976. N asalidad en Aguaruna, Série Lingüistica Peruana, 15. Pucallpa: Instituto Lingüistico de Verano. 1981, ‘Bosquejo fonologico del proto-shuar-candoshi’, R evista del M useo Nacional, 45.323-77. 1985. ‘The genetic classification o f Resigaro’, U A L , 51.222-31. 1989. 'O n proposing deep genetic relationships in Amazonian languages: the case o f Candoshi and M aipuran Arawakan languages’, paper presented at the Society for the Study of Indigenous Languages o f the Americas. 338 M ary Ruth Wise 1990a, ‘A ccent in A guaruna’, pp. 161-84 o f Amazonian tinguistics: studies In Lowland South American languages, ed. Doris L. Payne. Austin: University o f Texas Press. 1990b. ‘Some widespread grammatical forms in South American languages’, pp. 75-87 of Amazonian lingmstics: studies in Lowland South American languages, ed. Doris L. Payne. Austin: University o f Texas Press. Payne. Doris L. 1984. ‘Evidence for a Yaguan-Zaparoan connection’, 1984 Work Papers of the SÍL University o f North Dakota Session, 28.131-56. 1985a. ‘Degrees of inherent transitivity in Yagua verbs’, UAL, 51.19-37. 1985b. ‘A spects o f the gram m ar o f Yagua: a typological perspective’, Ph.D. thesis, University o f California at Los Angeles. 1986. ‘Basic constituent order in Yagua clauses: implications for word order universals’, pp. 440-65 o f Handbook o f Amazonian languages, vol. L, ed. D. C. Derbyshire and G. K. Pullum. Berlin: M outon de Gruyîer. 1987. ‘Noun classification in the Western Amazon', Language Sciences, 9.2 M 4 . 1990. The pragmatics o f word order: typological dimensions o f verb initial languages. Berlin, New York: M outon de Gruyter. Payne. Doris L. and Payne, Thom as E. 1990. ‘Yagua’, pp. 249-474 o f Handbook o f Amazonian languages, vol. II, ed. D. C. Derbyshire and G. K. Pullum. Berlin: M outon de Gruyter. Payne, T. E. 1983. ‘Yagua object clitics: syntactic and phonological misalignment and another possible source o f ergativity’, pp. 173-84 o f Papers from the Parasession on the Interplay o f Phonology, Morphology, and Syntax. Chicago: Chicago Linguistic Society, 1985. ‘Participant coding in Yagua discourse’, Ph,D. thesis. University o f California at Los Angeles. 1993, The twins stories: participant coding in Yagua narrative. University o f California Publications in Linguistics, 120. Berkeley and Los Angeles: University of California Press. Peeke, C. 1954. ‘Shimigae, idioma que se extingue’, Peru Indígena, 5.171-8. 1962. 'Structural summary o f Z áparo’, Studies In Ecuadorian Indian Languages, 1.125-216. Peeke. C. and Sargent, M. 1953. ‘Pronombres personates en andoa’, Peru Indigena, 5.12:103-12. Peeke, C.; rev, by M. R. Wise and S. H. l.evinsohn. 1991. Bosquejo gramatical del Záparo, Cuadernos Etnolingiiisticos, 14, Quito: Institute Lingiifstico de Verano. Pike. K. L. 1961, ‘C om pound affixes in O caina’, Language, 37,570-81. Powlison, E, 1971, ‘The suprahierarchical and hierarchical structures o f Yagua phonology’. Linguistics, 75,43-73, Powlison, P, S, 1962, ‘Palatalization portm anteaus in Yagua (Peba-Yaguan)’, H'ord, 18,280-99. 1965, ‘A paragraph analysis o f a Yagua folktale’, UAL, 31.109-18. 1969. ‘Yagua mythology and its epic tendencies', Ph.D. thesis. Indiana University 1985, Yagiia mythology: epic tendencies in a new world mythology. International Museum o f Cultures Publication 16. Dallas: SIL, 1995. -Diccionario Yagua-Castellano, Série Lingüistica Peruana, 35. Lima: Institute Lingiiistico de Verano. Powlison, P. S. and Powlison, E. 1976. ‘Yagua’, pp. 77-118 o f Loos 1976, vol. IT Rich, E 1963. ‘A rabela phonemes and high-level phonology’, pp.193-206 o f Studies in Permian Indian languages 1, SIL Publications in Linguistics, 9. Norm an: SIL and University o f Oklahoma, Rich, R, 1975. ‘Sufijos verbales y apuntes gramaticalcs sobre el idiom a arabela’, Datos EtnoLingiiisticos 18 (microfiche). Lima: Institute Lingiiistico de Verano. 1981. ‘Técnicas de narraciôn en arabela con referenda a tiempo y aspecto’, pp. 15.3-84 o f Wise and Stewart 1981. 12 Sm all language fam ilies and isolates in Peru 339 1999. D iccionan o A rabela-C asieltan o, Serie Lingüística Peruana, 49. Lima: Institute Lingüístico de Verano. Sargent, M. 1959. ‘Vocabulario íáp a ro ’, pp. 43-8 o f E studios de las lenguas H m o ra n i (A u ca), S M m igaey Z áparo. Quito: M inisterio de Educaciôn an d Institute Lingiiistico de Verano. Soares. M. R 1995. ‘Nùcleo e coda: a sikba ein T ikuna’, pp. 195-263 of Estudos fon olágicos clad linguas indlgenas B rasileim s, ed . L. Wetzels. Rio de Janeiro: Editera Universidade Federal de Rio de Janeiro. Stark, X E. 1990. ‘Linguistic trade-in: language change patterns among the Arabela’, MA thesis. University o f Texas at Arlington. Thiesen, W. 1969. ‘The Bora signal drum s’. Lore, 19.101-3. 1975. ‘Un breve informe de ta morfología bora’, D a ta s E tno-Lingülsticos 18 (microfiche). Lima: Institute Lingiiistico de Verano. 1976. ‘Bora’, pp. 7-24 of Loos 1976, vol. I. 1996. Oram ática del Idionm B ora, Série Lingüística Peruana, 38. Yarinacocha, Pucallpa: Instituto Lingiiistico de Verano. 1998- D iccio m rio B ora-C astellan o, C aslelkino-B ora. Lima: Instituto Lingiiistico de Verano. Thiesen, W. and Thiesen, E. 1975. ‘Fonemas del b o ra’, D a to s E tn o-L m gäístkos 1 (microfiche). Lima: Instituto Lingiiistico de Verano. Tripp, R. 1976. ‘Los verbos am arakaeri’, D a to s E lno-Lingiiislicos 33 (microfiche), Lima: Instituto Lingiiistico de Verano. 1976, ‘Sustantivos verbales y frases de sustantivos verbales en am arakaeri’, D a to s EtnoLinginsticos 50 (microfiche). Lima: Instituto Lingiiistico de Verano. 1995. Diccionario A m arakaeri-C asletlan o, Série Lingüistica Peruana, 34. Pucallpa: Instituto Lingiiistico de Verano. Trudell, B. 1993. Beyond the bilingual classroom: literacy acquisition among Peruvian Amazon com m unities, SIL and the University of Texas at Arlington Publications in linguistics, 117. Dallas: SIL and University of Texas at Arlington. Tuggy, J. 1966. Vocabulario Candoshi de L oreto. Série Lingüistica Peruana, 2. Pucallpa: - Instituto Lingiiistico de Verano. 1976. ‘Candoshi: datos fonéticos’, D atos E tno-LingOisticos 53 (microfiche). Lima: Instituto Lingiiistico de Verano. 1981. ‘A discourse phonology o f Candoshi’, Discourse Phonology, Research Papers o f the Texas S IL , 10.5^28. 1992. ‘Algunos aspectos de la m orfofünémica del candoshi’, pp. 322-37 o f E studios EtnoLingiHsticos II, ed. S. Parker, Docum ento de Trabajo, 23. Pucallpa: Instituto Lingiiistico de Verano. Tuggy, S. 1982. ‘Las seciiencias temporales y lôgicas en candoshi’, pp. 37-75 o f Conjunciones y otros nexos en très idiomns A m azônicos, ed. M. R. Wise and H. Boonstra, Série Lingüistica Peruana, 19. Pucallpa: Instituto Lingüistioo de Verano. Turner, G. D. 1958. ‘Alternative phonemicizing in Jivaro’, U A L , 24.87-94. 1958. ‘Jibaro phonology and m orphology’, Ph.D. thesis, Indiana University, 1992. Una breve gram ática d el Shuar, Cuadernos Etno-Lingüisticos, 19. Quito: Instituto Llngüistico de Verano. Velie, D. 1975. ‘Com parative word list of C ahuarano’, in L istas com parativas de paial>ras iisuales en idiom as verm ku los de la selrn, ed. M. Jakway. Lima: Instituto Lingüistico de Verano. Villarejo, A. 1988. A si es la s e lm , 4th edn. Iquitos: Centro de Estudios Tcologicos de la Amazonia de Iquitos. Walton, James. 1977. ‘Participant reference and introducers in M uinane clause and para­ 340 M ary Ruth Wise graph’, pp. 45-65 o f Discourse grammar: studies in indigenous languages o f Colombia, Panama, an d Ecuador, p a r t 3, ed. R. E. Longacre and F. Woods, SIL Publications in Linguistics and Related Fields, 52(3). Dallas; SIL. Walton, Jatnes and Janice P. Walton. 1967. ‘Phonemes o f M uinane’, pp. 3 7 ^ 7 of Phonemic system s o f Colom bian languages, ed. V. Waterhouse, SIL Publications in Linguistics, 14. Norm an: SIL and University of Oklahoma. 1975. Una g rm m lic a d e la kn gua Muinane. Bogotá: Ministerio de Gobierno. Wipio, G.; rev. by A. Paati and M. Jakway. 1996. Diccionario A guaruna-Castetlano. C astellano-A guarm ta, Série Lingülstiea Peruana, 39. Lima: Institute Lingiiistico de Verano. Wise, M . R. 1993. ‘A lgunas interrogantes en la clasiflcaciôn de las lenguas indigenas sudamericanas’, paper presented at the 10th International Congress o f the Asociacion de la Lingiiistica y la Filologia de la América Latina. 1996. ‘Las lenguas záparo - familia que se extingue’, paper presented at the 11th International Cotjgress o f the Asociacion de la Lingiiistica y la Filologia de la América Latina. Wise, M. R. and Stewart, A. (eds.) 1981. Cohesion enfoque en textos y discursos. Série Lingiiistica Peruana, 17. Pucallpa; Institute Lingiiistico de Verano. 13 Other small families and isolates A L E X A N D R A Y. A tK H E N V A L D A N D R . M . W. D IX O N The other small families and isolates are grouped in this chapter, according to the coim tiy in which most o f their speakers are located; some spill over into a nearby nation. Table 13.1 summarizes the languages, estimated num ber of speakers and approximate locations. These locations are given on map 12. (A number o f other lan­ guages, now extinct, on which little or no inform ation is available, are not listed here.)' 1 BRAZIL There are about 170 indigenous languages in Brazil. They belong to about 20 lan­ guage families and are spoken by approximately 155,000 people (A. D. Rodrigues 1986). Languages from the larger families are discussed elsewhere in this volume C arib in chapter 2, A rawak in chapter 3, Tupi in chapter 4, Tupi-G uarani in chapter 5, M acro-Je in chapter 6, Tucano in chapter 7, Pano in chapter 8, M aku in chapter 9, N am b iq u ara in chapter 10 and A raw á in chapter 11. Languages o f the C hapacura family, the Y anom am i dialect cluster, and isolates Trum ai, M uraPiraha, Jabuti, M aku and A ikaná are spoken exclusively in Brazil. Their typologi­ cal properties are briefly discussed here. LI Yanomami The Yanomami (also know n as Y anoam a, Yanomami-W aicá, X Írianá or G uah arib o ) are one of the largest groups o f non-acculturated tropical forest Indians in northern Brazil and in southern Venezuela (M igliazza 1972; Tovar and De Tovar 1984: 1 6 3 ^ ; Lizot 1988: 489). Y anom am i constitutes a dialect continuum Special thanks go to Nadia Pires, for providing information on Jabuti, Henri Ramirez for data on Yanomami, Hein van der Vbort for data on Koaia and lone Vasconcelos, for infor­ mation on Aikaná. §13.1.2, on Trumai, is a condensation, by the authors, of materials written by Raquel Guirardello. We are grateful to Paul Frank, Francisco Queixalos, Terry Malone, Janet Barnes and Tim Curnow for commenting on a previous draft of this chapter. 342 Alexandra V. Aikhenvald and R M. W Dixon M ap 12 Smtill language families and isolates o f Brazil, Bolivia, Colombia and Venezuela with approximate locations which could be regarded as making up a single language. There are Four main dialect groups as set out in table 13.1. Local groups o r villages identify themselves with names ending in -iheri ‘dwell­ ers of, village o f ’. This m orphem e is suffixed to names o f m ountains, or rivers, or places nam ed for p articular vegetation o r animals, e.g. fam a-ta-i-theri (tapir-placeriver-people) ‘X am atauteri’ (a group o f Y anomami: Ramirez 1994), am a-tha-iheri (aiw a-m ountain.place-people) ‘a group o f Yanam {on the Quaim i River, upper U raricaá)’ (M igliazza 1972: 32-3). 13 Other sm all fam ilies and isolates 343 Table 13.1 M inor languages and isolates ( Languages fo r which no linguistic information is available are m arked with *) BRAZIL 1 Yanomami dialect cluster la Yanomami (or Yanoam i, Yanomae), c. 15,000. There is a northern subdialect in the Padam o fiver basin; a central one on the Ocamo, Manaviche and upper Orinoco rivers; and a southern one south o f the Orinoco up to the headwaters o f the M arauia and Cauaburi. 1b Yanomam, c. 3.000. There is a northern subdialect around the upper Uraricuera and lower Parima rivers; a central one on the upper Parima river and the headwaters o f the Orinoco; a southern one in the C atrim ani and Demeni river basins; and pos­ sibly a fourth on the Upper Ajarani river. 1c Yanatn (or Ninam), c. 360. Us southern subdialect is spoken on the M ucujai river, and the northern variety is located on the upper U raricaá and Paragua rivers. 1d Sanum a (or (T)santma), c . 500 in Brazil and c. 1,500 in Venezuela. One subdialect is on the C aura river up to the Brazilian border; another is on the Erevato-Ventuari rivers; and the third is in the Auaris river basin. 2 Trumai, c. 50. Spoken in the Xingu reserve, on the Curisevo river, a tributary o f the Xingu river. 3 M ura-Piraha, c, 250. Maici river (off the M armelos river, off the M adeira river), Ainazonas. 4 Jabuti (Quipiu, or Jeoromitxi), c. 60 people in the indigenous area o f G uaporé, Rondônia (the nearby language Arikapii, with just a few speakers, may be closely geneti­ cally related). 5 Chapacura family 5a Wari’ (or O ro W ari, Pakaa Nova, Pacaas Novos), c. 1,800. Along the tributaries of the Pacaas Novos river and other tributaries o f the M anioré river, Rondônia (dialects include 'O ro N ao' and ‘Oro Eu’). 5b Oro Win, c. 40, A round the headwaters o f the Pacaas Novos river, Rondônia. 5c U rupá, c. 150. On the U rupá river, Rondônia. 5d Torá, c. 250, On the M armelos and Paricá rivers, and on the Posto Cabeça d ’anta, Amazonas. 5e More, spoken by a few people in Rondônia and in Eastern Bolivia. A few more extinct languages belonging to this family are mentioned by Loukotka (1968: 160-2). 6 M áku, 3 speakers in 1964, Around the Uraricuera river, Roraima, 7 A ikaná (or Tubarào, H uari, M asakâ; Kasupá and M undé), c. 120. To the east of the U pper Pimenta Bueno, Rondônia. 8 Koaia (Kwaga. Arara), spoken by a few people who live together with the Aikaná. 9 *lrantxe (Mynky, M ünkü), c. 190. On the Upper Juruena, Mato Grosso. 10 *Sapc (Caliana), 5 speakers in 1964. On the Upper Paragua and its tributary Caroni; South Venezuela/Braziî border (Migliazza 1978a). 11 *U ruak (Awaké), 5 speakers in 1964. On the U pper Paragua and Uraricaá rivets, Venezuela/Brazil border (Migliazza 1978a). 12 ?K atukina, a family which includes Kanam ari, c. 600; K atukina d o Biá, c. 250; T xunhuá-djapá, c. 30; and Katawixi, c. 10. A round the U pper Juruá, Jutai and Javari, Am azonas (A.D. Rodrigues 1986). 344 Alexandra Y. Aikhem ald and R. M. W. D ixon Table 13.1 (cont.) BOLIVIA 13 Tacana family 13a Araona, c. 80. Around the headwaters o f Manuparf river. 13b Cavineña (or Kavina, Kavincño), c. 2,000. South-east o f Riberalta, along the Beni river, 13c Tacana, c. 3,500. On the Beni and M adre de Dios rivers (sonne o f the Tacana dialects were called Tiatinagua - Girard 1971: 19). 13d Ese Eja (or Ese?ejja, Chama, Tiatinagua, Huarayo, Tambopata-Guarayo; different names may correspond to different dialects - Girard 1971: 19-20), c. 600-50, on the Beni and M adre de Dios rivers; (and c, 250-400 in the regions o f Tam bopata and Heath rivers around M aldonado in Peru). 13e Reyesano (San Borjano; also known as Sapibocona, M aropa; possibly Warisa G irard 1971; 18-20; these names may correspond to different dialects), probably extinct. Around 1,000 metnbers o f the ethnic group reside in Beni Department, around San Borja. Other Tacana languages, possibly extinct, include (Girard 1971:17-20; Loukotka 1968; 1 7 4 -6 ); 13f Arasa, formerly spoken on the M arcopata and Arasa rivers; 13g Mabenaro, formerly spoken on the M anuripi river. 14 Cayuvava (Cayubaba, Cayuwaba), nearly extinct. Formerly spoken west o f Mamore river and north o f Santa Ana (there are reported to be 900 people in the ethnic group). 15 Itonam a (M achoto, Sramo), c. 100 (all bilingual in Spanish). Itonam a river. 16 Movima, c. 1 ,0 0 0 . Around Santa Ana on the Yacuma river. COLOMBIA 17 Guahibo family 17a G uahibo (or Sikuani), c. 20,000. Eastern M etá and Vichada and across the border in Venezuela. 17b Guayabero, c. 800. On Guaviare river. 17c Cuiba (or Cuiba-W ámonae), c. 2,000. Casanare, Arauaca, Vichada and across the border in Venezuela. 17d Macaguán (o rJ itn » ),c . 180. In the D epartm ent o f A rauca, bordering Venezuela. !7e Playero, c. 150-60. Arauca river, between A rauqita and Arauca. 18 .Sáhba-Piaroa putative family 18a *Piaroa, c. 80 in Colombia and c. 12.000 in Venezuela. South bank o f the Orinoco river, inland from the Paguasa river to M anipiari, Amazonas. 18b Sáliba, c. 2,000, The Eastern Plains along the Meta and Casanare rivers and across the border in Venezuela. 19 Andoké, c. 75-100. The tributaries o f the Caquetá. 20 ‘ Puinave, c. 2,000. Initida river (and c. 240 across the border in Venezuela). VENEZUELA 21 Yaruro (or Llaruru, Fumé, Yuapin), c, 2,000-3,000. On the Orinoco, M eta and Apure rivers, in the states o f Amazonas and Apure. 22 Warao (or G uarauno, Guarao), c. 18,000. On the delta of the Orinoco river. 23 *Hoti (or Hodï, Yuwana, Waruwaru, Chicano), c. 300-500. On the Kaima river in Amazonas, and on tributaries o f the Asita and M anapiare rivers. 13 Other sm all fam ilies and isolates 345 The dialects are lexically very close, S anum a being the most divergent. The average num ber o f lexical cognates on a 200-word list is 70-80%; the dialects share 95% o f their gram m atical morphemes. The Y anomami live in a m ountainous jungle area o f difficult access, away from main rivers. They live in small com munities each o f 20 to 350 people, spread over a total area o f about 200,000 square kilometres. M eans o f subsistence are hunting, fishing, gathering and slash-and-burn agricul­ ture. The Y anomami are know n as fierce, dangerous people and are feared by their neighbours (C hagnon 1992; Lizot 1988). They are reported to practise ritual can­ nibalism; when an im portant person dies, their body is burnt, then the bones are crushed in a pestle and mixed with banana mash. A feast is m ade to which people come from distant Y anomami villages. At this feast, men and children consum e the mash an d take hallucinogenic drugs. The Yanomami are constantly engaged in intervillage warfare; they are known to raid neighbouring villages, sometimes killing all the men, and capturing women and children. The first contact with E uropeans is reported to have been after 1800 (M igliazza 1985; 27). A few missions have been established in Yanomami-speaking com m u­ nities, starting from 1950. Since 1970, their territories have been under constant threat o f being invaded by non-indigenous settlers. In recent years the Y anomami on the Veiiezuelan-Brazilian border have suffered the effects o f a massive gold rush (starting from 1987, see G om ez 1990:10). The num ber o f Yanomami has fallen drastically during the past few decades. The Y anom am i are mostly m onolingual, with a certain degree o f polyglossia in different Yanomami dialects (M igliazza 1972). A ritual lingua franca, called ‘wayamo’, is spoken between people from different villages, mostly on festive occa­ sions (1972: 47-58). A bout a q u arter o f Sanum a speakers are bilingual with Yekwana (or Dekwana), a C arib language. Nowadays, men tend to know some Portuguese or Spanish. The inform ation available on Yanomami languages varies in quality and in quanti.ty M igliazza (1972) provides an excellent description o f the Yanomami dialect continuum , cast in a generative framework. D ifterent varieties o f Y anom am i are described by Ram irez ( 1994) and Lizot ( 1996); there is a good gram m ar o f Sanuma by Borgman (1990); and a short study o f Y anam by G om ez (1990). (All the exam­ ples are from Y anomami unless indicated otherwise.) 1.1.1 Phonology There are seven vowels - 1, e, a, o, it plus high-central i and mid-central a The con­ sonant system is given in table 13.2. In Yanam , u has fallen together with i. There is regressive vowel assimilation 346 Alexandra Y. Aikhenvaldand R M. ÍV. Dixon Table 13.2 Yanomami consonants In all four groups: stop fricative affricate nasal flap p, t, k s, h th m, n r glide w, y Yanomami also has; Yanomam also has; Sanuma also has: Yanam also has: J f, J J tj ts across m orphem e boundaries. Vowels are weakened in a non-accented word-final position and may be deleted. Syllable structure is (Cj)(C,)V. The possibilities at C jC , a repr, kr, mr, hr. Prim ary stress is assigned to the penultimate syllable o f a word, and secondary stress to every alternate syllable before that. Every m orphem e is either intrinsically nasal, o r intrinsically oral (Migliazza 1972: 157-9). A nasal root spreads its nasality onto an intrinsically oral suffix pro­ vided that it does not start with a stop or a fricative. There is no nasality spread onto a clitic. Roots consist o f two moras. Lexical items which contain three or more m oras are interpreted as containing a tw o-m ora root and one o r two fossilized der­ ivational affixes. The following morphemes are longer than two moras; their third syllable, which starts with a stop and thus ‘blocks’ nasalization from spreading, is treated as a suffix: ârëpo ‘term ite’, m öhöti ‘idiot’, m âhëko ‘knee’ (Ramirez 1994: 68-70). 1.1.2 Word structure and word classes Yanomami is suffixing and predom inantly head-m arking with elements of depen­ dent-m arking. It is highly polysynthetic. Open classes are nouns and verbs; adjec­ tival concepts are expressed with stative verbs. Pronouns and demonstratives are similar to nouns in that they take the same case markers. There are five dem onstra­ tives, the choice o f which depends on distance from the speaker and from the hearer, and on the visibility o f the object. 1.1.3 Nom inal categories T here is a distinction between alienable and inalienable possession. Alienable pos­ session is m arked by simple juxtaposition of terms, e.g. S anum a kamisa sai-a (my house-sg) ‘my house’ (Borgman 1990: 128). Inalienably possessed body parts are obligatorily incorporated into the verb. They are positioned before the verbal root. 347 13 Other sm all fam ilies and isolates a fte r the cross-referencing markers, (la ) shows the incorporated noun Aa ‘head o f ; (lb ) is ungram m atical (Ram irez p.c.). (la ) Jam a ja-he-va-m a tapir Isg-head-eat-PAST ‘I ate tapir’s head.’ (lb ) *J'ama he ja va-ma If a noun functions as a predicate, the body p a rt is encliticized to it, as in (2). (2) Jam a-he tapir-head ‘This is a tapir's h ead.’ Possessive relations can also be m arked with independent pronouns o r 1st and 2nd person possessive suffixes (1st - p , 2nd -ha). 3rd person possessor is m arked by a clitic -e, or by incorporating the possessed noun, w ith a suffix -pi, onto the verb. Case m arkers are cliticized to the last word in an NP. They are -ni ‘ergative (with anim ate nouns); instrum ental (with inanim ate nouns)’; -fo ‘associative w ith’; and a general oblique case with the following forms: -ha ‘inanimate, non-peripheric (near)’, -i-ha ‘animate, non-peripheric (near)’, -ha-m i ‘inanimate, peripheric (far)’, -i-ha-mi ‘animate, peripheric (far)’. This is used to express a variety o f meanings (locative, directional, addressee, agent o f causative, com parative see Ram irez 1994: 193-200; and the com parison o f case markers in M igliazza 1972: 331). In (3), -/Aft m arks the beneficiary. (3) väro-pa-iha hiima ja-e-kipi-hipa-ke-ma-he dog lsg-OBL-DL-give-FOC-PAST-3pl.OBL ‘I gave the two dogs to the men.’ man-PL-BENEFIC(ARY 1.1.4 C la ssifiers There is a rich system o f verbal classifiers; some dialects are reported to have over 100, Classifiers categorize the S, as in (4), or the O, as in (5); they are obligatory and ap p ear im mediately before the verb root (R am irez 1994; 127). (4) kori una-ya-i crest CL:CREST-fly-DYNAMIC ‘The crest (bird) flew away.’ (5) mau u-koa-t-he water CL:UQUiD-drink-DYNAMic-3pl.ERG T h e crests (birds) drink water.’ kori-una-pa-ni Crest-CL;CREST-PL-ERG 348 Alexandra Y. Aikhenvald and R M. W Dixon In (5) the verb includes the ‘liquid’ classifier, referring to the O argum ent. In (4) the verb indudes a classifier referring to the S argum ent; this is a ‘unique’ classifier used just for ‘crest’ (a type o f bird). In verbiess clauses, classifiers are postposed to the predicate n oun, as in: (6) parava-u river-CLXiQUiD ‘This is a river.’ Some examples o f classifiers are ko ‘ro und’; si ‘objects with large surface, palm trees’; ma ‘h ard objects’; k i ‘snakes’; p a ‘turtles’; u ‘liquid’, etc. Classifiers can co­ occur with the plural m arker -ps, but not w ith the singular m arker -a. They are also used on nouns, e.g. kori-ima-ps (crest-CL;CREST-PL) ‘crests’. Most classifiers originate in body-part terms. 1.1.5 Grammatical relatio n s All Yanomami dialects have ergative case m arking. In addition, the verb always cross-references A and O o f a transitive, and S o f an intransitive clause. In (7), suffix -he cross-references an N P in the A function (marked with the erga­ tive case), vâro-ps, while the enclitic -a cross-references the O, absolutive Jama ‘tap ir’ (Ram irez p.c.). (7) vâro-pa-iii^ Jam a^ a-nia-m a-he tapir so;ABS-shoot.arrow-PAST-3pl.ERO ‘T he men shot arrows at a/the tapir.’ man-PL-ERG In (8), the absolutive (‘tap ir’) is unm arked, and is cross-referenced on the verb with absolutive a. (8) Jam a^ tapir a-toku-m a S0:ABS-run-PAST ‘T he tapir ran away.’ Cross-referencing m arkers indicate person and num ber (singular, dual or plural) of A, and ju st num ber for O or S. The distinction between inclusive and exclusive for 1st person pronoun has been lost in Yanam and Yanomam. Oblique constituents (most often the beneficiary) can also get cross-referenced on a transitive verb. In (3) the oblique marker he, formally identical with the 3pl erga­ tive marker, cross-references the addressee (‘men’) on the verb (further discussion is in Ramirez 1994: 315). Yanam (G om ez 1990: 48ff., 129) is reported to have a split-ergative system depending on th e semantics of NPs. In agreement with the N om inal Hierarchy, there 13 Other sm all fam ilies and isolates 349 is a n o m i n a t i v e - a c c u s a ti v e s y s te m f o r firs t a n d s e c o n d p e r s o n s i n g u la r p r o n o u n s . T h e e r g a t i v e m a r k e r is o b l i g a t o r y w ith t h i r d p e r s o n f r e e p r o n o u n s a n d w i t h n o u n s . 1.1.6 Verb structure Verbs may be intransitive, transitive, am bitransitive o f type S = A or am bitransitive o f type S = 0 . Verbal m orphology is extremely complicated. There are m ore than twenty positional classes o f proclitics (cf. Ram irez 1994: l(X)ff.) and over twenty positional classes o f suffixes. Example (9) shows a com plex verb, with four proclit­ ics and eight suffixes. (9) Jam a e-ki-pata-ra-hore-pi-pra-m a-re-hari-no-ve-i ta p ir OBL-DI .-AUGMENTATIVE-TOPIC-rUSh.ofî-DLrDISCONTINUOUSCAliS-TELIC-DIRECTIONAL-ACCOMPLISHED-AFFlRMATIVEEYEWITNESS ‘m a k i n g tw o h u g e t a p i r s r u s h a w a y ’ Proclitics are preposed to the verb but postposed to the noun in a verbless clause. They include pronom inal cross-referencing, classifiers, num ber o f S /0 (dual, plural, collective), incorporated body parts, some evidentiality, m odality, aspect, topic and focus markers, a dim inutive and an augmentative. T h e proclitic kài marks a com itative type o f applicative. The verb ro o t hu- ‘walk’ is intransitive, white kcii-lm- ‘walk w ith’ is transitive, as in: (10) (kam ija-ni) H iterawa ja -k ii-h u 4 Hiterawa Isg -co M iT -w a lk -D Y N A M ic ‘I am walking-with H iteraw a,’ Isg-ERG Verbal suffixes are used to m ark number, causative, tense-aspect (perfective, dynam ic, continuous, iterative, habitual, etc.), direction and negation. An intransitivizer -0 is used to form reflexives, e.g. nia- ‘sh o o t’; nia-o ‘shoot [arrow] at oneself’. C ausative -ma can b e used on any verb (transitive o r intransitive). I f this suffix is added to a-transitive verb, the causer takes the ergative case, the O rem ains O, and the underlying A is m arked with the peripheral clitic {-iha). Example (11) shows a double causative (Ram irez 1994: 247). (11) Hiterava-iha ih i r u H iterava-P E R iP H E R A L c h ild ja-e-hiim a-rara-m a-m a-re-m a Isg-O B L -dog-run-C A U S-C A U S -T E L icPAS'F ‘I let H iterava m ake the dog o f a child ru n .’ Y anom am i has a four-term evidentiality system. ‘Eyewitness’ is shown by a suffix, ‘deduced’ and ‘reported’ by proclitics, and ‘assum ed’ by a com bination o f 350 Alexandra Y. Aikhenvald and K M. W. Dixon proclitic plus suffix. Sanuma (Borgman 1990; 165-73) has a three-term system, with ‘eyewitness’, ‘verification’ and ‘supposition’, while Y anam ju s t has ‘eyewitness’ and ‘non-eyewitness’. Different evidentiality m arkers are used in the three dialects, perhaps indicating a relatively late development o f evidentials (also see Migliazza 1972; 268). 1.1.7 N oun incorporation and verb compounding Any noun in S or O function can be incorporated into the verb to m ark a topically continuous participant which is not in focus (Ramirez 1994; 385). Com pare (12), without incorporation, and (13), where the direct object, ‘axe’, is incorporated. Incorporation does not affect the transitivity o f the verb. (12) (kam ija-ni) sipara ja-puhi-i 1sg-ERG axe ‘I want a n /th e axe.’ (13) (kam ij3-ni) Isg-ERCi Isg-want-DYNAMic ja-sipara-puhi-i Isg-axe-want-DYNAMiC ‘I w ant [it], the axe.’ Verb com pounding is very productive. C om pounded verbs consist o f an indepen­ dent verb root followed by a bound verb root in a modifying function, e.g. kö ‘do again’ in ( 14) (Ram irez 1994; 330-47). Bound verb roots cannot form a predicate by themselves. (1 4 ) ja -p a -n a k a -4 -k ô -o Isg -P L -call-D Y N A M ic-d o .a g ain -iN T R ‘I c a lle d t h e m a g a i n .’ 1.1.8 Syntax The unmarked constituent order is AOV, SV. There are no copula verbs. Yanoniami has considerable freedom in the order of core and oblique nominal constituents within a clause. In Sanuma, constituent order displays ergative characteristics. Using X to indicate a peripheral constituent, the preferred order is AXOV, XSV, i.e. S and O pattern together, occurring between X and V. Demonstratives, quantifiers, numerals and classifiers precede the head noun, while adjectival modifiers follow it. Subordination is usually marked w ith verbal suffixes; some o f these suffixes indi­ cate switch-reference. Relative clauses follow the head. In Yanomam, Yanam and Sanuma a nom inalizer can also m ark the predicate o f a relative clause or of a com ­ plement clause. T hus (M igliazza 1972; 86); 351 13 Other small fam ilies and isolates ( 15) w àro-n Jam a m an -E R G /k a -w e i t a p i r kill-R E L w a re -m a e a t-c o M P L ‘T he m a n w h o k ille d t h e t a p i r a t e i t . ’ Sanum a also has a relative pronoun I This dialect is reported to have internally headed relative clauses; the m ain clause contains a pronoun coreferential w ith the common argum ent. See (16), from Borgman (1990: 133). There are no restrictions on the functions o f a coinnion argum ent. (16) p pa iilu kökö lo-le] î kökö hu mai kite my son 3 p i Sit-PRES r e l 3 p i go n e g ‘My sons w ho are sitting here [lit. ‘they’] will n o t go.’ pu t The type o f m arking o f a relative clause depends on its predicate. If the predicate o f a relative clause is an active verb, the relative pronoun can be om itted, and just -wi ‘relativizer, nom inalizer’ is used. If the predicate o f the relative verb is a stative verb, the predicate takes -i ‘relativizer’ as in: (17) [ ip a hoose a my y o u n g e r .b r o t h e r 3sg AUGMENTATIVE ta ll Î a REL 3sg to n ih a sa te Isg 3sg p a ta la p e to to -k i ke give-F O C i m m . pa s t e p i-i] in t e n s iv e - r e l z r ‘I g a v e it t o [lit. ‘h i m ’] m y y o u n g e r b r o t h e r w h o is t a l l ’ 1.2 Trumaj by Raquel Guirardello The Trum ai say th a t they cam e from the south-east into the U pper X ingu region (see ch ap ter 15) w ithin quite recent times. They provide an unusual example o f (tem porary) language revival. Trum ai num bers were reported to be 43 in 1928, 25 in 1948, only 18 in 1952, then 21 in 1963 and over 100 (including many children) by the early 1990s. However, only about h alf (approxim ately 50) o f the ethnic Trum ai speak the language today. There has been a great deal o f m arriage with people from o th er X ingu tribes, allied to the fact th a t the o th er X ingu people con­ sider Trum ai ‘to o hard’ a language to learn. Owing largely to contact w ith a nearby governm ent post, m ost j'oung people have switched to Portuguese. It is not likely that the language will continue to be spoken for m ore th a n another couple o f gen­ erations. The first vocabulary o f Trum ai was taken down by von den Steincn (1886). M onod-Becquelin (for example, 1975,1976) attem pted som e gram m atical analysis. G uirardello (1992) is a workman-like preliminary statem ent o f the phonology and m orphology which will be refined in G uirardello (forthcom ing). 352 Alexandra Y. Aikhenvald and R. M. W. Dixon Table 13.3 Trumai la b i O ' b ila b ia l ps d e n ta l p p a la to d e n ta l a lv e o la r a lv e o la r p a la ta l v e la r g lo tta l ? t t k v o ic e le s s e je c tiv e s t o p s f t’ k’ v o ic e d s to p d v o ic e le s s a f f r i c a te ts , t s ’ fri I la te r a l fric a tiv e ♦ ta p r a p p ro x im a n ts 1.2.1 w Phonology Syllable structure is (C)V(C). Stress goes on the last syllable o f a word. There are six vowels: /, e, a, o, u plus high central i. The consonants are set out in table 13.3. Notable features are the four ejective stops and affricates and the lateral fricative (these are n o t found in any other X ingu language). There is a language death phe­ nom enon - some young speakers make no distinction between plain and ejective stops and affricates. 1.2.2 Features o f the grammar There are open classes o f noun (including pronouns and demonstratives), verb, adjective and adverb and closed classes o f auxiliary, postposition, num ber/quan­ tifier, particle, subordinate clause m arker and interjection. Pronouns are analysable into root: ha (!sg and In.sg.excl), k a (In.sg.incl), hi (2) or in- (3); plus zero (sg), a (dl) or wan (pi). 3sg has distinct m asculine and feminine forms. Demonstratives ‘this’ and ‘th a t’ are based on 3rd person pronouns. Trumai is a dependent-m arking language. An N P in S and O function takes no m arking and there are postpositions for ergative (A function), dative, locative, allative, ablative, instrum ental and comitative (the ergative, dative and locative markers are enclitics). Verbs fall into four classes: plain intransitive (obligatory S argument), plain transitive (A and O), extended intransitive (S and dative) and extended tran­ sitive (A, O and dative). Verbs in the extended intransitive class include ‘eat’, ‘drink’, ‘see’, ‘like’ and ‘talk w ith’. There are two verbs ‘kill’ - fa which is extended intran­ sitive and disi which is transitive. Trumai has little morphology. There are no TAM suffixes to the verb, reference to time being expressed through adverbs. There are post-verbal particles marking 353 13 Other small fam ilies and isolates oegation, causation and emphasis. N egation is shown by anuk in copula and tak in non-copula clauses. There is a 3rd person enclitic -nl-e (referring to an argum ent in S or O function) which is attached to the right o f the verb phrase; interestingly, this cannot co-occur with a negative marker. For alienable possession the possessor bears the genitive suffix -k (a )te and pre­ cedes the possessed, e.g. Kum aru-kte tahu ‘ICumaru’s knife’. Inalienable possession can be shown ju st by apposition, e.g. dinoxo k u f ‘the girl’s hair’, ha k u j 'm y hair’. When a third person (anaphoric) possessor pronoun is employed, inalienably pos­ sessed nouns subdivide into two classes. T here is a prefix t(si)- on kin terms, e.g. tsipine ‘his/her cousin’; and there is an enclitic -afci (or -ea preceding a postposition) on bo dy-part nouns and also nam es for clothing, home, shadow, nam e and pet, e.g. ku f-a ke ‘his/her hair’, kiij-ea letsi ‘with his/her h air’. T he basic clausal constituent order is SV, AOV, with a dative NP following the verb- If the S or O N P does n o t immediately precede the verb, a particle ke m ust be included after the verb. In contrast, an A o r dative N P can be moved around w ithout any post-verbal m arking. W ithin an NP, quantifiers and possessors precede and adjectives follow the head. W ithin coordinate constructions, coreferential dele­ tion can operate on an S = 0 o r S = A basis, suggesting a pragm atic (rather than a syntactic) pivot. T he causative o f a transitive clause has a m ost unusual structure, in which both the causer and the A of the underlying verb take ergative m arking (the ergative enclitic is -is after Isg a nd -(e /a )k elsewhere). (18) Alaweru-k N a m e -E R G hai-ts Isg-E R G axos child+ abs disi ka beat caus ‘Alaweru m ade m e beat the child.’ 1.3 IVJura-Piraha The M ura tribes had a tum ultuous history, and speak an unusual language. U nfortunately, there are m any points o f uncertainty concerning both. Some nineteenth-century sources state that the M ura originated on the Andes in Peru an d travelled down the A m azon to escape the yoke o f civilization (one source says: o f the Incas; another states: o f the Spanish). O ther sources m aintain that the M u ra originated on the lower M adeira river. A fter the original tribes on the A m azon were eliminated by the Portuguese, through enslavement and disease, the M ura expanded and are said to have extended from the Trom betas river up to the border with Peru. Estimates o f their population at the time also vary - from 30,000 or 40,000 to 60,000 o r more. T he M ura were a fierce tribe, using guerrilla tactics o f am bush to terrorize other tribes and also Portuguese invaders. They made peace with 354 Alexandra Y. Aikhenvaldand R. M. K Dixon Table 13.4 Mura-Pirahä consonants b ila b ia l v o ic e d s to p p v o ic e le s s s t o p b f ric a tiv e a lv e o - p a la ta l t v e la r k g lo tta l ? g s h the Portuguese in 1784 (partly, it is said, out o f fear of the M undurukù - see chapter 4 - a n even fiercer people) but still indulged in raiding and kiUing into the nineteenth century. {Historical inform ation from Rodrigues and Oliveira 1977, who list ail sources.) There were a num ber o f dialects o f M ura which appear (from the word lists col­ lected - M artius 1867: 20-1; N im uendajii and Valle-Bentes 1923: 218-2S; Nimuendajii 1925:160-6; 1932:93-106) to have been fairly close. It seems that only one of these survives - Pirahä, spoken by about 100 people on the Maici river, a tributary o f the M armelos which itself flows into the M adeira. The P irah i have retained their identity and language partly through m aintaining an antipathy towards most aspects o f the European-based culture that has surrounded them. Three Sum m er Institute o f Linguistics teams have worked on Piraha - Arlo and Vi Heinrichs (from 1960 to 1966), Steven and Linda Sheldon (1967 until the late 1970s) and Daniel and Keren Everett (sporadically from 1978). They provide different accounts o f several aspects o f the language. 1.3.1 Phonology Heinrichs (1964) put forward a system o f eight consonants - see table 13.4 together with three vowels (front i, central a and back o) and three tones (high mid - and low ^’). Everett states that vowels may optionally be nasalized following ? or h. N ote th at the allophones o f Ibl include a bilabial nasal and also a bilabial trill. Heinrichs also described an unusual allophone o f /g/ - a type o f double flap in which the tongue tip hits the alveolar ridge and then (coming out o f the m outh) the lower lip. (Later, D. Everett 1982 described these sounds, but w ithout reference to H einrich’s initial account.) D. Everett (1986) states that women articulate hi as [h], always before /i/ and sometimes elsewhere. Sheldon (1974) accepted Heinrichs’ phonological analysis and described rules o f tone assimilation (e.g., the first vowel o f an adjective assimilates to the tone o f the final vowel o f a noun which im mediately precedes it) and metathesis (ai—>ial-a, and oi-¥iol-o, in certain circumstances). Everett ( 1986) states that he at one time recog­ 13 Other sm all fam ilies and isolates 355 nized four tones and then two (an acute accent indicating high tone and no m ark indicating low tone). He also suggests that k can be regarded as underlying hi, thus reducing the num ber o f consonant phonem es to seven, which he says is the small­ est num ber for any language in the world. 1.3.2 Features o f the grammar The m ost com m on constituent order is A(X)OV, S(X)V where X indicates periph­ eral NPs. N Ps are not m arked for core functions but there are suffixes: -oal-ai ‘instrum en­ tal’, and -o, ‘locative, general oblique’, A directional suffix -?io ‘to ’ may follow the locative/general oblique. Sheldon (1988; 174-5) gives the pronouns as: ti-’ Isg gi'?ai^’ 2sg hi^ 3sg ?i-’ 3sg feminine ? i‘k 3 si’ 3 anim ate aquatic non-hum an ?a^ 3 inanim ate anim ate non-hum an He gives plural forms for 1st, 2nd and 3rd person hum an, involving aiU?so^ added to the singular forms (hi^ai'li^so^ is used for both masculine and fem inine 3pl). A pparently no num ber distinction is m ade for 3rd person non-hum an, Everett (1986; 280R'.) has ju st tiie first four forms from Sheldon’s list plus 3 non­ hum an ?M (which he suggests may be a m orphophonological variant o f ?isi ‘anim al’). H e states that there is no plural m arker for pronouns, ?aitiso being a dis­ course particle ‘also, then’. (Everett quotes N im uendajii 1948 as stating that the pronouns in P iraha were borrow ed from N heengatù o r Lingua G eral, the old lingua franca o f the region. In fact N im uendajii’s rem arks applied to M ura (now extinct), not to Piraha, and he simply stated that three Lingua G eral pronouns were in ‘regular use’ by the M ura, n o t that they had replaced the original M ura forms.) It seems that pronouns can function as either free o r bound (with essentially the same forms) but details as to how these are used are lacking. A lthough n o explicit inform ation is given, it appears that some verbs can be arabitransitive, either o f type S = 0 (e.g. F oabcan m ean ‘die’ o r ‘kill’) o r o f type S = A (it appears that ?ohai ‘see, look’ can be used transitively or intransitively), Everett treats words with an adjectival m eaning as verbs. As w ith many A m azonian languages the verb takes a wide range o f suffixes. Sheldon (1988) describes over thirty suffixes, in ten order classes, while Everett 356 Alexandra Y. Aikhenvald and R. M. W. Dixon {1986) h as nineteen order classes. There is no tense marking but Everett lists the following aspects; perfective, imperfective, telic, atelic, continuative, iterative .and ingressive. O ther suffixes refer to distance (proximate/remote), certainty, desire and frustrated initiation or term ination o f an action. A lthough neither author uses the term , there also appears to be an optional system o f evidentials: ‘observation’, ‘hearsay’ and ‘deduction’. Everett states that there are no valency-changing derivations and no formal marking for causative (one must use a periphrastic construction with the verb ?ibiib ‘order, cause to d o ’), reciprocal or reiexive (one just says ‘1 hit m e’ or ‘he hit him’ which is am biguous between the ‘he’ and ‘him ’ being coreferential' or not). He briefly describes verb com pounding (involving directional verbs) which may be a type o f verb serialization. And there are several copula verbs; ?aagá ‘to have/to be’, ?iigá ‘to be/to have tem porariiy’ and xa i ‘to be/to do’. There is no formal m arking for possession. One simply apposes possessor (with or without a pronoun) and possessed, e.g. Paitá hi ?itohoi ‘Paitá he testicles’ for ‘Paitá’s testicles’. The nominalizer -sai (which replaces aspect m arking on the verb) is widely used. It can form a n oun from a complete clause, e.g.; (19) g a h io p i- 6 ? a b a ip - i- s a i a e ro p la n e w ater-L O C s íí- e p e n t h e t ic -n m l z r ‘h y d r o p l a n e ’ (lit. ‘s i tt i n g - o n - w a t e r p l a n e ’) A com plem ent clause from English may be translated with a -sai nominalized clause (this is a typical A mazonian trait), e.g.: (20) hi ob-áa?áí 3sg see/know -iN T E N S iFiE R kahai a rro w kai-sai m ake-N M LZR ‘H e really knows h o w t o m ake arrows.’ (21) ti ?og-i-bai gi?ai kahai kai-sai Isg want-PROX-iNTENSiFiER 2sg m ake-N M L Z R a rro w ‘I r e a lly like y o u r a r r o w - m a k in g ’ OR ‘I re a lly lik e y o u t o m a k e arrows.’ Temporal clauses are marked by a suffix -sol-ao ‘when, during, after’. Negation is m arked by verbal suffixes -sahai ‘don’t ’ and -hiab ‘n o t’. The latter also has a free form, hiaba, used for negating NPs and nominalized clauses. There is a coordinator piáii ‘and, also’ (generally used with two clauses that have the same verb) and also hoagá ‘but’. In a relative clause the argum ent in com mon with the main clause must be in a core (subject or object) function. Everett states that a relative clause can be marked 357 13 Other small fam ilies and isolates Table 13.5 Jabuti consonants voiceless stop voiced stop voiceless affricate voiced affricate voiceless fricative voiced fricative nasal tap bilabial dental p b ps bz t d P m alveolar alveo-palatal velar glottal tJ ds n r by the relative pronoun go ‘w hat’ o r by hi?, which also functions as the m arker o f a polar interrogative. 1.4 Jabuti Jabuti (or Q uipiu, called by its speakers Jeoromitxi) and Arikapti are both spoken around the R io Branco in the G uaporé region o f Rondônia. On the basis o f a com­ parative vocabulary of about twenty words, alm ost half o f them apparently cognate between the two languages, Loukotka (1963: 49-50) suggested that Jabuti and Arikapii make up a small language family. On the basis o f the vocabulary in Rivet (1953), he assigns M asubi (now extinct) to the same family, although here there are fewer cognates. The only substantial inform ation on Jabuti is Pires (1992) who reports that there are about sixty speakers (including some children), who live together with speakers o f M akurap (Tupari subgroup o f Tupi - see chapter 4). Arikapii has yet to be investigated. 1.4.1 Phonology The consonant system, set out in table 13.5, is unusual in having bilabial affricates (the only instance reported from A m azonia). There are seven oral vowels, /, e, a, o, u, high central unrounded ii and m id-central 3, plus ju st four nasal vowels, /, ë, ä and Ö. Syllable structure is (C)V. Stress is n o t contrastive and can vary in position but generally falls on the final syllable o f a word. 1.4.2 Features o f the grammar Jabuti is predom inantly agglutinating, m ostly em ploying suffixes. Affixation and com pounding are widely used in derivation. T he open word classes are verbs, nouns and adjectives^ (and possibly also adverbs). T he language is head-m arking. There is a set o f m ulti-purpose prefixes. They can occur on a noun, m arking the 358 Alexandra E Ä ikhem ald and K M . W. Dixon possessor; on a postposition, m arking the postpositional argum ent; on an intransi­ tive verb, m arking the S; or on a transitive verb, marking tlie O. A number distinc­ tion is m ade only for I st person: Isg is 0,1 p! is hi-, 2nd person is a-, while 3rd person is i- or e- with nouns, postpositions and transitive verbs, but 0 w ith intransitive verbs (here falling together with Isg). Free pronouns are Isg hU, Ipl hirii, 2 aje and 3 na. Interestingly, a transitive verb m ust be preceded either by an N P (which can be a free pronoun) in O function, or by a pronom inal prefix m arking O. In contrast, every transitive verb bears a pronom inal prefix indicating its A , w hether or not there is an explicit N P in A function. A postposition is either preceded by an NP, or takes a pronom inal prefix, e.g. a-paka ‘with you’, a-ri ‘for you’. Post-verbal particles m ark negation, teose, aspect and (at least with intransitive verbs) causative. The constituent order is basically SV, AOV; A and S can be moved into post­ verbal position but O always immediately precedes the verb. W ithin an N P a dem onstrative precedes and an adjective follows the head. An adjective can func­ tion as N P head and must then take the 3rd person prefix i- (unlike nouns, verbs and postpositions, adjectives cannot take I st and 2nd person prefixes). As in many other Amazonian languages, nominalizations are employed where other languages would have complement or adverbial clauses. In the following exam­ ples, from Pires (1992: 132, 87), the post-verbal particle a m arks the norninalization. (22) hü^ [a-hihi a Isg 2sgO-kill NMLZR ]q tohe w ant ‘1 w ant to be your m urderer.’ (i.e. i want to m urder you’) (23) hilj Isg 0 -jenona 1sgS-arrive [a-rukiihil 2S-leave a ] nmlzr ‘1 arrived just as you left.’ It is likely th at Jabuti originally had no numbers. In the contact situation ‘one’ is given as h o t (we have no information about its original meaning). For 'two', Je-bo is used, involving the verb root -bo ‘be equal’and the reflexive-type prefix je-. For ‘more than tw o’ the phrase hönö là may be used, literally ‘know n o t’, i.e. ‘can’t say how many’. (O f course, Portuguese numbers are now used as loans, in addition.) 1.5 Chapacura family Languages o f the C hapacura (Txapacuran) language family, one o f the smallest in southern A m azonia, are spoken in the G uaporé valley and along the tributaries o f the M adeira river, in the west of the state o f Rondônia and in the south o f the state o f A m azonas (Brazil), and in adjacent regions o f Bolivia. 359 13 Other sm all fam ilies and isolates Table 13.6 W ari’ consonants voiceless stop labialized stop voiceless fricative labialized voiceless fricative voiceless affricate nasal glottalized nasal bilabial dental p t alveolar palatal velar k k* glottal 7 h h* I/ rti m’ n n’ flap r glide w y T h e genetic relationship o f C hapacura languages has been know n since d ’O rbigny (1839); see atsoC réqui-M ontfort and Rivet (1913) (the alternative name used for the family in earlier sources was Pawumwan). The first data on Torá go back to 1716 (N imiiendajii 1925, and see N im uendajù and Valle-Bentes 1923); the first data on U rupá were collected at the end o f the nineteenth century. T he extinct languages o f the family include Jarii (possibly a dialect o f U rupá - N im uendajù 1925: 139) and C hapacura (or Tapacura, also know n as Hiiachis ~ Cham berlain 1910:184). There are only short word lists in these languages. A superficial com par­ ison o f word lists shows that they are quite close (maybe as close as Tupi-G uarani, o r as Romance). The only gram m atical description is that o f W ari’ by Everett and Kern (1997), from which all inform ation below is taken. 1.5.1 P h o n o lo g y The consonant system is given in table 13.6. There is also a complex sound made up o f a voiceless apico-dental plosive followed by a voiceless labiodental trill (see Ladefoged and Everett 1996; Everett and Kern 1997: 1, 396), This occurs only before loi an d /y/. It is reported to contrast with /t/ in one dialect (although only occurring in about twenty-five words) but is an allophone o f h i in other dialects. There are six vowels, Í, e, a, u plus two front unrounded vowels, high open y and mid-close Ö. Syllable structure is CV(C). There are elem ents o f vowel harm ony (not necessar­ ily across m orphem e boundaries; usually contiguous to r). Stress falls on the last syllable o f the word (Everett and Kern 1997: 416). 1.5.2 F e a tu re s o f th e g ra m m a r All C hapacura languages are agglutinating with some degree o f fusion, headm arking, with a nom inative-accusative profile. T he only open classes are nouns and 360 Alexandra Y. Aikhenvald and R. M. W: Dixon verbs; modifiers seem to be a subclass o f nouns. There are no core case markers, A/S and O are cross-referenced through verbal inflectional m arkers {verbal enclitics which contain reference to tense, aspect and m ood). Cross-referencing is obligatory with definite objects, and optional with indefinite objects. Objects are not cross-ref­ erenced when fronted to sentence-initial position in relative clauses, inform ation questions and clefted sentences. In W ari’, gender assignment is predom inantly semantic with a certain degree of opacity (which is characteristic o f languages o f southern A m azonia). It goes along the following lines: • feminine; hum an females, collective nouns, mixed groups o f female and male individuals; « masculine: hum an males, animals and culturally significant objects (natural phenom ena, insects, fish, etc.); ♦ neuter; most inanim ate objects, newly introduced objects/animals/ plants; loans, nominalizations, mixed groups o f neuter and masculine objects. Typically for an A m azonian language, gender is not m arked on the head noun. G ender agreement is m arked on modifiers, possessed forms, and through verbal cross-referencing (Everett and Kern 1997: 294-300). A set o f bound pronominal clitics cross-reference the person, num ber and possessor (if third person), e.g. capija-in wao' (m outh-3neiit basket) ‘rim o f the basket’; capija-con worn (mouth3masc cotton) ‘hem o f the skirt’; capija-cain narima {mouth-3fem woman) ‘w oman’s m outh’; wijima-in xirim (smallness-3neut house) ‘small house’ (lit. ‘the house’s smallness’). The possessor is the head o f an NP. There is one multi-purpose preposition which cross-references the person, num ber and gender o f its argument. Pronouns distinguish three persons, and three genders for 3rd person. There is singular/plural distinction in 1st and 2nd person and in 3sg.masc and 3sg.fem but n o t in 3sg,neut. 1pi distinguishes inclusive and exclusive. All these oppositions are m aintained in reflexive/reciprocal pronouns. Verbs distinguish realis and irrealis, m arked w ith inflectional clitics which are portm anteau with person, number, gender, tense, aspect and m ood. Tense distinc­ tions are; recent past, past/present and future. Irrealis fonns are also used as im per­ atives. Reali.s forms are used in prohibition. Clausal negation can be achieved either with the negative main verb 'om ‘not exist’, taking a com plem ent clause, or by a postverbal modifier 'akiva or m a’. D ouble negative m eaning results in an emphatic positive reading. There are a few aspectual particles; serial verbs are used to express habitual aspect. Iterative aspect is m arked by total reduplication o f the verb root (Everett and ICern 1997; 323-8). Some verbs have suppletive forms used when the S or O argum ent is plural, while 361 13 Other small fam ilies and isolates other verbs m ark plural (o f S or O) by partial reduplication o f the initial CV and infixation o f - m (1997: 33 7 -9 ). There is no m orphological causative mechanism . There appears to be an (im per­ sonal) passive; any verb can b e passivized/intransitivized. (24) w iric a m co q iierec EMPH:3sg.fem iN F L E C T iO N A L :m a sc /fe m .R E A L !S ,P A S T /p R E S see wa PASSV ‘She is the one who was seen.’ (25) ‘irawin ^ com a fte rn o o o sin g ‘ac wa travel P assv tara 3sg.FUT.REALis ‘Then the a fte rn o o n will be su n g in .’ Basic clausal constituent order is VXS, VOXS, where X is an indirect object. Verb com pounding (or ro o t serialization) is very productive. A typical directional com pound is: (26) p an ’ corom m am a fa ll enter go (pl.S) pin ‘awi nana c o m p le te ly com pletely 3 :r e a l i s . p a s t / p r e s ‘They all fell [fall-enter-go] [into the water].’ Verb com pounding is used to form causatives o f switch-subject type on intransi­ tive verbs only, e.g.; (27) ju c cam ara’ p u sh fall (pKS) 3sg:REALis.PAST/pRES'3neut n a -in too X ijam m eta l m ale .n a m e ‘Xijam k n o c k e d do w n (p u sh -fa ll) th e c a n s.’ A causative o f a transitive verb can be achieved by direct speech: ‘he said: do it’ for ‘l ie m ade me do it’. Serial verb constructions are frequently lexicalized. Subordination is marked on the verb. Relative clauses are marked w ith subordi­ nating proclitics. Only core argum ents can be relativized. Nominalization is also used to m ark subordinate clauses. 1.6 Máku* This language isolate had only three speakers in 1964, living near the U raricuera river, R oraim a, northern Brazil. The M áku, according to their own stories, origi­ nally lived around the M aluw aka m ountains between the upper Padam o and ^ This should not be confused with the M akii family - with diflferent stress - discussed in chapter 9. 362 Alexandra Y. A ikhenm ld and R. M. W. Dixon Table 13.7 M ákii consonants voiceless stop voiced stop voiceless affricate voiceless fricative nasal lateral glide bilabial alveo-dental p b t d ts s n r m w palatal J velar glottal k ? x h y C unuciinum a rivers in Venezuela. They used to be regular traders travelling from the upper O rinoco to the Rio Branco in Brazil (M igliazza 1965, 1966). T heir migra­ tion from Venezuela to Brazil was caused by constant attacks by Yanomama. N um bers fell drastically in the period from the 1930s to the 1950s, due to diseases. The M áku were first mentioned by K och-G riinberg (1913: 458). A short descrip­ tion o f M áku phonology and elements o f m orphosyntax are found in Migliazza (1965, 1966 an d 1978a - which also contains a short word list), unfortunately cast in opaque tagmem ic formulas. The consonants of M áku are in table 13.7 (M igliazza 1978a). There are four vowels (/', e, a, u) existing in both oral and nasal varieties, plus two additional oral vowels, high front rounded y and high-central i. Stress is not contrastive (its typical position is on the last syllable); vowel length is contrastive but occurs only in an initial C(V) syllable in a polysyllabic word. Syllable structure is (C)(C)V(C). M áku is head-m arking, highly polysynthetic and predom inantly suflFixing. There are no genders or classifiers. Pronouns distinguish first person exclusive and inclu­ sive. There is a very complex system o f tense-aspect oppositions. 1 .7 A ik a n á and K oaia A ikaná (also know n as Tubarào, H uari, M asaká, K asupá and M undé) is spoken in southeastern R ondônia by about 120 people including some children (Hein van der Voort, p.c.). There is only a little inform ation available - ñeldnotes by H arvey Carlson (1986) (H inton 1993 is based on this), and papers by Vasconcelos (I993a,b, 1996) which are prelim inary studies o f the phonology and some aspects o f morphology of A ikaná. O lder sources - H anke (1956) and Becker-Donner (1955) - contain just a few words. The co n sonant system, in table 13.8, includes a dental fricative (a rarity in the 363 13 Other small fam ilies and isolates Table 13.8 Aikaná consonants bilabial voiceless stop voiced stop voiceless affricate voiceless fricative voiced fricative nasal tap glide dental b t d m ö n alveolar alveo-palatal velar k glottal ? ts s h ji r w y A m azon). There are oral and nasal forms o f four vowels, i, e, a, ii plus a fifth oral vowel, front rounded u. It is unclear w hether there are contrastive tones. A ikaná is predom inantly suffixing with some prefixes. There is no inclusive/exclu­ sive distinction in pronouns. There are two num bers, and no genders or noun classes; anim acy is distinguished in interrogatives {bari ‘who’, tara ‘w hat’). M ost adjectival meanings are expressed through verbs; colour concepts are expressed by nouns. Possession is m arked with suffixes, and with genitive forms o f personal pro­ nouns; these can co-occur. G ram m atical relations are m arked by cross-referencing on the verb: subject (A/S) is m arked with suffixes or a com bination o f prefixes and suffixes, while object is m arked with suffixes. There is an accusative case on NPs. There is body-part incorporation, verba! classifiers, and possibly incorporating serial verbs. Verbal classifiers appear in the first position in the verbal word; they refer to the S, or the O, characterizing it in term s o f its nature (‘liquid’, ‘fibrous’, ‘metallic’), or its structure (‘powder-like’, ‘consisting o f coarse p articles’, e.g. seeds or kernels); there is one classifier covering anim ates and humans. Initial CV redu­ plication indicates repetitive action. There is probably a benefactive valencyincreasing derivation. T he verb is clause-final and there are postpositions. K oaia (also known as Kwaza) is spoken by about 25 people who live together with the A ikaná. The only inform ation available comes from preliminary studies by van der V oort (1997a, b) (a short word list is given in L oukotka 1963). T he consonant system is given in table 13.9. There are oral and nasal variants of seven vowels, /, e, c, t, a, o, ii. Syllable structure is CV(V). It is unclear w hether stress (on the last syllable o f the stem) is contrastive. G lides w and y are allophones o f u and i. K oaia appears to be predom inantly suffixing and com bines dependent-m arking with head-m arking. The verb takes an obligatory pronom inal suffix for S a nd A and a further suffix, which appears to be optional, for O. Only 1st and 2nd person dis- 364 Alexandra Y. A lkhem ald and R. M. W. Dixon Table 13.9 Koaia consonants bilabial voiceless stop implosive voiced stop fricative affricate p 6 alveolar t s ts approximant apico-alveo-palata! c | J t/ velar glottal k 7 h I nasal tap m n r tinguish singular and plural; there is also an inclusive/exclusive distinction. There appear to be verbal classifiers; the same m orphemes are used as agreement markers on nominal modifiers. There is a complex system o f tenses (im m ediate past, proxi­ m ate past, rem ote past) and aspects, and also directional markers on the verb. C onditional and concessive are also marked on the verb. Clausal constituent order appears to be free. There are about 35 indigenous languages spoken in Bolivia by around 70,000 people. Besides languages o f the m ajor families - Arawak, Pano, Tupi-GuaranI, as well as Q ueohua and A ym ara - languages spoken exclusively, o r alm ost exclusively, ill Bolivia include those belonging to the Tacana family (§2.1) and a num ber o f iso­ lates (§2.2). Here we will consider only languages spoken in the northern parts of Bolivia,^ which fall w ithin Amazonia. 2.1 T h e T a c a n a fa m ily The T acana family consists o f four or five living languages spoken mostly in north­ west Bolivia and in the adjacent areas o f Peru, between the rivers M adre de Dios and Beni (table 13.1). Loukotka (1968: 175-6) and G irard (1971: 20; based on M ason 1950) m ention a few more languages o f the Tacana family, all o f them extinct. ’ The other isolates - outside Amazonia - include: Canishana, formerly spoken in the Bolivian Lowlands on the Mamoré river (probably extinct); Leco, spoken by a few people east of the lake Titicaca, some in the Apolo region; Yuracaré, spoken by 500-2,000 people in Beni and Cochabamba departments: Chiqiiitano, spoken by 20,000 to 42,000 to the east of Santa Cruz (also see Ibarra Grasso 1982; M. R. Key 1979). 365 13 Other sm all fam ilies and isolates Table 13.10 Cavineña consonants bilabial voiceless stop voiced stop labialized stop voiceless fricative voiceless affricate nasal flap lateral p b dental t d alveo-palatal palatal c ^ velar k glottal 7 k* ra n s Is ji r I J tj" h T he genetic unity o f the Tacana family was recognized by Brin ton (1891, 1892); further classifications o f the Tacana languages were provided by Cham berlain (1913), an d later by C réqui-M ontfort and Rivet (1921-3), Schmidt (1926) and M ason (1950). Schuller (1933) first suggested that Tacana was related to Pano. This hypothesis was investigated by Key (1968,1979) who dem onstrated P ano-T acana relationships based on a phonological com parison. G irard (1971) reconstructed 116 lexical items for proto-Pano-T acana and established regular phonetic correspondences. While work rem ains to be done in term s o f gram m atical com parison and an overall reconstriiction o f the proto-language, sound foundations have been laid. M aterials available for T acana languages vary in quantity and in quality. There ate a num ber o f excellent-quality papers and a sketch gram m ar o f Cavinefla by Elizabeth Cam p, a few older sources and a short gram m ar (cast in non-user-friendly tagmem ic framework) are available for Tacana and Ese Eja; for A raona there are a few articles and a short gram m ar by Pitm an ( 1980). F or the rest, there are only word lists. 2.1.1 P h o n o lo g y Cavinefia (Key 1968: 19) has the largest phonological system - see table 13.10. It includes a series o f alveo-palatal phonem es missing from other T acana languages. T acana also has interdental voiced stop d, while Ese Eja lacks is. A ccording to Key (1979: 85), Ese Eja has a series o f implosive stops; there is also considerable allophonic variation, e.g. Ini can be realized as [ndj, [1], [nl] or [d]. All Tacana languages have four vowels, i, e, a, o (with « as an allophone). Stress is n o t contrastive (it falls on the penultim ate syllable of a root in Cavinefla and Tacana). Ese Eja has a com plicated system o f stress assignment which is 366 Alexandra Y. Aikhenvald and R, M. W. Dixon different for nouns and verbs. T he com m on syllable pattern is CV (VC structures are found in Tacana and only rarely in non-loans in Cavineña). A raona has a male-fem ale speech distinction ( / i n female speech corresponds to s in male). 2.1.2 Features o f the grammar All Tacana languages are head-m arking and polysynthetic, with long strings of verbal suffixes to express tense, m anner o f action and modality. They mostly use suffixes; a few prefixes occur only with verbs (Key 1968: 29), There are no genders, or classifiers. Valency-changing derivations are marked by verbal prefixes and suffixes- Tacana languages appear to have auxiliary verbs. Cases are used for m arking gram m atical relations on all types o f N Ps in Tacana and Cavineña; in A raona and Ese Eja case m arkers are apparently used only with pronouns. Basic clausal constituent order is SV, AVO. A t least some Tacana languages show ergative properties. Cavinefla has a very unusual pattern o f split ergative m arking o f gram matical relations with cases. It combines split-ergativity conditioned by (a) the semantics of nouns, (b) m ood and polarity, and (c) whether a clause is main or subordinate (Cam p 1985). Ergative (-m ) is obligatory with nouns in A function, but may be om itted from a pronoun in A function. ‘If the overt noun is ergative, then the absolutive pronoun is the object [...] if the noun is absolutive, then the absolutive pronoun is the subject’ (Cam p 1985; 44). In (28) both the pronom inal A (‘we tw o’) and the direct object (‘cows’) are m arked for absolutive case. In (29) the non-pronom inal A (‘cows’) is marked with ergative case. (28) T u ja y a -ts e w aka k*ana b a -ts a -tjin e um ada so 1-DL + ABS cow PL+A BS s e e - a r r i v i n g .o - P A S T m any ‘So (29) w e tw o saw m a n y c o w s c o m in g ’ A -ya ya-tse iyak'^a waka-ra do-PRES 1-DL + ABS nOW COW-ERG ‘T he COWS will get us now.’ Ergative m arking is obligatory with pronouns if the sentence is negative, poten­ tial, counterfactual o r intentional. Example (30) illustrates the ergative marking on a pronom inal A in a negative clause. (30) pake m e -ts e -ra e-k "e u ts e k * a a p p a re n tly 2 - d l- e r g 1 -g e n g ra n d c h ild + ‘A pparently y o u t w o d i d n ’t t a k e g o o d c a r e o f m y g r a n d c h i l d . ’ abs n a ru -t/in e am a c a re .f o r - P A S T not 367 J3 Other small fam ilies and isolates Table 13.11 Cayuvava consonants voiceless stop voiced stop voiceless fricative voiced fricative nasal tlap glide bilabial alveolar alveo-palatal velar P b 1 d s c j 1 3 Ji r k P m n h w Ergative m arking on personal pronouns is also obligatory in subordinate clauses, as shown in (31). (31) i-ke ni m i-ra ara-kara a-ya, duhu-k^e 1-affix+ A8S if 2 - er g eat- w a n t d o -p re s e n t take-lMPER i-ke espere hiruru 1-affix + ABS s tre a m edg e ‘If y o u w ant to e a t m e, ta k e m e to th e e d g e o f th e s tre a m .’ Personal pronouns are m arked for ergative case if they appear sentence-initially, as in (32), or sentence-finally (in fact, this is encountered only rarely): (32) e -ra bak*e a -k * a re 1-ERQ c o n tr a s t do-REM.PAST said’ Cavineña is unusual in that it appears to have an antipassive-like valencyreducing derivation, m arked w ith reduplication o f the verb stem; then A becomes S (i.e. is m arked with absolutive case), and O is om itted. 2.2 Isolates A list o f isolates spoken in n o rth ern and northeastern Bolivia is given in table 13.1. The materials available for these languages are o f mixed quality. For Cayuvava there is a short gram m ar by H arold Key (1967); for Itonam a there is a short gram m ar and a dictionary by C am p and Liccardi (1965, 1967), and for M ovim a there is a short gram m ar and a dictionary by Judy and Judy (1962, 1965). U nfortunately, both gram m ars are executed in a non-user-friendly tagmem ic model. T h e Bolivian isolates have rather diverse phonological systems. Syllable struc­ ture is (C)V. The consonants o f Cayuvava are in table 13.11 (H . Key 1961; 1967: 368 Alexandra Y. Aikhenvald and R M. W. Dixon Table 13.12 Itonam a consonants bilabial dental alveo-palatal voiceless stop voiced stop glottalized stop voiceless fricative voiced fricative voiceless affricate glottalized affricate nasal palatal velar glottal ? ty f la p lateral glide Table 13.13 M ovima consonants voiceless stop voiced stop labialized stop voiceless fricative voiced fricative nasal preglottalized nasal lateral flap lateral glide bilabial dental P b t d alveo-palatal palatal velar glottal c k ? k" h s P m ?m n ?n r 1 w y 15). Cayuvava has /, e, a, o, ii plus high central i, open-m id front «■ and mid-close back 0. Itonam a (Liccardi and Grimes 1968; C am p and Liccardi 1965) is unusual in that it has a series o f glottalized stops, and also palatalized />»-560 table 13.12. There is a typical A m azonian system of six vowels: i, e, a, o, u plus high central i. The consonantal system of M ovima is less complex (Judy and Judy 1965) - see table 13.13. It is unusual in having preglottalized nasals. There are just five vowels i, e, a, 0 and ii. Cayuvava has contrastive nasalization o f vowels and contrastive stress. N asalization, in addition, is reported to function as a kind o f honorific device: a person o f lower social or economic status addresses another one o f ,a higher rank with a prominence o f nasalization on all the vowels of the utterance (Key 1979: 83). 13 Other small fam ilies and isolates 369 All the Bolivian isolates are head-m arking, polysynthetic and nom inativeaccusative. Cayuvava has a m ost com plicated morphology. Its striking property is the existence o f six prefix positions on verbs - these include passive-reflexive, the­ matic, cross-referencing of person-num ber o f the subject, tense and aspect. N ouns take derivational prefixes, and discontinuous circumñxes for m arking possession. There are no genders o r classifiers. There are only two open classes - nouns and verbs; a nom inal modifier can be derived from any word class. C onstituent order is apparently free. M ovim a appears to be predom inantly suffixing w ith ju st one prefix position (occupied by a cross-referencing pronoun). In M ovim a, suffixes are used for caus­ ative, benefactive, malefactive, reflexive and a num ber o f tense-aspect, m odality and directional categories (e.g., Judy and Judy 1965: 188-9). Ito n a m a has pre­ fixes, suffixes and infixes (1965: 293); besides cross-referencing, m any o th e r cate­ gories - such as causative, benefactive, directional - are m arked w ith prefixes to the verb. An unusual property o f M ovim a is its system o f numeral classifiers. I t also has three genders (masculine, feminine, neuter or inanim ate) in personal pronouns and cross-referencing (1965:202). Preliminary fieldwork by G rinevald (1996) shows that the classifiers include -poy ‘quadruped anim als’, e.g. fox, crocodile, tapir, etc.; -mo ‘biped anim als’, e.g. rooster, duck, owl, etc.; and -ba ‘fruit’, e.g. papaya, guava, orange (cf also Key 1979: 67-8). For other native M ovima nouns, the last syllable is repeated on the numeral as an agreement device, e.g. -d’o for chad'o ‘plate’; -mas for d ’imas ‘hay’; -pi, for sukapi ‘belt’. For borrow ed nouns, the last two syllables are repeated if a noun consists of more than two syllables, e.g. -misa for kam isa ‘shirt’, and -pato for zapato ‘shoe’ (both loans are from Spanish). If a loan consists o f just two syllables, the reduplicated last syllable is repeated, e.g. -sasa for m esa ‘table’, -yaya for siya ‘seat, chair’. Ito n am a distinguishes m asculine and feminine genders in first, second and third person singular pronouns and in dem onstratives (Cam p and Liccardi 1965: 331-2; Key 1979: 40). It has a very complex system o f at least seventeen classifier suffixes based on animacy, shape and position o f object; classifiers com bine with adjectives, verbs and demonstratives. Itonam a may also have body-part incorporation (Key 1979:41). 3 COLOMBIA There are about 66 indigenous languages in C olom bia. They belong to 22 language families and are spoken by approximately 500,000 people (Landaburu 1994b). There are also two creoles o f A fro-European origin in north-w est Colom bia. In 370 Alexandra Y. A íkhem ald and R. M . Wf Dixon addition, Lingua G érai A m azônica (N heengatá), a creole o f Tupí-G uaraní origin, is spoken on the border with Brazil. Languages from the larger families are discussed elsewhere in this volume - Carib in chapter 2, A raw ak in chapter 3, T upi-G uarani in chapter 5, Tucano in chapter 7, M akù in chapter 9, and W itoto plus the isolate Ticuna in chapter 12. Languages of the G uahibo and Séliba-Piaroa"* families and the isolate, A ndoké, are spoken exclu­ sively in Colombia.-^ Their typological properties are briefly discussed here.^ The m ajority o f the indigenous peoples of Colom bia are slash-and-bum agricul­ turalists. Only the G uahibo peoples were nom adic hunters and gatherers, like the M akii peoples in Colom bia and Brazil (see chapter 9). G uahibo constitutes an unusual example o f a large nom adic community. G uahibo and Sáliba are currently being studied by scholars from the Centro C olom biano de Estudios en Lenguas Aborfgenes in the U niversidad de los Andes in Bogotá, and by members o f the SIL (see Queixalos forthcoming, and n.d., on Sikuani/G uahibo; K err 1995 on Cuiba; Estrada 1996, and M orse and Frank 1997, on Sáliba). There is a large but opaque gram m ar o f A ndoké by Landaburu (1979), which also contains a few texts. There is virtually no gram m atical inform ation on Piaroa. H uber and Reed (1992) give an overview o f phonological systems o f most Colom bian languages, accompanied by 200 w ords in each language. 3.1 Phonology The G uahibo languages, Sáliba and A ndoké have a voice distinction in stops. There is typically one liquid with a rhotic allophone. C uiba is typologically unusual in that it has no rhotic o r lateral; there is, however, a trilled r as an allophone o f d (K err 1995; 19). This phonological oddity may have an areal explanation.^ Sáliba has two rhotics and a lateral. * Mure evidence is needed to prove the genetic link between Sáliba and Piaroa: see Landaburu (1994b: 370) and Estrada (1996: xxv-xxvi.). * There is no adequate information on Puinave which is possibly another isolate (although it has been suggested that it may be distantly related to Makii). * We do not discuss, in this volume, language families whose members mostly or entirely fall outside Am azonia - Chibchan, Paez, Choco, Barbacoan, Kuna (all of which have - almost certainly mistakenly - been grouped together as ‘Macro-Chibchan’), ^ The dental stop is realized as a lateral flap in north-west Am azonia (Tucano-Tariana region) and also further to the south-west.*/- is an allophone o f in proto-W itoto (Aschinann 1993: 96). Resigaro, a language o f the Arawak family, which came under strong areal influ­ ence from W itoto languages, has no r; according to Allin (197.5). r may have once been an allophone o f d. Other r-less languages, besides Cuiba, are concentrated to the west o f the Vaupés: Hupda (Makii), and Aguaruna in the adjacent regions o f Peru (there is here a lateral fricative. / , which is an allophone o f t before /i). Allophonic variation between a dental stop and a rhotic stretches further to the west (it is also attested in the Barbacoan lan­ guage Awa Pit: Curnow 1997). 13 Other small fam ilies and isolates 371 Table 13.14 Cuiba consonants voiceless stop aspirated stop voiced stop voiceless fricative voiceless affricate nasal glide bilabial alveolar p p’' b t t'' d s ts n m w palatal velar glottal k kl- ? h tjy Table 13.15 Sâliba consonants voiceless stop voiced -Stop labialized voiceless stop labialized voiced stop voiceless fricative voiced fricative voiced alfricate vibrant flap lateral nasal glide bilabial alveolar P b t d <1> P s palatal velar glottal k ? g k« g" X h d3 m w r r ! n y T he consonant systems o f C uiba (G uahibo; H uber and Reed 1992; K err 1995), Sáliba and A ndoké are given in tables 13.14^13.16. M ost G uahibo languages have six vowels (i, e, «, o, a plus high central #); Sáliba has ju st five, lacking i; in each language there is a nasal vowel corresponding to each oral vowel. A ndoké has a larger system with nine oral and five nasal vowels, as in table 13,17(Landaburu 1979: 45^6). Typical syllable structure is (C)V; C uiba and Sáliba can have a nasal in coda posi­ tion, and C uiba also perm its stops. The G u ah ib o languages and Sáliba have contrastive stress (E strada 1996: 3). In G uahibo, stressed vowels tend to have high pitch (K ondo and K ondo 1967; 95). A ndoké appears to have three tonal distinctions: high, middle and low (Landabvsru 1979: 48-51). This property is shared with two neighbouring W itoto languages. Bora an d O caina, which distinguish two levels o f tone (see chapter 12). 372 Alexandra Y. Aikhenvald and R. M. W. Dixon Table 13.16 Andoké consonants voiceless stop voiced stop voiceless fricative voiced affricate flap nasal glide bilabial alveolar velar glottal P b t d s k 7 dS r n h Table 13.17 Andoké vowels front mid back front mid 3.2 Morphology 3.2.1 W o rd s tru c tu re a n d ty p o lo g ica l profile All o f these languages com bine head-m arking and dependent-m arking properties. They are agglutinating with polysynthetic tendencies, the G uahibo languages being the m ost polysynthetic of all. Typically for N orth A m azonian languages, there is one prefix position and numerous suffix positions. Some G uahibo languages are unusual in that nouns have more extensive suffix possibilities (up to six in Cuiba K err 1995: 267) than verbs. The prefix position on a verb is occupied by a cross-ref­ erencing pronoun, valency-increasing marker, o r by a m arker o f plural for S or O argum ent. Interestingly, in Sáliba verbal cross-referencing affixes are prefixes only on vowel-initial roots, while for consonant-initial roots they are infixes or the first suffix (Paul F rank p.c.). In A ndoké the cross-referencing pronouns are all suffixes (Landaburu 1979: 130). In G uahibo, nouns may take a possessive prefix. The G uahibo languages are unusual for northern A m azonia in that they have suppletive forms o f verbs relating to the num ber o f the S or O argum ent, e.g. Cuiba bexuba ‘kill (sg O y ju te b a ‘kill (many O )’. Adjectives are an open class, except in G uahibo where they are said to be a small­ ish class o f ab o u t twenty members (Francisco Queixalos p.c.). 13 Other sm all fam ilies and isolates 3.2.2 373 Nom inal m orphology Free pronouns and cross-referencing pronom inals distinguish three persons and, generally, two numbers. G uahibo languages are unusual in that they distinguish sin­ gular, dual, paucal and distributive plural, and also first person inclusive and exclu­ sive (F ran d sco Q ueixalos p.c.). In some G uahibo languages (e.g. Cuiba), there are dual forms o f nouns, tonic pronouns and verbs, but n o t o f cross-referencing bound pronouns; in o th er languages (Guayabero: T obar Ortiz 1994: 516-17), dual is m arked on all types o f nouns, pronouns and verbs. All th e languages discussed have gender in h ere th ird person pronouns, p lu s a large and com plicated system o f classifiers. G uahibo languages distinguish three genders (feminine, masculine, inanim ate) in independent 3rd person singular and dual pronouns, but these are neutralized in the plural. There are no gender distinc­ tions in bo u n d pronouns. Sáliba is reported to m ark three genders on 3rd person pronouns and adjectives: feminine, masculine and inanim ate (Suarez 1977; 21). G u ah ib o languages have complex, systems of classifiers. The same m orphem es are used as noun class agreem ent m arkers on adjectives, as in (33); with numerals, as in (34); and with deictics, as in (35). Adjectives, num erals and deictics with classifiers can be used headlessly. Examples (33)-(35) are from C uiba (ICerr 1995: 243ff.). (33) peru-nae o ld -C L :W O O D E N T H IN G S ‘old (canoe)’ (3 4 ) ca e -b o o n e -C L :n o u S E ‘o n e ( h o u s e ) ’ (35) barapo-bo this-CL:nousE .‘this (h o u se )’ In C uiba possessive constructions, classifiers are used independently o f w hether the noun is alienably or inalienably possessed, ju s t like in neighbouring Tucano and N orth A raw ak languages o f the U pper R io N egro region: (36) (K err 1995: 134) piya-nae jera 3sg.P0Ssv-CL:w00DEN THINGS canoe ‘his/her canoe’ Also as in East Tucano languages and Tariana, classifiers in G uahibo can be con­ sidered an open class due to the existence o f repeaters (nouns used to classify them ­ Alexandra Y. A ikh en va ld a n d R M. K D ixon 374 selves); however, only inalienably possessed nouns can be used as repeaters (1995: 132). Sáliba also has a system o f at least thirty classifiers. T he same set of morphemes goes on numbers, adjectives, dem onstratives, possessives and even interrogative pro­ nouns. Some classes distinguish a singular and a plural form. Interestingly, numbers ‘one’ and ‘tw o’ also have different forms for anim ate and inanim ate referents (Estrada 1996: 106-7). In G uahibo languages and in Sáliba (Paul F rank p.c.), the classifier suffixes used with nouns have an individuating effect: a noun is unm arked for num ber if used w ithout a classifier. If a classifier is added, the noun has a singulative reading aod can additionally be marked for number, e.g. C uiba (K err 1995: 154) tsocom ‘wild fruit’, tsocoma-ru (fruit-C L:FR U IT) ‘(one) wild fru it’, tsocoma-ru-n (friiltC L:FR U IT-PL) ‘(several) wild fruits’. This individuating effect o f noun classifiers is rem iniscent o f Tucano languages and Tariana o f the Vaupés area, and of Resigaro and W itoto-Bora. It may be an areal feature o f a region em bracing n o rth ­ eastern Peru, south-central Colom bia and north-w estern Brazil. A ndoké also has a complex agreement system. Verbal cross-referencing markers have three n oun classes. Class 3 divides into four subclasses cross-referenced on the verb in the same way, but marked differently on pronouns and noun modifiers. The class assignment is far from semantically transparent. The classes are: (i) C lass 1 includes inanim ate objects, mostly artifacts, but also some anim als, some mythical beings, and plants. (ii) Class 2 includes mainly long and hard objects. (iii) Class 3 includes m ost anim ates and also round and hollow objects. W ithin this class, the following subdivisions are made: — Class 3a contains round and hollow objects; — Class 3b contains m asculine animates; — Class 3c contains fem inine animates; Class 3d contains anim ate collectives. G uahibo, Sáliba and A ndoké have an unusually large num ber o f oblique cases com pared to o th er A mazonian languages (such as Tucano). There is one ‘multi­ purpose’ oblique case (e.g. C u ib a -tha ‘locative, directional, com itative’, or A ndoké -a ‘locative, indirect object m arker’), and a num ber o f other cases with more spe­ cific meanings, e.g. C uiba -xae ‘because o f ’, -xaeba ‘because o f (absence of)’, -nexa ‘purpose’, -yainwa ‘for fear o f’ (K err 1995: 107, 170); A ndoké -má ‘comitative, instrum ental’, -fa ‘privative’, -taa ‘benefactive’, -aka ‘maiefactive (to the detrim ent of)’ (Landaburu 1979: 166-7); Sáliba -di ‘dative’, -da ‘ablative-allative’, -na ‘inessive’, -gi ‘com itative’, -digi ‘privative’ (Estrada 1996: 92-5). 13 Other sm all fam ilies and isolates 3.2.3 375 Grammatical relations G ram m atical relations are m arked by cross-referencing. Only G uahibo languages have some traces o f split ergativity, o f an active-stative type. In Cuiba, verbs divide into five conjugations which take different tense markers and show different vocalic alternations (K err 1995: 3Iff.). Different cross-referencing suffixes are used for AtS^ in positive and negative clauses (Guayabero - Tobar Ortiz 1994: 521-2; Cuiba - Kerr 1993: 31). A nother set o f cross-referencing suffixes is used for the majority o f stative used for O, and also for (37) verbs (ICerr 1995: 59-60). A special set o f prefixes are of some verbs which refer to physical states, e.g. Cuiba: (K err 1995: 51) xam ne-jéita-me you lsg + 0-look.for-2sg + A ‘You looked for m e.’ (38) (K err 1995:71) ne-siiyaba 1sg + S^-have.diarrhoea ‘I have dia rrh o ea.’ 3.2.4 Verbal morphology There are usually several valency-increasing verbal derivations. In G uahibo the prefix ka- functions as an applicative and is also used to form causatives o f in tran ­ sitive verbs, while any verb can be causativized with an auxiliary (Francisco Queixalos p.c.). In G uahibo languages the same m orphem es can be used as appli­ cative markers on verbs, and as case-like m arkers on pronouns and inalienably pos­ sessed nouns. A privative applicative verbal prefix is illustrated in (39), from C uiba (K err 1995: 107): (39) ta-xant-iyo Is g -c h ild -F E M tsi-se-ta-n PR IV .W IV E . A P P L I C - C o o k - 1Sg-t>AST ‘1 c o o k e d f o r i n y d a u g h t e r w i t h o u t h e r h e l p . ’ If the N P prom oted to be a core argum ent is a pronoun or an inalienably pos­ sessed noun, the applicative m arker is suffixed to the NP, as in (40) (1995: 107): (40) ne-tsi Isg-PRiVATiVE.APPLic xane 3sg-(-eat ‘He ate w ithout me.’ A ndoké has two causatives, and also a benefactive applicative (L andaburu 1979: 204-5). Alexandra Y. Aikhenvald and K M . W. Dixon 376 There are also a num ber o f valency-decreasing derivations; G uahibo has passive, middle, and a curious derivation whereby the addressee o f a ditransitive verb becomes an oblique, if hum an (Francisco Queixalos p.c.). In Cuiba the reflexive-reciprocal derivation can also be used to em phasize the identity o f the agent, similar to English j e ^ i n I did it myself. A ndoké appears to have just one general intransitivizer (Landaburu 1979: 205). All the languages have rich systems o f m ood and modality, as well as directional suffixes on verbs. G uahibo languages have a restricted system o f evidentiality (reported vs non-reported), G u ah ib o languages have unusual incorporation patterns. Typically for A mazonian languages (e.g. Weir 1990, on N adëb), only inalienably possessed nouns can be incorporated. C uiba has two patterns o f noun incorporation: (i) inalienably possessed nouns, in S function, can be infixed into verbs of physical state (K err 1995: 104), e.g. atane ‘h u rt’, a-cobe-tane 'his hand h u rts’. (ii) with other verbs, an NP in S, O or an oblique function can be incorpo­ rated (but not infixed) into a verb. Typically for A m azonian languages, incorporated nouns are preposed to the verbal root. Unlike m ost other A m azonian languages - but tike N adëb (M a k á family) - C uiba perm its incorporation o f more than one (up to three) arguments (K err 1995: 278-9); here incorporation does not aflect the transitivity o f a verb, e.g. (41) and (42). However, in G uahibo noun incorporation is a valency-reducing derivation (Francisco Queixalos p.c.). (41) (K err 1995:280) cobe-fifina-n h an d-tired-lsg ■ ‘My hand is tired’ or ‘I am hand-tired.’ (42) (K err 1995:279) n a -m a x i-p é ri-n a -d o b ô b a -m e R E F L -a rm -s k in -h a ir-ta k e off-2sg ‘You take off the hair o f the skin o f the arm .’ 3.3 Syntax Clausal constituent order tends to be free. There is no verb serialization. G uahibo languages have som e verb com pounding (K err 1995: 171). A reduplicated main verb plus pona ‘go’ has an intensive m eaning ( 1995: 215). 13 Other sm all fam ilies and isolates 377 G ualiibo languages m ark relative and com plem ent clauses by nom inalizations (1995:119). Cuiba also h as full relative clauses m arked by a relative pronoun (1995: 247). G uayabero has a relative clause marker w hich goes on the com m on argum ent within the relative clause if it is also in subject function in the main clause. G uahibo and A ndoké have a large num ber o f suffixes which m ark the p rrfica te o f a subor­ dinate clause. Subordinating suffixes often indicate the tem poral relationship between main and subordinate clause (simultaneous, preceding o r following). These languages do not have switch-reference constructions. 4 VENEZUELA "In Venezuela there are about 38 indigenous languages spoken by around 60,000 people. Languages from the larger families are treated elsewhere in this volume C arib in chapter 2, and A raw ak in chapter 3. Lingua G eral A mazônica, o f TupiG u aran i origin, is spoken in the south. There are also three isolates - see table 13.1. The Yaruro, W arao and H oti are basically m onolingual nom adic hunters and gath­ erers. However, according to M itrani (1988: 164) and Heinen (1988: 611), the Y aruro and W arao have for some tim e practised slash-and-burn agriculture and in the 1930s the W arao com m enced rice cultivation. There are a few papers on Yaruro (Obregôn M iiñoz 1981, and references therein). For W arao, there are a num ber o f papers by O sborn (1966a,b, 1967), by RomeroFigeroa (1986a,b), and a short gram m ar by Romero-Figueroa (1997). There are alm ost no linguistic m aterials on H oti (only a w ord list in M attéi-M üller, Reid and Henley, 1994). 4.1 Phonology Y aruro has, a large phonem ic s y ste m -s e e tables 13-18 and 13.19. (Obregôn M uñoz 1981). It is unusual for the region in having a velar fricative (as well as a glottal fric­ ative) and a velar nasal, in having voice distinction for all the stops, and in having some fricatives. The consonant system in W arao is much simpler, as set out in table 13.20 (Rom ero-Figeroa 1997: 105). There are no voice distinctions in stops or fricatives {[d] is an allophone o f [r] in word-initial position). The only unusual phonem e is a labialized velar stop. There are five vowels [i, e, a, «, o), all with nasal counterparts (O sborn 1966a: 109). There is no contrastive length. Syllable structure is (C)V. Stress is on the penultim ate syllable o f a word. 378 Alexandra Y. Aikhenvald and R. M. W D ixon Table 0 .1 8 Yaruro consonan is voiceless stop voiced stop voiccless affrioate voiced alïricate voiceless fricative voiced fricative nasal flap lateral glide bilabial apical palatal P b t d ts dz s 6 n r 1 0 f V m velar glottal k ? g J 3 Ji w h X 0 y T able 13.19 Yaruro vowels front mid back i e * a i a 0 u 0 a T a b le 13.20 Warao consonants bilabial voiceless stop labialized stop voiceless fricative i^asai t1ap glide 4 .2 alveolar P t m s n palatal velar glottal k k" h r w y Features o f the grammar Warao is head-m arking, mildly polysynthetic and nominative-accusative. Its basic constituent o rd er is SV, OAV. There is no clear-cut distinction between nouns and adjectives. It is predom inantly suffixing, with few prefixes. D irect object crossreferencing m arkers are prefixed to the verb, while subject cross-referencing markers, plus tense-aspect, m odality and directional morphemes, are suffixes. 13 Other sm all fam ilies and isolates 379 There is a special verbal suffix indicating singular o r plural num ber o f the subject (S or A) argum ent. Possessive prefixes on nouns an d object cross-referencing pre­ fixes on verbs are alm ost identical. There are four oblique cases (dative, locative, allative and ablative). Tliere are also a num ber o f postpositions {causal, comltative, instrum ental). There are no genders o r classifiers. W arao has an agentive passive marked with completive past tense and perfective aspect on the verb; the dem oted A may be introduced with the agentive postposition; com pare the active clause in (43) w ith the corresponding passive in (44) (R om ero-Figueroa 1997: 93). (43) rihaw aratum a^ [ka-ina kokotukajQ non-a-e a n c e s to rs I p l.p o s s v - l a n d a ll m a k e -P U N C T U A L -P A S T ‘[Our] (44) a n c e s to r s .m a d e a ll in o u r l a n d . ’ [ka-ina kokotuka]j [rihaw aratum a aisialg^, Ipl.possv-land all ancestors AG nona-i-ha m a k e - C 0 M P i .- P E R FV ‘All in our land was m ade by [our] ancestors.’ Causative is marked with a prefix (e-) on transitive and intransitive verbs. When a transitive verb is causativized, the underlying A o f the causativized verb becomes the O o f the new causative, while the underlying O becomes oblique (e.g. acquires dative m arking) (R om ero-Figueroa 1997: 94). (45) w a h a b u -m a b a re -tira ^ ka e -n a h o ro -a -e v e n is o n - D A T iv E f a th e r - F E M A L B Ipl.O C A U S -eat-P U N C T U A L -P A S T ‘The nuns m ade us eat v e n is o n .’ There is no verb serialization; an auxiliary verb is used to express certain m ood meanings (e.g. prohibitive, potential and negative potential, and interrogative). Subordination is m arked with suffixes on the verb and with particles. W arao has an elaborate system o f speech styles: ‘council style’, ‘legend style’ and usual style. Besides lexical ditTercnces, the ‘council style’ dilTers From o th er styles in more frequent use o f some m orphological characteristics, e.g. suffixal pronouns and m orphological causatives o f ditransitive verbs. BIBLIOGRAPHY Allin, T. R. 197.“). ‘A gram m ar of Resigaro’, Ph.D. dissertation. University o f St Andrews. Aschmann. R. P. 1993. P ivlo Witotoan. SIL and University o f Texas at Arlington. Publications in Linguistics, 114. Dallas: SIL and University of Texas at Arlington. 380 Alexandra Y. Äikhenvald and R. M. W. Dixon Becker-Donner, E. 1955. ‘Nottzeti über einige Stänwne an dcn rechten Zuflüssen des Rio Gv& çon', ArchivJ'fír Välkerktmde, 10.275-343. Borgman, D. M. 1974. ‘Deep and surface case in Sanuma’, Linguistics, 132,5-18. 1990. ‘Sanum a’, pp. 17-248 o f Handbook o f Amazonian languages, vol. II, ed. D. C. Derbyshire and G. K. Pullutn. Berlin: M outon de Gruyter. Borgman, D. M. and Cue, S. L. 1963. 'Sentenœ and clause types in Central Waica (Shiriana)’, UAL, 29.222-9. Borgman, D. M., Cue S. L., Albright, S., Seeley, M. and Grimes. J. 19C5. ‘The Waioan lan­ guages’, Anthropological Linguistics, 7.1-4. Brinton, D. G. 1891. The American race. New York: Hodges Publisher. 1892. ‘Studies in South American native languages', Proceedings o f the American Philosophical Society, 30.45-105. Camp, E. 1982. ‘Referentes de movimiento y ubicaciôn en el discurso narrative en Cavineña’, Revista Latinoamericana de Estudios EtnolinglHsticos, 2.81-122. 1985. ‘Split crgativity in Cavineña’, UA L, 51.38-58. Camp, E. and Liccardi, M. 1965. ‘ítonam a’, pp. 223-383 o f Gramáticas estructurales de lenguas bolivianas, vol. Ill, ed. E. M atteson. Riberatta: Instituto Lingiiistico de Verano. 1967. Itonama: Castellano e Inglé.^, Vocabulários Bolivianos, 6. Riberalta: Institute Lingüístico de Verano. 1989. Diccionmio Cavineiki-Castellana CasteUano-Cavineña con bosquejo de la gramatica Cavinelia. Dallas, Tex.: SIL. Carlson, H. 1986, ‘The Aikana phonology and m orphology’. Ms. Chagnon, N. 1992. Yaiwmamö. 4th edition. Fort Worth, Philadelphia: Harcourt Brace College Publishers. Chamberlain, A. E 1910. ‘Sur quelques familles linguistiques peu connues ou presque incon­ nues de l’Amérique du Sud’, Journal de la Société des Américanisles de Faris, n. s., 7,179^202. 1913. ‘Linguistic stocks of South American Indians with distribution m ap’, American Anthropologist, 15.236-47. Créqui-M ontfort, G. de and Rivet, R 1913. ‘Linguistique bolivienne. La fiimille linguistique capacura.’ Journal de la Société des Américanisles de Paris, n. s., 10.119- 71. 1921-3. ‘La famille linguistique Takana’, Journal de la Société des Américanistes de Paris, n.s., 13.91-102,281-309; 14.141-82; 15.121-67. Curnow, T. }. 1997. ‘A gram m ar o f Awa Pit (Cuaiquer): an indigenous language o f south­ western Colom bia’, Ph.D. thesis, Australian National University. d'Orbigny, A. D. 1839. ‘Nation C hapacura’, pp. 288-90 o f L'homme américain de l’Amérique méridionale, considéré sous ses rapports physiologiques et moraux. 2 vols. Paris. Estrada. H. R. 1996. La lengua Sâliha. Safjtafé de Bogota: Colcultura. Everett, D. 1982. ‘Phonetic rarities in P ira h i’, Journal o f the International Phonetic Association, 12.94-6. 1986. ‘Piraha’, pp. 200-325 o f Handbook o f Amazonian languages, vol. 1, ed. D. C. Derbyshire and G. K. Pullum. Berlin: M outon de Gruyter. Everett, D. L. and Kern, B. 1997. Wari’: the Pacaas Noms language o f Western Brazil. London: Routledge. Girard, V. 1971. Proto-Takanan phonology. University of California Publications in Linguistics, 70. Berkeley and Los Angeles: University o f California Press. Gomez, G. G. 1990. ‘The Shiriana dialect o f Yanam (Northern Brazil)', Ph.D. thesis, Columbia University. Grimes, B. F. (ed.) 1988. Ethnologue: languages o f the world, l lt h edition. Dallas: WyclitTe Bible Translators, Inc. 13 Other small fam ilies and isolates 381 GrinevaJd, C. 1996. ‘Noun classification’, paper presented at Third Australian Linguistic Institute, Canberra, July 1996. Guirardello, R. 1992. 'Aspectos d a morfossintaxe d a lingua Trumai (isolada) e de seu sistema de m arcaçâo de caso’, M A thesis, Universidade Estadual de Campinas. forthcoming. ‘Trumai reference gram m ar’, Ph.D. thesis. Rice University, Houston, Texa& Hanke, W. 1956. ‘Beobachtungen über den Stamm der H uari (Rio Corum biara), Brasilien’, Archiv far Völkerkunde, 11.67-82. Heinen, H. D. 1988. ‘Los W arao’, pp. 585-692 of Los Aborigenes de Venezuela, vol. II. Caracas; Fundacién La Salle. Heinrichs, A. L, 1964. ‘Os foneraas do M ura-Pirahä’, Boletim do Museu Paramse Emilio GoeMi, 2Ï.X-9. Hinton, L. (ed.) 1993. (ms.) ‘A ikana modules: a class report based on field notes o f Harvey Carlson’. Berkeley: University of California. Huber, R. Q. and Reed, R. B. 1992. Comparative vocabulary: selected words in indigenous lan­ guages o f Colombia. Santafé de Bogota; Asociaciôn Institute Lingüîstico de Verano. Ibarra Grasso, D. E. 19S2. Lenguas indigenas de Bolivia. Cochabamba: Universidad Mayor de San Simon. Judy, R. and Judy, J. E. 1962. M ovim ay Castellano, Vocabulários Bolivianos, L Cochabamba, Bolivia: Instituto Lingüîstico de Verano, 1965. ‘M ovima’, pp. 131-222 o f Gramáticas estructmales de lenguas bolivianos, vol. Ill, ed. E. M atteson. Riberalta; Instituto Lingüîstico de Verano. Kerr, I. 1995. Gramatica pedagogics del cuiba-wámonae. Bogotá: Asociaciôn Instituto Lingüîstico de Verano. Key, H. 1961. ‘Phonotactics o f Cayuvava’, UAL, 27.143-50. 1967. Morphology o f Cayiivam. The Hague; M outon. Key, M. R. 1968. Comparative Tacanaphonology. The Hague: M outon. 1979. The grouping o f South American Indian languages. Tübingen: G unter N arr Verlag. Koch-Grlinberg, T. 1913. ‘Abschluss meiner Reise durch Nordbrasilien zuni Orinoco, mit besonderer Berucksichtigiing der von mir besuchten Indianerstämrae’, Zeitschrft fiir Ethnologie, 45.448-74, Kondo, V. and Kondo, R. 1967. ‘Ouahibo phonemes’, Phonemic Systems o f Colombian Languages. SIL Publications in Linguistics and Related Fields, 14.89-98, Ladefoged, P. and Everett, D. L. 1996. ‘The status o f phonetic rarities’, Language, 72.794-800. Landaburu, J. 1979. La langue des Andoke. Amazonie colombienne. Paris: Centre N ational de la Recherche Scientifique. 1994a. ‘Deux types de prédication, avec ou sans sujet: quelques illustrations colombiennes', pp. 639-63 o f I.andabuin 1994b. Landaburu, J. (ed.) 1994b, Estnitwas sintûclicas de la predicaciôn: lenguas Amerindias de Colombia. Bulletin de Vlmtitut français d'Etudes Andines, 23: 3. Liccardi, M. and Grimes, J, 1968. ‘Itonam a intonation and phonemes', Linguistics, 38.36-41, Lizot, J, 1988. ‘Los Yanomami’, pp. 479-584 o f Los aborigenes de Venezuela, vol. H. Caracas: Fundaciôn La Salle. ms. Nouveau dictionnaire de la langue yanomami. 1970. Diccionario Yanömami-Mspañol. Caracas; Universidad Central de Venezuela. 1996. Introduciân a la lengua yanomami. Caracas; Vicariato Apostôlico de Puerto Ayacucho. Loukotka, C. 1950. ‘La parenté des langues du bassin de la M adeira’, Lingua Posnaniensis, 2. 123^ . 1963. ‘Docum ents et vocabulaires inédits de langues et de dialectes sud-américains’, Journal de la Société des Américanistes, U.S. 52.1-M. 382 Alexandra Y. Aikhenvald and R M. W D ixon !968. Classification o f South American Indian languages. Los Angeles: Latin Arnerican. Center, UCLA. Martins, K. F. P. von. 1867. Ziir Sprachenkimde: Wortersammlung Brasilianischer Sprachen, Giossaria lingtiarim Brasiliensimn. Glossarios de diversns lingoas e dialeclos, que fallao os Indios do império de Brazil, Beiiräge ziir Ethnographie tind Sprachenkimde Sûdamerikas, zumalBrasiliens, vol. II. Leipzig; Friedrich Fleisclier. Mason, J. A. 1950. ‘The languages of South American Indians’, pp. 57-317 o f Handbook oj South American Indians, vol. VI, ed. J. H. Steward. Bureau o f American Ethnology, Bulletin 143. Washington, D C.; Smithsonian Institution. Mattéi-Müller, M.. Reid, H, and Henley, P. 1994. ‘Cultural and linguistic affinities of the for­ aging peoples o f Northern Amazooia; a new perspective’. Ms, Migliazza, E. C. 1965. t'onologia Mákii, Bolesim do Miiseu Paraense Emilio Goeldi, Antropologia, 25. 1966. Esboço sintätico de um corpus da lingua Máku, Boletim do Museu Paraense Emilio Goeldi, Antropologia, 32. 1972. ‘Yanomama grammar and intelligibility’, Ph.D. thesis, Indiana University. 1978a. ‘M akü, Sapé and Uruak languages; current status and basic lexicon’, Anihropological Linguistics, 20.133-40. 1978b. ‘Yanomama diglossia’, pp. 561-80 o f Approaches to language, ed. W. McCormack and S. Wurm. The Hague: M outon. 1985. ‘Languages of the O rinoco-A mazon region: current status’, pp. 17-139 o f South American Indian languages: retrospect and prospect, ed. H. M. Klein and L. Stark. Austin: University o f Texas Press. Migliazza, E. C. and Grimes, J. 1961. ‘Shiriana phonology’, Anihropological Linguistics, 3.31-4L M ilrani, P. 1988. ‘Los Fume (Yaruro)’, pp. 147-214 of Los Abon'genes de Veneiuela, vol. II. Caracas: Fundacion La Salle. Monod-Becquelin, A. 1975. Im pratique linguistique des Indiens Trumai. Paris: Centre National de la Recherche Scientifique. 1976. ‘Classes verbales et construction ergative en Trumai’, Atnerindia, !.l 17-43. Morse, N. L. and Frank, P. S. 1997. Lo nms importante es vivir enpaz: Los sálibas de los Llanos Orientales de Colombia. Santafé de Bogotá: Editorial Alberto Llevas Camargo. Nimuendaji'i, C. 1925. ‘As tribus do Alto M adeira’, Journal de la Société des Amiricanistes de Paris, n. s, 17.137-72. 1932. ‘W ortlisten aus Amazonien’, Journal de la Société des Américanistes de Paris, 24,93-119. 1948. ‘The M ura and Piraha’, pp. 255-69 o f Handbook o f South American Indians, vol. Ill, ed. J. H. Steward. Bureau o f American Ethnology, Bulletin 143. Washington, D .C: Smithsonian Institution. Nimiiendaju, C. and Valle-Bentes, E. H. do. 1923. ‘Documents sur quelques langues peu connues de l’Am azone’, Journal de la Société des Américanistes de Paris, n. s., 15.215-22. Obregôn Muñoz, H. 1981. ‘En torno al alfabeto de las lenguas indigenas de Venezuela a propôsito de la reducciôn del pumé a la escritura’, Revista Latinoamericana de Estudios Etnolingiiisticos, 1.25-46. Osborn, H. A. 1966a. ‘Warao I: phonology and morphophonemics’, UAL, 32.108-23. 1966b. ‘W arao II: nouns, relationals and demonstratives’, UAL, 32.253-61. 1967. ‘Warao: verbs and suffixes’, UAL, 33.46-64. Ottaviano, J. and Ottaviano, L fl. 1965. ‘Tacana’, pp. 3 0 9 ^1 7 o f Gramálicas estruclurales de lenguas holiviaitas, vol. III, ed E. M atteson. Riberalta: Instituto Lingiiistico de Verano. Pires N. N, 1992. ‘Estudo da gramâtica da lingua Jeoromitxi (Jabuti); aspectos sintâticos das clâusulas matrizes’, MA thesis, Universidade Estadual de Campinas. 13 Other sm all fam ilies and isolates 383 pitman. D. 1980. Bosquejo de la gramáíica Araona. N otas Lingüisticas, 9. Cochabamba: Instituto Lingüistico de Verano. Queixalos, F. forthcoming. Grammaire Sikimni ( Colombie), n.d, Diccionario Sikuani-Español. Lenguas aborigems de Colombia. Diccionarios. IJniversidad de los Andes, Santafé de Bogota, Colombia. Ramirez, H. 1992. IniciaçSo à lingua yanomami M anaus: Befvedere. 1993. Hapa tê p ê rë kuonowei - mitologia yanomami. M anaus: Inspetoria Saiesiana da Amazonia. 1994. ‘Le parler yanomam i des X am atauteri’, Ph.D. thesis, Université de Provence. Rivet, P. 1953. ‘La langue Masubi', Journal de la Sociétédes Américanistes de Paris, 42,1^9-25. Rodrigues, A. D. 1986. Lingiias brasileiras. Para o conhecimento das Unguas indigenas. Sao Paulo: Ediçoes Loyola. Rodrigues, I. and Oliveira, A. E. de. 1977. ‘Alguns aspectos da ergologia M ura-Piraha’, Boletim do Museu Paraense Emilio Goeldi. 65.1-47. Romero-Figueroa, A. 1986a. ‘Subjective and objective relative clauses in W arao’, Journal o f the Atlantic Provinces Linguistic Association, 8.94-111. 1986b. ‘W arao comparatives’. Studies in Language, 10.97-108. 1997. A reference grammar o f Warm, Lincom Studies in Native American Linguistics. Munich: Lincom Europa. Schmidt, W. 1926. Die Sprachfamilien iind Sprachenkreise der Erde. Heidelberg: Carl Winter. Schuller, R. 1933, ‘The language of the Tacana Indians (Bolivia)', Anthropos, 28.99-116, 463-84. Sheldon, S. N. 1974. ‘Some m orphophonem ic and tone perturbation rules in M ura-Pirahä', UAL, 40.279-82. 1988. ‘Os sufixos verbais M ùra-Pirahà’, pp. 147-75 o f Série Lingmstica, 9. vol. 2, ed. J. E, Grimes. Brasilia: SIL. Shoemaker, J. M. and Shoemaker, N. K. 1965. ‘Ese’ejja’, pp. 181- 306 o f Gramáticas eslructuraies de lengua.% bolirianas, vol. Ill, ed. E. M atteson. Riberalta: Instituto Lingüistico de Verano. Sleinen, fC. von den. 1886. Durcix zentral-Brosilien, Expedition zur Erforschung des Schingu im Jahre 1884. Leipzig: F. A. Brockhaus. Suarez, M. M. 1977. La lengiia Salim. Caracas: Centro de Lenguas Indigenas, Instituto de Invesîigaciones Histôricas. Universidad Catôlica Andrés Bello. ' Tobar O rtiz, N. 1994. ‘Relaciôn y determinaciôn en el predicado guayabero’, pp. 509-36 of Landaburu 1994b. Tovar, A. and De Tovar, C. L. 1984. Catâlogo de las lenguas de America del Sur M adrid: Editorial Gredos. Tracy, F. V. 1966, ‘The phonology and outline gram m ar o f the Aikamtheli dialect o f Shiriana. With notes on other dialects', MA thesis. University o f Pennsylvania. Vasconcelos, L 1993a. ‘M arcador de transitividade em A ikana’, paper at W orkshop on G ram m atical Categories, Campinas, Universidade Estadual de Campinas. 1993b. ‘The Aikana phonology: a preliminary analysis’. Ms. 1996. ‘Algumas considetaçôes sobre a morfologia aikana’, Boletim da Associaçâo Brasileira de LingOistica, 19.71-8. Voort, H. van der. 1997a. ‘Enkele problemen in de analyse van het Kwaza (Koaiá)’, a seminar at the Institute for Functional Research into Language and Language Use, Amsterdam. 1997b, ,‘Alguns aspectos da Lingua Koaiá, presentemente denominada Kwaza’, Boletim da Associaçâo Brasileira de IJngüistica , 20.35-54. Weir, E, M. H. 1990. ‘Incorporation in N adëb’, pp. 321-66 o f Amazonian linguistics. Studies in l.owland South American Indian languages, ed. Doris L. Payne. Austin: University o f Texas Press. Map 13 Languages in the Içaiia-Vaupés Basin witli approximate locations 14 Areal diffusion and language contact in the Içana-Vaupés basin, north-west Amazonia A L E X A N D R A Y. A I K H E N V A L D I AREAL DÍFFU SION IN NO RTH -W EST AM A ZON IA ' To define a linguistic area, a convincing num ber o f diagnostic traits m ust be estab­ lished which will distinguish this area from others, and which can be proved not to be genetically inherited. D istinguishing areal diffusion from genetically inherited phenom ena is particularly im portant for A m azonian languages - see chapter 1 above. The region around the U pper Rio Negro (the m ajor northern tributary o f the A m azon) contains a multiplicity o f languages. W ithin the U pper Rio N egro area, the com bined Vaupés and Içana river basins constitute a well-defined linguistic area; and within this the Vaupés basin is a clear linguistic sub-area (A ikhenvald 1996a). I shall here consider the V aupés-Içana basin in north-w est A m azonia as an example o f a linguistic area, which displays a convincing num ber o f structural fea­ tures shared by languages o f three genetically unrelated families - N orth Arawak, East T ucano and M akii. These features are not found in Arawak and M aku lan­ guages spoken outside the area, and thus can be considered as diagnostic for areal diffusion. In some cases we are able to establish the direction o f diffusion. T he languages spoken in the Içana and Vaupés region are described in §2, together with the cultural setting and language attitudes. In this section I also present historical evidence and argum ents in favour o f shared cultural patterns. Linguistic characteristics o f the Vaupés - due to the intensive areal convergence in the context o f obligatory m ultilingualism - are described in §3. Properties shared by the languages o f the Içana-V aupés area are discussed in §4. T he history ' I am very grateftil to Janet Barnes and Terry M alone for their comments. Alexandra Y. A ikhenm ld 386 o f linguistic contacts and m igrations in the Içana-Vaupés region is sum m arized in §5. 2 LIN G U IST IC SITUATION 2,1 Languages spoken IN THE IÇANÂ-VAUPÉS BASIN T he Içana and Vaupés basins and adjacent regions, within the U pper Rio Negro basin (see m ap 13), contain the following genetic groups o f languages: (a) N o rth A raw ak languages spoken on the Içana and its tributary, the Aiari (Baniwa/K urripako), on the Vaupés (Tariana) and on the Upper R io N egro (W arekena and, formerly. Bare); (b) several languages of the East Tucano family spoken on the Vaupés: Tucano, Tuyuca/Yuruti,^ G uanano/Piratapuya, Desano/Siriano, C arapana/T atuyo, M acuna, Barasano/Taiwano, W aimaja/Bará; and one C entral Tucano language, Cubeo, spoken both on the Q uerari river (a tributary o f the Vaupés which constitutes the northern border o f the Tucano area) and the U pper A iari (see chapter 7); (c) three M akii languages: Dâw, H upda-Y uhup and K akua-N ukak (see chapter 9). There is n o p ro o f o f genetic relationship between any o f these language groups. The N orthern branch o f A rawak on the Vaupés is represented here by T ariana. A lthough today over 1,500 people identify as Tariana (Rodrigues 1986), the lan­ guage is spoken by only about 100 people (none o f them children). O ther N o rth Arawak languages currently spoken in the U pper Rio Negro area are: (i) dialects o f Baniwa o f Içana (also known as Kurripako); spoken by 3,000-4,000 people on the Içana and its tributaries and in the adjacent regions o f Colom bia and Venezuela (see the lists o f dialects in N irauendajii 1950/5, Rodrigues 1986). (ii) W arekena, a dialect o f Baniwa o f O uainia (mainly spoken in Venezuela); spoken by a few dozen old people on the Xié river (see Aikhenvald 1998); (iii) Old W arekena (also called Warena, o r G uarequena); spoken by a few ^ Pairs o f names separated with I are mutually intelligible and can be considered dialects. Terry M alone (p.c.) gives the following percentages of shared lexicon (based on the list of over 300 lexical items found in Huber and Reed 1992): Tuyuca and Yuruti. 91%; Guanano and Piratapuya, 94%; Desano and Siriano. S9%: Carapana and Tatuyo, 96%. 14 Içana- Vaupés linguistic area 387 Table 14,1 The region o f Brazilian Vaupés: languages and their speakers people language speakers Tucano Piratapuya G uanano Dcsano Cubeo Tuyuca Tariana 4,500 4,500 1,232 200 1,000 200 1.800 150 3,000 300 723 700 1,500 100 old people on the Xié river, and by a few hundred people in Venezuela where they moved early this century (Aikhenvald 1998); (iv) Bare; once the m ost im portant language along the U pper Rio Negro, b u t now alm ost extinct (see Aikhenvald 1995a).-* The N o rth Arawak languages o f the Içana and Vaupés and the adjacent regions o f the U pper Rio Negro basin fall into three distinct genetic groups; Warekena, Bare a n d Baniwa-Tariana. E ast Tucano languages are closely related, but n o t mutually intelligible. The Tucano live along the Paca, Papuri, Vaupés and Tiquie rivers. In the Brazilian Vaupés, Tucano is gradually gaining ground as a lingua franca o f the area. Table 14.1, based on G rimes (1988), gives an idea o f the num bers o f those who identify themselves as belonging to E ast Tucano groups and to Tariana, and o f those who actually speak the language (see chapter 7). Sem i-nom adic M akii (the ‘untouchables’ o f the region) display a num ber of cul­ tural divergencies from East Tucanos and Tariana - see §2.2. There is one M akii language, N adëb, spoken by a few hundred people outside the Içana-Vaupés basin (see chapter 9 and Weir 1984). It does not belong to the Vaupés geographical and linguistic area. The data on N adëb are crucial for distinguishing between areal and genetic features in the M akii languages o f the U pper Rio Negro area. Lingua francas spoken in the region are Lingua G eral, or N heengatii, and Tucano. Lingua G eral is spoken in the whole region o f the U pper Rio Negro (see Rodrigues 1986, Bessa Freire 1983, M oore, Facundes and Pires 1994), On the Vaupés, it is only understood by older people. Lingua G eral is a creolized version o f Tiipinam bá (Tupi-G uarani family) which was spread from the east coast of Brazil by white m erchants and missionaries. It was the lingua franca o f the whole A m azon region from the late seventeenth century up to the m iddle of the nineteenth century, and the influence o f it can still be seen in a few loan words in Tariana and ^ A n u m b e r o t h e r Arawak languages of the U pper and Middle Rio Negro region became extinct during the nineteenth and twentieth centuries, e.g. M anao, Wiriiia, Wainuma, Yumana, Passe, Yabaana, M andawaka (Mawaca), Am arizana (Koch-Griinberg 1911), Bahwana. See map 2 and table 3.1 in chapter 3. 388 Alexandra Y. Aikhenm ld other languages o f the Vaupés. It was gradually replaced by Tucano as a lingua franca from the early twentieth century, as a result of the language policy o f Catholic missionaries and the civil authorities. Indo-European languages are represented by Portuguese and Spanish. They are also gaining ground as lingua francas o f the region, especially am ong younger people. The Y anomami - semi-nomadic hunters and gatherers who live in the jungle around the m inor tributaries o f the U pper R io Negro and sometimes travel as far westwards as the Vaupés region - are not culturally integrated with other peoples of the U pper R io Negro, and do not form a part o f the Içana-Vaupés linguistic area (see m ap 12 in chapter 13). They rarely interact w ith other tribes, being feared and avoided. Tlieir status can be com pared to that o f Gypsies in the Balkans, who, although present in the region, do n o t participate in linguistic and cultural exchange. 2.2 Cultural setting and language attitudes M ultilingualism used to be - and to a great extent still is - a cultural norm over the whole Vaupés region. A m ong East Tucano peoples and Tariana, m arriage is exogamous. There are strict m arriage rules, which are governed by language affiliation (see Sorensen 1967; A ikhenvald 1996a). M arrying someone who belongs to the same language group is considered akin to incest. Jackson (1974: 62). referring to the linguistic attitudes o f the Bará, reports being told: ‘My brothers are those who share a language with me’, and ‘We don’t m arry our sisters.’ Consequently language - which is acquired through patrilineal descent - is a badge of identity, together with the patrilineal descent. A n Indian always identifies w ith their father’s tribe and lan­ guage. The main unit, both socially and linguistically, is a speech com m unity identified by one’s father’s language."* Each settlem ent - w hether longhouse o r village - is multilingual because o f these exogamous m arriage rules. An individual generally knows between three and ten other languages o f the region, including his m other’s language which would often also be his wife’s language, and in addition Portuguese and/or Spanish. Since language identity is a badge o f ethnical identity, languages even the closest ones - are kept strictly a p a rt (see Sorensen 1972: 82). This creates a ■' In most cases a spcech community is identical to a phratry, the members o f which can inter­ m arry However, ‘the Cubeo, Piratapuya and M acuna are exceptions, in that each language represents more than one phtatry, so that, for example, Cubeos can m arry Cubeos if the marriage partner is a member of an exogamous phratry’ (Malone 1987: 27), See also Grimes (1985). Each phratry is divided into endogamous sibs. There are minor linguistic differences between sibs (Terry Malone, Janet Barnes, p.c.). 14 Içana- Vaupés linguistic area 389 very strong im pedim ent to lexical loans, in contrast to other multilingual situations, e.g. the village o f K upwar (G um perz and W ilson 1971: 161-2), or m ultilingual Jewish com m unities (Weinreich 1964). This unusual linguistic diversity is accom panied by a lesser degree o f cultural diversity. T he cultural homegeneity is corroborated by (a) shared m eans o f subsis­ tence, food, and ways of life; (b) shared stories and rites, and traditional social struc­ ture; wfith a minimal division o f ‘labour’ between tribes (see G alväo 1979 for the differences in pottery and basket-weaving techniques between Tucano- and A raw ak-speaking groups). However, each group - a nd apparently each subtribe do have their own versions o f origin myths. The traditionally nom adic M akii peoples are not included in the m arriage network; however, they are integrated into the econom ic exchange system, their main functions being as th e traditional producers o f arrow poison, and as skilful hunters. They have no agriculture, and depend on o th er peoples o f the region for the staple food, manioc. They are considered inferior by East Tucano and Tariana and are denom inated ‘underlings’ (see M artins 1994, Silverwood-Cope 1990, and chapter 9). They do not interm arry with either E ast Tucano or Tariana. However, they share a num ber of linguistic characteristics with oth er languages of the Vaupés region. They also share some im portant m yths w ith the Tucano and the Tariana. Unlike the Tucano and the Tariana, they exhibit no inhibition against lexical loans. As will be shown below, there is a large degree o f gram m atical and phonological convergence am ong the indigenous languages o f the Vaupés. The im pact o f these convergence phenom ena is espedally instructive in the case o f Tariana, since they are discernible as the result o f ‘foreign’ influence. T he main difference between the Içana basin and the Vaupés basin is the lack of com pulsory linguistic exogamy in the Içana. Baniwa cohabit with C entral-Tucanospeaking C ubeo on the Aiari river, a tributary o f the Içana, and on the Q uerari, a trib u tary o f the Vaupés which borders the Içana basin. T he Cubeo and the Baniwa are in co n stan t contact which results in diffusion o f cultural and linguistic features. M arriage tends to be exogamous between different descent groups and dialectal groups o f the Cubeo and the Baniwa. T he C ubeo have assimilated Baniwa dances and chants (see G oldm an 1979, G om ez-Im bert 1996). Some o f the Cubeo-speaking groups may originally have been Baniwa.* Nowadays there are no - or alm ost no - regular contacts between the East Tucano o f the Vaupés an d the T ariana, on the one hand, and the Baniwa on the other. * W ithin a larger language gruiip, all phratries are ranked, and there is a ranking o f sibs within a phatry. There is a tendency for lower-ranked sibs to be those whom one would suspect were originally speakers o f non-Tucano languages, in m ost cases Maku (see §1 of chapter 9 in this volume; Janet Barnes and Terry Malone, p.c.). Alexandra Y. Aikhenvald 390 However, the Baniwa share a num ber o f cultural siinilarities w ith the indigenous groups of the Vaupés which may be indicative of older contacts and of diffusion. These similarities include the structure o f the kinship system (described in Oliveira 1975, G alv io 1979), patrilineal descent, and the hierarchical organization o f descent groups. There are also a number o f traditions and myths shared by the Baniwa and the Tucano, e.g. the myth o f a woman going to the ‘beyond’ with the help of an agouti (C. Hugh-Jones 1979); stories about the cunning turtle and the tapir; the story o f the evil spirit’s son who turned into a deer; the cult o f the magic flute, Jurupary (S. O. Hugh-Jones 1979); burial o f the dead inside the house; infanticide o f twins; and the festival of Dabukuri - an offering accompanied by drinking manioc beer and ritual dances (cf Reichel-DoImatolT 1986). There are also a num ber of linguistic fea­ tures shared by Baniwa, Tariana and East Tucano languages but not found in other N o rth Arawak languages. These will be considered in §2.4. 2.3 Historical evidence H istorical and traditional evidence shows that neither the Tucano nor the Tariana constitute the autochthonous population o f the Vaupés, A ccording to N imuendajii (1982), the original inhabitants o f the Vaupés area were M akii tribes, and East Tucano tribes then entered the area from the west. The Tariana are the m ost recent arrivals. They probably came into the Vaupés basin from a tributary of the Içana river - probably the A iari (around the end of the six­ teenth century according to Briizzi 1977, fCoch-Griinberg 1911 and Nimuendajii 1982). This description o f the origin o f the Tariana is reflected in their myths, according to which all o f them originated from a ‘hole’ at the waterfall o f Apui, on the Aiari river, where at the time they lived together with the Baniwa, and with the Cubeo. N um erous stories about the wars between the Tariana and the G uanano, and between the Tariana and the Desano, provide rich ethnohistorical evidence for the Tariana invasion o f the Vaupés (see Briizzi 1977). The Desano, as the inhabitants o f the main rivers, were am ong the first to have suffered from the invasion o f the Tariana. I was told by my teachers o f Tariana that, as the result of the D esanos’ defeat, they were designated the ‘younger brothers’ o f the Tariana, and this is why the Tariana do not m arry them. We can reconstruct the following historical scenario for the linguistic situation in the Vaupés during the past 500 years. (i) Before 1500; East Tucano tribes moved from the west into the Vaupés area, which was previously inhabited by M akù tribes (cf chapter 9). T he East Tucano established dom inance over the M akù. The spread o f Tucano-speaking peoples to 14 Içana-Vaupés linguistic area 391 the Içana basin may have occurred at about the sam e time. A ccording to the oral tradition o f the C ubeo (G om ez-Im bert 1996), they cam e to. the Q uerari from the Içana tributaries a few hundred years ago, chased by B aniwa-speaking tribes. Some o f the C ubeo speakers on the Q uerari are descendants o f A raw ak-speaking tribes. It is hard to establish the exact age o f Baniw a-C ubeo contacts in the Içana basin; however, it was sufficiently long ago to prom ote the areal diffusion discussed below. T h e Baniwa o f Içana were at this time living together with the Tariana in the Içana basin. (ii) A round 1600: the Tariana moved from the tributaries o f the Içana r iv e r-th u s splitting from the Baniwa o f Içana - to the Vaupés region, in which Tucano tribes were already established. This m arked the beginning o f contact between East T ucano and Tariana. (ill) A round 1750-80: the first contacts with Portuguese took place, which started the spread o f Lingua G eral as a lingua franca. (iv) A round 1900; Tucano started to gain ground as a lingua franca o f the area, with some Tariana settlem ents beginning to use mainly Tucano (see KochG riinberg 1911). This tendency increased with the establishm ent o f perm anent Salesian missions in the Vaupés in 1925, and resulted in the growing endangerm ent o f indigenous languages o th er than Tucano in the Vaupés region, and the growing obsolescence o f Lingua Geral. T h e main consequence o f the spread o f Tucano in the Brazilian Vaupés is the gradual underm ining o f the identification between language and tribe. Language has gradually ceased to be an emblem o f tribal identity, and the m ajority o f lan­ guages other than Tucano have become endangered. T he discrepancy between the num ber o f those who belong to a tribe and those who actually speak the language (see table 14.1) is particularly m arked in the case o f Tariana. The spread o f Tucano is also leading to the gradual disappearance o f one o f the m ost fascinating m ulti­ lingual areas of the world, and the areal phenom ena associated w ith it. This does not seem to be the case in the C olom bian Vaupés. Tucano, however, is also the prestige language there; when the Tucano are present, speakers o f other languages use Tucano. T he language which seems to be disrupting the Tucano socioiinguistic situation there is Spanish (A rdila 1989; Terry M alone p.c.). 3 THH VAUPÉS REGION AS A LIN G U IST IC AREA 3.1 General observations H ere we consider the properties o f languages spoken in the Vaupés region which characterize it as a linguistic area. Alexandra Y. Aikhenvald 392 As pointed out by Thom ason and K aufm an (1988: 96ff.). a long-term m ultilat­ eral Sprachbund seems to prom ote gradual isom orphism in all aspects o f the struc­ ture o f languages except for the phonological shape o f morphemes. A famous example of the sam e surface structure for several languages spoken in the same lin­ guistic area com es from Indo-Aryan and D ravidian languages o f Kupwar (G um perz and Wilson 1971). In the case o f numerous multilateral linguistic areas, such as the Balkans (Thom ason and K aufm an 1988:97), the direction o f diffusion is easily established. In the case o f the Vaupés, it can also be established for certain features. The following characteristics o f the Vaupés as a linguistic area should be borne in mind: • the fact that there are several E ast Tucano languages and only one A raw ak language has provoked, by and large, unilateral diffusion pat­ terns: from East Tucano into the A rawak language, Tariana;* • there is a recent tendency of Tucano to develop dom ination over other languages, hence the destruction o f multilingualism in the community; • language identification and the perceived link between tribe and lan­ • while considering areal diffusion patterns and structural convergence guage drastically limit the extent o f lexical borrowing; between the languages o f the area, one has to take into consideration language obsolescence phenom ena for some languages. The Vaupés linguistic area (for some useful com m ents see Briizzi 1967,1977) has multilateral multilingualism characterized by linguistic diversity and a comparative lack o f cultural diversity. As already irientioned, the languages belong to three genetically unrelated families - N o rth A raw ak, Tucano and M akù - and their speakers do n o t accept language-mixing. However, people tend to identify the lan­ guage with its forms and the prohibition on ‘language mixing’ only relates to forms, not to gram matical categories o r structural patterns. As a result there is a great degree of diffusion o f structural patterns but little borrow ing o f actual forms. The idea that the Vaupés region constitutes a linguistic area was first suggested * There are a few examples o f Arawak influence on East Tucano languages. Carapana (Metzger 1981: 151 ff.) and Tatiiyo (Elsa Gom ez-lmbert p.c.) have an atlributive prefix kawhich displays striking similarity to the relative-attributive prefix kii- in Tariana and Baniwa of Içana. The ‘foreigniiess’ o f this prefix in C arapana and Tatuyo is confirmed by the generally suffixing character o f Tucano languages. Tanimuca-Retuama (Retuarä), a West Tucano language spoken in Colombia, apparently lost its complicated system of clas­ sifiers, switch-referenee markers and evidential oo the verb, under the areal influence of another N orth Arawak language, Yuciina, which has only a gender opposition (Gomezlm bert 1996). I( also acquired prefixes, possibly under Arawak influence. According to Terry Malone (p.c.), the speakers o f Tanimuca could originally have been a North Arawak group who switched to a Tucano language (see also chapter 7). 14 Içana-Vaupés linguistic area 393 by Sorensen (1967), with reference to the C olom bian side, where only E ast Tucano languages are spoken. Sorensen pointed o u t the existence o f a bundle o f m orphosyntactical isoglosses which constitute an East T ucano ‘profile’ o f language (1972: 82-3). However, whether this East Tucano profile is due to areal diffusion patterns or to the com m on genetic origin o f East Tucano languages remains a problem which goes beyond the scope o f the present discussion. (The solution o f this problem w ould involve a full reconstruction o f proto-E ast-Tucano and com parison o f it w ith proto-W est-Tucano and proto-Tucano.) T he im portant difference between the Brazilian Vaupés region and the C olom bian Vaupés region from the point o f view o f areal diffusion phenom ena is the fact th at a non-East-Tucano language, Tariana, is still spoken in Brazil. Thus, the East-Tucano-like features in Tariana (and also, North-A rawak-like features in East Tucano) enable us to establish a direction o f diffusion in the area. For the analysis I will use the following kinds o f d ata: • my own field d ata on Tariana and on other N orth Arawak languages o f the wider region (Baniwa o f Içana, W arekena o f Xié, Bare); protoA raw ak reconstruction and internal reconstruction (see Payne 1991; A ikhenvald 1994a, b, 1995a, 1996a, b, in prep.); • a num ber o f descriptions o f East Tucano languages (not all o f equal quality), and my own field data, and also some data on proto-Tucano reconstruction and descriptions o f East Tucano and West Tucano lan­ guages which do not participate in the Brazilian Vaupés as a linguistic area, as well as the d ata on proto-W est-Tucano. The analysis of areal and inherited features o f Tariana includes a fair am ount o f internal reconstruction, and synchronic com parison with Baniwa o f Içana and East Tucano languages. A com parison o f Tariana w ith geographically close and geneti­ cally related N orth Arawak languages is extremely useful, especially if we are able to determ ine which properties Tariana shares with East Tucano languages, and which are shared by other N orth Arawak languages, b u t absent from Tariana (also see A ikhenvald 1996a, on innovations and language attrition phenom ena in Tariana). C o n tact between Tucano and M akii languages (Dâw, H upda, Yuhup) in the Vaupés appears to be much older than contact between Tucano and T ariana - see §2.3. N ote that one would expect mainly unilateral diffusion patterns from East Tucano to M akii, since a fair proportion of M akii know at least one East Tucano language, o r more, but East Tucano peoples seldom have com petence in a M akù languagk To establish areal diffusion patterns in the M akii languages spoken in the Vaupés 394 Alexandra Y. Aikhenvald and U pper Rio Negro regions, data from N adëb, a M akù language spoken in the M iddle Rio N egro area (based on M artins 1994; Weir 1984, 1990; see also chapter 9 above) are o f crucial importance. The materials available on M akii languages spoken in the Vaupés and U pper Rio Negro areas are not extensive. However, they reveal that, in m ost cases, areal diffusion is more superficial (com pared with the T ariana-T ucano interaction). This is what would be expected, since the M akù are accorded an inferior social status and are not fully integrated into the m ultilingual socio-cultural community. F urther work is needed to determine possible areal diffusion features in the M akù languages spoken in C olom bia {Kakua and Nukak). I shall now consider shared phonological characteristics o f the languages o f the Vaupés (§3.2), their gram m atical structure (§3.3), syntax and discourse techniques (§3.4) and semantics (§3.5). 3.2 Phonological characteristics A num ber o f phonological characteristics o f Tariana are shared with East Tucaiio languages, but not with the genetically related N orth Arawak languages from outside the Vaupés area. Some phonological characteristics o f M akù languages are similar to East Tucano languages, but differ from Nadëb, another language o f the same family. They can be e.xplained through areal diffusion from Tucano languages. The following characteristics are shared by Tucano, Tariana and M akù languages. 1 N asalization is a w ord-prosodic feature in Tariana and in H updaY uhup a nd K akua (see chapter 9). T he same phenom enon is found in all East Tucano languages, but n o t in other N orth Arawak languages o r in N adëb (see Aikhenvald 1996a).^ 2 Pitch accent and intonation patterns in Tariana and in the M akù lan­ guages display striking similarities to Tucano, hence the ‘Tucano’ accent in Tariana. Daw and H upda-Y uhup have two tones, and K akua has four (see chapter 9). Tones are absent from N adëb. Pitch accent is attested in Baniwa (see §4); however, unlike the languages of the Vaupés there is here no tendency tow ards developing a tone distinc­ tion. ’ The origin o f the direction o f nasal prosody (from left to right o r from right to left) remains problematic. Left to right is the pattern attested everywhere; however, Terry Malone (p.c.) reports that some traces of the opposite direction (right to left) can be found in Desano, Siriano, Barasano, and also in one West Tucano language, Koreguaje, Note that in Choco languages, spoken in the Andes, nasalization operates from left to right; this may be indic­ ative o f areal diffusion between Choco and Tucano (Malone p.c.). 14 Içana- Vaupés linguistic area 3 395 There are two phonem es with low functional load in Tariana which are also attested in Tucano languages and in all M akii languages, but are absent from other N orth Arawak languages. The vowel i appears in two m orphem es: -pi ‘augm entative’ and ihmeni ‘to m oan’; and o appears in the fem inine m arker on kinship terms: -Co (cf. Baniwa -Cu-, Tucano -Co) (where C stands for a consonant). The following features o f Tariana result from East Tucano influence. A In Tariana, inherited phonem es absent from Tucano - aspirated stops an d aspi­ rated glide - have a reduced functional load with respect to other phonem es and when com pared to the status o f the same phonem es in other N orth A raw ak lan­ guages (cf. Aikhenvald 1995a for Bare; Taylor 1990, Aikhenvald 1996b, for Baniwa).® They also tend to be in free variation with non-aspirated stops and glide respectively. There is a tendency to use just í instead o f s and tf; s and t f occur in all N o rth A raw ak languages, and Tucano languages have only i. Diflerentiation in vowel length, which is absent from Tucano, also has a low functional load in Tariana. There is significant variation between long and short vowels in Tariana, e.g., T ariana u:n i-uni ‘w ater’, Baniwa u:ni. As is typical for a language contact sit­ uation, Tariana is losing the phoneme.? not found in its neighbours. B Syllable patterns in Tariana are: (C)V, (C)Vh; the latter alternates with (C)hV. Unlike in oth er N orth A raw ak languages, h in an unstressed w ord-initial or m or­ phem e-initial syllable can occupy a postvocalic position, thus creating VC-like syl­ lables, e.g., e h k u á p i - h e k u á p i ‘day’ (cf Baniwa h e k w á p i ‘day’). There are several cases where a syllable CVh occurs word-niedially (independently o f stress pattern) in slow to norm al register. T he onset o f the following syllable m ust be a voiceless stop t o r k , e.g. - m h tá ‘to think, to reason’, -b u h tá ‘conditional’, r n a r á ta h k a ‘a kind o f wave’, k a r á h t a ‘lung’. In rapid speech h disappears: n m r á t a k a ‘a kind o f wave’, -b u tá ‘conditional’, etc. All these words have a N orth A raw ak origin. T he occurrence o f CVh syllables in Tariana resembles CVh syllables in Tucano. N ote that in East Tucano languages CVh and/or CV? are the only instances o f consonantal coda. C Tariana has several phonological processes which are not found elsewhere in N o rth A raw ak but are characteristic o f Tucano; • }’> ii contiguous to nasal vowels. In word-initial and word-medial position Baniwa d z and K urripako y correspond to Tariana il, if the following or the preceding syllable contains a nasal consonant, e.g. Tariana ñ a m a , Baniwa d z m n a , • Only O iianano has been analysed as having phonemic aspirated stops (Waltz and Waltz 1967). In other Tucano languages, there is a tendency to slightly aspirate voiceless velar sujps before non-front high vowels (Terry Malone p.c.). 396 Alexandra Y. Äikhenvald K urripako yam a ‘tw o’; Tariana -ñami, ia n iw a -diam i, K urripako -yami ‘to die, to be terminally ill'; Tariana pamima, Baniwa pamudziia, K urripako pam uya ‘middle’ (cf. also Tariana ñamu, W arekena o f Xié yanm-du ‘evil spirit’). T ariana y corresponds to Baniwa y in other positions, e.g., Tariana, Baniwa maliye ‘knife’. Tariana / corresponds to Baniwa, K urripako y in loan words from Lingua G eral, e.g. Tariana, Baniw ayaiana ‘a white person’; Tariana, Baniwa ya ia ki ‘caxiri, m anioc wine’. • y > d y word-initially, e.g., Tariana yu m ‘a kind o f m osquito’ realized as • d > r contiguous to a front vowel (optional), e.g., Tariana -pidena- [dyuru] in rapid speech. -pirena ‘rem ote past inferred’. The fact that these phonological processes are found in other S outh American lanr guages, as well as in languages o f other areas o f the world, does not go against their areal character in Tariana, since they are absent from other N orth A rawak languages; the same holds true for nasalization as a word-prosodic property, and for glottal stop. D T he glottal stop, absent from other N o rth Arawak languages, sporadically appears in Tariana word-finally (but is not contrastive). There is a glottal stop phonem e in Desano, Siriano, Tucano, G u anano and Piratapuya, but in Barasano and Tatuyo the glottal stop occurs as a sentence-final prosodic feature (not as a phonem e) (Jones and Jones 1991; Whisler and Wliisler 1976). 3.3 Grammatical structure 3.3.1 Typological profile and word structure East Tucano languages are head-m arking, with elements o f dependent-m arking. They are agglutinating and suffixing. Arawak languages are head-m arking, and agglutinating with elements o f fusion. They are suffixing with the addition o f one prefix position in a word. N orth Arawak languages tend to have few suffix positions in verbs, Tariana is basically headmarking, with a few elements o f dependent-m arking, one prefix position, and a large num ber of suffix positions in verbs. The M akii languages o f the Vaupés com bine elements o f head-m arking and dependent-m arking. They are predom inantly suffixing, with very few relics o f a pre­ fixing morphology. In contrast, N adëb is predom inantly head m arking, with an unusually large num ber o f prefix positions for an A m azonian language. The almost com plete loss of prefixing m orphology in M akii languages other than N adëb can be considered the result o f areal diffusion from Tucano, with its predom inantly suffixing structure (see chapter 9 on Makii). 397 14 Içana-Vaupés linguistic area 3.3.2 N o m in a l m o rp h o lo g y ( i ) C a se-m a rk in g Core-case m arking on personal pronouns w ith an anim ate re fe re n t-a n d on nouns, dependent on their topicality - is the m ost striking areal property shared by East Tiicano languages, Tariana and M akii in the Vaupés. Generally, Arawak languages do not have case-m arking. Tariana has the follow­ ing case-m arking system which shows similarities to E ast Tucano languages (see Aikhenvald 1994b; on case-m arking in Tucano languages, see Kinch 1977, Gralow 1980),’ • Personal pronouns with an anim ate referent have an obligatory oppo­ sition between subject case (used to m ark A/S) and non-subject case, used to m ark O and dative argum ent. • O ther constituents (including third person pronouns with inanim ate referent) have an optional topic m arker (Tariana -nuku. East Tucano -re) which goes on any non-subject constituent provided it is the topic. In East Tucano, case m arking is obligatory with personal pronouns and highly desirable with personal names. N ote that the non-subject topic marker is optional in (1) but obligatory in (2)-(4). Firalapiiya (my fieldwork) (1) yei oha-gi-(re) Lenita-re Isg write-RELATIVE-(TOPICAL.NON.A/s) Lenita-TOP!CAL.NON.A/S ou -’u give-PAST T gave the pen to L enita.’ (2) yei Isg bt?»-re Lenita-re ou-’u yOU-TOnCAl..NON.A/S Lenita-TOPICAL.NON.A/S give-PAST ‘I gave you to Lenita’ (talking to a baby) T a r ia n a (3) nuha pa-dana-ni-na-nuku 1sg iMPERS-write-TOPic. a dv a n c in g - c l :vertical - topical .n o n .a/ s Lenita-nuku nu-a Lenita-TOPiCAL.NON.A/s Isg-give ‘I gave a pen (lit. a vertical writing device) to 1 N ote that in chapter 7 the case m arking system in Tucano is described as a systein o f 'spec­ ifiers’ relating to the correlation between the choice o f case and the discourse function o f a 398 Alexandra Y. Aikhenvald (4) nuha Isg pi-na Lenita-nuku 2 s g -0 Lenita-TOPiCAL.NON.A/s nu-a Isg-give *I gave you to Lenita’ (talking to a baby) The M akii language D âw has a m arker with similar function. An object is m arked with -îy’ if it is topical, and follows the V -c o m p a re (19) and (20) in chapter 9. T he m arking can be om itted when the order is AVO; it is obvious who did what to whom in (21) o f chapter 9: ‘A jaguar killed the deer’ - deer don’t kill jaguars. A technique for m arking A/S when in focus - or contrastive - is shared by Tariana, some East Tucano'® and some West Tucano languages, and Daw. In Tariana A/S is m arked if it is in focus, as in (5), o r is consequential, as in (6). The m arker in Tariana is hom ophonous with the instrum ental (see Wheeler 1987; 124, 127ff.). (N ote that in (3), the subject (A) is not in focus.) Taricma (5) nuha-ne pa-dana-ni-na-nuku 1-FOC.A/S IMPERS-write-TOPIC.ADVANCING-CL:VERTlCALtopical .no n .a/ s Lenita-nuku nu-a L e n ita - T O P iC A l..N O N ,A /s ‘I (a n d n o t so m eb o d y Is g - g iv e else) g a v e a p e n ( l i t . a v e r t i c a l w r i t i n g d e v i c e ) to L e n ita .’ (6) ka-wana ka-hña-ne pima-ka RELATfVF.-call RELATIVE-eat-FOC.A/S Cry-REC.PAST.VISUAL ‘The evil spirit (the one who yells and eats) cries’ (and this is om inous) In Dâw, A/S is m arked with an enclitic-td i if it is in conirastive focus and is agentive. See (22) in chapter 9 on M akù. N adëb has no core case marking of any sort, in agreement with its predom inantly head-m arking tendencies. ( ii) L o c a tiv e cases a n d p o stp o sitio n s The languages o f the Vaupés region tend to have one peripheral case-marker or post­ position, M ost East Tucano languages have one peripheral case-marker -/)#(Barasano -hi) 'locative/directional, elative’ which can co-occur with the definite/topical non­ subject marker -re if the corresponding constituent is topical." Tariana has a corre“ Tucano m arks the focused A/S with -pe, and Desano and Siriano with -gapi. " The equivalent form in Desano and Siriano is -ge. 14 Içana- Vattpés linguistic area 399 spending suffix -je; e.g., Tucano Bogota-pi-re, T ariana Bogota-se-nuku(^o%otk-\-.OCTO PICAL.NO N .A/S) ‘to/from/in Bogota’. Daw, H upda-Y uhup and K akua have just one locative marker (D -rid, K k -p ‘to, from, in’) (see chapter 9). N adëb, which does not belong to the area, distinguishes ablative and locative (Weir 1984; 96). The following properties are shared by T ariana and Tucano languages, but n o t by Baniwa o r Maki'i. ( Hi) A lien a b le a n d inalienable po sse ssio n In m ost N orth Arawak languages inalienably possessed nouns include all kinship nouns an d body parts and a few other items, e.g. ‘hom e’, together w ith deverbal nominalizations. Inalienable possession is m arked with the help o f possessive pre­ fixes, as in (7). In Arawak languages other than T ariana, alienably possessed nouns have a special possessive suffix, as in (13) (see A ikhenvald 1994b). In Tariana, inalienably possessed nouns are restricted to body parts and ju st a few kinship nouns (a smaller set than in other N orth A raw ak languages). Inalienable possession is marked by possessive prefixes. Alienable possession is shown simply by juxtaposition. Tariana (7) n u -h a d o or nuha h a -d u a Is g -m o th e r or Isg p a re n t-F E M ‘m y m o t h e r ’ C om pare this with the same lexical item as inalienably possessed in Baniwa: (8) h n ad ua< (underlying) ñu-ha-duá Isg-parent-FEM ‘my m other’ In Tucano inalienable possession is m arked by juxtaposition, e.g.: (9) y i’i Isg pa-co p a re n t-F E M ‘m y m o t h e r ’ All Tucano languages use the genitive m arker ya plus a classifier to m ark alien­ able possession, e.g. Tucano: ^(10) m i’i 2 sg ya possv ‘y o u r h o u s e ’ w i’i house 400 Alexandra Y. Aikhenvald Like Tucano, Tariana uses the possessive m arker ya with cross-referencing pre­ fixes plus a classifier as an alternative possessive construction for alienably and in­ alienably possessed items; the head noun is then usually om itted. In Baniwa classifiers are n o t used in possessive constructions {unless possession is predicative - s e e §4.3). Tariana (11) pi-ya-ku (am a-ku) 2sg-P0SSV-CL:EXTENDED (hammOCk-CLIEXTENDED) ‘your ham m ock’ Tucano (12) m t’i ya-gî 2sg po ssv -c l (pü-gî) : LA RG E ( h a m m o c k - C L : LARGE) ‘your h am m ock’ Baniwa (13) nu-pieta-ni 1sg-ham m ock-rossv ‘my ham m ock’ In both N adëb (Weir 1984:84-6) and D âw (M artins 1994:40,140), alienable and inalienable possession are m arked with juxtaposition of nouns: Dâw (M artins 1994: 46) (!4) dâw pat person hair ‘person’s h air’ For predicative possession, a possessive m arker which combines with classifiers is used in Dftw, b u t not in N adëb; this property is shared with Baniwa and will be considered in §4.3. (iv ) G ender In o th er N orth A raw ak languages, a gender opposition (feminine anim ate vs the rest) is found in verbal cross-referencing m arkers and demonstratives.'^ Classifiers are n o t used with demonstratives. Tariana and Tucano have a gender opposition in verbal cross-referencing markers, personal pronouns and kinship terms.'^ However, Among the M aku languages, Kakua distinguishes masculine and feminine genders in third person singular pronominal prefixes, independent pronouns and demonstratives, probably due to the Tucano influence (see §5 o f chapter 9). In Tucano there is also gender marking on adjectives, demonstratives, numbers and nominalized verbs. 14 Içana- Vaupés linguistic area 401 dem onstratives have special anim ate and inanim ate forms and they also com bine with classifiers. The head noun is frequently om itted. This is illustrated for Tariana in (15) and for Tucano in (16). In Baniwa, shown in (17), the dem onstrative takes a non-fem inlne form , but no classifier. Tariana (15) h a -k u (am a-ku) D EM : i n a n - c l : e x t e n d e d (h a m m o c k -C L : e x t e n d e d ) ‘t h i s h a m m o c k ’ Tucano (16) a ti-g î (p ü -g î) d e m ; i n - c l :l a r g e ( h a m m o c k -C L :l a r g e ) ‘th is h a m m o c k ’ Baniwa (17) hliehc pieta D E M + N .F E M ham m ock ‘this ham m ock’ (v ) C lassifiers a n d repealers Baniwa, Tariana and Tucano languages have noun classes, and numeral and pos­ sessive classifiers (see A ikhenvald 1994b; Barnes 1990, for Tuyuca). Baniwa and Tariana are the only N orth Arawak languages which use classifiers in possessive constructions. However, there are a few significant differences between Baniwa, on the one hand, and Tariana and Tucano, on the other. In Tariana and Tucano there is a special classifier for annnate nouns; in Baniwa anim ate non-hum an nouns are classified according to their shape: Tariana (18) a p i/a ;p i h a n u -ite snake b ig - C L :A N iM ‘a b i g s n a k e ’ Tucano (19) p in o p h a i-g i snake b ig - C L :A N lM ‘a b i g s n a k e ’ Baniwa {20)'^ a ;p i n ia k a -k h a y sn a k e big-CL:cuRViLiNEAR ‘a b i g s n a k e ’ 402 Alexandra Y. Aikhenvald Baniwa has a closed, albeit large set o f forty-four classifiers, including a residual classifier for unclassifiable items (see Aikhenvald 1996b). Tariana and Tucano lan­ guages have very large sets o f classifiers which can be considered ‘open’ since just about any noun w ith an inanim ate referent can be used to m ark agreement to clas­ sify the same noun (this phenom enon is known as ‘repeaters’ - see Aikhenvald forthcoming: chapter 10). Tariana and Tucano use ‘repeaters’ for otherwise unclas­ sifiable nouns. In East Tucano and Tariana, ‘repeaters’ (i.e. classifiers which look segmentally identical to the noun) lose their stress, whereas the corresponding independent noun will always retain its stress (Barnes 1996 for Tuyuca; Ram irez 1997 for Tucano). Tariana (21) pam'si ha n ii-p an isi house big-CL:HOUSE ‘a big h o u s e ’ Tucano (22) w i’i house p h a i-n '-w i’i big-SG.MASC-CL:HOUSE ‘a b ig h o u s e ’ Baniwa (23) p a n tj h ouse m a k a -d á ji big-CL:ROUND ‘a big (round) house, a big burrow ’ ‘R epeaters’ can be used as ‘ad hoc’ classifiers in Tariana and Tucano (see Aikhenvald 1994a, for discussion). The use o f a ‘repeater’ in Tariana is illustrated in (24). Panisi ‘house’ is used as an agreement marker on the adjective m at fa ‘good’, as an indicator o f the special im portance o f the newly introduced referent ‘house’. Otherwise, a classifier -dapana ‘habitat’ would be used, as in (25). Tariana (24) nuhua m a tja-panisi-m ha panisi nu-na I good-house-PRES.NON.viSUAL h o u se Isg -w an t “ ‘I w a n t a real Iv go o d h o u s e ” [the m an said to th e m ag ic rin g ].’ (25) kayu diha di-sata-ka diha depita so he 3sg.n.fem -ask-SEQ he n ig h t+ a d v h ik u -p id a n a p an isi m a t F a-d ap an a th u y a appear-REM.PAST.INFERRED h o u se eood-CLIHABITAT all 403 14 Içana-Vaupés linguistic area ajia-dapana-pidana hiku extst- c l :h a b it a t - r e m . past .in f f .r r e d appear diha-dapana-se-pidana naha he-CL:HABITAT-LOC-REM.PAST.lNFERRED they mat fa-dapana-se kahwi na-ka:-niki gOOd-CL:HABlTAT-LOC wake up 3pl + gO-SEQ-COMPL ‘A fter he had asked Hke this, the same night a good house appeared, in which there was everything, they woke up in this (house), a beautiful one.’ Classifiers are widely used in anaphoric and discourse-backgrounding functions. The head noun is often om itted in Tucano and Tariana (it is om itted from just before the second occurrence o f hiku in (25)). This does not happen in Baniwa. However, classifier systems in Baniwa and T ucano-T ariana still display a num ber o f shared properties not found in other neighbouring N o rth Arawak languages. We shall return to these in §4.4. Tariana and Tucano have portm anteau plural + classifier m orphem es, absent from N o rth Arawak. ( vi) P lural a n d p lu r a l ag reem en t Tariana - unlike other N orth A raw ak languages - and Tucano distinguish between anim ate and inanimate plural. Plural agreement is obligatory in a noun phrase. In both Tariana and East Tucano languages, inanim ate plural morphemes are not used for less than four objects. There are many suppletive plural forms, especially for kinship nouns. 3 .3.3 G ra m m a tic a l re la tio n s The m arking of gram m atical relations is uniform am ong the languages o f the Vaupés. Tucano languages have straightforw ard nominative-accusative patterns in crossreferencing and case-m arking (see §2 o f chapter 1 above). T ariana has lost the proto-A raw ak m orphological split-ergativity, which is m arked w ith cross-referencing affixes in Baniwa o f Içana and Warekena (for further discussion o f cross-referencing affixes and enclitics in Baniwa of Içana, Warekena, Bare and Tariana see Aikhenvald 1995b). In these languages: — prefixes cross-reference A (in a transitive clause) and (in an active intransitive); — suffixes and enclitics cross-reference O in a transitive clause and a stative intransitive). (in 404 Alexandra Y. Aikhenvald Tariana has n o cross-referencing enclitics, but it has acquired a norninativeaccusative case-m arking pattern under areal pressure from Tucano. O ne o f the striking differences between N adëb and the M akii languages o f the Vaiipés (D aw an d H upda) lies in the m arking o f gram matical relations. N adëb has one o f the m ost com plicated ergative-type systems in the world {Dixon 1994: 134, 178, 233; and ch apter 9 above). D iw and H upda are consistently nom inativeaccusative (see §2 o f chapter 9). This may be accounted for by areal pressure from the predom inantly suffixing East Tucano la n g u a ^ s . Unlike N adëb, Dâw and H upda have strong verb-final ten­ dencies. East Tucano languages and Tariana are also predom inantly verb-final, This drift tow ards AOV/SV typology due to areal diffusion may have led to the loss of constituent order beitig used to m ark gram matical relations (as it does in Nadëb). D âw and H upda have acquired accusative patterns o f case-m arking which are sim ilar to East T ucano, and to Tariana. 3.3.4 Verbal morphology and predicate structure (i) Verb serialization A m ong the languages o f the Vaupés basin, Tariana and Tucano have a typologically unusual com bination o f verb serialization and verb com pounding. Serial verb constructions can consist o f a sequence o f two or more verbs. Verb com pounding results in the creation o f aspect- and valency-changing m arkers (Aikhenvald ma G om ez-Im bert 1988). The M akii languages o f the U pper Rio Negro have productive verb com pound­ ing which results in the creation o f various aspect markers. This property is shared with East Tucano a nd Tariana, but not with the genetically related N adëb. It is also the means whereby M akû languages are gaining ‘new’ m orphology to make up for the préfixai m orphology they lost under areal pressure from East Tucano suffixingtype languages. T he following features are shared just by E ast Tucano and Tariana. (1) U nlike o th er N orth Arawak languages, Tariana is like East Tucano in distinguishing eyewitness and non-eyewitness evidentiality, immediate and d istant future, recent and rem ote past, conditional m ood and several imperatives. (2) Tucano and Tariana have suffixal negation, a negative copula, and (3) Tucano languages have a num ber o f com pound verbal forms which special negative words T do not know ’, ‘there is nothing’. consist o f a participial form o f the main verb and an auxiliary. In relaxed speech by younger people in Tariana, serial verb constructions 405 14 Içana-Vaupés linguistic area are replaced by a sequence ‘relative form o f the main verb plus auxil­ iary-like stance verb’, following the Tucano pattern o f a ‘participial form of the m ain verb plus auxiliary’ (Sorensen 1972) 3.4 Syntax and discourse technique The m ost striking feature o f syntactic organization o f the languages o f the Vaupés is a strong verb-final tendency. This is found in East Tucano, Tariana and in M akû. O ther N o rth Arawak languages show a verb-m edial tendency; N adëb is predom i­ nantly verb-initial (with constituent order used to m ark gram matical relations: see §10 o f chapter 9). The following properties are shared by T ariana and Tucano, but not by M akû. (1) T ariana and Tucano have long chains o f juxtaposed clauses, and use relative verbal forms (‘participles’) to m ark the predicate o f a subordinate clause. O ther N orth Arawak languages prefer sequencing verbal clitics (similar to medial verbs in Papuan languages; c f Aikhenvald 1995a, 1998, on sequencing clitics in Bare and W arekena). Tariana still m akes limited use o f a sequential clitic -ka, to m ark the predicate o f a com plem ent clause (and this is used m ost often in narratives told by old speakers).'"* Thus, Tariana com bines both ‘Tucano-like’ and ‘N orth-A raw aklike’ strategies o f com plem entation. (2) Tariana - unlike o ther N orth Arawak languages - and Tucano have a complex system o f switch-reference. This is illustrated in (26) and (27) from Tariana. In (26), -nisawa is used to m ark diflerent subjects, and in (27) -sita marks the same subject. (26) k a v -di-ni-nisaw a d ih a k u isi-se d i-w a so-3sg.n.fem -do-AFTER:D S he shore-LOC 3 sg .n .fem -en ter d i-a -p id a n a d i-w a d i-a 3sg.n.feni-go-REM.PAST.iNFERRED 3 sg .n .fe m -e n te r 3 sg .n .fem -g o ‘A fter h e (th e m a n ) d id this, h e (th e o tte r) c am e u p on th e sh o re a n d e n te re d (w a te r) a n d d ived aw ay.’ (27) duhua du-m ara-pidana du-nu-sita sh e 3sg.fem-float-REM.PASr.iNFERRED 3sg.fem-come-AFTER:SS d u -n u -k a d u in a 3sg.fem-come-DECL 3sg.fem + lo o k .fo r d h e w i-n u k u 3sg.n.fem + pip-TOPiCAL.NON.A/s ‘A fter she floated dow nstream , she cam e to lo o k for the p ip (o f a p e a ch p a lm f ru it) .’ A subordinating encütiç -ka is also found in four East Tucano languages (Carapana, Tatuyo, Waimaja and Yuniti); it is not clear whether or not it was spread by contact. 406 (3) Alexandra Y. Aikhenvald O ther N o rth A raw ak languages - like m ost other S outh A m erican Indian lan­ guages - avoid sentences with two full NPs, especially when one o f them is a free pronoun. Free pronouns are mainly restricted to em phatic function. Tariana, simi­ larly to East Tucano languages, makes wide use o f personal pronouns, as in (26) an d (27). (4) Unlike other N orth Arawak languages, Tariana and Tucano tend to use a ‘resuming’ phrase: ‘so he/she/they did’ at the end o f every paragraph: Tucano tojo (so do-3sg.masc), Tariana kay di-ni (so 3sg.n.fem-do) ‘then, thus, and so’ [lit. ‘so he did’] are used to point out a consequence, or the result o f an action. 3.5 Semantics Tariana has undergone a number o f lexical shifts under East Tucano influence. In Tariana, one w ord - k e ii - is used to refer to both ‘sun’ and ‘m oon’, similar to nnihipii Tucano. O ther N orth Arawak languages have two distinct words for these, e.g. Baniwa o f Içana k e y ‘m oon’, kcmmi ‘sun’. Tariana iri ‘blood’ has acquired the meaning o f ‘sap ’ (hence: rubber, plastic) under the influence o f East Tucano Æ ” O ther N orth Arawak languages have different words for ‘blood’ and ‘sap’, e.g. Baniwa o f Içana iri ‘blood’, dzeka ‘sap, rubber, plastic’ (see Aikhenvald 1996a). D aw and H upda-Y uhup also use ju st one term for ‘m oon, sun’; whereas N adëb has two different words (see chapter 9 on Makii). 4 PROPERTIES SHARED BY LANGUAGES OF THE 1ÇANA AND VAUPfiS REGION We can now look at the larger area consisting o f the Içana and the Vaupés basins. Besides cultural similarities w idespread across this area, there are a number o f lin­ guistic properties shared by Baniwa and the languages o f the Vaupés discussed above, but absent from other N orth Arawak languages. Table 14,2 summarizes these properties. Only one o f these properties - pitch accent - is also shared by the M akii languages o f the region. 4.1 Pitch accent Pitch accent (a low -high distinction) is a property o f all the languages spoken in the Vaupés. It is also found in Baniwa and K urripako (e.g. Valadares 1994), but is D i s t i n c t w o r d s f o r ‘b l o o d ’ a n d ‘s a p ’ a r e f o u n d i n W e s t T u c a n o l a n g u a g e s S i o n a , S e c o y a a n d K o r e g u a j e , a n d i n C e n t r a l T u c a n o T a n i m u c a - R e t u a m a a n d in C u b e o , 407 14 Iça m - Vaupés linguistic area Table 14.2 Properties shared by languages o f the Içana- Vaupés N o r th A ra w a k la n g u a g e s o u ts id e V aupés P r o p e rtie s s h a re d E ast T a r ia n a T ucano p itc h a c c e n t yes D aw . Içan a th is a re a B a n iw a W a re k e n a , B a re yes no H upda yes yes to p ic - a d v a n c in g v e rb a l d e riv a tio n yes yes no yes no p o s s e s s iv e c la s s ifie rs yes yes no y e s (w ith no p o s s e s s iv e p r e d ic a te s ) p o s s e s s iv e -ya- to yes yes no yes no yes yes no no no yes yes yes yes no w h ic h c la s s if i e r s a r e a tta c h e d c la s s ifie rs w ith d e m o n s tra tiv e s se v e ra l ty p e s o f c la s s ifie rs absent from Warekena and Bare, N orth A raw ak languages spoken outside the area (Aikhenvald 1995a, 1998). Topic-advancing verbal derivation 4 .2 Tariana and Baniwa have an argum ent-m anipulating derivation which creates an alternative construction type. There is no difference io transitivity value o r gram ­ m atical relations; it is ju st th a t a different argum ent is focused on in the basic and alternative construction types (Di.Kon and Aikhenvald 1997: 91 4). This derivation, m arked w ith suffix -ni on the verb, has the following properties. (i) T he fli-derivation m arks the advancem ent o f any non-subject topical constiifuent to subject; this constituent cannot take a non-subject case m arker (-na for pronouns; -mtkit for everything else). (ii) The original A/S^ is always cross-referenced on the predicate; the full (iii) The ni-derivation can be formed on intransitive active (S^) verbs and N P is often om itted. transitive verbs (transitivity is m aintained), but not on stative prefixless (S ) verbs. 408 Alexandra Y. Aikhenvald (iv) The verb includes a classifier suffix which cross-references just the NP moved into surface subject slot. (v) O ther argum ent N Ps (besides the original subject and the N P pro­ m oted into the surface subject slot) are m aintained. (vi) The main discourse function o f topic advancement is to m ark a con­ stituent which is more topical than the underlying subject. Example (28) is a plain transitive clause; the verb is underlined. (28) n u -n i-se-k a h a -p u n a -n u k u pi-na Isg-make-RES-DECL d e m : in a n - c l :st r e t c h - t o p ic a l . n o n .a/ s 2sg-0 ‘I m a d e this ro ad for y o u .’ Topic advancernent o f a direct object is illustrated in (29), and that of a locative in (30). (29) h a -p u n a n u -n i-n i-p u n a -se -k a DEM:INAN-CL:STRETCH lsg-make-TOPlC.ADVANCINO-CL:STRETCHRES-DECL pi-na 2sg-0 ‘This road was made by me for you.’ (30) kuphe flsh h ip a-n i-p u n a 2pl + take-TOPIC.ADVANCING-CL:ROAD ‘(It is) the place w here you get fish.’ The advanced constituent has five of the six criteria! properties o f ‘subject’: (i) equi-NP deletion; (ii) subject position in constituent order: AOV/SV; (iii) pronominalization; (iv) agreement in classifier; (v) case-marking. Cross-referencing is the only subject property retained by the original subject, e.g. Isg in (29). Tucano has a sim ilar ‘focus-changing’ derivation marked with a suffix -no(') (West 198Ü: 59; Terry M alone p.c.‘*). The suffix precedes the negative Hi and tense-aspect morphemes. This derivation applies to transitive and to intransitive verbs. In East Tucano languages gram m atical subjects are expected to be topical; in the rare cases where a non-subject (most often, the object) is more topical than the subject, the -no derivation m ust be used (Terry M alone p.c.). The constituent advanced to subject by a -no derivation triggers subject agreement in gender/number on the verb, and cannot be m arked by the case morphem e -re ‘definite/topical non-subject’. S i m i l a r d e r i v a t i o n s a r e f o u n d in o t h e r E a s t T u c a n o l a n g u a g e s ( T e r r y M a l o n e p .c . ) . 09 14 Içana-Vaupés linguistic area Example (31) illustrates an intransitive clause; in (32) the -m derivation has been applied, w ith a locative constituent being advanced to subject. (31) ci he B o g o tá -p i-re s ih a -w ï Bogotá-LOC-TOPiCAL.NON.A/s trav el-3 sg .m asc ‘H e w ent to Bogotá.’ (32) Bogotá sija-no’-w* Bogotá travel-TOPIC.ADVANClNG-3sg.JNAN ‘Bogotá was where he w ent.’ (lit. ‘Bogotá was gone to’) The following examples illustrate -no with a transitive verb. Example (33) is a plain transitive clause. The direct object is more topical than the subject in (34), and then the -no derivation has to be used. (33) na wa’i-re they fish-TOPiCAL.NON.A/s ba’a-wâ eat-3pl ‘They ate fish.’ (34) (West 1980: 59) w a’i ba’a-no’-wî fish eat-T0PlC.ADVANCING-3sg.INAN ‘The fish was eaten.’ The agent is frequently om itted, but can be retained, as in; (35) topi beans diakhi only m ahsa b a ’a-no-wt pe o p le eat-TOPic.ADVANCiNO-3sg.iNAN ‘Only b e a n s a re e a te n by p eo p le.’ 4.3 Possessive classifiers and -ya- possessive marlcer Tucano and Tariana use possessive m arker -ya- with classifiers to form possessive constructions (§3.3). In Tucano, these are restricted to alienable possession (see (10) above). Possessive constructions marked with -ya- are found in Tariana, where they can be used with both alienably and inalienably possessed nouns. Example (11) above shows an alienably possessed noun; (36) exemplifies an inalienably possessed one: (36) nu-ya-da (nuhwida) 1Sg-POSSV-CL:ROUND ( Is g + h ea d -C L : ROUND) ‘my (head)’ 410 Alexandra Y. Aikhenvald Baniwa a l s o h a s p o s s e s s iv e m a r k e r -yal-dza^’’ w h ic h is u s e d ex c lu siv e ly to mark predicative possession o f alienably and inalienably possessed nouns. Verba! classifi­ ers are obligatory in this construction. (37) h iie h ë p ie ta n u -d z a -k u this:SG .N .FEM ham m ock Isg-possv-CL ‘This h a m m o c k is m in e .’ (a lie n a b le ) (38) h jie h e h n u w i- d a this:SG .N .FEM head-C L:R O U N D n u -d z a -d a lsg-POSSV-CL:ROUND ‘This h e a d is m i n e .’ ( i n a lie n a b le ) This construction is n o t found in any other N orth Arawak language. It is found elsewhere in Tucano. However, in Baniwa it is restricted to possessive predication. This suggests the possibility o f unilateral diffusion from Tucano into Baniwa and Tariana within the V aupés-Içana contact area.'* 4.4 Complex systems of classifiers East Tucano languages, and Baniwa and Tariana, are unusual in having extensive systems of classifiers in which the same or alm ost the same sets o f classifier m or­ phemes are used in a num ber of distinct m orphosyntactie environments. All o f these languages use classifiers as agreement markers on adjectives and numerals. The difference between the possessive classifiers in Baniwa and those in T ucanoTariana was described in §4.3. Unlike Baniwa, T ucano-Tariana use classifiers with demonstratives. Unlike Tucano, Baniwa and Tariana use classifiers with verbs in the form of topic-advancing voice, as discussed in §4.2. See table 14.2. O ther N o rth Arawak languages o f the region just have numeral classifiers (Warekena, and a few N orth Arawak languages in Colom bia), or else no classifier systems at all (Bare). The Makii languages o f this region also have several types o f classifiers: generic noun classifiers - an extremely rare type in the Amazon - and locative classifiers (see M artins 1994; A ikhenvald forthcoming). N adëb only has possessive classifiers o f a generic type. However, these are distinct from classifier types found in other lan­ guages; and may represent individual innovations in the M aku languages o f the area (see §6 o f chapter 9). The 'S D aw phonem e has dz is p o s s e s s iv e correspondent o f T a r i a n a v in s o m e dialects o f classifiers o f a different kind; see §6 o f chapter 9. a r e g u la r B a n iw a . 14 Içana- Vaupés linguistic area 5 411 CONCLUSIONS Languages o f the three genetically unrelated groups - N o rth A rawak, E astT ucano and M akii - spoken in the Vaupés area share a num ber o f phonological, gram m at­ ical and sem antic features. As I have also shown elsewhere (Aikhenvald 1996a), T atiana, the only Arawak language spoken in the Vaupés, shares an overwhelming num ber o f structural features with East Tueano languages, while also m aintaining some A raw ak gram matical structure. The m ajority o f features shared by the lan­ guages in the Vaupés are found in Tucano languages outside the Vaupés, and ate not attested in languages o f the Arawak and M akii families spoken outside this area. T he direction o f diffusion thus appears to be from East Tucano to Tariana, and to M akii. Areal diffusion from East Tucano to Tariana involves: (a) emergence o f new categories present in E ast Tucano but absent from A raw ak, e.g., case-m arking connected with topicality, evidentials, verb com pounding, switch-reference; (b) structural levelling o f Tariana to agree w ith East Tucano syntactic structures and discourse techniques, and also obsolescence a nd subse­ quent loss o f categories that are not present in E astT ucano languages. T he parts o f T ariana phonology and gram m ar which are m ost affected by areal influence from East Tucano languages are nasalization; accent; verbal and nom inal m orphological categories; the m arking o f gram m atical relations interacting with topicality; syntactic structures, such as clause chaining strategies, and discourse techniques. Tariana retains a considerable num ber o f m orphem es o f A raw ak origin and the predom inantly suffixing A raw ak word structure, with a limited num ber of prefixes. T his suggests that area! diffusion in this linguistic area (which has a very shallow time depth) has n ot affected word structure as it has affected the rest o f the gram m ar. As in other linguistic areas (see Sherzer 1976: 255), internal linguistic factors along with socio-cultural conditions - such as code-switching, and the exo­ gamie principle o f interm arriage which results iri obligatory multilingualism account for the diffusion o f linguistic traits. A ttitudes towards language (i.e., a strong inhibition against language mixing viewed in term s o f lexical loans) result in alm ost no direct lexical diffusion. W hat one does get is diffusion o f phonological and m orphosyntactic structures, but not o f forms. As has been pointed out by Sorensen (1972), East Tucano languages on the C olom bian side - where Tariana has never been traditionally spoken - form a lin­ guistic area sharing a fair num ber of linguistic traits. However, since these languages 412 Alexandra Y. A ikhem ald are genetically closely related, it is hard to distinguish areally diffused from geneti­ cally inherited features. Dâw, Hiipda and Yuhup - M aku languages spoken in the Vaiipés region in Brazil - show a few instances o f unilateral diffusion from East Tucano languages. Diffusion patterns from East Tucano languages into M aku include loss o f prefix­ ing (as com pared to the M aku language spoken outside the Vaupés, Nadëb); acqui­ sition o f core case m arking associated with topicality of the noun; and development o f verb com pounding which results in the creation o f aspect- and valency-chang­ ing morphemes. The contact between M akii and E ast Tucano peoples is much older than that between the Tucano and the Tariana, and this m ight explain the fact that - unlike in the case o f Tariana - areal diffusion has affected the word structure. However, the degree o f cultural integration between East Tucano and Makii is much less than th at between Tariana and E ast Tucano. This may account for fewer signs o f diffusion than in the case o f the Tariana-T ucano contact. This unilateral character o f diffusion is accounted for by the fact that the M akii used to know at least one Tucano language, but not vice versa. However, m ore work is needed on both Brazilian and, especially, Colom bian M akil languages such as K akua and N ukak (spoken outside the Vaiipés) to distinguish between further possibilities of areal diffusion an d genetically inherited developments in this family. Tariana and East Tucano languages have been in contact for no more than about 400 years. The settlem ent o f East Tucano tribes on the Vaupés goes further back (cf. N im uendajû 1982: 169-70); we do not have any exact dates. The other reason­ ably well-described linguistic areas o f the world, e.g. the Balkans, A ruhem Land in A ustralia (see Heath 1978, 1981), M esoamerica {Campbell, K aufm an and SmithStark 1986), S outh A sia (M asica 1976) and the linguistic areas o f N orth America n o rth of Mexico (Sherzer 1976), such as the north-w est coast, are considerably older than this. As I have argued elsewhere (Aikhenvald 1996c), a study o f types o f Tariana placenaraes shows that two o f these types o f nam es are predom inantly, m onolin­ gual - ‘historical’ nam es which refer to places where the T ariana used to live in the rem ote past, an d ‘m ythological’ names which refer to the adventures o f characters in origin myths. In contrast, placenames which refer to actual dwelling sites are multilingual, a nd are usually calqued into several languages. Even when ‘historical’ placenames also have nam es in languages other than T ariana they are never caique translations from one language into another. These properties o f ‘historical’ and ‘mythological’ placenames, unexpected in an environment o f obligatory multilin­ gualism, suggest th a t the Tariana might have arrived in the Vaupés from a predom ­ inantly m onolingual context, and that they have adopted multilingualism fairly recently. 413 14 Içm a-Vaupés linguistic area T he Vaupés linguistic area can be com pared - both in time depth and in degree o f moltilingualism - to K upwar village in India. A ccording to G um perz and Wilson ( 1971: 153), the coexistence o f U rdu, M arathi and K an n ad a in the region goes back about three or four centuries, when the U rdu-speaking M uslims arrived there. However, K annada-speaking and M arathi-speaking people have been in the region for m ore than six centuries. U nlike Kupwar, the T ariana-T ucano contact area has no t reached the same degree o f m orphosyntactic isom orphism between genetically unrelated languages, which m ay have been a partial consequence o f a ‘puristic’ ten­ dency to keep languages as m uch apart as possible. Language attitudes in the Vaupés, which exclude lexical borrow ing and lexical diffusion, are also different from th e situation in the K upwar (where occasional lexical borrowings do take place). T h e existence o f structural and even formal similarities shared by T ariana, East Tucano languages and Baniwa - but absent from other N o rth Arawak languages shows a certain am ount o f diffusion in an area which goes beyond the Vaupés into the basin o f Içana and its tributaries (see m ap 13).'* BIBLIOGRAPHY A i k h e n v a l d , A , Y. 1 9 9 4 a . ‘G r a m m a t i c a l r e l a t i o n s i n T a r i a n a ’, Nordic Journal o f Linguistics, 1 7 ,2 0 1 ^ 1 7 . 1 9 9 4 b , ‘C l a s s i f i e r s i n T a r i a n a ’, 1995a, Bare, Anthropological Linguistics, 3 6 ,4 0 7 - 6 5 , L a n g u a g e s o f th e W o rld / M a te r ia ls 100, M u n ic h : L in c o m E u r o p a . 1 9 9 5 b , ‘P e r s o n m a r k i n g a n i l d i s c o u r s e i n N o r t h A r a w a k l a n g u a g e s ’ , Studia Linguistica, 4 9 .1 5 2 - 9 5 , 1 9 9 6 a . ‘A r e a t d i f f u s i o n Linguistics, in N o r t h w e s t A m a z o n i a : t h e c a s e o f T a r i a n a ’, Moscou’ Journal o f Linguistics, 3 ,7 - 3 3 . 1 9 9 6 c , ‘ M u l t i l i n g u a l a n d m o n o l i n g u a l p l a c e n a m e s i n T a r i a n a ’, mastics, Anthropological 3 8 ,7 .1 - 1 1 6 , 1 9 9 6 h , ‘C l a s s i f i e r s i n B a n i w a ’, Names: a Journal o f ono­ 4 4 .2 7 2 - 9 0 , I 9 9 S , ‘W a r e k e n a ’ , p p . 2 2 5 - 4 3 9 o f Htmdhook o f Amazonian languages, v o l. IV , e d , D . C . D e r b y s h ir e a n d G . K . P u l lu m . B e rlin : M o u to n d e G r u y t e r . f o rth c o m in g . Classifiers: a typology o f noun categorization devices, O x fo rd ; C la r e n d o n P ress. T r a d itio n a l s to rie s o f th e T a r ia n a a n d B a n i w a i n d i c a t e t h a t t h e s e p e o p l e s h a d b e e n in c o n t a c t w ith th e E a s t T u c a n o g r o u p s b e fo r e th e T a r i a n a h a d m o v e d f ro m th e R a p id s of A p u i o n th e A ia ri riv e r (a t r i b u t a r y o f th e I ç a n a ) t o th e V a u p é s . A s to r y o f th e a c tu a l m o v e f ro m th e A p u i c a lle d myaka-dapana ( ‘o l d - C L : H A B ( T A T ’) ‘t h e d w e l l i n g o f t h e a n c i e n t s ’ d e s c r i b e s h o w th e T a r i a n a liv e d t o g e t h e r w ith t h e C u b e o b e f o r e th e y le f t A p u i. F u r th e r s tu d ie s a re n e e d e d to in v e s tig a te th e a n c ie n t c o n ta c ts b e tw e e n T u c a n o and B a n iw a In th e I ç a n a - V a u p é s a r e a . F o r th e tim e b e in g , o n e c a n h y p o th e s iz e t h a t t h i s c o n t a c t m u s t d a t e b a c k t o a t i m e p r e c e d i n g t h e c r e a t i o n o f t h e T a r i a n a - T u c a n o d i f f u s i o n a r e a in th e V a u p é s. 414 Alexandra Y. À ikhenm ld m s . ‘V e r b c o m p o u n d i n g a n d s e r i a l i z a t i o n i n T a r i a n a ’ . in p r e p .  gramm m o f Tariana. A r d i l a , O . 1 9 8 9 . ‘D i v e r s i d a d l i n g ü í s t i c a y m i i l t i l i n g ü i s m o e n l o s g r u p o s T u c a n o d e l V a u p é s ’, Forma y Ftindon, 4 .2 3 - 3 4 . ÍJA L, 5 0 . 2 2 5 - 7 1 . Amazonian linguistics: studies in Lowland South B a r n e s , X 1 9 8 5 . ‘E v i d e n t i a l s i n t h e T u y u c a v e r b ’, 19 9 0 , ‘C l a s s i f i e r s i n T u y u c a ', p p . 2 7 3 - 9 2 o f American Indian languages, e d . D o ri s L . P a y n e . A u s tin : U n iv e rs ity o f T e x a s P re ss. 1 9 9 6 . ‘A u t o s e g m e n t s w i t h t h r e e - w a y le x i c a l c o n t r a s t s i n T u y u c a ’, UAL, 6 2 .3 1 - 5 8 . B e s s a F r e i r e . J. 1 9 8 3 . ‘D a “ f a l a b o a ” a o p o r t u g u ê s n a A m a z ô t i i a b r a s i l e i r a ’, Amerindia, 8 ,3 9 - 8 3 . B r ü 'ö i , A . A . d a S . 1 9 6 7 , Obsermçôes granmticais da lingua Daxseyê ou Tucano. la u a r e te : C e n tr o d e P e s q u is a s d e Ja u a re te . 1977. A civilizaçâo indigena do Vaiipés. R o m e: L as. C a m p b e l l , L „ K a u f m a n , T . a n d S m i t h - S t a r k , T . 1 9 8 6 . ‘ M e s o - A m e r i c a a s a l i n g u i s t i c a r e a ’, Language, 6 2 .5 3 0 - 7 0 . D ix o n , R . M . W . 1994. Ergativity. C a m b r id g e : C a m b r id g e U n iv e rs ity P re ss, D i x o n . R . M . W . a n d A i k h e n v a l d , A . Y . 1 9 9 7 . ‘A t y p o l o g y o f a r g u m e n t - d e t e r m i n e d c o n s t r u c ­ t i o n s ’, p p . 7 1 - 1 1 3 o f Essays on language function and language type, e d . ,1. B y b e e , J. M a im a n a n d S. T h o m p s o n . A m s te r d a m : J o h n B e n ja m in s . G a l v à o , E . 1 9 7 9 . “A c u l t u r a ç â o i n d i g e n a n o r i o N e g r o ’, p p . 1 3 5 - 9 2 o f h i s edades. R i o d e J a n e i r o : P a z e t e r r a . I. 1 9 7 9 . The Ciibeo: Indians o f the Northwest Amazon. G o ld m a n . Encomro de sod- 2 n d e d itio n . U rb a n a : U n iv e rs ity o f I llin o is P re ss. G o m e z - l m b e r t , E , 1 9 R 8 , ‘C o n s t r u c c i < 5 n v e r b a l e ii b a r a s a n a y t a t u y o ’ , Amerindia, 1 3 .9 7 -1 0 8 . 1 9 9 6 . ‘W h e n a n i m a l s b e c o m e “ r o u n d e d ” a n d “ f e m i n i n e ” : c o n c e p t u a l c a t e g o r i e s a n d l i n g u i s ­ tic c la s s if i c a tio n in a m u lt ilin g u a l s e t t i n g ', p p . 4 3 8 - 6 9 o f Rethinking linguistic relativity, e d . J. J, G u m p e r z a n d S , C , L e v i n s o n . C a m b r i d g e : C a m b r i d g e U n i v e r s i t y P r e s s . G r a l o w , F. L . 1 9 8 0 . ‘ F o r m a s d e t e m a t i z a c i o n e n c o r e g t i a j e ' , Arikulos en Ungiiistica y campos afines, 8 . 1 - 3 6 . G r i m e s , B , F . 1 9 8 5 , ‘L a n g u a g e a t t i t u d e s : i d e n t i t y , d i s t i n c t i v e n e s s , s u r v i v a l i n t h e V a u p é s ’, Journal o f Multilingual and Multicultural Development, 1 9 8 8 . Ellmologue: languages o f the world. 1 1 t h ( e d .) 6 .3 8 9 ^ 0 1 . e d itio n . D a lla s : W y c lif f e B i b le T r a n s la to r s , In c . G u m p e r z , J. J. a n d W i l s o n , R . 1 9 7 1 . ‘C o n v e r g e n c e a n d c r e o l i z a t i o n . A c a s e f r o m t h e I n d c A r y a n / D r a v i d i a t i b o r d e r in I n d i a ' , p p . 1 5 1 - 6 8 o f guages, c d . H e a t h , J. 1978. Pidginization and creolization o f lan­ D . r iy m e s , C a m b r id g e : C a m b r id g e U n iv e rs ity P re ss. Linguistic diffusion in Arnhem Land, C a n b e r r a : A u s tra lia n I n s titu te of A b o ri g in a l S tu d ie s . Language, 5 7 . 3 3 5 - 6 7 , Comparative vocabulary: selected words in indigenous lan­ guages o f Colombia. S a n t a f é d e B o g o t a : A s o c i a c i ö n I n s t i t u t e L i n g û l s t i c o d e V e r a n o . H u g h - J o n e s , C . 1 9 7 9 . From the M ilk River: spatial and temporal processes in northwest Amazonia. C a m b r i d g e : C a m b r i d g e U n i v e r s i t y P r e s s H u g h - J o i i e s , S . O . 1 9 7 9 . The palm and the pleiades: initiation and cosmology in northwest Amazonia. C a m b r i d g e : C a m b r i d g e U n i v e r s i t y P r e s s . 1 9 8 L ‘A c a s e o f i n t e n s i v e le x ic a l d i f f u s i o n ’. H u b e r, R . Q . a n d R e e d , R , B, 1992. J a c k s o n . J. 1 9 7 4 . ‘L a n g u a g e id e n tity of th e C o lo m b ia n Explorations in the ethnography o f speaking, V a u p ê s I n d i a n s ', p p . 5 0 - 6 4 of e d . R . B a u m a n a n d J. S h e rz e r. C a m b r id g e : C a m b r id g e U n iv e r s ity P re ss. J o n e s , W . a n d J o n e s , P. 1 9 9 1 . Barasano synta,\. S t u d ie s in th e l a n g u a g e s o f C o l o m b ia , 2 . D a lla s : S I L a n d th e U n iv e rs ity o f T e x a s a t A r l in g to n . 415 14 Içana- Vaiipés linguistic area K in c h , R . A . Tucanos, 1 9 7 7 . ‘E l e n f o q u e te n iâ t ic o K o c h -G rü n b e rg , G e b i e t e ', Esludios v s e l e n f o q u e n o t e m à t i c o e n y u r i t i ’, 2 .1 2 9 - 7 5 . T. 1911. ‘A r u a k - S p r a c h e n N o rd w e s tb r a s ifie n s Mitteilungen der anthropologischen und der a n g re n z e n d e n G e if/f a c A n /î, 4 1 .3 3 - 1 5 3 , 2 0 3 - 8 2 . M a l o n e , T . 1 9 8 7 . ‘P r o t o - T u c a n o a n a n d T u c a n o a n g e n e t i c r e l a t i o n s h i p s ’. C o l o m b i a : I n s t i t y t o L in g ü is tic o d e V e ra n o . M s. M a r t i n s , S . A . 1 9 9 4 . 'A n á l i s e d a m o r f o s i n t a x e d a l i n g u a D f t w ( M a k ù - K a m â ) e s u a d a s s i f t c a ç à o t i p o l é g i c a ’, M A t h e s i s , U n i v e r s i d a d e F e d e r a l d e S a n t a C a t a r i n a , F l o i i a n o p o l i s , B r a z il. M a s ic a , C . P 1976. Defining a linguistic area: Somh Asia. C h ic a g o : U n iv e rs ity o f C h ic a g o P ress. M e tz g e r , R . G . Gramâtica popular del Carapana. 1981. B o g o ta : I n s titu to L io g ü is tic o d e V eran o . M o o r e , D . L . , F a c u n d e s , S . a n d P i r e s , N . N . 1 9 9 4 . ‘N h e e n g a t u ( L i n g u a G e r a l A m a z o n i c a ) , i t s Proceedings o f the Meeting o f SSILA July 2-4. JV93 and the Hokan-Penutian Workshop, July i . 1993, e d . M . L a n g d o n . h i s t o r y t w d t h e e f f e c t s o f l a n g u a g e c o n t a c t ’, p p . 9 3 - 1 1 8 o f B e rk e le y ; U n iv e rs ity o f C a lif o r n ia . N i m u e n d a j t i , C . 1 9 5 0 /5 . ‘ R e c o n h e c i m e n t o d o s R i o s I ç a n a , A y a r i e V a u p é s ’, Journal de la Société des Américanistes, n . s ., 3 9 . 1 2 5 - 8 2 , 4 4 . 1 4 9 - 7 8 , 1982. Textos indigenistas, Sâo Paulo: Ediçôes Loyola. O l i v e i r a . A . E . d e . 1 9 7 5 . Ä t e r m i n o l o g i a d e p a r e n t e s c o B a n i w a ’, Emilio Goeldi. Nova Série, Antropologia, Boletim do M meu Paraense 56 . B e lé n i. P a y n e , D o r i s L . 1 9 9 1 . ‘A c l a s s i f i c a t i o n o f M a i p u r a n ( A r a w a k a n ) l a n g u a g e s b a s e d o n s h a r e d l e x i c a l r e t e n t i o n s ’, p p . 3 5 5 - 4 9 9 o f Handbook o f Amazonian language.!, v o l. I l l , e d . D . C - D e t b y s h i r e a n d G . K.. P u l l u m . B e r l i n : M o u t o n d e G r u y t e r . R a m ir e z , H . 1997. A faia Tukano dos Yepä-imsa. Tomo I: Gramâtica. M a n a u s : In s p e to ria S a lc s ia n a . R e ic h e l- D o lm a to f f, G . 198 6 . Desana: simbolismo de los indios Tukano del Vaupés. B o g o ta : N u e v a b ib lio te c a c o lo m b ia n a d e c u ltu r a . R o d rig u e s , A . D . 1986, IJnguas hrasileiras: para o conhecimento das iinguas indigenas. Sâo P a u l o ; E d iç ô e s L o y o la . S h e r z e r , X 1 9 7 3 . ‘A r e a l l i n g u i s t i c s i n N o r t h A m e r i c a ’ , p p . 7 4 9 - 9 5 o f tics, 1976. Current trends in linguis­ v o l. X , e d . T . S e b e o k . T h e H a g u e ; M o u to n . An areal-typological study o f American ktdian languages north o f Mexico. N o rth H o l l a n d L in g u is t ic S e r ie s , 2 0 . A m s t e r d a m : N o r t h - H o l l a n d . Os Makit: povo caçador do Noroeste da Amazônia. S i l v e r v v o o d - C o p e , P. L . 1 9 9 0 . B ra s ilia : E d i t e r a d a U n iv e rs i d a d e N a c io n a l d e B ra s ilia . . S o r e n s e n , A . P ., ,lr. 1 967 ( 1 9 7 2 ). ‘M ïïh ilin g iia lis m Anthropologist, 6 9 . 6 7 0 - 8 4 in t h e N o r t h w e s t A m a z o n ’, American Sociolinguistics, e d . J. B. P r id e ( r e p r i n t e d in 1 9 7 2 , p p . 7 8 - 9 3 o f a n d J. H o lm e s . H a r m o n d s w o r t h ; P e n g u in ) . T a y lo r, G . 1 9 90. Introduçâo à Hngtia Baniwa do Içana. C a m p i n a s : E d i t e r a d a U n i c a m p . Language contact, creolization and genetic linguis­ T h o m a s o n , S. G . a n d K a u f m a n , T , 1 9 8 8 . tic.'^. B e rk e le y a n d L o s A n g e le s : U n iv e rs ity o f C a lif o r n ia P re ss. V a la d a re s, S. M. B. 1994. Ä s p e c to s da lin g u a K u rrip a k o ( K i t m a n d e n e ) ’, M A th e s is , U n i v e r s i d a d e F e d e r a l d e S a n t a C a t a r i n a , F l o r i a n o p o l i s , B r a z il. p h o n e m i c s ’ , p p . 2 5 - 3 6 o f Phonemic systems o f Colombian languages. N o r t n a n : S I L a n d U n i v e r s i t y o f O k l a h o m a . Languages In contact: findings and problems. T h e H a g u e ; M o u t o n . W a l t z , N . a n d W a l t z , C . 1 9 6 7 . 'G u a n a n o W e in r e ic h , U . 19 6 4 . W e ir , E. M . H. 1 9 8 4 . ‘A negaçâo e o titro s to p ic o s d a U n i v e r s i d a d e E s t a d u a l d e C a m p i n a s , B r a z il. g r a m â tic a N a d ë b ’, M A th e s is , 416 Alexandra Y. A ikhenm ld 'I n c o r p o r a t i o n i n N a d ë b ’, p p . 3 2 1 - 6 6 o f Amazonian linguistics: studies in Lowland South American Indian languages, e d , D o r i s L . P a y n e . A u s t i n ; U n i v e r s i t y o f T e x a s P r e s s . W e s t , B . 1 9 8 0 . Gramáiicapopular del Tucano. B o g o t á : I n s t i t u t e L i n g ü î s t i c o d e V e r a n o . W h e e l e r , A . 1 9 8 7 . Gantëya Bain: el pueblo Siona del rlo Putumayo, Colombia. Tomo 1, Etnologla. gramâtica, textos. B o g o t a : I n s t i t u t o L i n g ü î s t i c o d e V e r a n o . W h i s l e r , D . a n d W h i s l e r , J. 1 9 7 6 . ‘F o n o l o g i a d e l t a t u y o ’, Sistemas fonolágicos de idiomas Colombianos, 3 . 1 1 3 - 2 2 . 1990. 15 The Upper Xingu as an incipient linguistic area LU CY SEK I 1 TH E U PPER X IN G U AND IT S L A N G U A G E S ' When the Portuguese cam e to Brazil they lost no time in colonizing the coasts and m ajor rivers. The Xingu, a m ajor southern tributary o f the A m azon, is navigable for only 200 kilometres. A fter this there are innum erable rapids, which proved a barrier to the invader. As a consequence, the U pper Xingu area rem ained unm o­ lested. T h e Indian tribes living there were able, by and large, to maintain their tra­ ditional way of life. O ther tribes converged on this region, as a haven o f escape from the white m an ’s takeover o f Brazil. (In addition, some tribes were sent to the Xingu area by the Europeans.) C ontact with the outside world was only really established in the 1950s. As a result, the U pper X ingu region - which was, in 1961, declared a ‘national ■park’ by the Brazilian governm ent - is rem arkable for its linguistic diversity. C ultural traits diffuse more quickly than linguistic ones. The U pper Xingu has become culturally rather hom ogenous, due to close intertribal contacts, a network o f interm arriage and the sharing o f material culture. But linguistic diffusion is as yet in its early stages; what we have here is an incipient linguistic area. 1.1 Languages spoken There are currently 17 indigenous gtoups in the U pper Xingu and the adjacent areas. See m ap 14. Only 10 o f them^ have been there for more than 100 years. The languages spoken and the num ber o f speakers^ in the U pper Xingu are listed in ' I a m g r a t e f u l lo F r a n t o m é B , P a c h e c o , f o r i n f o r m a t i o n o n C a r i b la n g u a g e s . ^ S i n c e N a h u k w a a n d M a t i p u liv e t o g e t h e r i n o n e v i l l a g e , t h e y a r e s o m e t i m e s c o n s i d e r e d a s o n e g r o u p (S e k I to r t b o o m in g ) . ’ N u m b e rs o f s p e a k e r s in t a b l e s 15 .1 - 1 5 .3 a r e g iv e n a c c o rd in g to M e iid o n ç a e i a /. (1 9 9 5 ); d a ta o n Y a w a la p iti a re f ro m M u ji c a ( 1 9 9 2 ) . B a r u z z i, R o d r ig u e s , 418 L ucy Seki 419 15 Upper Xingu linguistic area table 15.1."* This table also contains inform ation about the territory where each group now lives and used to live before, and on the origin o f each group. There are 3 tribal groups speaking Arawak languages. W aurá and M ehinaku are dialects of one language. Yawalapiti shares 80 per cent vocabulary with W au rá-M ehinaku but the gram m ar is very dilferent, so that there is no m utual intelligibility and Yawalapiti has to be considered a separate language. Kuikiiro, K alapalo, N ahukw á and M atipu are m utually intelligible, and consti­ tute a single language o f the C arib family; this is m arkedly different from the other C arib languages spoken south o f the A m azon (F ranchetto 1995: 53). K am aiurá belongs to the T upi-G uarani subgroup o f the Tupi family, while Aweti is a separate, single-member family within Tupi (see chapters 4 and 5 on Tupi and Tupi-G uarani). iVlaterials available on X inguan languages vary in quantity and in quality A full gram m ar has been produced for K am aiurá (Seki forthcoming). There are good m aterials on Trum ai (G uirardello 1992, forthcom ing), Kuikuro (F ranchetto 1986, 1995) and W aurá (Richards 1973; 1977; 1988; 1991). Partial descriptions are avail­ able for M ehinaku (M edeiros 1990) and Yawalapiti (M ujica 1992); Aweti rem ains the least known o f all (there is ju st a description o f the phonology in Em merich and M onserrat 1972). This m akes difficult a full study o f gram matical and lexical diffusion within the area. Table 15.2 shows the language groups spoken adjacent to the U pper X ingu region; two of them belong to the Jê family, and one is Tupi. Their contacts with the peoples o f the U pper X ingu have been com paratively recent. 1.2 Fiistorical background The A rawak-speaking peoples are believed to have been the earliest settlers in the U pper X ingu region. Otiier groups began to arrive from the seventeenth century the C arib tribes, the Aweti, the K am aiurá (A gostinho da Silva 1993: 283). A ccording to Cowell (1973: 227), the K am aiurá were one of the last tribes to arrive in Xingu. The Trumai are the most recent arrivals. The G erm an explorer K arl von den Steinen established the first contact w ith X ingu peoples in 1884. He reported that the Trumai still m aintained cultural differences from their neighbours (see note 8). The X ingu culture area m ust have been established in the second half o f the eigh­ teenth, or in the early nineteenth, century (H eckenberger 1996).’ ■' E x t i n c t l a n g u a g e s in t h e X i n g u b a s i n in c lu d e : T s u v a a n d N a r a v ilte ( C a rib ) , K u s te n a ii ( A r a w a k ) , a n d A n u m a n ia a n d M a n its a w a (T u p i). ^ T h e a r c h a e o lo g ic a l e v id e n c e s h o w s t h a t th e o c c u p a tio n o f th e b a s in o f th e U p p e r X in g u s t a r t e d in th e e a r ly e le v e n th c e n tu r y . T h e s e ttle m e n t o f t h e L o w e r C u l u e n e c o n t i n u e d u n til a t l e a s t t h e e n d o f t h e t h i r t e e n t h c e n t u r y ( B e c q u e l i n 1 9 9 3 : 2 2 8 ) . H o w e v e r , it is n o t c l e a r w h a t g r o u p s w e r e in v o lv e d . 3 S ä S •Si S s as < -S III á e2 n j .S ,S :3 I3 I« I"2 00 ^ S I<U IG Jlf-^ !r a S ë 8 E •£ o .£ ü ë ■$■3 t U I S-. ^ II i l l is 3 t2 H aÖ i g- 9I M 2 j z N = II :s 5«i s ^ ,& j - | I á á ;2 (2 ,s ii H i S’ s 8 -S i 111 g sc Sd f1= ■S ■£ ~ 2 I t|.S < !^ 3 O 2i X á <L> o c •c B* r s? .r 2 s Si III I Æ 5 > Æ ■« „ Ä o w ä S e B S i |H!-Si2 s|1S '3 âS S -s a H o p i> ^ ^ T3 c o 2 "i I .§ g ffl ) 'Z ^ o> ■B ca û J ^ 3s ä 5?>3a - X — c o X :S I o ^ "S “ “ ä "So fSS f S l i s I fSs IS Upper Xingu linguistic area 1.3 423 The mobility o f groups, language loss and further contacts An im p o rtan t problem in the XIngu area is the frequent change o f location by tribal groups. D ue to intertribal hostilities and warfare, som e groups become weaker, they often end up living with an o th er group for protection. This frequently results in lan­ guage endangerm ent and loss. In the past, the K ustenaù (Arawak) cam e to live with the o th er Arawak groups, W aurá and M ehiiiaku; A num ania (Tupi) joined Aweti (Tupi); and M anitsawá (Tupi) joined Suyá (Jê). K ustenaít, A num aniá and M anitsawá are now e.xtinct. T he oral history o f K am aiurá provides another exam ple o f possible language loss in the past. These people say that the present-day fCamaiurá is the result o f five dialect (o r language) groups com ing together, for unknow n reasons; currently only one person is considered to be a ‘real’ K am aiurá. O ther K am aiurá, when asked, m aintain th at ‘they are stealing the K am aiurá language'. Recently, the N ahukw á cam e to live with the closely related M atipu (both Carib). In the Lower X ingu, the Tapayuna joined the M etuktire, both Jê, for reasons o f protection. The Trumai have changed their location several times during the past 100 years, moving from the lower Cukiene up to the northern parts of the upper X ingu. and then to the m iddle Xingu (G uirardello 1992: 12). T he m obility o f groups w ithin the U pper X ingu results in increasing contacts between them, and also in the integration o f groups which did not originally belong to the U pper X ingu. T he expansion o f white culture brought the peoples o f the U pper Xingu into closer contact with some of their neighbours. Suyá (Jê) are a case in point. They have never been part o f the U pper X ingu region, b u t as the result o f interactions with the X inguans, they have absorbed a num ber of cultural traits they did n o t have before, e.g. ham m ocks, canoes, m anioc (Seeger 1974, G alvao 1953). These ties have become stronger since 1959, when regular contacts w ith white people were established. As a result, the Suyá are considered m arginal representa­ tives o f the U pper X ingu culture area. Several groups which traditionally did not inhabit the Xingu basin were trans­ ferred to the Xingu N ational Park ^y the Brazilian governm ent to save them from extinction and to free their territories for exploration. Peoples who have been trans­ ferred into the Xingu basin since the 1950s are listed in table 15.3. In the long run, the presence o f new peoples in the Xingu area and their m obility w ithin the region m ust have an efTect on the m ore traditional inhabitants o f the U pper Xingu. 2 THE U P PER XIN G U AS A CULTURE AREA The peoples of the U pper Xingu share a num ber o f distinctive cultural and material traits developed as the result o f long-term contact between the groups, intertribal 424 Lucy Seki m arriage and economic exchaoge. The main m aterial traits include haystack-shaped houses arranged in a circle around a central plaza; dependence upon fish rather than game for protein and the lack of ferm ented drinks. All the peoples in the U pper Xingu have benches in the form o f animals fashioned from a single piece o f wood, whistling arrows, bark canoes and the bull-roarer. They use the spear-thrower in intertribal games and ceremonies. Men cut hair in a circular line above the ears, while women wear it cropped on the front in a straight line. X inguan women wear ulurt, a triangular bark pubic cover. Typical X inguan necklaces are made o f rectangular or disc-shaped pieces of periwinkle shells (G alvao 1953, Murphy and Q uain 1955). There are strong similarities in social organization and kinship systems. All the Xingu groups have extended families which are predom inantly patrilineal; a wife will norm ally move to her husband’s village, after an initial period in her own village, ‘Com m on’ people (cam am ) are difTerentiated from a small elite which con­ sists o f tribal leaders. The peoples of the U pper Xingu share cultural practices. Shamanism is much the same am ong all groups. Sun and M oon feature as the main creative forces in their mythology. They share the main festivities, for instance, kwaryp (a term of K am aiurá origin), the feast o f the dead; a m artial game called huka-huka; and the dance ritual o f jciwari. The X inguan peoples also have dances involving wooden o r straw masks, and ceremonial flutes which women are not allowed to see (Galvao 1953). Alongside cultural uniformity, each group has its own speciality. An im portant part o f the traditional life in the U pper X ingu is intertribal meetings (known as m oitara) for the purpose of exchanging products supplied by individual groups. The K am aiurá produce black bows, the Trum ai make traditional stone axes, Carib groups are responsible for necklaces m ade o f periwinkle shells, while the Waurá m ake ceramic pots. Trum ai and M ehinaku provide salt (which was traditionally extracted from aquatic plants). Aweti and M ehioaku traditionally played the role o f middlemen. 3 L IN G U ISTIC SITUATION In spite o f the cultural similarity, each group m aintains its identity - its language is its m ain distinctive feature. The idea o f a language as a symbol o f identity is sup­ ported by the restrictions on language use during ceremonies: one is not allowed to use a language/dialect o th er than one’s own in the situation o f intertribal com m u­ nication and ceremonies. Interaction takes place via a system o f non-verbal com ­ m unication shared by ail the participants (Em merich 1984).® * Basso (1973; 5) described the Upper Xingu as a ‘communication network’ rather than a ‘speech community’, i.e. ‘a system consisting of several kinds of linkages between individ- JS Upper Xingu linguistic area 425 M onolingualism is not rare am ong the peoples o f the U pper Xingu. M any people, however, know, or a t least understand, m ore than one language due to inter­ tribal marriages,'' Since, before a couple moves to the husband’s village, they tend to stay in the wife’s village for a while (G alväo 1953), b o th have an opportunity of learning each oth er’s language, and the children learn their m other’s and their father’s language. However, exogamy is not required (unlike in the Vaupés: see ch ap ter 14). G enerally speaking, the linguistic situation in the U pper X ingu can be characterized as that o f 'passive bilingualism ’ (Em merich 1984, Basso 1973). Yawalapiti and Trutnai are the only two m ultilingual groups. By the end o f the nineteenth century, the Yawalapiti were already a small group. They did not have a village o f their own but lived am ong the Kuikûro, W aurá, M ehinaku, Aweti and K am aiurá. In 1950, the m embers o f the expedition Roncador Xingu convinced them to build their own village (Cowell 1973: 227), and their numbers increased. This was also due to intertribal marriages, and resulted in the integration o f numerous speakers o f K am aiura and ICuikùro into the Yawalapiti com m unity (Viveiros de C astro 1977:69; Mujica 1992). As a result, the linguistic sit­ uation in the Yawalapiti village is more com plicated than elsewhere in the Xingu. The Yawalapiti language continues to be the m ark o f the tribal identity; however, every Yawalapiti knows another language. Mujica (1992) reports that m ost children have at least one non-Yawalapiti-speaking parent, and they prefer to speak a language other th an Yawalapiti. As the result o f this, Yawalapiti is endangered - of the 130 inhabitants o f the Yawalapiti village, only 13 are fluent in the language (1992; 7). Shnilarly, the Truniai* have been in decline since the end o f the nineteenth century (see M urphy an d Quain 1955: 91T.). In 1938, many were bilingual in K am aiura. Nowadays, all the Trumai speakers are m ultilingual, and they tend to know more uals and groups, in which verbal and non-verbal codes are present but not necessarily shared by the total set of participants. The use ol' these various codes results in intersect­ ing lines of communication by which any message can be ultimately transm itted to, and understood by, any local group or individual’ (1973: 5). According to Basso, there are two types o f coramimiciition. Interpersonal communication involves the use o f language, while a non-personal (ceremonial) communication is linked to non-verbal deviccs. The frequency o f intertribal marriages differs from group to group. The Kuikûro tend to m arry other Carib-speaking groups, o r Yawalapiti, or M ehinaku, or, rarely Waurá o r Aweti (Franchetto 1986), while Kartiaiura tend not to m arry K.uikiiro. The Trumai, the most recent arrivals to the Upper Xingu, also have certain differences from other peoples of the area; they do not take part in the kwaryp ceremony and do not have the same food prohibitions as other groups. According to Trumai traditions they came to the Xingu from the far southeast, where their ancestors did not know m anioc, and did not use bows or hammocks. They were said to have had long hair, like Suyá (Jê), to hunt tapir and to tie up their penises. The latter practice is said to have been abandoned after arrival in the Xingu area. Note that when von den Steinen first visited them, Trumai women did not wear the pubic cover iihiri (they do so now). This suggests that Trumai must have arrived in the Xingu area n ot long before von den Steinen visited them. 426 L ucy Seki Portuguese than do other X inguan peoples. Children tend first to learn languages other than Trum ai (dainiing that Trum ai is too hard). Trum ai is likely to become endangered in the near future (G uirardello 1992: 20). 4 IN C IPIE N T AREAL DIFFU SION I N THE U PPER XIN OU Given the short tim e depth and limited multilingualism, one would not expect to find many traces o f linguistic diffusion in the U pper X ingu. Since the Arawak peoples were probably the first to have setttled in the area being followed by C arib and Tupi, and then by Trumai - one would expect the great­ est areal diffusion to be between Carib, A raw ak and Tupi, and from these languages into Trum ai.’ The Jê groups listed in table 15.2, and Jiiruna (Tupi), are marginal to the U pper Xingu, and one would not expect to find much areal diffusion into these languages. There is evidence o f lexical borrowing, mostly from K aniaiurá into Trumai and into Yawalapiti. A ccording to M urphy and Q uain (1955; 8-9), ‘the songs o f two m ajor Trumai ceremonies, both o f which are practiced also by K am aiurá, show a preponderance o f ÍCamaiurá words and phrases’, while words o f C arib and Arawak origin ‘are frequent in other songs’. A m ong lexical loans from Kam aiurá into Trumai are Uiwawi ‘m at’, y'a ‘calabash’, yawj ‘pequi oil’, and wyrapy ‘vulture’. According to F ranchetto (1986; 126) Kuikiiro myths and ceremonial discourse are reported to bear an A raw ak influence. There is a certain am ount o f lexical diffusion from K am aiurá into Suyá, a m ar­ ginal m em ber o f the U pper Xingu culture area, e.g. Suyá k a m î‘m ixture o f water and manioc b read ’, parana ‘river’, wom i ‘corn’ from K aniaiurá kawî, parana, awasi (e f Seeger 1978: 161). Some languages o f the U pper Xingu share certain traits which are not found in genetically related languages outside the area and may have developed as the result o f areal diffusion. Four of these traits are; (A) The development o f the vowel phonem e i in Yawalapiti Vowel systems in all the Xinguan languages include ;, a, i and «.'® The following ch art shows vowel correspondences between proto-A raw ak, W aurá and Yawalapiti (Seki and Aikhenvald forthcoming). The phonological changes in * The Trumai, being the latest arrivals (and among the weakest tribes in the Upper Xingu), would not be expccted to be the source o f difTusion and borrowing. Interestingly, a Trutnai word misu ‘water’ is used in a ninnber of names o f rivers in the northern part o f the Upper Xingu, such as Auaiá Miçu, Suyá Miçu, Manissauà-Miçu. The vowel e is found in W aurá-M ehinaku, and Kuikùro, Awetí and Kam aiurá have both e and 0 . 427 15 Upper Xingu linguistic area Table 15.4 Vowel correspondences in Arawak proto-Arawak W aurá Yawalapiti *i ♦e ♦i i e } i i i pre-Yawalapiti have to be ordered as follows: (i) (ii) In Yawalapiti the two high vowels merged; then e shifted to L (See table 15.4.) The recreation o f the vowel t {after the original t developed into Í) was possibly due to areal pressure, since i is present in every other X ingiian language. (B) Shift p > h : from Arawak to C arib and Tupi'-G uarani A feature shared by genetically unrelated languages in the U pper Xingu region is the shift p > h . In Kuikiiro, p has become h in intervocalic position {Franchetto 1995: 55). K am aiurá (chapter 5) and proto-T upi-G uarani *pw has become hw or A. In Y aw alapiti,/>í- ‘2sg prefix’ becomes hi- if the following root begins with w orj>. This change is not found in other Tupi-G uarani o r C arib languages. It is n o t unfrequent in A raw ak languages outside the region. For example, p has becom e h in Pareci, the A raw ak language which is genetically and geographically closest to X inguan A raw ak, and also in a num ber of N orth A raw ak languages (e.g. Bahw ana, A chagua, Yavitero). One may hypothesize that the source of this diffusiooal feature could have been the Arawak ianguages. (C) CV syllable structure: from Arawak into C arib All A raw ak languages, including those spoken in the U pper Xingu, have CV syl­ lable structure. C arib languages spoken outside the U pper Xingu have CV and CVC syllables, while the Xinguan C arib language (w ith its four dialectal varieties) has only CV syllables. It may well be th a t this feature is the result o f Arawak influence. (D ) Loss o f genders in X inguan Arawak: a feature diffusing from C arib and Tupi into A raw ak The Arawak languages of the U pper Xingu have lost the gender distinction between masculine an d feminine in j;joss-referencing markers. W aurá-M ehinaku has also lost genders in independent pronouns. We may hypothesize that this loss could have occurred as the result of contact with Tupi and C arib languages in the Xingu, none o f which have genders. These tra its are indicative o f an early stage o f areal diffusion in the U pper Xingu region. T h e exact am ount and the character o f lexical borrowings and other 428 Litcy S eki diffusional features in the languages of the U pper Xingu require further studies, b o th descriptive and comparative. 5 CONCLUSIONS The U pper X ingu culture area is relatively young - 150-200 years old. The time d epth has n o t been long enough for a linguistic area to develop; this is the reason why, in spite o f cultural similarities and a certain degree o f multilingualism , ‘strong linguistic indicators o f long-term m utual influence are not abundant here’ (Cam pbell 1997; 340). However, the existing multilingualism , lexical borrowings a n d a few shared structural features are suggestive o f an incipient linguistic area. This illustrates well the problems which m ight be caused by assuming ‘that culture areas and linguistic areas will coincide’ (Cam pbell 1997; 340). T he Upper X ingu area is thus reminiscent o f the much-discussed case o f the G reat Plains region o f the U nited States - a recognized culture area, but n o t a linguistic one. Similarly to ths G reat Plains, the U pper X ingu area has developed recently; in both the U pper Xiogu a nd the G reat Plains there has not been sufficient time depth (and, possibly, social stability) for a true linguistic area to be established (D oug Parks p.c.; see also Sherzer 1973; 773; Bright and Sherzer 1976: 235). C ontacts with white people are now increasing in the U pper X ingu region. In the 1970s a m ajor highway was constructed through the N ational Park. Indigenous neighbours who did n o t originally belong to the U pper Xingu culture complex are undergoing strong influence from U pper Xingu culture. This goes together with the mobility o f tribal groups who freely move about the whole X ingu N ational Park. The scope o f what used to be the Upper Xingu culture area is thus increasing, and the situation is less and less stable. Thus, it is hard to predict whether recent social developments will lead to further areal diffusion, and what direction this diffusion m ight take. BIBLIO GRA PHY Agostinho da Silva, P. 1993. ‘Testemunhos da ocupaçâo Pré-Xinguana na Bacia dos Formadores do Xingu’, pp. 233-87 o f Coelho 1993. Baruzzi, R. G., Rodrigues, D. A., Mendonça, S. e colaboradores. 1995. 30 anos Xirtgii relatorio de aihidades da Unidade de Sm'ide e Medicina no Parque Indigena do Xingu. Sào Paulo: Eiæola Paulista de Medicina, Universidade Federal do Estado de Säo Paulo. Basso, E. B. 1973. ‘The use of Portuguese relationship terms in Kalapalo (Xingu Karib) /J Upper Xingu linguistic area 429 encounters: changes in a central Brazilian communications network’. Language in Society, 2.1-21. Becquelin, P. 1993. ‘A rqueologia X inguana’, pp. 223-32 o f Coelho 1993. Bright, W. and Shetzer, X 1976. ‘Areal phenomena in North-A merican Indian languages’, pp. 228-68 o f Variation and change in language. Essays by W. Bright, ed. A. S. Dil. Stanford: Stanford University Press. Campbell, L. 1997. American Indian languages. The historical linguistics o f Native America. New York: Oxford University Press. Coelho, V. P. 1993 (ed.) Karl von den Steinen: urn século de antropologia no Xingu, Sâo Paulo: E ditora da Universidade de Sâo Paulo. Cowell, A. 1973. The tribe that hides from man. London: The Bodiey Head. Dole, G. 1993. ‘Homogeneidade e diversidade no Alto Xingu vistas a partir dos Cuicuros’, pp. 375-403 of Coelho 1993. Emmerich, C. 1984. ‘A lingua de contato no Alto Xingu. Origem, forma e funçào’, Ph.D. Thesis, Universidade Federal de Rio de Janeiro. Eramerich, C. and M onserrat, R. M. F. 1972. Sobre a fonologin da lingua Aweti (Tupi), Boletim do Museo Nacional Antropologia, 25. Rio de Janeiro. Franchetto, B. 1986. ‘Falar Kuikuro. Estudo etnolingüístico de um grupo Karib do Alto Xingu’, Ph.D. thesis, Museu Nacional, Rio de Janeiro. 1995. ‘Processos fonologicos era Kuikuro: uma visâo auto-segmental’, pp. 53-83 of Estudos fonologicos das linguas indigenas brasileiras, ed. L. Wetzels. Rio de Janeiro: Editora Universidade Federal do Rio de Janeiro. Gaivâo, E. 1953. Ciiltura e sistema d e parentesco das tribos do Alto Xingu, Boletim do Museo Nacional Antropologia 14. Rio de Janeiro. Guirardello, R. 1992. ‘Aspectos da m orfossintaxe da lingua Trumai (isolada) e de seu sistema de m arcaçâo de caso’, MA thesis, Universidade Estadual de Campinas, forthcoming. ‘Trumai reference gram m ar’, Ph.D. thesis. Rice University. Heckenberger, M. 1996. 'W at and peace in the shadow o f empire: sociopolitical change in the U pper Xingu o f Southeastern Am azonia, AD 14OU-20Ü0’, Ph.D. thesis, University o f ■Pittsburg. ‘ Medeiros, M. C. 1990. ‘Posse nominal na lingua M ehinaku’, paper presented at the Annual meeting o f Asociacion de Linguistics y Filologia da America Latina, Campinas. Mujica, M. L 0 . 1992. ‘Aspectos fonologicos eg ram aticaisda lingua Yawalapiti (A ruak)’, MA thesis, Universidade Estadual de Campinas. Miinzel, M. 1971. ‘Medizinmannwesen und Geistervorstcllungen bei den Kam ayurá (Alto . _ Xingu Brasilien)’, Ph.D. thesis, Johann W. von Goethe Universitat von Frankfurt. Murphy, R . F. and Quain, B. 1955. The Trumai Indians o f Central Brazil. New York: J, J. Augustin Publisher. Oberg, K. 1953. Indian tribes o f Northern M ato Grosso. Publication 15, Institute o f Social Anthropology. Washington, D.C.: Smithsonian Institution. Richards, X 1973. ‘Dificuidades na análise de possessâo nominal na lingua W aurá’, Série Lingfiistica,\.l\- 2 9 . 1977. ‘Oraçôes em W aurá’, Série Lingüistica, 7.141-84. 1988. ‘A estrutura verbal W aurá’, Série Lingiiistica, 9.197-218. 1991. ‘Presentacion y gram âtica de los pàrrafos en la narrativa W aurá’, Revisla Latinoamericana de Estudios Etno-Lingiiisticos, 6 .157-82. Seeger, A. 1974. ‘Nature and culture and its transform ations in the cosmology and social orga­ nization o f the Suya, a Ge speaking tribe o f Central Brazil’, Ph.D. thesis, University o f Chicago. 430 Lucy S eki 1978, 'A identidade étnica como processo: os indios Suyá e as sooiedades do Alto Xingu’, Ámiário Äntropolágico for 1978, pp. 156-75. Seki, L. forthcoming. Gramálica da Lingua Kamaiwii. Campinas: Editora da Unicamp. Seki, L. and Aikhenvald, A. fortlicoming. ‘Arawak languages o f Xingu: a reconstruction’. Sherzer, J, 1973. ‘Areal linguistics in North America’, pp. 749-95 o f Current trends in linguiitics, vol. X, ed. T. Sebeok. The Hague: M outon. Viveiros de Castro, E. B. 1977. ‘Individuo e sociedade no Alto Xingu: os Yawalapiti’, MA thesis. Universidade Federal de Rio de Janeiro. INDEX OF AUTHORS A bbott, M. 2 4 -5 ,3 1 ,3 9 ,4 2 ,4 4 Adam , I,. 73 Adams Lichlan, P. 297 Adelaar, W. F, H. 19 Agnew, A. 317 Agostinho d a Silva, P. 419 Aikhenvald, A. Y. xxviii. 9, 65, 74, 77. 79, 83, 86-90, 92^3, 95,97-8. 100-1. 385-8, 3 93-5,397,399,4 0 i~ 7 .410,412 Albisetti, C. 171 Alicea, N. 307 Aliin, T. R, 77, 79, 83-4,93, 370 Alvarez, 1 77, 79,84,90, 100 Alves, P .M . n o Anchieta, J. de, see de Anchieta, J. Anderson, Lambert 318 Anderson. S. R. 179 Arciiia, O. 391 Arm ellada, C. de, see de Armellada, C. Aschm ann, R. P. 77, 312, 316, 370 Azevedo. G. M. C. 171 Baarle, P. van, see van Baarle, P. Baidi, P. 111 Baptistii. P, 77-8 Barbosa, E. A. 171 Rames. ,1. 2 1). 214, 219. 222. 252, 388 -9. 401 2 Barrientos, J. 131 Baruzzi, R. G. 417 Basso, E. B. 424-5 Becker-Donner, E. 362 Beeqiielin, P. 419 Bendor-Samuei, D. 152 Bcndor-Samuel, 1 T. 79,313 Bernardo de Nantes, R. P. Fr., 170 Bessa Freire, J. 387 Bontkes. C. 110, 121 Bontkes,W. 110, 119 Borges, M. d e N .d e O. F. 171 Borges, M .V, 171,177 Borgman, D. M. 345^ 6, 350-1 Boswood, J, 165, 171, 173 Braga, A. de O. 110 Brandâo Lopes, A. 255 Brandon, F. R. 110, 121 Braun, I. 110 Bright, W. 428 Brinton, D. G. 73,293,365 Brüzzi, A. A, S. 255,390,392 Bublilz,T. 171 Burgess, E. 82,95, 98 Burquest, D. A. 82, 100, 102 Burtch, B. 326, 331 B urtch,S.A .321 Burum, M. 110 Butler, N. E, 79, 82, 86, 99 Cabral, A. S. A. C. 129,131 Callow, J. C. 171 Camp, E. 366, 368-9 Campbell, L. 412,428 Captain, D. 74-5 Carlson, H. 363 Carlson. R, 42 Carson, N. 24-5 Caspar, F, 5, i !0, 115 Cathcart, M. 255. 258. 260. 263 Cavalcante, M. P. Í 71,179, 184 Chafl'atijon, X 72 Chagnon, N. .345 Cham berlain, A. F. 359, 365 Chandless, W. 5. 293^4 Chapm an, S. 294.301 Colbacchini, A. 171 Cole, R 10 Cook, D. M. 213, 215,218,220 Coudreau, H. 108 Cowell, A. 419,425 Créqui-M ontfort, G. de, see de CréquiM ontfort.G . Criswell, L. L. 213, 215, 218 432 Index o f authors Ctofts, M. HO, U 4, 116-20 Crowell, T, 171, 174, 176 Curnow, T. J. 370 d ’Orbigny, A. D. 359 Davis, I. 165, 172, 17S Day, D. 237 de A n c h ie ta ,! 125, 166 de Armellada, C. 25 de Créqui-M ontfort, G. 359, 365 de Goeje, C. H. 2 4 ,6 1 ,7 4 ,8 9 de la Grasserie. R. 227 de Kinch, P. 217.220^4 de Mattos, R. 171, 180 de Oliveira, A. E. 390 deSüuza, S. M, 171, 180 de Souza, T. C. 24,29, 33,60 de Tovar, C. L. 14^15,74-5, 341 Del Viena, D. 255 Derbysliire, D. C, 3, 15, 24-5, 28, 32, 38, 43,50, 5 3 ,5 5 -6 ,6 1 ,7 5 ,8 1 , 100,294, 301 Dietrich, W, 107, 110, 126-7, 131, 133^4, 137, 147 Dixon, R. M. W. xxviii, 15-16, 90, 295, 298, 302, 306 do Valle-Bentes, E, H. 354,359 Dobson, R. 141 Dooley, R, A. 110, 119 Dourado. L. G, 171, 18! Durbin, M. 25,27 Eastm an, E. 318,326, 333 Eastman, R. 3IS, 326, 333 Eberhand, D, 270 Edwards, W. F. 24-5 Ehrenreich, P. 12, 15, 171, 177 E kdahl.E . M .7 9 ,8 1 ,8 2 ,9 9 ,101 Emmerich, C, 24. 27, 110, 112, 419, '424^5 Estrada, il. R. 370-1,374 Everett, IX L. 354-6. 359-60 Eynde, E. van den, see van den Eynde, E. Facundes, S. 82, 96, 387 Fargetti, C. M. 110 Faron, L. C. 7 Fast, D. 321,323, ,326^8 Fast, G. 321,323, 32fr « Fast. R. 321,323,326-8 Ferguson, J. 212 Fortune. D. L. 171, 176-7 Fortune. G. 171, 177 Franclietto, B. 24, 27,29, 31, 36,40, 419, 425-7 Frank, P. 370, 372, 374 Franklin, G. 255,257,265 G a b a s ,N .J r ! 1 0 - l l, 113,115-16 Galiicio, A. V. 110 Galvâo, E. 252, 390,420,423-5 Giacone, A. 255 Gibson, M. 307 Gildea, S. L. O. 23-5,28, 30-4, 36-8,45, 48-9, 55, 57-8, 60-1 G irard, V. 25, 227, 343, 364-5 Gnerre, M. 312 Goeje, C. H. de, see de Goeje, C. H. Goldman, I. 389 Gomez, G. G. 345, 348 G om ez-Im bert, E. 212, 224, 389, 391-2, 404 Gonçalves,_C. H. R .C . 116 Gonzalez Ñáñez, 0 . 89 Graham, A. 110, 1J9 Graharo, S. 110, 119 Gralow, F. L. 397 Grasserie, R. de la, see de la Grasserie, R. Green, D. 9, 74, 83, 8 8 -9 ,9 3 ,9 5 ,9 7 Green, H. 89 Greenberg, J, H. 12, 15-16, 165-6, 327 Grimes, B. F. 25-6, 133, 387-8 Grimes, J. E .8 I, 101,368 Grinevald, C, 369 Gudschinsky, S. C. 165, 170-^1, 173, !76, 179, 181 Guedes, M. 171 Giiérios, R. F. M. 165, 170-1 Guirardello, R. 351,419,423,426 Giimperz, J. J. 389,392,4 ( 3 Hall,J. 171, 199 Hall, K. L.24--7, 29,34 Ham, R 171-2 Hanke, W. 110,362 Flarrison, C. 110, 119, 151, 155 H art, G. 313 H art, H . 313, 320-5,327, 330, 332,334 Hawkins, R. E. 24-5, 27, 38, 50 Heath, J. 412 Heckenberger, M. 419 Heinen, H. D. 377 Heinrichs, A, L, 354 Hemming, J. 6,107 Henley, E 24-5,377 H e n ry ,! 171 Hill, X H. 2+-5. 28-9 Hill, T. E. 2 4-5,28-9 Hinton, L. 362 Hoeller, A. 139 Index o f authors Hoff, B. 1 23-6, 28, 30, 37-8, 41, 50, 53, 55, 75 Huber, R .Q . 255, 370-1,386 Hugh-Jones, C. 390 Hugh-Jones, S. 0 . 390 Ibarra G rasso, D. E. 364 Jackson,X 252, 388 Jackson.W . S. 24, 27 Jefferson, K. 171 Jensen, C. 61,134-5, 142,144, 147-8,156-7 Johnson, O. E. 212,219 Jones, P. 207, 2 i 2 ,215-16, 218-19, 222. 224, 396 Jones, W. 207,212, 215-16,218-19,222, 224,396 Judy, J. E. 367-9 Judy, R. 367-9 Juliao, M. R. S. 139 Jusayù, M. A. 92-3 Kakumasu, J, Y. 9, 139 Kaufman, T. 1 2,23,25,60,73,166,392, 412 Kensinger, K. M. 227 Kern, B. 359-60 Kerr, I. 370 6 Key, H. 367 Key, M. R. 227, 364-6,368-9 Kinch, P. de 217, 220-1 Kinch, R. A. 217,220-1,397 Kindberg, W. 91 Kliimpp, D. 88,98 Klumpp, I 82, 100, 102 Koch-Grünberg, T. 72, 251-3, 362, 387, 390-1 Kochn, E. 24-5,29 Koehn, S. 24-5, 29, 42 ■Rondo, R. 371----Kondo, V. 371 Koopman, L. 171 Krause, F. 177 Kunike,H. 171, 177 LacluiiK.G. 17), 199 Ladefoged, P. 359 Landaburu,X 369-72, 374-6 Landin, D. X 110, 118 Lapenda,G . 171, 185 Larsen, T. W. 195 Larson, M. L. 308, 332, 334 Lathrap, D. W. 75 Lauriault, E. 237 Leach, I. 317 433 Leavitt, C. W. 24, 27 Leite, Y. 144, 157 Lemle, M- 134, 140 Lévi-Strauss, C. 269 Levinsohn, S. H, 212,219-20,255,258, 263 Liccardi, M. 367-9 Lima, S, T. R 171 Lizot, 1 3 4 1 ,3 4 5 Longacre, R, E. 223 Loriot. X 237 Loukotka.C , 14-15,65,74,107, 165, 170, 343-4. 357, 363-4 Lowe, L 278 Lyon. P. X 312 Maia, M. A. R. 171 Malone, T. 207,209, 2 U , 214, 386, 388-9, 391-2, 394-5,408 Mamiani, L. V. 170, 185, 194,197 Marlett, S. 297 Martins, S. A. 255, 389,394, 400,410 Martins, V, 255-6 M artius, K. F. P. von, see von Martius, K. R P. Masica, C. P. 412 Mason, X A. 12, 14-15, 74,165, 364-5 Mattéi-Muller, M. 24-5, 377 M atteson, E. 13-15, 73, 75, 82, 86, 88, 92, 95,98, 100-1 Mattos, R. de, see de Mattos, R. McLeod, R. A. 171,178,199 Medeiros, M. C. 81, 419 Meer, T, H. van der, see van der Meer, T, H. Meland, D. 171, 174-5, 180, 185 Meléndez, M, A, 84,98 Mello, A. A .S. 133 • M endonça, S. 417 Métraiix, A. 15, 165 Metzger. R. G. 212. 216, 218-19, 392 Migliazza. E. C. 25Í, 34Í-3, 345-7, 350, 362 Miller, M. 211.215 Miner, K. L. 45 Mitchell, V. 171,199 M ithun, M. 95 Mitrani, P. 377 Monod-Becquelin, A. 351 Monscrrat, R. M. E 110, 112, 294,419 M ontes Rodriguez, M. E. 318 Moore, B. X 255, 257,265 Moore. D. L. 110,112-14, 117, 119-21, 387 Morse, N. 212, 370 Mosonyi, J. C. 25 Mujica, M. L O, 72,417, 419,425 Mullen. P. J71 434 Index o f authors Munroe, P. 90, 93 Münzei. M. 420 Murphy, R. F. 420,424-6 Newton, D, 139, 144 Nimuendajii, C. 107-g, 165, 170,252-3, 354-5,359,386, 390,412 Noble. G. K, 15,65, 73,75 Obregon Muñoz, H. 377 Oliveira. A. E. de, jee de Oliveira, A. E. Oliver, J, R. 69, 72, 75 Olza Zubiri, J. 92 -3 Osborn, H. A. 377 Ott. R. 94,99 Ott, W. 94, 99 Otterloo, R. V. 24-^5 Palácio. A. R 171,173, 175, 180,187 Parks, D. 428 Pat te, M. 89 Payne, David L. 68, 73^-4, 7 7 -8 2 ,8 5 ^ . 88, 90, 93,96,98, 312, 314-15, 327 Payne, Doris L. 13-14, 24, 34, 42, 55-6, 61, 312, 315, 319-21,323-4, 328-32, 334 Payne, Judith 76. 79, 90, 94 Payne. Thom as E. 24,45, 61, 320-1, 323-4, 334 Peckhani, I 24^5 Peeke. C, 331 Pereira, D. G. 170, ÍS6 Pet, W. J. A. 92-3, 97-8, 100 Pike,E. G. 317 Pinto, E. 171 Pires, N. N. 357-8, 38? Pitman, D. 365 Popjes, J. 171, 181, 186, 196-7 Popovich, F. 170, 173, 176, 179 Popovich, H. 170, 173, 176, 179,197 Povvlison, P S. 3! 5, 320 Pulkim, G. K. 3 Quain, B. 420, 424-6 Queixalos, F. 370, 372-3, 375- 6 Ramirez, H. 73, 84. 94, 99, 342, 345-50, 402 Reed. J. C. 88 Reed, R. 255, 370-1, 386 Reichei-Dolmatoff, G. 390 Reid, H. 377 Reinoso, A, G. 98 Ribeiro, E, R. 171 Rich, F. 3 ! 7 - IS Rich, R 3 2 2 , 328-31,333 Richards, I 81, 88,95,419 Rivet, R 5, 13, 15, 253, 357, 359, 365 Rodrigues. A. D. 12,14-15, 61,73, 107-8, 110-11, 115, 125, 127-9, 132-4, 143, 147, 165, 170-1,179, 182, 199, 343, 386-7,417 Rodrigues, C. L. R. 110 Rodrigues, I. 354 Roniero-Figeroa, A. 377 Rouse, I. 65. 69 Rowan, O. 82,95.98 Ruiz de M ontoya. A, 126 Salazar, G. 25 Salser, J. K. 212,219 Santos, L. C. 171 Santos, M. 78 Schauer, J. 100 Schauer, S. 100 Schmidt, M. 171 Schmidt, W. 165, 365 Schuller, R. 12,365 Schultz, H. 171 Scbestyén, E. 170 Seeger, A, 423.426 Seki, L. 2, 9 ,7 4 ,7 7 .1 2 1 , 141, 170,417,415 426 Shafer, R. 74 Shaler, D .9 t-2 Sheldon, S. N. 354-5 Shell, O. A. 227 Shephard G. Jr, 83 Sher2er,J.411-12,428 Silva, A. O, 294 Silva, T. C. 170 Silvcrwood-Cope, P. L. 255, 389 Sluyters, W, 170 Smith-Stark, T. 412 Soares, M. F. 144, 318 Sorensen, A. R 2Î0, 388, 393. 411 Souza, S. M. de, see de Souza, S. M. Souza, T C. de. see de Souza. T, C. Spearman. A. M, 7 S ta r k ,! E, 312 Sieinen, K. von den, see von den Steinen, K. Steward, J. H. 7 Storto, L. 110-13 Stout, M. 171, 179 Strom, C. 207,212-13,219-21 Stute, H. HO Suarez, M. M. 373 Suzuki, M. D, S. 110, 120 Sw adesh,M . HO Swift, K, E. 93 435 Index o f authors Tastevin, C. 5, 13, 15, 253 Taylor. D. M. 72^5,84, 88, 93,95 Taylor, G. 395 Tcixeira, J. B. 171 Thiesen, W. 316, 319, 324-5. 328-9, 333 Thom ason, S. G. 392 Thom son, R. 171, 179 Tobar Ortiz, N. 373, 375 Tocantins, A. M. G, 108 Tovar, A. 14-15, 74..5, .341 Tovar, C. L. de, see de Tovar, C. L. Tracy. F, V. 88, 95 Tremaine, S. 171 Tripp, R. 325 Trudell, B. 308 Tuggy, S. 325 U rban. G. 171 Valadares, S. M. B. 406 Valenti, D. 72, 74 Valenzuela, P, 73 Valle-Bentes, E. H. do, see do Valle-Bentes, E. H, van Baarle, P. 88 van den Eynde, E. 312 van der Meer, T. H. 110, 114. 120-1 van der Voort, H. 362-3 Vasconeelos. I. 362 Venlurelli.J. 171 Vigna, D. del 255 Viveiros de Castro, E. B. 425 Vogel, A. R. 302 von den Steinen, K. 72-3, 108, 351 von M artius, K. F. PT2. 108, 171, 354 Voort, H. van der, see van der Voort, H. Wallace, A. R. 72 Wallace, R. 24. 27 Waller, H. 171 Wallin. R. 77-8 Waltz, C. 395 Waltz, N. E. 207,218,225, 395 Weinreich, U, 389 Weir, E. M. H. 251. 255,257-63,265, 376, 387, 399-400 West, B. 210, 214-17,408 Wetzels, L. 170, 179 Wheatley, J. 24 Wheeler, A. 207, 224, 398 Whisler, D. 396 Whisler, J. 396 W iesemann, U. 171-2,185 Williams, J. 24-5 Wilson, P. J. 82, 98-9,101 Wilson, R. 389,392,413 W ipio,G .315 Wise, M. R. 74-6,82, 8 5 -8 ,9 0 -4 ,9 6 , 100-1, 326-7, 331 Zam poni, R. 90 INDEX OF LANGUAGES AND LANGUAGE FAMILIES A’ukré, see Kayapô A-*l?uh'n?e-'ka‘ta-'ki2le%u2 269 A’làn'te-îsu2 269 Achagua 70. 72, 427 Aché, see Oimyaki Achiial, see Achuar Acliuar 307, 312,321, 323. 325, 327-8. 331 Achuar-Shiwiar 309 Aguajun, see Aguaruna Agiiarima 307-9, 312-14, J32, 334, 370 Aikaná 341, 343, 362-3 Ajurii, see Wayorô Akawaio 24-5 Akroá 167 Akuriyô 26 Akwáwa 131 Algonquian 11 Almosan-Keresioiiau 327 Amacacore, see Iquito Amanaye 132 Amapari 132, 141, 143-t, 149,157 Atnarakaeri 307, 311-12 Amarizana 71, 387 Amawaca 229-30. 233, 240 Amondawa 132 Amuesha 68. 73-^, 90-1 Anambé 132, 139. 145 Anauya 70 A ndoa 309 Andoké 344, 370-2, 374-7 Anodöb, see Nadêb Anumania 419, 423 Añun, see Paraiihano Apalat 24-5,43, 140 Apiaká 132 Apinajé 167,171-2 Apolista 67 Apuriná 13, 68 Arabela 307,309, 312, 317-18, 320-2, 328-31,333 Araiku71 A ranâ, see Krenák A raona 344, 365-6 A rara 23,26, 109, 343 Arasa 344 Arasairi 311 Arawá ramily 292-306 Arawà language xxv, 5, 10,12-15,61, 292-1, 296-7, 301, 303 Arawak family xxiv, 1,4 -5 ,1 2 -1 5 , 17-19, 23, 25. 65-102, 211-12, 222, 252-2, 260, 312, 364, 370, 385-6, 389, 391^2, 396-7, 399,411,419-20,423,426-7 Arawak language, see Lokono Araweté 131 A razaire229 Arekuna 24-5 Arikapii 343, 357 Arikém 108-12 Aroä 69 A r u á 109 Aruan. see Aroà Ashaninca 13, 68 Ashaninka, see Ashaninca Asheninca 68, 80 Asiirini 138 Asiirini o f Tocantins 131.133,138,140, 142-3, 145. 147, 149-50, 153-4, 156, 159 Asiirini o f Xingu 131 ASorai, see Wapishana Atsawaca 229 Aueti, see Aweti Auetö, see Aweti Aushiri 309 Aitslronesian 11 Auxira, see Aushiri Avá 128, 130-1 Awa Pit 370 Awake 20, 343 Aweti 107-10, 112-14, 119-20,419-20. 423-6 Index o f languages and language fam ilies Ayanene 7 0,96 Aymara 9, 364 Bahwana 71. 73,81,427 Bakairi 23-4 Bakuén, see Krenák Banawá xxv, 293-4, 296-7 Baniwa 6 5 ,7 5 ,8 i - 2 , 251-2, 386-7, 389-91, 394-6, 399-403,406-7,410,413 Baniwa o f G uainia 65,70,386 Baniwa of Içana 65, 70. 72, 74. 82,96,251, 386,391-3,403,406 Baniwa o f Yavita 70 Bara 207, 209, 386, 388 Barasano 207, 209,211-12, 215-16; 218, 222,224, 286, 394, 396 Barbacoan 370 Bare 12, 65. 70. 386-7, 393, 395, 403,405, 407 Bauré 67 Beiço de Pau, see Tapayuna Black Carib, see Oarifuna Bora 74, 76, 307, 310, 312, 316-20. 324-5. 328-9. 333. 371 B ora-M uinane 310, 312 Borôro 165-6, 168-9. 173-6, 182-3, 188, 191-3, 196 Boruborà, Me Puruborà Botocudo, se€ Krenâk Bùe, see M urui Witoto Cabiyari 70 Cabre 70 Cacataibo. see Cashibo Cahuapana xxiv, 307-9. 313.318. 320-1, 324-27. 329, 334 Caiiana, see Sape Camannawa 229-30,245 Cam pa 13,68,72,89-9) Canam ari 229, 233 Candoshi 307, 311-12, 318, 320, 325 .Candoshi-Shapra 310 Canela 165,187.190, 192.195,197 Canela Apanyekrâ, leeT im bira Canela Ram kokamekrà, see Timbira Cangiti. see Apurinà Canishana 364 Canoeiro, see Avá C anpo piyapi, see Chayahiiita Capanawa 228-30, 233, 237-9. 241, 243, 245-8 Caquetio 69, 72 Caquinte 68, 93 C arapana 207, 209,212, 216, 218-19 Carib xxvi, 1 ,4-5, 12, 17,22-61,74-5, 151, 181,258,345,417,419^20, 422^7 437 Caribbean Arawak 74,84 Cariff, see O arifuna Carijona 24-5 Cariyai 7Î Carnijo, see Yatê Carru, see Baniwa o f Içana C arütana, see Baniwa o f Içana Cashibo 229-30,243 Cashinaw~d xxiv, 228-30 Catapolitani, see Baniwa of Içana Cavineña 344, 365-9 Cawarano 309 Caxiniti 67 Cayuvava 344, 367 Central T ucano 207, 211, 219, 386, 389,406 Chacobo 229-32,240-1 Cham a, see Ese Eja Cham icuro 68 Chandinawa, see Sharanawa Chané (Arewak) 67, 79, 81, 88 Chané C rupi-G uarani) 129^30 Chapacura 10, 14, 341, 343, 358-61 Chapara 310 Chayahuita xxiv, 307-9,313, 320-7, 330, 332,334-5 Chebero, see Jebero Chibcha 12,370 Chicano, see Hoti Chikào, see Txikâo Chimane 227 Chiquitano 364 Chiquito 166 Chiriana, see Bahwana Chiriguano 127, 130, 133, 159 Chiripà, see Nhandéva Chiripuno. see Arabelÿ Chitonawa 229-30 Chiwaro, see Jivaro Choco 370, 394 C h o lo n 3 ll Chontaquifo 68 Cinta-larga 109 Cofân-M akû 251 Comabo, see Cashibo Conibo, see Shipibo Coroado 168 Coropô, see Koropô Cubeo 209.212, 218-19, 222, 3 8 ^ 9 1 ,4 0 6 Cuiba 261. 344,370-7 Culina. see Kulina Culino 229 Cuniba 68 Cushitineri 68 Dâw 2 52-5,257, 259, 386, 394. 3 9 ^ ^0 0 , 4 0 4 .406-7,410,412 438 Index o f languages and language fam ilies Dekwana 24-5, 34, 345 Deni 292-305 Desana. see Desano Desano 207,209, 211-12,215, 252, 386-7, 390, 394, 396, 398 Digiit, see Gaviâo Dravidian 11, 392 Dukaiya 312 Dutch 23 Dzawi-nai 70, 96 Dzubukuá 168, 170 East Tucano 74, 78, 207-9, 211, 219. 223, 257, 265, 373,384-90, 392-8,402-8, 410-43 Easterti Boroto 168, 171 Etluriit 207 Emeriilon 132, 151 Etiiivvenê-nawê, see Salutnâ English xxiv, x.xvi-xxvii, 10-11, 20, 23,44, 57, 93, 126, 157, .356, 376 Equatoriai-Tucano 15 Ese Eja 344, 365-6 Ese?cjja, see Ese Eja Etwet, see Krenák French 7, 23 Fulnio, see Yatê Gae, see Andoa Galibi. see Carib Garifiina 69, 72, 74 Gavtäo 109-14, 118-21 Gaviâo Parakatejé, see Timbíra Gaviâo Piokobjé, jee Timbira Gaye, see Andoa Ges-Tapuya, see iè Gorotire, see Kayapo Guaharibo 341 G uahibo4, 85,261,344, 370-7 Guaicurii 4 Guajà 132 G uajajàra 131, 138--9, 141, 143-5, 147, 149-54,156, 160 Guajibo, see Guahibo Guajiro 69, 72, 74 Giiamo 14 G uana 79, 81, 88 Giianaiio 386-7, 390, 395-6 G uané 67 G uarani 19, 125-31, 150 Guarao, see Warao G uarauno, see Warao Guarayu 129-30, 139, 141, 144, 154 Giiareqiiena 70, 386 Giiariba, see Nadëb G uarû 70 G uatô 165, 168-9, 171, 173-5, 180, 182-3, 186-90,192-3, 198 Guayabero 344, 373, 375,377 Guayaki 128-30 Guerén 167 H a ’hâiT’te^su* 269 Ha-lot^te^su* 269 Kàhâhàe, see Pataxô Haliti, see Pared Hatnacore, see Iquito Harakm bet 14, 74, 307, 311-12, 318, 320, 324-5, 327,329 Hate, see Harakmbet H ianákoto 26 Hivaro, see Jivaro Ilixkaryana 24-5, 3 7 ,40,43 Hodï, see Hoti Hohôdene 70, 77, 96 Honikoin, see Cashinawa Hora. see Jorá Hoti 20, 344, 377 Huachipairi 68 Huachis 359 Huarayo, see Ese Eja H uari, see Aikaná Huitoto, see Witoto H upda 252-4, 259, 370, 386, 394, 399,404, 406-7,412 latê, see Yatê Ibo’tsa3IO Ignadano 67,81 Iköro, see Gaviâo Ikpeng422 Iftapari 68, 81 Indic 11 Indo-Aryan 392 Indo-European 11, 73, 388 Iñeri 69, 75 Ingain 167 Ingariko 26 Inonawa, see Sharanawa Ipeka, see Baniwa o f Içana Ipurinâ, see Apuriná Iquito 307, 309, 312, 318, 324, 326, 329, 333 Irantxe 20, _M3 Irish Gaelic 11 Isconawa 225. 245 Island Carib 69, 72, 74-5 Itogapuk, see Káro Itonam a 344, 367-9 Itucali, see Urarina Izoceno, see Chané Jabuti 166, 341, 343, 357-8 .laikô 167, 171 Index o f languages and language fam ilies Jamamadi xxv, 12, 293 4, 296^7, 300, 303-5 Japreria 25 Jaraw ara xxv, 293 4 , 297^306 Jari 132, 138, 144 Jaru 359 Javaé, see Karajá Javvaperi, see Waimir i-Atroari J ê 4 , 12. 18,61,165, 167-9, 172,178, 180, {83. 185,187, 190, 192, 195-7,419, 423,425-6 Jebero 308-9, 313, 325 ■leoromitxi, see Jabuti Jitn a, see Macaguán Jivaro xxiv, 307, 309. 312-15, 318, 320-1, 324 7, 329, 332, 334 Jorà 128-30 Jiidja, see Jum na Juma 132 Juruna 108-10, 112-14.119,421,426 Ka’lu’whaPko’te^su^ 269 Kabori, ire Nadeb Kabori, see Nadëb K a ija n a 7 i Kaingátig 165, 167. 171-2, 178, 180-5. 193, 196 Kaiwá 128. 131, 144 K akua 252-5, 260, 386, 394, 399^00, 412 K a la p alo 2 3 ,26,60,419-20 Ka1iBa,s<;eCarib Kam a, see Daw K am aitirá 72, 132, 141-2, 419-20, 42.Î-7 K am akà 165, 167-70,188 Katnayurá, see Kamaiurá 168 •68 Í343 Kandoshi, see Candoshi K annada413 K anoé 20 Kapong, see Akawaio Kapoxo 167 Kijrajá 165, 168, 171, 176-8, 182, 186, 188, 192 K araraô, see Kayapö K arihona, see Carijoiia K aripuna 132 K aripuná 229 Kariri 165-6, 168-70,173, 175, 184-5. 188-94.196. 198 K aritiána 109^ 15,118-20 K aro 109-10, 112-16, 119-20 Kashiiyana 24-5 K asupá, .MS Aikaná Katawixi 343 K atukina 20. 229, 343 K atukina do Biá 343 439 ; Cavineña see Caviiieña 132 Kaxariri 229 Kaxuyana, see Kashuyana Kayabi 131,136, 139,141-2, 148-9, 151, 156, 160,422 Kayapo 165, 167,169, 171, 178-9, 189-90 Kayapwe, see Zaparo lCepkiriwát 109 KPtäu-'lhu^ 269-70 Kikreturn, see Kayapo K in ik m a o 6 7 ,7 9 ,81,88 Kipeá 168, 170, 173,175, 180, 182-3,185, 188-97 Kiriri, see Kipeá Kisambaeri 311 Koaia 343. 362-4 Kokám a 129,131 Kokamiya, see Kokáma Kokraim ôro, see Kayapo Koreguaje 20 9 -1 3 ,2i5, 2 1 8 -Í9 ,222, 224, 394,406 Koropo 165, 168 Kotoxé 167 Kovare[ka], see O tuke K-ozarini, see Pared K raho, see Timbira Krekmùn. .see Krenák Kren-akarôrc. see Panarà Krenák 165, 167-70. 188, 192 Krenakore, see Panará Krenjé, see Timbira K rinkati, see Timbira K.ubenkrankén, see Kayapo Kubeo 252 Kuikuro xxvi, 23-4, 35-6, 40,419-20, 425-7 K-ulina 5. 13 4 5 , 292-305 Kumanaxo, see Kapoxo Kum andene 70, 77, 96 Kuna 370 Kurripako 65, 70, 74 ,7 7 ,9 6 ,2 5 1 , 386 Kuruáya 108-9 K u ry m in a[k al,.w O tiik e Kustenau 12. 67,419, 423 Kuyawi 252-5 Kwaza 343, 363 Latin 11 Layana 67 Leco364 Lingua G eral 7, 72, 127, 130, 252, 370, 377. 387, 391,.396 Llagiia, see Yagua L la ru ru ,s e e Yaruro Lokono 69. 72-5, 89,258 440 Index o f languages and language fam ilies M abenaro 344 M acaguán 344 M achiguenga 13, 68 M achoto, see Itonam a M aco-Piaroa 251 M acro-Chibchan 370 M acro-Jê 12, 15, 18. 165-201 M acro-Karib 12 M acro-Tucano 15 M acto-Tupi 12 M acuna 209,212, 224, 386, 388 M acuxi, see Makushi Madi 13, 292-8. 300-5 M adiha, see Madija M adija 294 M ahibarez, see Nam biquara M aina. see Achuar-Shiwiar M aipure 14,70. 72-3 M akiritare, see Dekwana Miiko, see Cotan-M akii Makoni 167 M a k ù x x v ,4 -5 ,9 , 12, 17-18, 95, 172, 251-65, 385-7, 389-90, 392-400, 404-6.410-12 M áku 251, 341, 343, 361^2 M akunadöbo, see Nadêb M akurap 109-10. 112 43, 357 M akushi 24-5. 35-7,40.43 Malalt 165,167 M amainde 269-71, 274-6. 279-^80 M ambyuara, see Nam biquara M a E ao7l,387 M andawaka 70. 387 Maniteneri, see Piro Manitsawá 108-9,419,423 Manx 11 Mapidian, see Mawayana Mapoyo 26 M arathi 413 Marawan 69 M ariaté 71 Marinawa, see Sharanawa M aropa, see Reyesano M arubo 229 M asaká, see Aikaná M asakali, see Maxakah’ M asakará 167 Mashko-Piro 68 Mastanawa 229 Masiibi 357 Matipii 26,417,419-20, 423 M atoinahä, see Canamari M atses227.229-31 Maué, see Mawé Mawaca 387 Mawakwa. see Mawayana Mawayana 69 Mawé 107-10. 113-14, 119-21 M axakali 165, 167-70, 173, 176, 179,182, 186, l8 8 - m 192, 194, 197 M axineri, 165,167-70,173,176, 179, 182, 186,188-90, 192, 194, 197, see a/so Piro M ayan 11-12 M ayna, see O m urano M ayoruna 68,229 M babaram 11 Mbyá 127-8, 130,138, 140-1, 143,153,158, 160 M ehinaku 67,419-20,423-7 M ekéns 109-10 Mêkîsedje, see Suyá Menién 167 M enkrangnoti, see Kayapo M entuktire, see Kayapo M erritne 165 M etuktire 421,423 M m w a W itoto 310 M inyàyirùn, see Krenák M iraña310,312 M ondé 108-10, 112, 114, 117-18 M ongoyô 167 M onoxô 167 More 343 M orique 68 M oronawa 229 M oviraa 344, 367-9 Moxo 67,73 M uinane 310,312. 316-17 M uinane Witoto 310, 312 M undé, see Aikaná M undurukû 95. 107-10, 112-20 MunicW 307,311 M unichino, see Munichi M iinkü, see Irantxe M ura 353-5 M ura-Pirahâ 4,341, 343, 353-7 Murafo. see Candoshi-Shapra M urui W itoto 307,310, 316-17, 321-2, 326, 331 Mynky, see Irantxe Nadêb 9 ,4 5 ,2 5 2 -5 .2 5 7 , 260, 376. 387, 394, 396, 3 9 8 ^ 0 ,4 0 4 - 6 ,4 1 0 , 412 N ahua, see Yoranawa Nahukw a 26,417,419-20,423 N akarothe 269-70 Naknyanùk, see Krenák Nakpié, see Krenák Nakrehé, see Krenák Nakyapniâ, see Krenák N am biquara 12, 269-91 Index o f languages and language fam ilies Naraviite419 Nawa, see Yoranawa Ne^su^ 269 N handéva 127^8, 130 Nheengatii 7, 127, 130,140, 371, 387 Ninam, see Yanam Nipode 310, 317 Nolcaman 229 Nom atsiguenga 13,68,79 N orth Arawak 9,69, 74-81,84^8,90-5, 97^101,258,260,264, 373, 385-7,390, 392-6, 399-401,403-7,410-11,413, 427 Ntogapid, see Káro Niikak 252-5, 386, 394,412 Nyepnyep, see Krenák Ocaina 74, 307, 310, 316 -17, 371 Olayé 165. 168, 171, 173, 181, 188 Olayé-Xavánte, see Ofayé Old G uarani 126^7, 130 Old Irish 11 Old W arekena 386 Omawá, see Kokánia Otnurano 309 Opaie, see Ofayé Opayé, see Olayé Orejôn 209-10.219 Oro Eu, íee W arf Oro Nao, see Wari’ Oro Wari, see Wari’ Oro Win 343 Otanabe, see Munichi Otí 166 O tíike !66, i68 Pacaas Novos, íee W ari’ Pacawara 229 Pacü, see Baniwa o f Içana Paez-3Î0 Pahenbakebo, see Capanawa Paiooneca 67 Paitér, see Surui Pajonal Carapa 68, 94 Pakaa No¥a. see W ari’ Palikur9, 69,74 ,8 1 .9 5 ,2 5 8 Pananahuri 309 Panará 167, 169, 171,181,190,422 Panare 24-5 F'ânerâ 207 Panháme, see Kapoxé Pano family 12, 61, 95, 227-48, 364-5 Pano language 227, 229 Panobo, w e W ariapano Pápiwa 207 Papuan 405 441 Paraguayan G uarani 127-8,130 Parakanä 131 Paraná, see Kaingáng Parauhano 69,89 Pareci 67, 81,427 Pareci-Cabixí 67 Pareci-Saraveca 67 Pared-X lngu 67,80, 83, 86, 90, 93 Parintintín 132, 136, 138-9, 141, 148,151 Parquenahua, see Yoranawa Pasé71 Passe 387 Patam ona 26 Pataso, see Pataxo Pataxé 165, 167 Pauliene, see Baniwa o f Içana Paum arî 12, 292-306 Pauna 67 Paweté 132 Pawishiana 26 Pawumwan 359 Pe b a 3 1 0 ,3 l2 Peba-Yagua 307, 310,312, 319,321, 327 Pedra Branca 168 Pemong 24-5 Perene 68 Peruvian Arawak 10, 93 Piapoco 70 Piaroa 344, 370 Pichis 68 Pinche, see Taushiro Pirahà 354-5 PiratapLiya 207, 209, 214, 386-8, 396-7 P ire 13,68 Piro-Apurinà 68, 80,90-1 Pisamira 209 Playero 344 Pojitxà, see Krenák Portuguese xxiv-xxv, 1 -2,7, 19-20, 23, 72, 125, 127, 140, 157,177, 184, 251, 254, 271,293, 310, 345, 358, 388, 391,426 Potén, see Krenák Poturu, see Z o ’é Poyanáwa 229 Pre-andîne Arawak 10, 90, 312 Pre-Jamamadi 14 Proto-Am erindian 14 Proto-Arawâ 12-14, 295-6.298, 300, 304 Proto-Arawà-Arawak 12 Proto-Arawak 12-14, 74-5, 77, 79-81, 84-5, 88,93. 393,403,426-7 Proto-Ashaninka 13 Proio-Carib 26,31-2, 37, 45, 60 Proto-East-Tucano 393 Proto-Jê 165. 172,178 Proto-Lokono-Guajiro 74 442 Index o f languages and language fa m ilies Pruto-M acro-Jê 199 Proto-M adi 13-14 Proto-M akii 255-7, 264 Proto-Newiki 13-14 Proto-North-Arawak 77 Proto-Pano 227,230 Proto-Pano-Tacana 365 Proto-Piro-Apuriná 13 Proto-Shiiar-Candoslii 312 Proto-Tueano 209, 211, 393 Proto-Tiipari 111-12 Proto-Tupi 107-8, 110-11, 129 Proto-Tupi-Guarani 61, 111-12, 119, 133-4, 141, 145, 147-8, 150-1, S53-4 ProiO“Tuyuca 214 Proto-West-Tucano 393 Proto-Western-Newiki 14 Proto-W itoto 77 Proto-Xingu 74, 77 PuGi-Uma, see iqiiito Puittave 20, 251,344.370 Pumé, see Yaruro Puri 165-7, 170,188 Puruborá 107-9,112, 114 Quechua 9, 19, 72, 74, 307, 309-10 Quipiu, see Jabuti Qiiiturran, see fquito Ram arám a 108-10,112 Remo 229 Resigaro 70, 73-4, 312, 370. 374 Retuama 392, 406 Retiiara, see Tanimuca Reyesano 344 Rikbaktsá 165, 168-9, 171, 173-4, 185, 188 Roiimaina, see Om urano Sabanes 269-70 Sabiiyá 168 Sakirabiàt 109 Sáliba 344, 370-4 Sáliba-Piaroa 251, 370 Sáliva-Maco 251 Salumà 26, 67 San Borjaiio, see Reyesano Sanskrit 11 Saniima 343, ,345-6, 350 Sape 20, 343 Sapibocona, íe í Reyesano Sapiteri 311 Sapoyá 168 Saraveca 67 Sateré-Mawé, see Mawé Scots Gaelic 11 Secoya 209-10, 212,219,406 Seeptsa, see Cholon Semigae, see A ndoa Shanindawa, see Sharanawa Sharanawa 227-31, 233-4, 245 Shebayo 69, 72 Shimacu, see Urarina Shimigae, see Andoa Shipibo 229-30,232-3, 237,241,245-6 Shuar 307, 309,312 Sikiana 26 Sikuani, see Guahibo Siona 209-11,219, 264,406 Sipája, see Xipáya Siriano 207, 209, 211-12, 386, 394, 396, 398 Siriono 128-30 Stuci 70, 96 Sorowaha 292-7, 300-4 South Arawak 67, 77-8, 80. 83, 85 -6, 90-1, 93^ . 100-1 South-Western Arawak 68, 77, 80, 86 Spanish xxiv, 1-2, 7, 19, 23,65, 72.93. 127, 131,229,252,293, 3 0 7 -i l , 345,369, .388, 391 Sramo, see Ítonarna Surui 109-10, 113-14, 118-21 Surui o f Tocantins 131 Suyá 167, 169, 171,421-3,425-6 Tacaoa61,227, 344, 364-7 Tain o 6 5 ,6 9 , 72 Taiwano 207, 209 Takunyapc132 Tanibopata-Guarayo, see Ese Eja Tanimuca 207, 209-13, 219-20, 222, 392, 406 Tapacura 359 Tapayuna 167, 169, 422-3 Tapieté 129-30 Tapirapé 131, 142-3, 145, 150, 157 Tapuya-zé, see Jê Tariana 12-13. 65, 70, 72, 74-5, 81, 252, 2f4. 370, 373-4. 384, 386-407, 409-13 Tatuyo 386. 392, 396,405 Taushiro 307, 311-12, 318, 329 Taveri. see Canam ari Tembé 131 Tenetehára 131, 141 Tenharim 132 'lerêna 67. 81-2, 88 Tiatinagua. see Ese Eja Ticuna 307, 311,318,329 Timbira 165,167, 178, 180-3, 186-7,190, 192-3, 195-7 Tinganeses. see Cholon Tiriyo, see Trio Torá .343. 359 Index o f languages and language fam ilies Toyoeri 3 11 Trinitario 67 Trio 24 Trumai xxvii, 341, 343, 351-3, 419-20, 423-6 (T)sanima, sec Sanuma isiiaaw i, see Chayahuita Tsuva419 Tu%a^'l7una’ko-’ie2su2 269 Tubarào, see Aikaná Tucano tamily 1. 5. 9-10, 12, 72. 85, 94, to i. 207-25. 252. 254-6. 258, 263-5, 370. 373-4, 386^06,408^10,412^43 Tucano language 209-10, 212, 215-47 Tukuna, see T icuna Tupari 5-6, 108-10.112-18 Tupixxv, 1,4. !2, 15, 17, 19.61,95, 107-22. 127-30, 165, 181, 269, 419-20, 42.3, 426-7 Tupi Austral 127, 130 Tupi do M achado 132 Tupi-G uarani xxv, 6-7, 9, 15, 19, 25, 61, 65, 95, 107-10. 114, 120, 125-61, 166, 170, 364. 387,419-20. 422,427 T u p in am bá6-7. 111. 125-30. 134 6, 138-9. 141-2, 149-54. 156, 160-1, 170, 252, 387 Turiwára 132 Tutxinawa 229 Tuyuca 207, 209-14, 216-17, 220-5, 252, 386-7,401 Txapacuran, see Cliapacura TxikSo 23-4,422 Txukaraniae 421 Txunhua-djapá 343 Ucayali 68 Umutiiia 168, 171 Uralic 11 U ra rin a 3 il,3 2 9 Urdu 413 Uro-Chipaya 14 Uru-eu-wau-wau 132 Uruak 343 U rubu-K aapor 132, 139-40, 144, 147-8, 151, 156 Urukii. see Káro Urunii 109 U rupá 343. 359 W a% a’li»te%u^ 269 Wa-’suh's?u^ 269 Wachipaeri 3 11 Wai Wai 24-5.43, 45.69 Wai-'su’ 269 Waik^ti^te-su^ 269 443 Waimaja 207,209,212, 386,405 Wainiare 67 W aimiri-Atroari 24-5 Wainuma 71, 387 Wambisa 307, 309,312 W anano 207, 209-11, 214. 218, 223 Waninnawa, see Camannawa W apishana 69 Waraicù, see Araikù W arao 344, 377 9 W arekena 65. 386-7,403,405,407,410 W arekena o f Xié 70, 393, 396 Warena, see Old Warekena W ari’ 343, 359-60 W ariapano 227, 229-30, 241 Warikyana, see Kashuyana Warisa, see Reyesano W aniwaru, see Hoti W aura 67,419-20,423-7 Wayampi 132-3, 135-6, 138,140-1, 143-5, 147, 149-52, 154, 157,160 W ayana 24 Wayoro 109-10 Wayyu, see Guajiro West Tiicano 207.209, 219, 264,392-4, 398, 406 Western Bororo 168 Wirafed 132 W irina71,387 W itoto xxiv, 74, 76-7, 79, 307, 312, 316-21, 324-7, 329, 331, 370-1,374 W itoto-0caina310,312,321 Xakriabá 167 Xam atauteri 342 Xambiwá 168 Xavánte 167, 171, 178, 184, 193, 195, 199 Xerénte 167, 171, 178, 180 Xetá 128, 130 Xetebo, see Shipibo Xikrin. see Kayapô Xingu 67, 75 .7 9 ,8 1 . 101.419 Xipáya 107-10, 118 Xirianá 341 Xiriwai, see Nadëb Xivaro. see Jivaro Xiwila, see Jebero Xokléng 167, 171,178, 185 Yabaana 70, 387 Yabuti, see Jabutí Yagua 307-8, 310, 312, 315, 318-21, .323-6, 328-32, 334 Yam eo310,312 Yanñaka 229 Yaminawa 227-32, 245 444 Index o f languages and language fam ilies Yanam 342-3, 345-6, 348, 350 Yanoama 341 Yanoami, see Yanomami Yanomae, see Yanomami Yanomam 343, 346, 348, 350 Yanomama 362 Yanomami xxv, 61, 95, 341-51, 388 Yanomami 342-3, 349-50 Yanomarai-Waicá 341 Yaruro 377^^8 Yatê 165, 168-71, 173-5. 180, 182, 185-6, 188, 192-3, 195 Yava,see Yagua Yavarana 26 Yiivitero 70 Yawalapiti 67, 72, 417,419-20, 425-7 Yawanawa 228-30 Yawarete-tapuya 96 Yegua,se e Yagua Yekwana, see Dekwana Yiporok, see Krenàk Yoranawa 227-3i Yuapin, see Yanito Yucuna 70, 392 Yudya, see Jum na Yuhup 252-4, 386, 394, 399, 412 Yukpa 25,60 Yumana 71, 387 Yuqui 6-7. 128-9 Yuracaré 364 Yuruna, see Juruna Yuruti 209, 212, 217, 220-1, 386,405 Yuwana, see Hoti Zapara, see Zaparo Zaparo 307, 310,312,317-8, 320-1, 327-9, 331 2c, see Jê Zo'é 128, 132 Zorô 109 SUBJECT INDEX absolutive 61, 96, 155-7, 187, 194, 241-^2, 320, 348, 366-7 accusative 9-10, 32, 35-7, 60-1, 90, 118, 121, 187.241-2,254, 263 active verbs xxvi, 32, 81, 83, 86-7, 117, 146-7, 154-5, 157-60,213, 351,407 adverbialization 57-8 ambitransitive verbs xxvi, 8!, 86,91,259, 300, 349, 355 applicative xxvii, 9. 80, 85, 91-2, 262, 301, 321,327,333,349,375 aspiration 77,79,271,295,313 auxiliary verbs 39, 55, 60, 89, 97-8, 213, 216,235,247, 301, 366, 379,405 body parts 9 ,4 1 ,4 5 . 82. 85.95, 116.149, 152,218, 243, 298,330, 346-9. 353, , 363. 369, 399 causative xxvli, 9 ,4 4 ,8 0 , 86,90-2,98, 120, 158, 195-6.244,247. 260-1, 301, 318, 327-8 347,349, 353, 356, 358, 361,369, 375, 379 classifiers xxviii, 1 ,8-10.42, 80, 82-4, 86-7, , !16, 185,218-19,221,225,258,280-!, 290-1, 319, 324, 3.32, 347-50, 362-4, 366. 369, 3 7 3 ^ . 379, 399^403, 407-10 collective num ber 38-42, 53, 84, 183, 259, 349, 360, 374 complement clauses 99, 101, 121, 265, 306, 350, 356, 358, 360, 377,405 complementizers 121 completive aspect 38, 93, 214-15, 260, 301-2, 326, 379 conditional clauses 52,278 conditional mood 94,216,246, 277,281, 364, 395, 404 continuative aspect, see continuous aspect continuous aspect 38-40, 119,140, 260, 301, 349, 356 contraexpectation tnood 216, 265 coordination 99, 101,263,288 dependent-marking xxvii, 10, 80, 180, 254, 346, 352, 363, 372, 396 desiderative m ood 58,9.3-4,216, 245, 326 detransitivizers 31,44, 159 different subject marking xxviii, 101, 197, 224, 405 ditransitive verbs xxvii, 154,158,243, 286-7, 327, 376, 379 dubitative m ood 93-4, 274, 318 durative aspect 93, 301, 326 ejectives3l3, 352 epenthesis 31, 136 ergativity 1, 25,32, 36, 55, 59-61, 118, 155, 157, 187, 193-5, 241-2, 254, 320, 348, 366 split-ergativity 9, 32,60, 87, 89-90,98, 263, .348, 366, 375,403 evidentials 53, 93,213-14,216, 222 evidentiality xxvi, 1,8, 87,119, 214, 261, 274,277-9, 282, 286, 288, 302, 349-50, 376,404 feminine, see gender gender 8. 10, 80,83-4, 96-7,100,116, 148, 177, 185, 192^3,214, 221, 258, 298-9, 302, 319, 360, 362-3, 366, 369, 373, 379,400-1,408,427 glottalization 28, 76, 79, 3Î3 habitual aspect 93,260, 325,349, 360 head-m arking xxvii, 8, 10, 80, 114, 180, 254, 273, 346, 357, 359, 362-3, 366, 369, 372,378, 396, 398 hierarchy, person/agentivity 33,36, 155 imperfective aspect 274-6, 356 implosives 210,317, 365 incompletive aspect 38, 245, 326 incorporation 9-10,45, 92. 159, 254, 261-3, 265,324, 363, 369, 376 446 Subject index noun incorporation 9, 45, 93, 220, 243, 261.350, 376 inverse 31 -2, 34 iterative aspect214-15, 301, 349, 356, 360 kinship terms 41,82, 183. 218, 258, 330, J95, 399-400,403 iabializalion 29, !4 0 -l, 318 laryngealization 271 masculine, see gender metathesis 28, 31, 77, 79, 136, 138, 314-15, 354 nasalization 8-9, 78-9, 134-7, 143-5,153, 179.211-12,231,240^1,255,271, 313-14,346.368,394, 396,411 negation 48-9, 52, 58, 96, 93, 154, 222, 244, 264-5,301, 349, 353, 356, 358, 360,404 negative imperatives 96. 276 prohibitives 96, 260,264-5, 379 nominalizations 10,49, 51, 56- 7, 121, 144, 159-60, 183, 194,246, 260,277,280, 287,290-1,334,358 nominalizers 40, 45,48-50, 98, 143, 218, 223, 281. 290,350-1,356 nominative 90, 147, Í87, 241-2,254. 263, 320-1, 349, 359. 369, 378.403-4 reciprocal 44, 85, 88, 92-3, 119-20, 154. 159, 195, 237,243-4, 260,282, 328, 356, 360, 376 reduplication 80-1, 86. 119,183-4, 260,302, 304, 323, 325,360-1, 363, 367 reflexive 37, 44, 88, 9 t-3 , 119-20,154, 159, 182, 195,237,243-4, 260, 283, 303, 328, 349, 356, 358, 360, 369, 376 relativization 265, 287-8, 332-3 relative clauses xxvi, xxviii, 57. 95, 9 9 -HW, 152,222.236, 245, 265,287, 332-3,350-1,356, 360-1,377 S,xxvi, 32-5, 82, 86-90, 97-8, lOt-2, 117, 147, 155, 158-60,375,403,407 same subject marking xxviii, 88, 101,197, 224,405 serial verbs xxviii, 97-8, 147, 157,213, 360-1, 363,404 S xxvi, 32-5, 86-90, 98, 100-1, 117, 146. 155, 158, 160, 375,403,407 stative verbs xxvi, 32, 81,83, 86,89,91, 95, - 117, 146-7,149, 151, 154-5,160, 213, 221,257, 284,300-1, 346, 351,375, 407 subordination 265, 333-5 subordinate clauses 287 switch-reference xxviii, 10, 99,101, 197-8, 22.3-4, 227, 237-9, 265, 333-5, 350, 377, 392,405,411 oblique-topicalized verbs 156 palatalization 29, 138-40, 318 perfective aspect 32, 39-40, 93, 274 6, 326, 349, 356, 379 pitch accent 79, 114,212, 256, 313-14, 316-18,394,406-7 pivots49, 51.60, 90,99, 101, 121-2, 195, 263,298, 301-2,304-6,353 possession 10, 2 5,29,40-2, 82-3, 150-3, 182,190. 200,257-8, 3.S6, 363, 369, 400.410 alienable 8, 40. 82, 114. 116 -17, 190-1, 258,289, 299, 303, 330, 346, 353, 399-400,409-10 inalienable 8,40, 82, ! 14, 117, 190, 258, 289, 299, 330, 346, 3 99^ 00,409-10 possessives 10, 41. 82-4, 95-6, 99, 115-16, 182, 219, 235, 241.265. 299,303, 330, 347, 372-4, 379, 3 9 9^ 01, 407, 409-10 possessive clauses 99 possessive pronouns 218, 248, 289, 291, 321 progressive aspect 89, 93, 100, 216, 326 tone 79, 114, 180, 209, 230, 253, 256, 271-2, 313-14, 316, 318, 354-5, 363,371, 394 contrastive lexical tones 10 topic advancing 407-10 topic marker 335, 349, 397 topiealization 56, 85, 93, 262 valency-changing derivations 9, 90-3, 260-1,356.366 decreasing, see reducing below increasing xxvii, 87, 90-1, 101. 363, 372, 375 reducing xxvii, 90, 9S, 260, 367, 376 valency-changing devices, see valencychanging derivations valency-changing proces.ses, see valencychanging derivations verb compounding 98, 216, 259-60, 350, 356, 361,376,404,411-12 vowel assimilation 256, 298, .345, 354 vowel harmony 79,81, 232, 359 vowel length 28-9, 313-14, 362, 395 vowel shift 112, 144 -5 THE AMAZONIAN LANGUAGES Paperback Re-issue The Amazon Basin is arguably both the least-known and the most complex linguistic region in the w orld today. It is the home o f some 300 languages belonging to around 20 language families, plus more than a dozen genetic isolates, and many o f these languages (often incompletely documented and mostly endangered) show properties that constitute exceptions to received ideas about linguistic universals. This book provides an overview in a single volume o f this rich and exciting linguistic area. The editors and contributors have sought to make their descriptions as clear and accessible as possible, in order to provide a basis fo r further research on the structural characteristics o f Amazonian languages and their genetic and areal relationships, as well as a point of entry to im portant cross-linguistic data fo r the w ider constituency o f theoretical linguists. C a m b r id g e UNIVERSITY PRESS www.cambridge.org