John Benjamins Publishing Company

John Benjamins Publishing Company
This is a contribution from The Diachronic Typology of Non-Canonical Subjects.
Edited by Ilja A. Serzant and Leonid Kulikov.
© 2013. John Benjamins Publishing Company
This electronic file may not be altered in any way.
The author(s) of this article is/are permitted to use this PDF file to generate printed copies to
be used by way of offprints, for their personal use only.
Permission is granted by the publishers to post this file on a closed server which is accessible
to members (students and staff) only of the author’s/s’ institute, it is not permitted to post
this PDF on the open internet.
For any other use of this material prior written permission should be obtained from the
publishers or through the Copyright Clearance Center (for USA: www.copyright.com).
Please contact [email protected] or consult our website: www.benjamins.com
Tables of Contents, abstracts and guidelines are available at www.benjamins.com
part i
Rise of non-canonical subjects
or subject-like obliques
© 2013. John Benjamins Publishing Company
All rights reserved
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish
anticausative constructions
Exploring subjecthood*
Olga Fernández-Soriano & Amaya Mendikoetxea
Universidad Autónoma de Madrid
This paper examines the notion of subject and subjecthood by analysing the
properties of a construction found in Spanish, as well as in a variety of genetically
quite different languages, in which a non-selected dative argument is added to an
anticausative construction and may be interpreted as accidental or unintentional
causer of the event. In particular we explore three hypotheses: (i) the addition
of a dative argument to a typical anticausative structure requires the projection
of a cause predicate in the syntax, in line with the analysis proposed by Schäfer
(2008), among others; (ii) the dative argument is introduced by a high applicative
phrase (Pylkkänen 2008) and behaves as a subject with respect to certain syntactic
properties; in particular, it has properties akin to those of a quirky subject of the
type found in languages such as Icelandic or Georgian (Sigurðsson 1996, 2002a, b),
and (iii) the subject properties displayed by the noncore dative mostly follow
from the fact that it participates in the first (initial) subevent of the predicate, as
is the case for external arguments in general (Harley 1995). This explains why the
unintentional causer construction cannot be found with a subclass of change of
state verbs, namely those expressing internally caused eventualities, which lack a
causative predicate, a fact that has either gone unnoticed or remained unexplained
in other analyses of the construction.
* This paper was partially funded by research projects EDU2008-01268, FFI2008-01584
and FFI 2011–23829 (Spanish Ministry of Science and Innovation). Previous versions were
­presented in the Seminario de Lingüística Teórica (CSIC, Madrid) and the Societas Lingüística
Europaea, 43rd Annual Meeting at the workshop: ‘Subjects and Transitivity in Indo-European
and beyond: a typological diachronic perspective’. We thank the audiences at both events for
their comments and suggestions. We are also grateful to the editors and to an anonymous
reviewer for comments on the initial manuscript. All remaining errors are ours.
© 2013. John Benjamins Publishing Company
All rights reserved

Olga Fernández-Soriano & Amaya Mendikoetxea
1. Introduction
The notion of ‘subject’ is central to the study of grammar. The prominent role of
subjects in grammatical description is recognized by all theoretical frameworks and
formalized in different ways. In the generative tradition, the Extended Projection
Principle (EPP) (Chomsky 1981) requires all clauses to have a subject, but no equivalent requirement is established for objects, or other grammatical relations. However,
despite the centrality of this notion, it is not possible to provide a uniform definition
of subjects. Instead, subjecthood comprises a heterogeneous cluster of properties, so
subjects are often defined by a combined set of morphosyntactic and behavioural
properties (agreement, structural position, raising, Case, binding…). The study of
constructions in which those properties fail to converge on a single verbal argument
is crucial for our understanding of the notion of subject and subjecthood. This is the
case for Spanish anticausative constructions with dative arguments, which are the
central concern of this paper.
Canonical subjects are those in which all properties converge. Thus, the DP
John in English and its equivalent Juan in Spanish occupy the initial position,
are external arguments (as opposed to the glass and el vaso, which are internal
arguments), ­semantically they may be interpreted as agents, are associated with
(abstract) nominative Case, and trigger verbal agreement:
(1) a. John broke the glass.
b. Juan rompió el vaso.
It is well-known that sometimes a particular nominal argument does not necessarily
display all the properties above; i.e. there seem to be ‘degrees of subjecthood’. Thus,
together with structures like (1), verbs like break/ romper are associated with structures like those in (2), in which the syntactic subjects the glass and el vaso are internal
arguments, semantically interpreted as affected themes, while displaying nominative
Case and triggering verbal agreement:
(2) a. The glass broke.
b. El vaso se rompió.
The structures in (1) and (2) illustrate the two variants of the causative alternation.
Argument alternations of this type are useful in order to explore the notion of subjecthood, but perhaps of more interest is the type of argument structure variation
involving the addition of noncore arguments with subject-like properties, in which the
characteristics of subjecthood appear to be scattered across more than one nominal
argument. This paper examines a construction found in Spanish, as well as in a variety
of genetically quite different languages, in which a non-selected dative argument is
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions
added to an anticausative construction and may be interpreted as accidental or unintentional causer of the event, as illustrated in (3).
(3)a.AJuansele
haroto el vaso.
toJuanse cldat hasbrokentheglass
‘Juan has (unintentionally) broken the glass.’
b.APedrosele
quemóla comida.
toPedrose cldatburnedthefood
‘Pedro has (unintentionally) burned the food.’
The verbs involved are transitive verbs like romper ‘break’ and quemar ‘burn’, which do
not typically select for a dative argument as part of their lexical properties. In ­canonical
transitive contexts, these verbs are constructed with a nominative and an accusative
argument, as shown for romper in (1b). In this, they contrast with other biargumental
verbs such as gustar ‘like’, belonging to the subclass psych-(ological) verbs including,
among others, molestar ‘bother’ or asustar ‘frighten’, which select for dative experiencers, as well as for nominative theme arguments:
(4)a.A Juanle
gustan losheladosnom.
toJuancldatlike-3pltheice creams
‘Juan likes ice cream.’
b.A Maríale
molesta elruidonom.
toMaria cldat bother-3sgthenoise
‘Noise bothers Maria.’
Thus, the verbs romper ‘break’ and quemar ‘burn’ are associated (i) with transitive
causative structures like (1b), where the subject Juan is interpreted as being directly
responsible for the event denoted (though Juan may act either intentionally or unintentionally); (ii) with anticausative constructions like (2b), in which the sentence is
interpreted as an event that takes place ‘spontaneously’, without the intervention of
an external cause, and (iii) with dative argument constructions like (3), in which the
referent of the dative may be interpreted as indirectly responsible for the event and acts
unintentionally: it is an accidental causer. Additionally, constructions like (3) may also
have an affectedness reading: the dative DP is interpreted as the benefactor or malefactor of the event described (see Schäfer 2008 and Rivero 2003).1 For a sentence like (3b),
this interpretation could be paraphrased as ‘the food burned and Pedro is somehow
affected’. This reading is obtained in anticausative constructions but also in transitive
. An inalienable possession meaning is also possible. See Fernández-Soriano and
­Mendikoetxea (2011) for this interpretation.
© 2013. John Benjamins Publishing Company
All rights reserved


Olga Fernández-Soriano & Amaya Mendikoetxea
(5a) and passive contexts (5b), with which the unintentional causer interpretation is,
however, blocked.2
(5)a.Le
quemaron lacomida aJuan.
cldat burned-3plthefood toJuan
‘They burned the food and Juan is affected.’
b.Le
fueronreparadoslos bienesdañados aldemandante.
cldat were repaired thegoodsdamagedto.theplaintiff
‘The damaged goods were repaired and the plaintiff was affected’.
Though the affectedness reading can also be obtained in anticausative constructions
such as (3), the most salient interpretation of these sentences is that in which the dative
argument is related to the cause of the burning or the breaking.3 This is the reason for
the contrast in (6): while the dative element (al chef) can only be given the affectedness reading and cannot be considered as the accidental or unintentional causer of
the burning in a transitive context such as (6a), both interpretations are found in the
anticausative constructions in (6b), and hence the incompatibility with an additional
causer (el pinche ‘the scullion’), which, however, can be added in (6a):
(6)a.Al chefle
quemaron la comida:fue el pinche.
to the chef cldat burned-3plthefood: wasthescullion
‘The food was burned and the chef was affected: it was the scullion.’
b.#Alchefsele
quemó
la comida:fue el pinche
to the chef se clDAT burned-3sgthefood wasthescullion
‘The chef (unintentionally) burned the food: it was the scullion.’
According to Rivero (2003) the ‘causer’ reading for the dative arises because
­anticausatives like (2b), as opposed to transitives like (1b), lack formal external
­arguments. Though we agree with the basic idea behind this statement, this i­ntuition
will be reformulated in the analysis that follows. We will argue that in order to p
­ rovide
a principled account of constructions like those in (3), it is essential to look at the
lexical semantics of the verbs involved. Verbs like quemar and romper denote externally caused change of state eventualities and are commonly analyzed as involving
two s­ubevents: a causing subevent and a result or change of state (COS) subevent
(Levin & Rappaport Hovav 1995). Crucially, the dative argument, in the accidental
causer i­ nterpretation, always participates in the first (initial) subevent of the predicate,
. The accidental causer interpretation is not possible with unergative verbs either.
. Some authors derive one reading from the other (Cuervo 2003). In fact the affectedness
reading, as mentioned, is not excluded and can always be obtained in the cases we analyze.
The opposite does not hold, that is, the ‘causer’ reading, which is the object of our study, is
­restricted, as we will show, to a particular class of verbs: alternating se-marked externally
caused change of state verbs.
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions
as it is the case for external arguments in general (in the sense of ­Harley 1995; Borer
2005; Ramchand 2008), and is not an argument of the COS subevent. We take this as
an indication that the added dative behaves like a subject in a way that will be specified
in what follows (sec. 3 & 4 below). Lack of an external argument is not enough to trigger the unintentional causer reading, as this reading is blocked with internally caused
verbs of change of state such as palidecer ‘turn pale’, oscurecer ‘darken’ or adelgazar ‘get
thin’, that are analyzed as pure (monoeventive) unaccusatives (see Section 4).
In this paper we explore the following hypotheses: (i) structures like (3) are the
result of adding a non-selected dative argument to a typical anticausative structure.
Such an addition requires the projection of a causative predicate in the syntax, in
line with the analysis proposed by Schäfer (2008), among others; (ii) the ‘added’ dative
argument is introduced by a high applicative phrase (Pylkkänen 2008) and displays
most of the (syntactic) properties associated with subjects; in particular it seems to
behave as a quirky subject of the type found in languages such as Icelandic or ­Georgian
(see Sigurðsson 1996, 2002a, b), and (iii) the subject properties displayed by the noncore dative mostly follow from the fact that it participates in the first (initial) subevent
of the predicate, as is the case for external arguments in general (Harley 1995). Regarding (i), the fact that these structures take as their basis the anticausative variant of
the causative alternation, explains the presence of se and the occurrence of a ­typically
postverbal internal argument which, nevertheless, triggers verbal agreement and
shows nominative case. As for hypothesis (ii), as we will see, despite what agreement
facts indicate, the dative element in the high applicative phrase has subject properties.
Finally, hypothesis (iii) explains why the unintentional causer construction cannot
be found with a subclass of change of state verbs, namely those expressing internally
caused change of state eventualities, which lack a causative predicate, a fact that, to our
knowledge, has either gone unnoticed or remained unexplained.
Our proposal accounts for structures like (3) in Spanish, but also for equivalent
constructions with parallel interpretations in other Romance languages like in Italian
(7), as well as in German (8), Greek (9) (with a genitive, instead of a dative, argument),
and some Slavic languages, as illustrated in (10a) for Polish and (10b) for Bulgarian
(see Schäfer 2008; Rivero 2003):
(7)AMariosiruppe ilportacenere
(pererrore).(Italian)
toMariosibroke-3sgtheashtray
by mistake
‘Mario (accidentally) broke the ashtray (by mistake).’
(8)DemHans zerbrachversehentlich die Vase.(German)
thedatHansbroke unintentionallythevase
‘The vase broke and Hans caused this unintentionally.’
(Schäfer 2008: 44)
(9)Tu Bentu
kaikei supa.(Greek)
thegenBenhegenburntthesoupnom
‘Ben involuntarily caused the soup to burn.’
(Schäfer 2008: 71)
© 2013. John Benjamins Publishing Company
All rights reserved


Olga Fernández-Soriano & Amaya Mendikoetxea
(10) a. Jankowizłamałysię okulary.(Polish)
Johndat brokenfem-plReflglassesfem-pl
b. Na Ivan muse sčupixaočilata.(Bulgarian)
Johndat hedat Reflbroke3plglasses.the
‘John accidentally boke the glasses.’
(Rivero 2003: 471)
There are, of course, syntactic differences between the structures in (7)–(10) and the
Spanish structures in (3). As noted, the added argument is in the genitive Case in
Greek, as opposed to the other languages, where it is in the dative Case. In Italian,
as well as in Polish and Bulgarian, the presence of a reflexive clitic is obligatory (like
in Spanish), though lexically restricted (see Schäfer 2008), but there is no reflexive
­element in Greek, nor in German.4 In addition, clitic doubling of the dative is obligatory in Spanish and Bulgarian, but not in Polish or Italian. As suggested by Schäfer
(2008), these differences derive from independent properties of the different languages.
In what follows, we first look at the causative alternation, as it is crucial to understand its properties in order to provide a principled analysis of the structures in (3)
(sec. 2). We then examine the syntactic properties of the dative argument and provide evidence that it occupies a high position in the structure (sec. 3); we analyze
dative argument constructions as the result of adding a noncore argument in a high
applicative phrase to an anticausative construction (hypothesis (i) and (ii)). ­Section 4
discusses our hypothesis (iii) by providing evidence that the dative argument participates in the initial subevent of a causative construction, where it is interpreted as
­effector, showing some of the semantic restrictions on dative subjects. ­Section 5
contains the conclusions.
2. The causative alternation
Hypothesis (i) above states that the ‘dative subject’ construction in (3) is the result
of adding a dative argument to anticausative structures like that in (2b) (repeated in
(11b) below), which are, in turn, related to transitive structures like (1b) (repeated in
(11a) below). What this means is that the relation between (11a) and (11c) is mediated
through (11b), the anticausative variant of the causative alternation:
(11) a. Juanha roto el vaso.Transitive
Juanhasbrokentheglass
b.Seha roto el vaso.Anticausative
se hasbrokentheglass
‘The glass has broken.’
. In fact, as Schäfer (2008) shows, the accidental causer dative we are analyzing only appears
in German with non (reflexively) marked verbs.
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions
c. AJuansele
haroto el vaso. Dative subject
toJuanse cldathasbrokentheglass
‘Juan has accidentally broken the glass.’
In order to understand the syntactic and semantic processes involved in the derivation of (11c), it is first necessary to look at the properties of the causative alternation.
This alternation is crosslinguistically widespread and has been the topic of intensive
research in linguistic theory. Under the common assumption that the two variants
of the causative alternation (transitive (11a) and unaccusative (11b)) are derivationally related, a recurrent question is to determine which one is basic and which one
is derived. Two views can be identified: (i) alternating verbs are basically monadic
predicates and the transitive variant is derived via a process of causativization (Hale &
K
­ eyser 1986, 1993; Harley 1995; Pesetsky 1995, among others), or (ii) alternating verbs
are basically dyadic predicates and the anticausative variant is derived via a process
of decausativization or detransitivization (Grimshaw 1982; Chierchia 2004; Levin &
Rappaport Hovav 1995; Reinhart 2002; Koontz-Garboden 2009, among others). It is
not our intention to review these two approaches here (see Schäfer 2009 for a detailed
overview). In what follows, we first look at anticausativization as decausativization and
then argue in favour of the common base approach, according to which there is no
direct derivational relationship between the causative and the anticausative variants
of the alternation.
2.1 Anticausativization as decausativization
Reflexive markers, like those found in Romance and Slavic languages, are common
devices to mark the anticausative form. Under a derivational approach the presence of
a morphological marker is taken to indicate that the anticausative is derived from the
causative by means of addition of the reflexive form se, which is commonly analyzed
as a marker of this derivational process (see among many others Marantz 1984; Burzio
1981, 1986; Zubizarreta 1987): 5
(12) a. Juan rompió el vaso
>
Juan broke the glass
El vaso se rompió
The glass broke
b. Pedro quemó la comida > La comida se quemó
Pedro burned the food The food burned
. See Haspelmath (1993) for a thorough description of the ways in which languages mark
the causative/ anticausative distinction. Koontz-Garboden (2009) and Schäfer (2009) also offer
an updated overview of this topic and its relevance for theoretical analyses of the ­alternation.
© 2013. John Benjamins Publishing Company
All rights reserved

 Olga Fernández-Soriano & Amaya Mendikoetxea
According to Levin and Rappaport-Hovav (1995) (L&RH, henceforth), verbs entering the c­ ausative alternation, such as romper ‘break’ and quemar ‘burn’ in (12), are
those expressing externally caused changes of state; they “imply the existence of an
‘­external cause’ with immediate control over bringing about the eventuality described
by the verb: an agent, an instrument, a natural force, or a circumstance” (L&RH: 92)
(see Mendikoetxea 1999 for an overview of those verbs in Spanish). In a projectionist, lexicalist approach, like that advocated by L&RH, anticausatives derive from transitive structures through a lexical operation of existential binding which suppresses
the external argument in the mapping between the lexical semantic representation
(LSR) of the verb and its lexical-syntactic representation or Argument Structure (AS),
thereby preventing the expression of that argument in the syntax, as shown in (13):
(13) LSR:
[[x-do something] cause[become[y ⟨BROKEN⟩]]
↓
Lexical binding 0
Linking rules
↓
AS
⟨y⟩
[L&RH: 108]
The anticausative structure will then be interpreted as asserting that the central
(become) subevent came about via some causing event, whose nature is unspecified:
“what characterizes the class of alternating verbs is a complete lack of specification of
the causing event” (L&RH: 107) (see also Guerssel et al. 1985; Hale & Keyser 1987).
This is reflected in the fact that a wide variety of subjects are allowed with these verbs:
e.g. for romper ‘break’, the external argument can be realized as an agent or an instrument, natural force or cause, as shown in (14):
(14) Juan/ el vendaval/ la piedra rompió el cristal.
‘Juan/ the whirlwind/ the stone broke the glass.’
That is, the external arguments of the alternating verbs may either be agents or causes of
an event (Schäfer 2008: 3.4). A human subject like Juan in (14) may act with or ­without
intentionality. According to Schäfer (2008), intentionality presupposes ­agentivity, and
agentivity presupposes the feature [+human]. Instruments presuppose agentivity, but
not intentionality. As for natural forces, they are not human and they are, therefore,
non-intentional. As stated by Koontz-Garboden (2009: 82):
“A root can specify of its arguments more or less entailments, from highly
articulated specification to rather serious underspecification. Some verbs are
highly specified and take only an agentive causer, while others are underspecified
and can take agents, instruments, natural forces, etc.”
Following Van Valin and Wilkins (1996), Koontz-Garboden calls this underspecified
role effector. In Section 4, we focus on the properties of this effector argument.
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions
Crucially, underspecification of the causing event is also responsible for the fact
that the cause argument may be left unexpressed, so that the sentence expresses an
externally caused eventuality which comes about spontaneously, without the volitional
intervention of an agent. Note that verbs expressing externally caused eventualities
with specified causing events (i.e. denoting events that require agentivity) lack not
only structures like (11b) but also dative subject constructions like (11c), as shown in
(15b, c) and (16b, c), which we take as evidence that (11c) is the result of adding a Juan
(and le) to (11b):6
(15) a. Pedro/el cuchillo cortó la carne.
‘Pedro/the knife cut the meat.’
b. *La carne se cortó (ella sola).
The meat se cut (by itself)
c.*A Pedro se le cortó la carne (accidentalmente).
To Pedro se clDAT cut the meat (accidentally)
(16) a. Un arquitecto famoso construyó el museo.
‘A famous architect built the museum.’
b. *El museo se construyó (él solo).
The museum se built (by itself)
c.*A un arquitecto famoso se le construyó el museo (accidentalmente).
To a famous architect se clDAT built the museum (accidentally)
Within this typology of external arguments, the dative argument is usually human and
necessarily unintentional, which explains why non-human causers are disallowed in
the dative construction ((17b) vs. (17a)). We will come back to this issue in Section 4
below.
(17) a. A Juan se le rompió el cristal.
To Juan se clDAT broke the glass
b.*A la piedra/al vendaval se le rompió el cristal.
To the stone/the whirlwind se clDAT broke the glass
2.2 Common base approaches and the reflexivization analysis
In recent years, the idea that there is a direct derivational relationship between the two
variants of the causative alternation has been questioned. Instead, it has been argued
that both are derived from one source, e.g. a category neutral verbal root in Alexiadou
et al. (2006a, b) (see also Pylkkänen 2008; Schäfer 2009 for a review of different “common base” proposals). Under this approach, causatives and anticausatives do not differ
. The adverb accidentalmente ‘accidentally’ signals the intended meaning. If the dative is
interpreted as affected, (15c) and (16c) are grammatical sentences.
© 2013. John Benjamins Publishing Company
All rights reserved

 Olga Fernández-Soriano & Amaya Mendikoetxea
in the number of events involved. For Pylkkänen (2008) both causatives and anticausatives involve a root which expresses a resultant state predicated of the theme, but in
anticausatives this root combines with a become projection (18a), while in causatives
it combines with a cause projection, as well as a voice projection which introduces
the external argument (18b). This is under Kratzer’s (1996) hypothesis (following
Marantz 1984) that external arguments are not introduced by the verb itself but by a
non-eventive voice head:
(18) a.
vPbecome
vbecome
b.
√Root
VoiceP
Voice
vPcause
vcause
√Root
Alexiadou et al. (2006a, b) propose that both causatives and anticausatives have
the same event decomposition: they both involve the verbal head cause, but differ
in whether voice is projected or not (see also Kratzer 2005). That is, the causative
­alternation is really a Voice alternation, with the event decomposition in (19):
(19) a. The vase broke
b. John broke the vase
=
=
[cause [the vase broken]]
[John [Voice [cause [the vase broken]]]]
These proposals express in the syntax the lexicalist hypothesis that anticausatives are
inherently causative. As such, they can license PPs introducing causative adjuncts
involving non-human causers or causing events, which according to Alexiadou et al.
(2006a, b), modify the causative event and are thematically licensed via adjunction to
vcause, as shown in (20a) for English, (20b) for Spanish and (20c) for German (from
Schäfer 2009), while agentive by-phrases are ruled out, due to the absence of voice:
(20) a. The ship sank because of the explosion/ the storm/*by the enemy.
b.El barco se hundió a causa de la explosión/por la tormenta/*por el
enemigo.
‘The ship se sank because of the explosion/ by the storm/ *by the enemy.’
c.DieTür öffnete sich durch einenWindstoß/*durchMaria.
the dooropenedreflthrougha
blast-of-wind/byMary
‘The door opened because of a blast of wind/*by Mary.’
As pointed out by Schäfer (2009), the claim is not that there is an implicit causer
­argument in anticausatives, but that there is a causative event: an event leading to the
resultant state of the theme (but see Kallulli 2006, 2007 for a different explanation).
This causative event is unspecified, both for the causative and anticausative, as argued
in Section 2.1 above.
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions 
We assume that this view of anticausativization is essentially correct (see Schäfer
2008, 2009 for arguments in favour). There is only one verb (or one verbal root),
romper, which basically expresses an unspecified causing (COS) event and which is
projected in the syntax with a cause predicate, as in (19). For simplicity purposes,
we associate the external argument with the cause predicate, without an additional
functional projection (though the analysis could be reformulated to include a voice
projection introducing this argument). When the external argument is realized as a
DP like Juan, el vendaval ‘the whirlwind’ or la piedra ‘the stone’, we have a canonical
transitive structure like (21) with vP as the syntactic realization of the cause predicate
and the verbal head expressing the resulting COS predicate:
(21) Juan/el vendaval/la piedra rompió el cristal.
‘Juan/the whirlwind/the stone broke tha glass.’
[vP DPext arg v [= cause ] [VP romper +DPtheme]]
The anticausative has the same syntactic structure as (21), but the external argument is
realized as a null pronominal (pro), whose interpretation is that of the DPtheme, which
is obtained through an operation defined as ‘clause internal control’ in Mendikoetxea
(2000), after obligatory externalization of the DPtheme to a position higher than pro.
From that position the theme argument can control the reference of the empty pronominal. This proposal is in line with analyses of anticausativization as reflexivization,
as advocated by Chierchia (2004), Reinhart (2002), Reinhart & Siloni (2005) and, more
recently, Koontz-Garboden (2009). Under this approach the structure in (22a), a true
reflexive, and the anticausative construction in (22b) are derived by means of the same
rule: a lexical operation of reduction, which turns a transitive verb in the lexicon into
an intransitive verb in the syntax. According to Chierchia (2004: 29), this operation
identifies the two arguments of a relation, thereby reducing it to a property in which the
external and internal argument are set to be identical [closest Montague-style equivalent: λ[wash (x) (x)]), with the reflexive clitic as a marker of the reduction operation:
(22) a. María selavó.
María
sewashed
‘Maria washed herself ’.
b.Elbarcosehundió.
theboat sesank
‘The boat sank’.
This is also the end result of the clause internal control analysis, which according to
Mendikoetxea (2000), applies in reflexives and anticausatives in (21). The meaning of
the anticausative is a reflexive form of the causative predicate, but with anticausatives,
the causing factor is not an action, as it is in true reflexives like (22a), but it must be
understood statively: in (22b) the boat has or comes to have a property that causes
© 2013. John Benjamins Publishing Company
All rights reserved
 Olga Fernández-Soriano & Amaya Mendikoetxea
its sinking (Chierchia 2004: 37) (but cf. Piñón 2001 and Folli 2002, against this interpretation). That is, what characterizes the two constructions in (22) is that the theme
­argument participates in both subevents: the causative event and the COS event.7
We assume that the clitic heads its own projection above vP (see Mendikoetxea
2008). The element se is not an anaphor, nor does it have anaphoric properties. Its
presence is required for syntactic reasons. In particular, under an analysis in which
nominal elements check (or agree with) the features of functional heads (in line with
Chomsky 2000, 2001), Mendikoetxea (2008) argues that se is required to check the
person feature of T, as the element pro is defective and cannot perform that role.8
(23) Se rompió el cristal.
‘The glass broke.’
[TP DPtheme [CLP se [vP pro extarg v [= cause ] [VP romper +DPtheme]]
Though this analysis is simply sketched here (see Mendikoetxea 2000, 2008 for
­justification), some evidence that it is on track is given by the facts in (24) and (25).
Bare NPs are only possible in Spanish in a VP-internal position. Anticausatives like
those in (24) do not allow bare NP themes (see Masullo 1992; Mendikoetxea 2000;
Cuervo 2008), suggesting that the DPtheme must be externalized, as opposed to ‘pure’
. Reinhart’s (2002) analysis is inspired by Chierchia (2004). Sentences like (22) are derived
from transitive counterparts by means of a reduction operation which eliminates either the
external (22b) or the internal argument (22a). In her system, theta roles are decomposed into
two binary features: [± c], indicating whether the argument is responsible for causing the
event or not, and [± m], indicating whether the mental state of the argument is ­relevant for the
event or not. For the external argument to be reduced, it must be just [+c]: i.e. underspecified
for the contrast between agents and causers. As pointed out by Schäfer (2008: 119), neither
lexical binding in Levin & Rappaport-Hovav (1995), nor Reinhart’s (2000, 2002) operation
(also called ‘expletivization’) have been defined in a formal way and it is not clear how this
can be done within standard semantic systems of function-argument application (see also
Koontz-Garboden 2009 on this point), so these theories are simply descriptions of the facts,
under the hypothesis that verbs entering the causative alternation are basically dyadic.
. This analysis can account for the presence of se both in anticausatives and reflexives, as
in impersonal constructions like (i), which also contain a null pronominal in subject position
with arbitrary interpretation.
(i)a.Setrabaja muchoenestaoficina.
Sework-sgmuch in thisoffice
‘One works a lot in this office.’
b.Sevendencasas
Sesell-pl houses
‘Houses are on sale.’
See Mendikoetxea (2008) for a detailed analysis of these constructions and Mendikoetxea
(2012) for an overview of the different se-constructions.
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions 
unaccusative verbs like those in (25) that may be constructed with bare NPs. This will
become important when we look at the properties of the dative argument construction
in the following section.
(24) a. *Se cocieron patatas (ellas solas).
se boiled potatos (by themselves)
b.*Se fundió hielo (por si solo).
se melted ice (by itself)
(25) a. Vienenmujeres.
come women
‘Women come’.
b.Existenproblemas.
exist problems
‘Problems exist’.
In sum, the analysis adopted here establishes a relation between reflexives and anticausatives by claiming that they are both the result of clause-internal control, without
resorting to a reduction rule and without deriving the anticausative constructions from
the causative/transitive construction. In what follows, we will put forward a proposal
in which dative subject constructions are the result of the addition of a dative argument to a structure like (23) (hypothesis (i)), via the introduction of a high applicative
phrase (Pylkkänen 2008) between TP and the clitic projection (hypothesis (ii)), so that
the dative argument is associated with the first (initial) subevent and displays subject
properties: it shows the syntactic behavior of an external argument.
3. Subject dative arguments and syntactic structure
Previous accounts of the dative argument of an anticausative construction analyze this
element as an external argument with ‘quirky’ subject properties (Fernández-Soriano
1999, 2000), in contrast with benefactive or goals, which are also realized as dative
arguments, but remain VP-internal. Consequently, the theme argument behaves as an
internal argument and does not externalize, although it triggers verbal agreement. In
what follows, we provide evidence for these claims, before moving on to the analysis
itself in Section 3.2.
3.1 The syntactic properties of the dative argument
The first argument in favor of our proposal comes from word order and information structure. It can be shown that the internal (nominative) argument occupies
the postverbal (VP internal) position, where it is interpreted as informational focus,
whereas the dative DP behaves as the (most) external argument of the structure.
© 2013. John Benjamins Publishing Company
All rights reserved
 Olga Fernández-Soriano & Amaya Mendikoetxea
We take the unmarked constituent order to be the one chosen when the whole
­sentence is interpreted as new information in out-of-the-blue contexts (which can
be an answer to questions like ‘what’s up?’). This being so, the contrast in (26a) vs.
(26b, c) is explained if, as we claim, the dative is external to the VP and the theme
remains internal. This is only true for accidental causer datives; as can be seen in
(26d, e), goal or benefactive datives do not show the same behavior in these contexts:
(26) A:¿Algo que contar?
‘Any news?/ What’s up?’
B:a.A Pedro se le ha quemadola comida.
toPedrose cldat hasburned thefood
‘Pedro has (unintentionally) burned the food.’
b.??Se le ha quemado la comida a Pedro.9
c.??La comida se le ha quemado a Pedro.
d.#A Juan le han dado el regalo.
to Juan cldat have given the present
e.#A María le han roto el coche.
to Maria cldat have broken the car
Second, when appearing under raising verbs like parecer ‘seem’, the dative argument in
the subordinate clause raises to the matrix subject position, as in (27a), (28a). Raising
of the theme is impossible if the dative is present, as in (27b), (28b), which suggests
that it remains inside VP. In anticausatives without the dative it is the theme that raises,
as expected (29):
(27) a. AJuanparece habérsele
roto elcoche.
toJuan seemsto.have.se.cldat brokenthecar
‘Juan seems to have (unintentionally) broken the car.’
b.??El cocheparecehabérseleroto
a Juan.
the car seems to.have.se.cldatbrokento Juan
(28) a. AMaríapareceperdérsele
el niñocontinuamente.
toMaría seemsto.lose.se.cldat thekid all-the-time
‘María seems to lose the kid all the time.’
b.??El niñopareceperdérsele
a Maríacontinuamente.
the kid seems to.lose.se.cldatto Maríaall-the-time
. An anonymous reviewer notes that this sentence, as well as (27b) and (28b), might be
subject to dialectal variation, as they appear to be acceptable to some speakers. The problem
here, however, is to isolate the intended meaning (accidental causation) from other possible interpretations of the dative element, which are not excluded in these contexts. Focus
­intonation may also affect acceptability judgments.
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions 
(29) a. El cocheparecehaberse roto.
thecar seemsto-have.sebroken
‘The car seems to ha ve broken.’
b.La comidaparecequemarse.
thefood seemsto.burn.se
The food seems to be burning.’
c.Esteniñopareceperderse
continuamente.
thiskid seemsto.get.lost.seall-the-time
‘This kid seems to get lost all the time.’
In this respect, this dative contrasts with other dative arguments, such as goals, which
never block raising of the subject (30a), experiencers of psych verbs like m
­ olestar
‘bother’ (30b) and gustar ‘like’ (30c), which have been claimed to occupy a high
p
­ osition (cf. Belletti & Rizzi 1988), as well as benefactive datives (30d).
(30) a. Juan parece habérselo dicho a Pedro.
‘Juan seems to have told Pedro about it.’
b. Tu respuesta no parece haberle molestado a Juan.
‘Your answer does not seem to have bothered Juan.’
c. La obra parece gustarle al público.
‘The play seems to be pleasant to the audience.’
d.Juliaparecehaberle
estropeadoel cocheasu novio.
Juliaseemsto.have.cldatbroken thecar toherboyfriend
‘Julia seems to have broken her boyfriend’s car’.
A third piece of evidence which indicates that the dative is merged in a high position
in anticausative constructions comes from the scope of adverbial phrases like de nuevo
or otra vez ‘again’. It has been noted that these phrases are ambiguous in that they can
modify the whole causative event or just the (change of) state resulting from the event
expressed by the predicate (see von Stechow 1995, among others). This is the reason
why a sentence such as (31) has two possible readings:
(31) John has broken the car again.
i. It is the second time John has broken the car.
ii. It is the second time the car has been broken.
One possible structural correlation of this is that the adverb again can leave John
­outside its scope, if it modifies the VP, or inside its scope, if it modifies a higher node
including the external argument. Interestingly enough, the same ambiguity obtains
with respect to the dative argument in the sentences we are analyzing. As the glosses
indicate, the adverb can leave out of its scope only the dative, but not the theme or any
other internal argument.
© 2013. John Benjamins Publishing Company
All rights reserved
 Olga Fernández-Soriano & Amaya Mendikoetxea
(32) AJuan se le
haroto el cochede nuevo.
toJuanse clDAT hasbrokenthecar again
i. ‘It is the second time that Juan has accidentally broken the car.’
ii.‘The car was previously broken and now Juan has accidentally broken it
for the second time.’
As expected, this ambiguity does not obtain with other datives, such as goals or
­benefactives. In (33a) the interpretation in which the prize was given to someone else
is out: Cela must have received the Nobel Prize twice. For the same reason, John must
have had his car stolen twice in (33b):
(33) a. A Celale han dado el premioNobel de nuevo.
toCelacldathavegiventheprize Nobelagain
‘They have given the Nobel prize to Cela again.’
b.A Juanle han robadoel cochede nuevo.
toJuancldathavestolen thecar again
‘They have stolen Juan’s car again.’
Finally, the behavior of bare NPs can also indirectly provide evidence for the high
­position of causer datives. Subject bare NPs are always VP-internal, as in (25) above,
and are interpreted existentially (see Diesing 1992). As mentioned in Section 2, bare
NPs are not possible in anticausative constructions (34a) (and (24) above). These constructions do not allow for the existential interpretation of the theme argument, which
must be externalized as a consequence of the clause-internal control analysis described
in sec. 2.2 above. What is relevant here is that the presence of the dative argument
allows the theme to be realized as a bare NP, with existential interpretation (34b, c).
We take this as evidence that externalization of the theme argument is not required/
allowed when the dative argument is added in a high position in the structure.
(34) a. *Serompen vasos.
sebreak-3plglasses
b.AJuan se le rompen vasoscontinuamente.
toJuanse cldat break-3plglassesall-the-time
‘Juan (unintentionally) breaks glasses all the time.’
c.AJuan se le pierdencosas.
toJuanse cldatlose-3plthings
‘Juan loses things.’
Summarizing, from the facts presented above involving word order, raising ­contexts
and the scope of de nuevo ‘again’, we conclude that the non selected dative in
­anticausative se-structures behaves like an external argument, while the nominative
theme is an internal argument, which does not undergo raising and may be realized
as a bare NP. Our account for this fact will be that this element is introduced by a high
applicative head and behaves like a (quirky) subject.
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions 
3.2 Syntactic structure: Datives as quirky subjects
Having established that the dative argument has properties similar to those of external arguments, we now explore the hypothesis that anticausative constructions with
dative arguments are the result of adding a dative argument to a construction like (23).
In principle, there are two possible positions for the dative argument: (i) specifier of
v, where canonical agents are projected (Kallulli 1999), a position which is occupied
by pro in the anticausative construction (23) above, or (ii) specifier of an additional
(higher) projection: an applicative head (Cuervo 2003). As Schäfer (2008: 3.5–3.6)
observes, in dative anticausative constructions the dative can be integrated in the event
in many different ways. The terms ‘unintentional’ or ‘accidental’ causer do not capture
all facets of the construction: a number of subinterpretations can be distinguished
crosslinguistically. Thus, in (35), the girl can be an unintentional or accidental causer,
an involuntary or indirect facilitator (‘the girl let the doors open’) or an unexpected
causer (‘the girl managed to open the doors, unexpectedly’) and this depends on contextual/ pragmatic factors.
(35) Ala niña se le abrieron laspuertas.
tothegirl se cldat opened-3plthedoors
It is this polysemy that leads Schäfer to claim that the dative does not occupy the canonical ⟨Spec, v⟩ subject position. Canonical transitive subjects may act unintentionally or
accidentally as suggested by the adjuncts in (36), but the other two readings do not obtain.
(36) La niña abrió la puerta sin querer (al apoyarse).
‘The girl opened the door accidentally (by leaning on it)’
We conclude with Schäfer (2008: 110) that “the relation between the dative causer
and the event is much less constrained than the relation between canonical causers or
canonical agents and the event.” From this, it follows that the dative argument is not
introduced by v, but by an additional head.10
We assume, therefore, that the alternative proposal by Cuervo (2003) that dative
arguments in anticausatives are introduced by an applicative head is basically correct
(hypothesis (ii)). An applicative is an element (usually a morpheme) that increases the
valency of a verb by adding a noncore argument to it. In recent literature it has been
claimed that there are (at least) two types of applicatives (Pylkkänen 2008): high applicatives, which denote a thematic role (a relation between an event and an individual)
and combine with the VP by event identification, and low applicatives, which denote
a relation between individuals (e.g. transfer of possession) and are internal to the VP
. Another argument used by Schäfer (2008) concerns the fact that the unintentional
causer reading is licensed with pure unaccusatives that lack a position for the external argument, namely internally caused COS verbs, and, arguably, do not project v. We disagree with
Schäfer (2008) in this respect, as will be shown in Section 4.
© 2013. John Benjamins Publishing Company
All rights reserved
 Olga Fernández-Soriano & Amaya Mendikoetxea
(see Cuervo 2003 for a three-way distinction of applicative heads). Our proposal, in
line with Cuervo (2003) and Schäfer (2008), is that the noncore dative under study is
introduced by a high applicative head between T and the projection of the clitic.
(37)
TP
T′
AppIP
DP
→ Dative subject
Appl′
A Juan Appl
le
→ Anticausative
CLP
SE
→ Transitive
vP
CL
v′
pro
v
VP
V
rompe
DP
el vaso
Note that the transitive structure (Juan rompió el vaso) involves simply a vP projection,
with Juan in ⟨Spec, v⟩ ; the anticausative construction (el vaso se rompió) is represented
in (23): with pro (an unspecified cause) in the canonical subject position and the clitic
se heading its own projection; finally, the dative argument construction involves the
projection of an applicative phrase above the clitic projection, whose head is the clitic
le and whose specifier is occupied by the dative argument a Juan, which binds pro.
Syntactically, applicative heads introduce noncore arguments, like the dative argument in anticausative constructions. It is our contention that this element has properties
associated with what is known as quirky subjects: inherently (ACC, DAT or GEN) Case
marked, nominal arguments in subject position, such as those illustrated in (38) for
Icelandic (similar examples may be found in languages such as Croatian and Georgian):
(38) a. Hennileidduststrákarnir
herdat bored the boys.
‘She found the boys boring.’
b.Mig vantarpeninga.
meacclacks moneyacc
‘I lack money.’
© 2013. John Benjamins Publishing Company
All rights reserved
(Sigurðsson 2002a: 692)
(Zaenen et al. 1985: 454–455)
Non selected dative arguments in Spanish anticausative constructions 
Sigurðsson (2002a, b) distinguishes Icelandic quirky subjects from non-nominative
subject-like elements found in languages such as Russian and German, (39). While the
latter have several properties which make them similar to canonical subjects, quirky
subjects behave like ordinary nominative subjects with respect to several syntactic
diagnostics for subjecthood (raising, reflexivization, word order and so on).11
(39) Mir wurdegeholfenGerman
medatwas helped
‘I was helped (by somebody).’
(Sigurðsson 2002a: 694)
The distinction is based on the nature of the feature ‘person’. As is well known, the
finite verb in Icelandic quirky subject constructions may show third-person agreement
(singular or plural), but no 2p or 1p agreement, as illustrated in (40):
(40) a. Égveit að homunlíka
þeir
I knowthathedat like-3pltheynom
‘I know that he likes them.’
b.*Égveit að homunlíkið þið
I knowthathedat like-2plyou-plnom
‘I know that he likes you.’
c.*Égveit að homunlíkum við
I knowthathedat like-1plwenom
(Sigurdsson 2002a: 719–720)
‘I know that he likes us.’
In contrast, similar DAT-NOM constructions in languages like German or Spanish do
not show this person constraint, as illustrated in (41) for Spanish:
(41) Sé
queaél legustan
ellos/
gustas tú/
know-1sgthattohimcldatlike-3pltheynom/like-2sgyounom/
gustamos
nosotros.
like-1plwenom
‘I know that he likes them/you/ us.’
The structures in (41) contain the V gustar, a Psych Verb whose experiencer argument
is syntactically realized as a dative that shows certain subject-like properties, but does
not show person restrictions and is not a quirky subject (see (4) and (30c) above).12
. See, among many others, Thráinsson (1979), Zaenen et al. (1985) and Sigurðsson (1989,
1996).
. Rivero (2004) shows, however, that some DAT-NOM constructions in Spanish display
person restrictions of the type found in Icelandic, namely constructions with the verb
antojar(se) ‘fancy’, ocurrir(se) ‘come up with’ and olvidar(se) ‘forget’. She analyzes these cases
as instances of the Person-Case Constraint (Bonet 1991). According to Rivero (2004), it is the
presence of a reflexive clitic that triggers the person constrain. The presence of the reflexive
© 2013. John Benjamins Publishing Company
All rights reserved
 Olga Fernández-Soriano & Amaya Mendikoetxea
In Icelandic, the facts in (40) are accounted for by assuming that there is split
­person-number agreement. Under the assumption that 3p is non-person (see ­Benveniste
1966 and Kayne 1993, among others), agreement in (38) and (40a) involves only number
agreement. The dative enters a default (3p) agreement relation with the finite verb, and
hence the nominative argument is blocked from entering a person agreement relation
with the verb but it can still enter a number agreement relation. Thus the verb displays
default 3p and full number agreement. Additionally, the dative checks the Extended
Projection Principle (EPP) feature of T(ense) and thus qualifies as the clausal subject.13
Our contention is that this is also what happens in Spanish anticausative
­constructions with dative subjects. In (37), the dative triggers (null or default) person
agreement with the finite V and checks the EPP feature of T, while the theme argument
shows number agreement and checks the NOM feature:14
(42) [tp T[applp a Juan [app le [clp se [vp pro [v’ v [vp rompe [dp el vaso ]]]]]]]
epp
epp3p (def)/sg sg
nom
dat
nom
3p (def)
Interestingly, while anticausative constructions do not display person constraints (43),
when a dative element is added, these structures exhibit properties which are reminiscent of the Icelandic quirky Case constructions, as illustrated in (44) for the verb
hundir ‘sink’:
(43) a. El barcosehundió.
theboat sesank-3sg
‘The boat sank.’
b.(Yo)me
hundí.
(I) cl-1sg sank-1sg
‘I sank.’
clitic also distinguishes these verbs from gustar ‘like’ in (41). Both patterns of dative experiencer verbs are found in Old Spanish. See Rivero and Diaconescu (2007) for the diachronic
evolution of both types of constructions in Spanish and Romanian.
. As claimed by Sigurdsson (2002b), this is also what happens in English there constructions such as There have been three men arrested, in which expletive there checks the EPP
feature and enters a null person agreement relation with V, while three men triggers number
agreement.
. The feature composition and agreement relations in (42) are oversimplified for clarity of
exposition. T also contains subject agreement features that need to enter an agreement relation with a matching DP (see Chomsky 2000, 2001). Additionally, we have not said anything
about the feature composition and agreement relations of se, pro and v. These elements also
enter complex agreement relations. See Mendikoetxea (2008) for a proposal on this.
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions 
(44) a. AJuan se le hundió elbarco.
toJuandat se cldatsank-3sgtheboatnom
‘Juan (accidentally) sank the boat.’
b.*AJuan me
lehundíyo.
to Juandat cl-1sg cldatsank-1sgInom
The structure in (44b) violates the well known Person-Case constraint (Bonet 1991,
1994) (see Note 12 above): a universal constraint that blocks co-occurrence of a dative
clitic and clitics other than 3p, as shown in (45):
(45) *Te
le
hanrecomendado.
clacc-2sg cldat-3sghaverecommended
‘They have recommended you to him.’
There are two facts that indicate that this is not, however, just another instance of this
constraint. First, a weaker version of the constraint allows (in most dialects) combinations of 1p and 2p clitics in structures like (46a), so that the restriction only applies
with 3p dative clitics (Bonet 1991: 82, but see Ormazábal & Romero 2007). But, though
the combination of 1p and 2p clitics is also grammatical in anticausatives, (46b) cannot
receive the accidental causer/dative subject interpretation in (ii). Instead the sequence
can only be interpreted as containing an ethical dative, which, as is well-known, is not
affected by the constraint, or, interestingly, an affected (low applicative) dative.
(46) a. Te
me
hanrecomendado.
clacc-2sg cldat-1sghaverecommended
‘They have recommended you to me.’
b.Ami túte
mehundiste.
tomedat you clacc-2sg cldat-1sgsank
i. *I accidentally caused you to sink.
ii. You sank on me/You sank and I am affected by that fact.
Second, a similar situation is found in impersonal constructions with the clitic se in
Spanish, with no dative clitic, like those in (47) (see also D’Alessandro 2004: 117 for
similar constructions in Italian). Mendikoetxea (2008) takes the contrast between
(47a) and (47b) to indicate that agreement in impersonal constructions like (47a) is
agreement in number, and not in person, as we are also claiming for the anticausative
constructions here:
(47) a. Desdeaquíseven las montañas.
fromheresesee-plthemountainsnom
‘One can see the mountains from here.’
b.*Desdeaquísevemos nosotros.
from here sesee-1plweNOM
‘One can see us from here.’
© 2013. John Benjamins Publishing Company
All rights reserved
 Olga Fernández-Soriano & Amaya Mendikoetxea
What this seems to indicate is that the person constraint observed is not a subcase of
the Person Case constraint that applies between dative and accusative clitics, but it
has to do with processes that take place when an element enters multiple agreement.
Multiple agreement is only possible when the two DPs that enter an agreement relation
with V (or T) show no feature mismatch, as has also been claimed for Icelandic quirky
Case constructions (Taraldsen 1995; Anagnostopolou 2005). In (47), T(ense) (and V)
enters multiple agreement with the DP las montañas ‘the mountains’ and a default 3p
pronominal element in subject position (like pro in (37)); this is possible as long as
there is no feature mismatch between the two DPs (see Mendikoetxea 2008). As for the
dative subject construction, in (44a) the dative enters a null (3p) agreement relation
with v, which also enters a (number) agreement relation with the theme argument.
But when the theme argument is other than 3p a mismatch of features makes multiple
agreement impossible.
In sum, in addition to the properties specified in Section 3.1 above, we take the
person constraints observed as crucial evidence for the status of the dative a­ rgument
as a subject, like Icelandic quirky subjects. Both constructions show that the p
­ roperties
associated with canonical subjects may be scattered along more than one element:
the EPP feature and person agreement belong to the dative element, while NOM
Case and number agreement belong to the theme argument, and multiple agreement
with v/T ensures that all the features are properly checked. In the following section
we focus on the properties of dative subjects as effectors and explore the relation
between ­interpretation and subjecthood.
4. Th
e properties of dative subjects and event interpretation:
Datives as effectors
In this section, we explore our hypothesis (iii) that the subject-properties displayed
by the noncore dative mostly follow from the fact that it participates in the first
(­initial) subevent of the predicate. Semantically, applicative heads have very little content. They have been claimed to establish an abstract have-relation between the DP
in its specifier and its complement (Harley 1995, 1998, 2002; Cuervo 2003; McIntyre
2006; Schäfer 2008, among others). Depending on the complement they take, different (have‑)relations and specific interpretations are obtained. In the case of high
applicatives an argument is added to a COS event and the meaning obtained is that
the dative DP ‘possesses’ that event (Cuervo 2003). In other words, the applicative
head expresses an abstract have-relation according to which dative causers have the
COS event. On the other hand, as has been mentioned above, one property that has
been attributed to subjects is that of being event initiators. That is to say, external
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions 
arguments always ­participate in the first (initial) subevent denoted by the predicate
(Harley 1995; G
­ rimshaw 1990; Ramchand 2008). In our view, this is the crucial property of dative subjects, their accidental causer reading and the usually human nature
that has been attributed to them are in some sense byproducts of that property. We
explore both in turn.
4.1 Dative subjects and the cause/initial subevent
We have seen that the semantic role associated with the subject of verbs like transitive
romper is that of an underspecified causer, an effector (sec 2.1 above). Additionally,
we have taken the view that in the anticausative version the cause is not an action
performed by an agent but it is interpreted as a property of the internal argument,
which is somewhat responsible for the change of state denoted by the predicate (stative
causation, Brosseau & Ritter 1991). We propose that, if a dative argument is added in
a high position (external to VP), the structure ends up having two effectors, in line
with proposals that argue that an event can have more than one effector (KoontzGarboden 2009; Piñón 2001), as long as they are of different types.15 For the structures
under study we propose that an accidental causer type of effector is added to the
anticausative construction which already contains a (stative) causer. Furthermore,
the causing subevent is understood as being (accidentally) initiated by the dative
argument. These assumptions entail that for a dative argument to be interpreted as
unintentional causer and thus behave as a subject, the predicate must contain a cause
subevent. The first consequence of this proposal is that verbs projecting a COS event
but not a causative event cannot have this type of dative argument. This is the case for
internally caused COS predicates.
Internally caused COS verbs like palidecer ‘turn pale’, oscurecer ‘darken’, adelgazar ‘get thin’, envejecer ‘grow old’, crecer ‘grow’, florecer ‘blossom’, hervir ‘boil’, arder
‘burn’, enfermar ‘get sick’ and so on denote “causation initiated by, but also residing
in, the single argument and hence dependent on its properties” (L&RH: 94). These
verbs lack a causative predicate as part of their lexical s­ emantics and therefore do
not have transitive counterparts (Mendikoetxea 1999, 2000). We contend that unintentional causer/effector datives cannot be added to these predicates. Thus, the
. In particular Piñón (2001) claims that a single event cannot have two effectors of the
same type –i.e. it may not have more than one agent, more than one instrument, etc., but can
have one of each.
© 2013. John Benjamins Publishing Company
All rights reserved
 Olga Fernández-Soriano & Amaya Mendikoetxea
only possible interpretation for sentences like (48) is that of affectedness (i.e. low
applicative):16
(48) a. A Maríale adelgazó elniño.
toMaría cldat got-thin-3sgthechild
‘The child lost weight and Mary was affected.’
b.Ami padrele crecieronlostomates.
tomyfather cldat grow-3plthetomatoes
‘The tomatoes grew and my father was affected.’
In Fernández-Soriano and Mendikoetxea (2011) evidence is provided to show that
datives with internally caused non alternating COS predicates are not effector
datives (i.e. ‘dative subjects’). We summarize the arguments below.
Adjuncts such as por sí sólo/a, por sí mismo/a ‘by itself ’ or él/ella sólo/a ‘alone’
can appear with anticausative se-predicates, indicating that the subject argument
. Very little attention has been paid to these structures and their interpretation is
­controversial. Schäfer (2008), taking data from Rivero (2004) and Cuervo (2003), claims
that sentences like (ia) in Spanish and (ib,c) in Italian contain an unintentional causer dative
(we only give the translation provided for this meaning, but of course the affected reading is
­possible).
(i)a.A Juanle florecen losárboles.
toJohncldat bloom-3plthetrees
‘John causes the trees to somehow bloom (i.e. he is a good gardener).’
b.
A Francescae bollito fuoriil latte (per errore).
toFrancescaisboiledover themilk(by mistake)
‘Francesca accidentally caused the milk to boil over.’
c.AFrancosonoappassitetuttele piantein
giardino(per errore)
toFrancoare wilted all theplantsin.thegarden (by mistake)
‘Franco accidentally caused all the plants in the garden to wilt.’
(Schäfer 2008: 69)
For us, however, no unintentional causer reading is available in these examples. Together with
the (most salient) affected interpretation, the second meaning obtained in these cases, as the
glosses indicate, is one of ‘indirect’ causation: ‘Juan makes the trees blossom’, ‘Francesca let
the milk boil/ the plants wilt’. Schäfer (2008), in fact, acknowledges that no example could
be provided of accidental causer datives in Romanian or Albanian with internally caused
anticausatives. Cuervo (2003) is not very explicit about these constructions. Her claim is that
what she calls ‘unintentional responsibility’, a subtype of the unintentional causer reading,
is either unavailable or ‘less available’ in constructions with COS verbs that lack a transitive
counterpart (though no explanation is provided for this fact). We claim that the unintentional
causer interpretation is unavailable in sentences with internally caused COS verbs in general.
Whatever responsibility the dative may have in the event can only be contextually derived.
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions 
­ articipates in both the initial (cause) and the COS subevent and may be interp
preted as both theme (undergoer) and cause, as in (49). If, as we have claimed, dative
causers/ effectors participate in the causing subevent, the prediction is that if this
argument is added, this type of adjuncts is impossible. This is what the contrast
between (49) and (50) indicates:
(49) a. Elbarcosehundióporsí mismo.
theboat sesank by itself
b.Lapuertase abrió por sí misma.
thedoor seopenedby itself
(50) a. *(Amí)el barcoseme
hundióporsí mismo.
tome theboat se cl1stdatsank by itself
b.*AJuansele abrió la puertaporsí misma.
toJuanse cldatopenedthedoor by itself
Internally caused COS verbs do not normally allow for these adjuncts, a fact that has
been used to argue that they lack a causative predicate as part of their lexical semantics
(Mendikoetxea 1999, 2000): they are monoeventive, pure unaccusative predicates. In
some contexts, though, their presence can be forced, and then, the dative does not
block the presence of the por sí mismo constituent. Thus, in (51) the by itself adjuncts
indicate a contrast with the context in parentheses and are compatible with the datives,
which shows that the dative with internally caused COS predicates is not an effector
dative:
(51) a. ?Al jardinero
las flores le crecieron
por sí mismas
To the gardener the flowers CLDAT grew-3pl by themselves
(no tuvo que abonarlas).
(he did not have to fertilize them)
b.?A Miguel el pollo
le engordó
por sí mismo
To Miguel the chicken CLDAT got.fat-3sg by itself
(no tuvo que darle nada).
(he did not have to give it anything)
Furthermore, raising phenomena indicate that the dative argument with internally
caused COS verbs occupies a low position in the structure, and hence cannot undergo
raising (52b, c), as opposed to the dative argument with externally caused COS in
­anticausative constructions (52a) (and (27)–(28) above):
(52) a. AMaríaparece habérsele quemadola comida.
ToMaría seems tohave.se.cldatburn
thefood
‘María seems to have burned the food accidentally.’
© 2013. John Benjamins Publishing Company
All rights reserved
 Olga Fernández-Soriano & Amaya Mendikoetxea
b.#APinocho parecíancrecerle losmonstruos.
To Pinocchio seem to grow.cldatthemonsters
c.#AMaríaparecía engordarleelpollo.
To María seemed to get-fat.cldatthechicken
Moreover, internally caused COS verbs without se and with a dative cannot take bare
NP themes (53b, c). In this respect they contrast with the effector dative constructions under study (53a) (c.f. Section 3.1, example (34)).
(53) a. AJuan se le rompencosas.
ToJuanse cldatbreak-3plthings
‘Juan unintentionally breaks things.’
b.*A Juanle engordanpollos.
To Juan cldatget.fat chicken
We take this to show that in the case of internally caused COS verbs the dative does not
behave as a subject and remains in a low position, whereas the agreeing element displays subject properties (and hence cannot be a bare NP, like the subject of externally
caused COS verbs) (see (24) above).
A final argument has to do with the inferential relationship observed between
the causative and the anticausative structure of a given verb; a sentence such as Juan
rompió el plato ‘Juan broke the plate’ entails El plato se rompió ‘the plate broke’. The
consequence of this fact is that negation of the anticausative form is impossible if the
causative is asserted (#El plato no se rompió, Juan rompió el plato ‘The plate didn’t
break, Juan broke the plate’). However, as Koontz-Garboden (2009) points out, this is
not necessarily the case: there are contexts where the causative can be asserted while
the corresponding anticausative is negated.17 This can be taken to mean that the cause
argument is not eliminated in the anticausative version. In the cases under study, the
situation is very clear since contrastive pronouns are used to mark the arguments
involved. Interestingly enough, what the data show is that the contrast can be obtained
both between the effector dative and the agent of the transitive construction (i.e.
accidental causer vs. agent) (54a), and between the anticausative-se structure with no
dative and the one with an accidental causer dative (54b):
(54) a. El vasono se me rompió*(amí),lo rompistetú.
Thevasenotse cl1sgdatbreak-3sg tome, clacc brokeyou
‘I did not break the vase accidentally, you broke it.’
b.El vasono serompió,se terompió
ati.
Thevasenotsebreak, se cl2sgdat broke toyou
‘The vase did not break, you broke it accidentally.’
. We refer the reader to Koontz-Garboden (2009: 103) for the relevant examples.
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions 
In sum, what is negated in (54b) is that the theme is also the cause of the COS event
and, instead, it is asserted that the dative argument is involved in the causing event.
Again, we cannot obtain parallel structures with internally caused COS verbs. The case
of caer, without se (55c), is especially clear:
(55) a. La lecheno hirvió, #tehirvió
ati.
themilknotboiled,cl2sgdatboiledtoyou
b.El rosalno floreció,
#lefloreció
aJuan.
therose treenot blossomed,cl3sgdatblossomedtoJuan
c.El rayonocayó, #tecayó
ati.
Thebolt of lightningnotfell,cl2sgdatfell onyou
These facts support the hypothesis that anticausative externally caused se-predicates
involve a (underspecified) cause event and this fact allows them to take a high
­applicative node whose specifier is an argument, with dative case, interpreted as an
accidental causer. The dative element participates in the initial subevent of the predicate and behaves as an event initiator in the sense of Harley (1995) among others. This,
in our opinion, is the property which distinguishes accidental causer effector datives
from other (internal) noncore datives.
4.2 The semantic properties of dative subjects. More on ‘subjecthood’
Let us now look at the human restriction on the dative subject. The V caer ‘fall’ is
interesting for our purposes. One can assume that (56a) constitutes a clear case of
unintentional causer dative. If this is the case, it does not seem plausible to propose a
different analysis for (56b). But (56b), crucially, does not entail any (accidental) cause
to be attributed to the dative. The same line of reasoning can be extended to (56c) and
even to (56d) which contain non-human datives:
(56) a. AJuan sele cayó el libro.
toJuandat se cldatfell-3sgthebook
‘Juan (unintentionally) dropped the book.’
b.AJuan se le cayóel
pelo.
toJuandattoJuanse cldatfell-3sgthe hair
‘Juan lost his hair.’
c.Ala muñeca se le cayó
el
pelo.
tothedolldat tothedoll se cldatfell-3sgthe hair
‘The doll lost its hair.’
d.Alcepillo
se lehan
caído
to the hairbrushdat se cldathave-3pl
lospelos.
fellthe hairs
‘The hairbrush has lost its bristles.’
© 2013. John Benjamins Publishing Company
All rights reserved
 Olga Fernández-Soriano & Amaya Mendikoetxea
It has been claimed that the ‘human restriction’ for applicatives in general is actually
related to possession (as Schäfer 2008 notes). More specifically, as McIntyre (2006)
shows, non human affected datives are possible only if an inalienable possession
is established between the dative and the theme. We claim that this is also the case
with effector datives, that is, non human dative subjects are possible only if they
­participate in the first subevent of the predicate. In this case, they are not cause but
pure source (i.e. event initiators) and a relation of possession is established between
the dative and the internal argument which undergoes the change of state.18 This can
be seen in the following examples. In (57a) the dative is interpreted as the source of
the change of state undergone by the theme, which, in turn, is interpreted as possessed
by the dative DP (as in (57a), where the pot has a handle which falls off it). This makes
the accidental causer reading available (and preferred) and allows for the effector
dative. The same goes for (57b) and (57c):
(57) a. Ala olla se le salió/cayó elasa.
tothepotdat se cldatwent-out/fellthehandle.
‘The pot’s handle went out/fell’. ‘The pot lost its handle.’
b.Ala lavadora
se le estropeóel filtro.
tothewashing machinedat se cldatbroke thefilter
‘The filter of the washing machine broke.’
c.Aljuguete se le salió
unmuelleque teníadentro.
to the toydat se cldatwent-outa springthat ithad inside
‘A spring went out of the toy/The toy lost a spring that was inside it.’
In sum, being event initiators is a crucial property of subjects. As Harley (1995:78)
states it, “interpretations triggered by the subject […] involve forcing the type of event
denoted by the verb to match the type of the event of which the subject could be the
initiator”. So the subject status of noncore datives introduced in a high applicative head
seems to rely on their participation in the initial subevent of the predicate.
5. C
onclusions
In this paper, we have examined the properties of a type of argument structure
­variation that involves the addition of noncore arguments with subject-like properties,
in which the characteristics of subjecthood appear to be scattered across more than
one nominal argument. In particular, an analysis has been provided for the s­ yntactic
. Schäfer (2008) shows that in Agul the unintentional causer is not only the possessor of
the change of state event but also its source.
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions 
properties and semantic interpretation of constructions in which a dative element
is added to the anticausative (reflexively marked) variant of a transitive verb in the
­causative ­alternation, focusing on the interpretation of the noncore dative argument
as an unintentional or accidental causer.
Semantically, we have referred to these elements as effector datives, ­following
Koontz-Garboden (2009). These datives are added to verbs denoting an externally
caused change of state, which involves two subevents: a causing subevent and a
resultative subevent (see e.g. L&RH 1995). Crucially, these verbs have an underspecified causing event, this is why they participate in the causative alternation and
can be constructed with noncore effector datives. The cause predicate associated with them is syntactically projected as a vP node, both in the causative and the
anticausative construction. The dative argument is introduced by a high applicative
phrase (Cuervo 2003; Pylkännen 2008; Schäfer 2008) and has subject properties,
akin to those of an external argument, while the theme argument ­undergoing the
COS remains internal to VP, i.e. it is never externalized although it shows nominative Case and triggers verbal agreement. In fact, we have claimed that verbal
agreement is agreement in number with the theme argument, while the verb shows
default 3p features in agreement with the dative element, as in Icelandic quirky Case
constructions.
One of our main conclusions is that the subject properties displayed by the
­noncore dative are directly related to the fact that it participates in the initial subevent
of the predicate, like external arguments in general (Harley 1995). These assumptions
entail that for a dative argument to be interpreted as unintentional causer the predicate must contain a cause subevent. We have also explored some consequences of the
applicative nature of the node introducing effector datives, in particular of the ‘have’
relation that has been claimed to be established between the applicative and the COS.
The idea we have developed is that it is this ‘have’ relation and the ‘initiator/source’
nature of the event that is needed to have dative subjects.
References
Alexiadou, A., Anagnostopoulou, E. & Schäfer, F. 2006a. The properties of anticausatives crosslinguistically. In Phases of Interpretation, M. Frascarelli (ed.), 187–211. Berlin: Mouton de
Gruyter.
Alexiadou, A., Anagnostopoulou, E. & Schäfer, F. 2006b. The fine structure of anti-causatives.
In Proceedings of NELS 36, C. Davis, A.-R. Deal & Y. Zabbal (eds), 115–128. Amherst MA:
GLSA.
Anagnostopoulou, E. 2005. Strong and weak person restrictions: A feature checking analysis.
In Clitics and Affix Combinations [Linguistik Aktuell/Linguistics Today 74], L. Heggie &
F. Ordoñez (eds), 199–235. Amsterdam: John Benjamins.
© 2013. John Benjamins Publishing Company
All rights reserved
 Olga Fernández-Soriano & Amaya Mendikoetxea
Belletti, A. & Rizzi, L. 1988. Psych-verbs and Theta-theory. Natural Language and Linguistic
Theory 6: 291–352.
Benveniste, E. 1966. Problèmes de Linguistique Générale. Paris: Gallimard.
Bonet, E. 1991. Morphology after Syntax: Pronominal Clitics in Romance. Ph.D. dissertation, MIT.
Bonet, E. 1994. The Person-Case Constraint: A morphological approach. In The MorphologySyntax Connection [MIT Papers in Linguistics 22], H. Harley & C. Phillips (eds) 35–52.
Cambridge MA: MIT.
Borer, H. 2005. The Normal Course of Events. Oxford: OUP.
Brousseau, A.M. & Ritter, E. 1991. A non-unified analysis of agentive verbs. In WCCFL 10,
Dawn Bates (ed.), 53–64. Stanford CA: CSLI.
Burzio, L. 1981. Intransitive Verbs and Italian Auxiliaries. Ph.D. dissertation, MIT.
Burzio, L. 1986. Italian Syntax. Dordrecht: Reidel.
Chierchia, G. 2004. A semantics for unaccusatives and its syntactic consequences. In The
­Unaccusativity Puzzle: Explorations of the Syntax-Lexicon Interface, A. Alexiadou,
E. ­Anagnostopoulou & M. Everaert (eds), 22–59. Oxford: OUP (initially circulated as
Chierchia (1989), as a manuscript from Cornell University).
Chomsky, N. 1981. Lectures on Government and Binding. Dordrecht: Foris.
Chomsky, N. 2000. Minimalist inquiries. In Step by Step: Essays on Minimalist Syntax in Honor
of Howard Lasnik, R. Martin, D. Michaels & J. Uriagereka (eds.), 89–156. Cambridge MA:
The MIT Press.
Chomsky, N. 2001. Derivation by phase. In Ken Hale: A Life in Language, Michael Kenstowicz
(ed.), 1–52 Cambridge MA: The MIT Press.
Cuervo, M.C. 2003. Datives at Large. Ph.D. dissertation, MIT.
Cuervo, M.C. 2008. La alternancia causativa y su interacción con argumentos dativos. RLA.
Revista de Lingüística Teórica y Aplicada 46(1): 55–79.
D’Alessandro, R.A.G., 2004. Impersonal si Constructions: Agreement and Interpretation. Ph.D.
dissertation, University of Stuttgart.
Diesing, M. 1992. Indefinites. Cambridge MA: The MIT Press.
Fernández-Soriano, O. 1999. Two types of impersonal constructions in Spanish: Locative and
dative subjects. Syntax 2(2): 101–140.
Fernández-Soriano, O. 2000. Datives in construction with unaccusative se. Catalan Working
Papers in Linguistics 7: 89–105.
Fernández-Soriano, O. & Mendikoetxea, A. 2011. Non-selected dative subjects in anticausative
constructions. Archivio Glottologico Italiano XCVI(Fascicolo I C.M. 07.10.15): 87–128.
Folli, R. 2002. Constructing telicity in English and Italian. Ph.D. dissertation, University of
Oxford.
Grimshaw, J. 1982. On the lexical representation of Romance reflexive clitics. In The Mental
Representation of Grammatical Relations, J. Bresnan (ed.), 78–148. Cambridge MA: The
MIT Press.
Grimshaw, J. 1990. Argument Structure. Cambridge MA: The MIT press.
Guerssel, M., Hale, K., Laughren, M., Levin, B. & White Eagle, J. 1985. A crosslinguistic study of
transitivity alternations. In Proceedings of the 21st Annual Meeting of the Chicago ­Linguistics
Society, Vol 2: Parasession on causatives and agentivity, 48–63. Chicago IL: ­Chicago
­Linguistics Society.
Hale, K & Keyser, J. 1986. Some transitivity alternations in English. Lexicon Project Working
Papers 7. Cambridge MA: Center for Cognitive Sciences, MIT.
Hale, K & Keyser, J. 1987. A View from the Middle [Lexicon Project Working Papers 10].
­Cambridge MA: The MIT Press.
© 2013. John Benjamins Publishing Company
All rights reserved
Non selected dative arguments in Spanish anticausative constructions 
Hale, K & Keyser, J. 1993. On argument structure and the lexical expression of syntactic
­relations. In The View from Building 20, K. Hale & S.J. Keyser (eds), 53–109. Cambridge
MA: The MIT Press.
Harley, H. 1995. Subjects, Events and Licensing. Ph.D. dissertation, MIT.
Harley, H. 1998. You’re having me on: Aspects of have. In La grammaire de la possession,
J. Guéron & A. Zribi-Hertz (eds), 195–226. Paris: Université Paris X – Nanterre.
Harley, H. 2002. Possession and the double object construction. In Yearbook of Linguistic
­Variation, Vol. 2, P. Pica & J. Rooryck (eds), 31–70. Amsterdam: John Benjamins.
Haspelmath, M. 1993. More on the typology of inchoative/causative verb alternations. In
­Causatives and Transitivity [Studies in Language Companion Series 23], B. Comrie &
M. Polinsky (eds), 87–120. Amsterdam: John Benjamins.
Kallulli, D. 1999. Non-active morphology in Albanian and Event (De)composition. In Crossing
Boundaries, I. Kenesei (ed.), 263–292. Amsterdam: John Benjamins.
Kallulli, D. 2006. A unified analysis of passives, anticausatives and reflexives. In Empirical Issues
in Formal Syntax and Semantics 6, O. Bonami & P. Cabredo-Hofherr (eds), 201–225.
⟨http://www.cssp.cnrs.fr/eiss6/⟩
Kallulli, D. 2007. Rethinking the passive/anticausative distinction. Linguistic Inquiry 38:
770–7780.
Kayne, R.S. 1993. Towards a modular theory of auxiliary selection. Studia Linguistica 47:
3–31.
Koontz-Garboden, A. 2009. Anticausativization. Natural Language and Linguistic Theory 27:
77–138.
Kratzer, A. 1996. Severing the external argument from its verb. In Phrase Structure and the
­Lexicon, J. Rooryck & L. Zaring (eds), 109–137. Dordrecht: Kluwer.
Kratzer, A. 2005. Building resultatives. In Event Arguments in Syntax, Semantics, and Discourse,
C. Maienborn & A. Wollstein-Leisten (eds), 178–212. Tubingen: Niemeyer.
Levin, B. & Rappaport Hovav, M. 1995. Unaccusativity at the Lexical Semantics-Syntax Interface.
Cambridge MA: The MIT Press.
Marantz, A. 1984. On the Nature of Grammatical Relations. Cambridge MA: The MIT Press.
Masullo, P. 1992. Incorporation and Case Theory in Spanish. A Crosslinguistics Perspective.
Ph.D. thesis, University of Washington.
McIntyre, A. 2006. The interpretation of German datives and English have. In Datives and Other
Cases [Studies in Language Companion Series 75], D. Hole, A. Meinunger & W. Abraham
(eds.), 185–211. Amsterdam: John Benjamins.
Mendikoetxea, A. 1999. Construcciones inacusativas y pasivas. In Gramática Descriptiva de la
Lengua Española, Ch. 25, I. Bosque & V. Demonte (eds). Madrid: Espasa-Calpe.
Mendikoetxea, A. 2000. Relaciones de interficie: Los verbos de cambio de estado. In Cuadernos
de Lingüísitca VII, A. Bravo, C. Luján & I. Pérez (eds.), 125–144. Madrid: Instituto Universitario Ortega y Gasset.
Mendikoetxea, A. 2008. Clitic impersonal constructions in Romance: Syntactic features and
semantic interpretation. Transactions of the Philological Society 106(2): 290–336.
Mendikoetxea, A. 2012. Passives and SE constructions. In The Handbook of Hispanic Linguistics,
J.I. Hualde, A. Olarrea & E O’Rourke (eds), 477–502. Oxford: Blackwell.
Ormazabal, J. & J. Romero 2007. The object agreement constraint. Natural Language and
­Linguistic Theory 25: 315–347.
Pesetsky, D. 1995. Zero Syntax. Experiencers and Cascades. Cambridge MA: The MIT Press.
Piñón, C. 2001. A finer look at the causative-inchoative alternation. In Proceedings of SALT 11,
R. Hastings, B. Jackson & Z. Zvolenszky (eds). Ithaca NY: CLC Publications.
© 2013. John Benjamins Publishing Company
All rights reserved
 Olga Fernández-Soriano & Amaya Mendikoetxea
Pylkkänen, L. 1999. Causation and external arguments. MIT Working Papers in Linguistics 35:
161–183.
Pylkkänen, L. 2008. Introducing Arguments. Cambridge MA: The MIT Press.
Ramchand, G. 2008. Verb Meaning and the Lexicon: A First Phase Syntax. Cambridge: CUP.
Reinhart, T. 2000. Strategies of anaphora resolution. In Interface Strategies, H. Bennis, M. ­Everaert
& E. Reuland (eds), 295–325. Amsterdam: Royal Academy of Arts and Sciences.
Reinhart, T. 2002. The theta system. An overview. Theoretical Linguistics 28(3): 229–290.
­ naccusativity
Reinhart, T. & Siloni, T. (2005): Against the unaccusative analysis of reflexives. In The U
Puzzle: Explorations of the Syntax-Lexicon Interface, A. Alexiadou, E. ­Anagnostopoulou &
M. Everaert (eds), 288–331. Oxford: OUP.
Rivero, M.L. 2003. Reflexive clitic constructions with datives: Syntax and semantics. Formal
Approaches to Slavic Linguistics 11: 469–494.
Rivero, M.L. 2004. Spanish quirky subjects, person restrictions, and the Person-Case Constraint.
Linguistic Inquiry 35: 494–502.
Rivero, M.L. & Diaconescu, C.R. 2007. A diachronic view of psychological verbs with dative
experiencers in Spanish and Romanian. Plenary paper at ICHL 2007. 18th International
Conference on Historical Linguistics. Montreal, Quebec.
Schäfer, F. 2008. The Syntax of (Anti-)Causatives [Linguistik Aktuell/Linguistics Today 126].
Amsterdam: John Benjamins.
Schäfer, F. 2009. The causative alternation. Language and Linguistics Compass 3(2): 641–681.
Sigurðsson, H.Á. 1989. Verbal Syntax and Case in Icelandic. Ph.D. dissertation, University of
Lund.
Sigurðsson, H.Á. 1996. Icelandic finite verb agreement. Working Papers in Scandinavian Syntax
57: 1–46.
Sigurðsson, H.Á. 2002a. To be an oblique subject: Russian vs Icelandic. Natural Language and
Linguistic Theory 20(4): 691–724.
Sigurðsson, H.Á. 2002b. Icelandic non-nominative subjects: facts and implications. In
­Non-nominative Subjects, Vol. 2 [Typological Studies in Language 61], P. Bhaskararao and
K.V. Subbarao (eds.), 137–159. Amsterdam: John Benjamins.
Taraldsen, T. 1995. On agreement and nominative objects in Icelandic. In Studies in Comparative Germanic Syntax, H. Haider, S. Olsen & S. Vikner (eds), 307–327. Dordrecht: Kluwer.
Thráinsson, H. 1979. On Complementation in Icelandic. New York NY: Garland.
Van Valin, R.D. & Wilkins, D.P. 1996. The case for effector: Case roles, agents, and agency
revisited. In Grammatical Constructions, M. Shibatani & S.A. Thompson (eds), 289–322.
Oxford: Clarendon Press.
Von Stechow, A. 1995. Lexical decomposition in Syntax. In Lexical Knowledge in the Organization of Language [Current Issues in Linguistic Theory 114], U. Egli, P.E. Pause, C. Schwarze,
A. von Stechow & G. Wienold (eds), 81–118. Amsterdam: John Benjamins. Zaenen, A., Mailing, J. & Thráinsson, H. 1985. Case and grammatical functions: The Icelandic
passive. Natural Language and Linguistic Theory 3: 441–483.
Zubizarreta, M.L. 1987. Levels of Representation in the Lexicon and in the Syntax. Dordrecht:
Foris.
© 2013. John Benjamins Publishing Company
All rights reserved