PRONOMINAL VERBS IN SPANISH
Grant Armstrong
University of Wisconsin-Madison
Abstract. Pronominal verbs (PVs) are a morphological form class of intransitive verbs in Spanish
such as acatarrar-se (= to get a cold) and quejar-se (= to complain) that are obligatorily marked
with the reflexive, or SE, clitic. The main objective of this paper is to propose a solution to the
following puzzle: PVs represent a heterogeneous set of argument structure configurations but are
formally marked in the same way. In order to do so, I empirically motivate the existence of two
different argument structure configurations within the class of PVs, unaccusative and unergative. I
then show that what makes these configurations unique is that they contain a deficiency in Voice,
in the form of unchecked formal features. This deficiency has a different source in each
configuration: it is an unchecked categorial D-feature in unaccusatives and unvalued phi-features in
unergatives. The reason why these two configurations are marked in the same way is because, as
proposed in Pujalte & Saab (2012), the SE clitic paradigm is used as a general repair mechanism to
eliminate unchecked D and phi features on Voice post-syntactically.
Key words: SE clitic, pronominal verbs, argument structure, Distributed Morphology
1. Introduction
Pronominal verbs (PVs) are a familiar class of predicates among Spanish grammarians and
linguists that comprise a set of intransitive verbs that must be marked by the reflexive (or, SE) clitic
as shown in (1).
(1)
a. Pedro *(se) arrodilló
Pedro SE.3s knelt down
‘Pedro knelt down’
b. David *(se)
queja mucho de esta clase
David SE.3s complains a lot about this class
‘David complains a lot about this class’
The point of departure of this paper can be understood as follows: PVs represent a
heterogeneous set of argument structure configurations but are formally marked in the same way.
In subsequent sections, I outline why this is an interesting puzzle and then propose a solution to it.
In order to do this, I first demonstrate (in section 2) that PVs can be divided into two classes based
on a particular set of unaccusativity diagnostics. Examples like arrodillar-se (1a) are unaccsuative
PVs and those like quejar-se (1b) are unergative PVs. The idea that this division exists among PVs
is not novel (see Masullo 1992a,b, for instance), but I propose a representation of each class of
PVs using current syntactic approaches to argument structure that contributes new insights to our
understanding of this division. Beyond these two broad classes of PVs, though, lies the question
of whether a single grammatical function or semantic value can be assigned to SE in sentences like
(1). In sections 3-5, I argue in a step-by-step fashion that the answer to this question is negative
and then show that, in spite of this, there is still a way to account for why the same clitic marks
such different classes of intransitive verbs.
In section 3, I present an overview of the status of the reflexive clitic paradigm in current
theories of object clitics. There are basically three possible analyses that could be applied to any
use of SE: (i) an argument that moves to a functional projection, (ii) the reflex of agreement
between a functional head and an argument or (iii) a clitic that is inserted in a particular type of
morpho-syntactic configuration (Lidz 2001; Pujalte & Saab 2012). I argue that the latter idea is
1
the most adequate for dealing with the properties of PVs and proceed to motivate the particular
type of syntactic configuration that triggers the insertion of SE. The main claim is that the verbal
roots in each class of PVs are specified to appear with a defective Voice head. By defective, I mean
that the Voice head has a formal feature that is unchecked in the narrow syntax. Thus, the reason
that all PVs have the same formal signature is because SE is inserted post-syntactically in order to
eliminate this unchecked feature. What is crucial about this point is that deficiencies in Voice do
not necessarily arise for the same reason, and this allows for a plausible explanation of the puzzle
introduced above. PVs differ markedly in terms of argument structure properties because the
deficiency in Voice arises due to different reasons in each type of PV. These different types of
deficiencies are discussed in turn in sections 4 and 5.
In section 4, I motivate the idea that unaccusative PVs have an expletive Voice head. This
particular type of Voice has been argued for extensively in Schäfer (2008). He claims that this
Voice head introduces no new semantic content (no theta role), but it has an EPP (or, D) feature
that is checked by inserting reflexive or non-active morphology, thus capturing the form of marked
anticausatives cross-linguistically. In the case of Spanish, SE insertion is what is used to check the
EPP feature on expletive Voice. I show that establishing a tight link between unaccsuative PVs
like arrodillar-se (kneel) and marked anticausatives such as romper-se (break-SE) and abrir-se
(open-SE) is a welcome result since these two constructions exhibit nearly identical behaviors,
differing only in that unaccsuative PVs show extremely limited uses as transitive verbs while verb
roots in marked anticausatives can always appear in transitive constructions. If unaccusative PVs
are identical to marked anticausatives then a natural question emerges as to why they are virtually
excluded from transitive environments. Following work by Haspelmath (1993), Levin &
Rappaport Hovav (1995), Mendikoetxea (1999) and Schäfer (2008), I suggest that there is a lexical
semantic underpinning that explains the strong tendency of unaccusative PVs to only appear as
intransitive verbs, mainly, that these have an in-between status with respect to the notions of
external and internal causation.
The case of unergative PVs is treated in section 5. In these, it is argued that the deficiency
in Voice has a different source. It arises because the verb root must appear with a transitive Voice
head but cannot semantically license a DP argument that can value the unvalued phi-features of
this Voice head. Since transitive Voice in this case cannot enter into an AGREE relation with a goal,
a default mechanism must kick in in order to repair the unvalued formal features (Preminger 2011),
which is why SE is inserted. I suggest, following previous work by Campos & Kempchinsky
(1991), that this conflict between what the verb root can license semantically and what it requires
syntactically has arisen with verbs that had transitive uses in earlier stages of the language but have
since been reanalyzed as special types of marked intransitive verbs.
Section 6 concludes the paper by discussing the current proposal in light of recent shifts in
grammatical theory that have completely re-conceptualized the role of the lexicon. Of particular
importance is the claim that verbal argument structure alternations do not involve lexical
operations that apply to a stored base form of a verb. For instance, the causative alternation in
Schäfer (2008) is not the result of a detransitivization of a basic transitive entry. Instead, the two
forms abrir (open-transitive) and abrir-se (open-intransitive) are derived from a common base.
The common base approach to argument structure alternations, coupled with certain claims about
the status of the SE clitic supported here, provide a clear way of accounting for the existence of
PVs without having to propose that these verbs are all underlyingly transitive in some sense and
involve an obligatory reduction/absorption operation in the lexicon. In this particular theory, the
burden of accounting for argument structure alternations is shifted away from stored base forms
2
and lexical rules to the conceptual and formal properties of argument-licensing heads. It is argued
that this provides a better way of characterizing PVs from a conceptual and empirical perspective.
2. Classes of PVs
As mentioned in the introduction, while some authors have either suggested or explicitly proposed
that PVs can either be unaccusative or unergative, there has not been, to my knowledge, a
systematic application of a set of diagnostic tests that motivates this division. The main objective
of this section is to explain and apply some unaccusativity diagnostics to PVs and then propose a
representation of the two classes that accounts for the behavior in the tests.
2.1 Unaccusativity diagnostics
While Spanish lacks unaccusativity diagnostics such as auxiliary selection and ne-cliticization,
certain grammatical contexts have been identified in the literature that highlight differences
between unaccusative and unergative intransitive verbs. 1 One such context includes presentational
sentences that contain bare post-verbal subjects (Cuervo 2003, 2014a,b; Masullo 1992a;
Mendikoetxea 1999; Suñer 1982; Torrego 1989). One type of presentational sentence is used in
out-of-the-blue, past-tense contexts, as in (2).2
(2)
a. Cayeron hojas
fell
leaves
‘Leaves fell’
b. ??Jugaron niños
played children
Intended: ‘Children played’
Unaccusative verbs like caer (fall) admit bare post-verbal subjects in these contexts while
unergative verbs like jugar (play) do not. A potential way of capturing this is to claim that bare
nouns can only be internal arguments. However, this would fail to account for additional data in
which the subject of an unaccusative verb cannot be bare. As noted in Cuervo (2003, 2014a),
Mendikoetxea (1999) and Masullo (1992a), unaccusative verbs that are marked with SE such as
caer-se (fall down) and anticausatives such derritir-se (melt) do not admit post-verbal bare plurals
in presentational sentences. This is shown in (3).
(3)
a. ??Se cayeron hojas
SE.3p fall.PL
leaves
Intended: Leaves fell (down)
b. ??Se derritió nieve
SE.3s melts snow
Intended: ‘Snow melted’
1
It should be noted that it is controversial whether the unaccusativity diagnostics identified in the literature on Spanish
are indications of different syntactic properties or different semantic properties of intransitive predicates (Aránovich
2000). In this section, I attempt to show how two particular diagnostics can be used to separate unaccusative from
unergative verbs and then link these properties to different syntactic structures.
2
As noted in Cuervo (2003), Mendikoetxea (1999), Suñer (1982) and Torrego (1989), this test only distinguishes
between unaccusatives and unergatives in the very specific discourse context described above. Pre-verbal topics and
focus on the post-verbal subject itself obscure the distinction.
3
In order to account for this, Cuervo (2003) has revised a constraint originally formulated in Suñer
(1982) as “the naked noun phrase constraint.” This generalization is spelled out in (4).
(4)
The Naked Noun Phrase Constraint – Revised (Cuervo 2003: 90)
“An unmodified common noun cannot be the subject of a predicate under conditions of
normal stress and intonation” 3
Cuervo (2003, 2014a) argues that the internal arguments in (3) are in fact inner subjects of a small
clause predicate that is interpreted as a result of a change of location or state (fallen down and
melted, respectively) and thus have the same status with respect to (4) as the subject of unergatives
like (2b). Other unaccusative verbs like caer in (2a) license their arguments as complements and
these do not fall under the generalization in (4).
Turning to PVs, we see that they do not permit bare nouns in post-verbal position in
presentational sentences of the same type as (2).
(5)
a. ??Se arrodillaron peregrinos
SE.3p knelt down pilgrims
Intended: ‘Pilgrims knelt down’
b. ??Se
quejaron estudiantes (de la clase)
SE.3p complained students (about the class)
Intended: ‘Students complained (about the class)’
The unacceptability of the examples in (5) can be interpreted in three possible ways. It could mean
that all PVs are unergative like jugar (2b) or that they are all change of state/location unaccusatives
like caer-se and derritir-se in (3). A third possibility is that they divide into both classes but since
this diagnostic does not distinguish between unergatives and change-of-state/location
unaccusatives, another way of illustrating the differences between them must be found. Intuitively,
it seems that the third possibility is promising since arrodillar-se describes a change in position
that could easily be grouped with change-of-state/location verbs, but quejar-se describes an
activity. Evidence in favor of this hypothesis comes from absolute constructions and adjectival
participle formation.
Unaccusative verbs give rise to adjectival participles that can be the main predicate in
absolute constructions and predicative constructions with estar as shown in (6).
(6)
a. Caídas las hojas, salimos todos a limpiar el jardín
Fallen the leaves, we went out all to clean the garden
‘Once the leaves had fallen, we all went out to clean up the garden’
b. La nieve está derretida
The snow is melted
‘The snow is melted’
PVs like arrodillar-se give rise to adjectival participles that can appear in both constructions,
showing that these behave exactly like unaccusative verbs.
Note that “under conditions of normal stress and intonation” are meant to limit the generalization to the out-of-theblue contexts described in (2).
3
4
(7)
a. Arrodillados los peregrinos, empezaron a rezar
Knelt down the pilgrims, they began to pray
‘Once the pilgrims had knelt down, they began to pray’
b. Los peregrinos están arrodillados
The pilgrims
are knelt down
‘The pilgrims are kneeling down’
On the other hand, unergative verbs do not give rise to adjectival participles that can be used in
the same contexts.4
(8)
a. *Jugados los niños,
los dormimos
Played the children, them we put to sleep
Intended: ‘Once the children had played, we put them to sleep’
b. *Los niños están jugados
The children are played
Intended: ‘The children are played’
PVs like quejar-se behave like unergatives in lacking adjectival participles that can be used as
main predicates in absolute constructions or with estar.
(9)
a. *Quejados los estudiantes, se enojaron los profesores
Complained the students, SE.3p got mad the professors
Intended: ‘Once the students had complained, the professors got mad’
b. *Los estudiantes están quejados
The students
are complained
Intended: ‘The students are complained’
To sum up this section, I have shown that PVs fall into two classes based on standard
unaccusativity diagnostics in Spanish. In the next section, I present an analysis of these facts and
discuss certain consequences.
2.2 Structures for PVs
Based on the data in 2.1, I claim that PVs fall into two structural classes represented in (10). For
the moment, I will simply mark the Voice head in each structure with SE, holding off on an
explanation for why Voice is ultimately responsible for the appearance of SE and how it comes to
agree with the grammatical subject until section 3.
4
Note that unergative and transitive verbs do give rise to adjectival participles such as viajado (well-traveled) and
leído (well-read) but these do not appear in the same contexts as the more productive ones formed from unaccusative
and transitive change of state/location/position verbs (see Armstrong, to appear; Varela 2002 for discussion).
5
(10)
Pronominal Verb Structures
a. Unaccusative (arrodillar-se)
b. Unergative (quejar-se)
VoiceP
Voice
VoiceP
vP
DP
Voice´
SE
v
SC
Voice
SE
DP
(Prefix-)√Root5
vP
v
√Root
I will begin by discussing the primary motivations for analyzing unaccusative PVs as in (10a).
Following work by Cuervo (2003, 2014a,b) and Schäfer (2008), I claim that unaccusative PVs
have the same properties as other marked unaccusative and anticausative constructions such as
caer-se and derretir-se, shown above. The structure in (10a) reflects this and is motivated by four
pieces of evidence. The first is that all three constructions give rise to adjectival participles that
can be used as predicative adjectives with estar (see data above). I claim that this is so because
these particular types of adjectival participles are derived by merging an Asp head above vP whose
primary function is to pick out a resultant state in the event structure representation of the verb
(see Embick 2004; Kratzer 2000; McIntyre 2013 for details). Only if a root lexicalizes the resultant
state in a configuration like (10a) will it be able to give rise to an adjectival participle with a
resultant state interpretation that is required in a predicative construction with estar. Since the root
in (10b) does not lexicalize a resultant state, there is no corresponding adjectival participle that can
be formed with a resultant state interpretation.
A second piece of evidence that unaccusative PVs have the same structure as other marked
unaccusatives and anticausatives is that all of these constructions have telic interpretations as
evidenced by the in/for test.6
(11)
a. Se cayeron los libros en/*durante un minuto
SE.3p fell the books in/for
a minute
‘The books fell down in/*for a minute’
b. Se derritió la nieve en/*durante una semana
SE.3s melted the snow in/for
a week
‘The snow melted in/*for a week’
5
The reason why a separable prefix is shown here and not in (10b) is discussed below.
Note that time spans introduced by durante in Spanish and for in English are actually compatible with any lexical
aspectual class (see MacDonald 2008), but they give rise to different readings depending on whether the predicate is
a state, activity, achievement or accomplishment. There is an odd iterative reading associated with durante in (11a)
(the books fall down over and over for a minute). Ramchand (2008) has claimed that an incompletive reading is
available in phrases such as the snow melted for a minute (but did not melt completely), but this is not available in
(11b), a fact noted by Folli (2002) for similar examples from Italian. Finally, durante can also measure states. In (11c)
and its English counterpart the pilgrims knelt down for a minute, there is an interpretation in which the time span
introduced by durante/for measures the duration of the resultant state (= being knelt down). The intended reading in
(11) is the time in which the event culminates from action to resultant state. On this reading, all are compatible with
time spans introduced by en.
6
6
c. Se
arrodillaron los peregrinos en/*durante un minuto
SE.3p knelt down the pilgrims in/for
a minute
‘The pilgrims knelt down in a minute’
These facts are straightforwardly accounted for by structure (10a). The resultant state that is
lexicalized by the root represents a culmination in the change event described by the verb and thus
is always interpreted as telic.
A third piece of evidence that unaccusative PVs are essentially identical to marked
unaccusatives and anticausatives can be gleaned from how causers can be added to unaccusative
PVs. As discussed in detail by Haspelmath (1993), Levin & Rappaport Hovav (1995),
Mendikeotxea (1999) and Schäfer (2008), the roots that have marked anticausatives such as
romp(er) are classified as “cause-unspecified.” This generally means that they may appear in a
marked anticausative such as (12a), in a transitive construction with either an agent or an inanimate
causer (12b) and also take an oblique causer modifier in the anticausative construction (12c).
(12)
a. Se rompió la ventana
SE.3s broke the window
‘The window broke’
b. El niño / el viento de la tormenta rompió la ventana
The child / the wind from the storm broke the window
‘The child/the wind of the storm broke the window’
c. Se rompió la ventana con el viento de la tormenta
SE.3s broke the window with the wind of the storm
‘The window broke from the wind of the storm’
(direct causer)
(oblique causer)
While unaccusative PVs do not typically alternate with transitives, there are some verbs that have
transitive uses with a causative interpretation, as in (13a). These idiosyncratic transitive uses can
be easily captured by the structure in (10a). Verb roots like acurruc(ar) in (13a) may combine with
a thematic Voice head in which case they are interpreted as changes of state with an external cause,
which is generated in the spec of Voice. Moreover, oblique causers can also be introduced as
modifiers of some unaccusative PVs (13b). This follows from treating these on par with marked
anticausatives.
(13)
a. Ven
gatito, te
voy a acurrucar en mis brazos
Come here cat, 2s.ACC I will curl up in my arms
‘Come here cat, I’m going to curl you up in my arms’
b. El niño
se
acatarró con el cambio brusco de la temperatura7
The child SE.3s caught-a-cold with the change sudden of the temperature
‘The sudden change in temperature caused the child to catch a cold’
Finally, there is morphological evidence that unaccusative PVs constitute a form class that should
be represented along the lines of (10a). What is striking about the list in (14) is that nearly all of
7
I thank an anonymous reviewer for mentioning that the direct causer variant el cambio brusco de la temperatura
acatarra with the intended reading ‘the sudden change in temperature causes people to catch colds’ is ungrammatical.
I have adapted this observation in this particular example in order to show that oblique causers can appear with some
unaccusative PVs.
7
the verbs are decomposable into a prefix+nominal root base that gives rise to the verb. For instance,
the example used above, arrodillar-se, has a somewhat transparent decomposition of a-rodill-ar
(to become at one’s knees, = to kneel). In fact, the prefix a- is almost as indicative a marker of
these PVs as SE.8
(14)
acalambrar-se ‘to cramp up’, acalenturar-se ‘to become feverish’, acodar-se ‘to lean on
one’s elbows’, acurrucar-se ‘to curl up’, afiebrar-se ‘to get a fever’, (a)gangrenar-se ‘to
become gangrenous’, agolpar-se ‘to crowd together’, apoderar-se (de) ‘to take over’,
arrellenar-se ‘to sprawl out’, arrepentir-se (de) ‘to repent’, arrodillar-se ‘to kneel’,
desmayar-se ‘to faint/pass out’, enamorar-se (de) ‘to fall in love (with)’, enfurruñar-se ‘to
get angry’, ensimismar-se ‘to become engrossed/lose oneself (in something)’, privar-se
(de) ‘to be/become deprived of’, repantingar-se/repanchingar-se ‘to sprawl out’
Such a decomposition of these verbs is in line with Hale & Keyser’s (1993, 2002) denominal verb
analysis in which the prefix represents a conflated P of terminal coincidence and the idea expressed
in Mateu (2002) and Kayne (2009) that adjectives (in this case, resultant states) are actually derived
from a more basic P + N structure. I leave details for further work, but it is certainly not outlandish
to hypothesize that the state described by unaccusative PVs is one that is composed of more basic
components typically represented by a prefix and nominal root.
The main question that arises with respect to the analysis presented here is why
unaccusative PVs do not alternate more productively with transitive variants. Following Schäfer
(2008), I suggest that the primary reason has to do with the nature of the causation that brings
about the resultant state described by this class of verbs. For all unaccusative PVs, it is probably
not entirely accurate to say that they are cause-unspecified like verb roots that undergo the
causative alternation. Close observation of the list in (14) reveals that, even though these changes
may be initiated indirectly by an external cause, they are directly brought about by some internal
property of the internal (undergoer/theme) argument. For instance, a virus or some other bug may
initiate a set of events that leads to someone becoming feverish, but the state of having a fever is
directly caused by the body’s internal response to this external cause. To conclude, I claim that the
lack of alternations ultimately comes down to root semantics and how the causes that typically
give rise to the states described by the roots in unaccusative PVs. This will be discussed in more
detail in section 4.
I will now turn to the unergative structure in (10b). In each of the domains used to illustrate
the unique properties of unaccusative PVs, it can be shown that unergatives behave distinctly,
which I argue is a consequence of the different syntactic configuration in which the root is merged.
First, since there is no resultant state lexicalized by the root (represented as a small clause in (10a)),
no adjectival participles with resultant state interpretations are possible. It has already been shown
that this holds of unergative PVs like quejar-se. Second, since this structure does not encode a
resultant state, the expectation is for telicity to be variable depending on the meaning of the root
and the presence or absence of an idiosyncratically licensed oblique object (see Cuervo 2014a,
Ramchand 2008 and Schäfer 2008 for discussion of variable telicity outside of result state
configurations and MacDonald 2008 for a general treatment of the syntax of lexical aspect). As
8
Note that an open empirical question is if all prefixes have same status in approaches like Hale & Keyser (1993,
2002) and DM. I suspect that the answer is no. Some prefixes such as a- and en- can be argued to have locative content
and thus be linked to a conflated P. Others such as re- and des- do not obviously have such locative content and most
likely attach higher in the structure, perhaps to already built verb.
8
can be observed in (15), different verbs that belong to the class of unergative PVs may be either
telic or atelic.
(15)
a. Los estudiantes se
quejaron
de la clase *en/durante un semestre
The students SE.3p complained about the class in/for
a semester
‘The students complained about the class *in/for a semester’
b. Me deshice de todos los juguetes en/*durante una tarde
SE.1s got.rid of all the toys
in/for
an afternoon
‘I got rid of all the toys in/*for an afternoon’
A third difference between unaccusative PVs and unergative PVs can be seen in the types of
transitive uses that may apply idiosyncratically to certain verbs. As discussed above, unaccusative
PVs sometimes (not productively) have causative transitive uses in which an external causer is
added to the verb and the subject of the PV is rendered as the direct object of the transitive. For
unergative PVs, certain verbs present transitive uses in which an internal argument is added to the
PV as in (16). Since the specifier of Voice is occupied by the sole argument of the PV in (16a), an
external causer cannot be added to this position to create a causative. 9 On the other hand, an
idiosyncratically licensed internal argument, which is licensed by the root, is possible as in (16b)
(see Levin 1999 for the difference between structure participants and constant participants).
(16)
a. La bailarina se contonea al ritmo de la música
The dancer SE.3s sways to.the rhythm of the music
‘The dancer sways to the rhythm of the music’
b. La bailarina contonea sus caderas al
ritmo de la música
The dancer sways her hips to.the rhythm of the music
‘The dance sways her hips to the rhythm of the music’
Certain unergative PVs that take prepositional objects may alternate with transitive variants in
which the same argument is direct as in (17). It should be noted, however, that many such cases
are often associated with unpredictable changes in meaning and it thus might be a stretch to call
them alternations. The point here is that what changes in the shift from PV to transitive is the
internal argument – the external argument is what remains, from a configurational perspective,
constant.
(17)
a. Juan se despidió
de Jorge
Juan SE.3s said.goodbye DE Jorge
‘Juan said goodbye to Jorge’
b. Juan despidió a Jorge
Juan fired ACC Jorge
‘Juan fired Jorge’
9
Two exceptions seem to be the repetitive motion verbs columpiar-se (swing) and tambalear-se (stagger/wobble).
These verbs do have causative transitive uses but lack all the other properties of unaccusative PVs, so I have classified
them as unergative PVs. These special transitive uses could either be treated as very rare transitivized unergative
structures, which do exist in some languages such as Japanese (see Hale & Keyser 1993, 2002 for discussion of
transitive forms of laugh and cry) or it could be that these roots may appear in an independent transitive construction
that is not derived by adding an additional Voice head to an unergative base. I leave a definitive answer to this question
for further research.
9
Finally, it is probably not accidental that unergative PVs lack a clear morphological signature like
the prefixes of unaccusative PVs. The latter are uniformly associated with a structure that describes
a change of state, location or position that is typically lexicalized as a prefix with locative meaning.
The former are not associated with any particular sub-event that has a morphological reflex in the
language. The heterogeneity of meaning and form of unergative PVs can be observed in (18).
(18)
acordar-se (de) ‘to remember’, aprovechar-se (de) ‘to take advantage (abuse trust)’,
balancear-se ‘to wobble’, cerner-se ‘to hover’, columpiar-se ‘swing’, compadecer-se (de)
‘to take pity on’, comportar-se ‘behave’, confesar-se ‘confess’, contonear-se ‘sway’,
desgañitar-se ‘to yell vociferously’, deshacer-se (de) ‘to get rid of’, despedir-se (de) ‘to
say goodbye’, desquitar-se (de) ‘to get even (with)’, ensañar-se (con) ‘to take it out (on
someone one)’, incautar-se (de) ‘to sieze/confiscate’, jactar-se (de) ‘to boast (about)’,
lamentar-se (de) ‘to be sorry about’, mecer-se ‘rock’, menear-se ‘sway’, olvidar-se (de)
‘to forget’, portar-se bien/mal ‘behave well/poorly’, quejar-se (de) ‘to complain’,
tambalear-se ‘stagger/wobble’, vanagloriar-se (de) ‘to boast (about)’
2.3 Structural versus conceptual properties of PVs
In this section, I briefly discuss some potential areas of confusion that might arise from the
characterization of PVs argued for in section 2.2. The first concerns the connection of the notion
of agentivity to the structures in (10). Agentivity is often used as an intuitive test for separating
unaccusatives from unergatives. For instance, the sole argument of caer (fall) is a theme or
undergoer while the sole argument of jugar (play) is an agent. Given this distinction, it is tempting
to ascribe the property of agentivity to arguments of unergatives and not to those of unaccusatives.
There are, however, many cases in which this correspondence does not hold. Verbs that appear in
unaccusative environments such as absolute constructions can have clearly agentive subjects as
shown in (19).
(19)
a. Llegaron los manifestantes para reclamar al
gobernador
Arrived the protesters
to chide ACC.the governor
‘The protesters arrived in order to chide the governor’
b. El director se
fue deliberadamente
The director SE.3s went deliberately
‘The director left deliberately’
c. María se
sentó sabiendo que iba a estorbar a Juana
María SE.3s sat down knowing that she was to block ACC Juana
‘María sat down knowing that she would be in Juana’s way’
Moreover, verbs typically categorized as unergative can be non-agentive as in (20).
(20)
a. El niño estornudó
The child sneezed
‘The child sneezed’
b. No puedo dejar de toser
NEG I can
stop cough
‘I can’t stop coughing’
10
c. Estás sangrando
You are bleeding
‘You are bleeding’
These data simply show that the labels placed on syntactic argument positions in non-lexicalist
approaches like the present one (or, theta-roles in earlier lexicalist approaches) do not correspond
directly to thematic roles, which are purely semantic (see Bowers 2010: 6-10 for discussion). The
unaccusativity of the examples in (19) can be accounted for by treating the subjects as themes or
undergoers – they all undergo a change of location or position. The agentive properties that they
have can be derived by claiming that thematic roles are actually composites of syntactic positions
that can be linked to one argument (Ramchand 2008) or treating these constructions as reflexives
(Chierchia 2004; Cuervo 2014a; Koontz-Garboden 2009) in which the initiator/undergoer is coreferent with the theme of the resultant state. The examples in (20) can be handled by positing
that different types of external argument-introducing heads exist, which correspond to general
notions such as agents/causers, holders and sources, and that these vary according to the type of
event described by the verb (Kratzer 1996; Harley 1995). The distinction between syntactic
positions (and/or theta roles) and thematic roles applies equally to the subjects of PVs. Note that
many of the subjects of unaccusative PVs such as arrodillar-se (kneel) and apoderar-se (take
power/possession of) have agentive properties while not every subject of unergative PVs is
agentive (e.g. – cerner-se ‘hover’ or olvidar-se (de) ‘forget’), which is straightforwardly accounted
for under any of the ideas mentioned above.
A second aspect of the division made above that veers somewhat from traditional grammars
is that PVs that take prepositional arguments (= complementos de régimen, see Cano Aguilar 1999)
may fall into either class of PVs. I will treat these prepositional arguments as sisters of the root
itself, regardless of the position that the root occupies in the structure.10 The consequence of this
idea is that superficially identical verbs that belong to similar semantic fields may be underlyingly
different.
(21)
a. Los revolucionarios se
apoderaron del gobierno
The revolutionaries SE.3p took.control of.the government
‘The revolutionaries took control of the government’
b. Los agentes se
incautaron de una caja de drogas
The agents SE.3p confiscated of a box of drugs
‘The agents confiscated a box of drugs’
(unaccusative PV)
(unergative PV)
These verbs behave differently in that only apoderar-se has an adjectival participle that can appear
with estar as shown in (22a). This supports treating it as an unaccusative PV. Incautar-se, on the
other hand, fails this test (22b) and thus should be treated as an unergative PV.
10
This may be a controversial assumption (see Cuervo 2014a,b and Harley 2014 for ideas in favor of the projection
of roots in syntax and Alexiadou 2014; Borer 2005; Bowers 2010, among others that roots do not project). The relevant
distinction for the purposes of this paper is that PP arguments of PVs correspond to Levin’s (1999) constant
participants, which are licensed idiosyncratically by roots (see Cuervo 2003 for a similar idea outlined within the
framework employed in this paper).
11
(22)
a. Los revolucionarios están apoderados del gobierno
The revolutionaries are in control of.the government
‘The agents are in control of the government’
b. *Los agentes están incautados de una caja de drogas
The agents are confiscated of a box of drugs
Intended: ‘The agents are confiscated of a box of drugs’
I claim that this distinction follows from the two types of PV structures introduced in (10) coupled
with the idea that roots may (idiosyncratically) license arguments.
In this section, I have argued that PVs are derived from two types of argument structure
configurations, an unaccusative (with a resultant state) and an unergative. These two structures
provide a way of capturing observed differences in adjectival participle formation, distinct
interpretative tendencies in the idiosyncratic transitive uses of PVs in addition to different
morphological patterns that arise within the two classes. I have said nothing here about why or
how SE comes to be associated with these two structures, which ultimately gives PVs their
signature characteristic. In the next sections I will address these questions.
3. The status of the SE clitic in PVs
In order to address why and how SE comes to mark PVs, it is first necessary to discuss current
approaches to object clitics in Spanish and beyond in order to get a clear idea of the possible
analyses the could be applied to SE. In this section, I motivate treating SE in PVs as a special type
of clitic that is inserted post-syntactically in a particular syntactic configuration that involves a
deficiency in Voice, following work by Lidz (2001) and Pujalte & Saab (2012).
3.1 Background: the status of reflexive clitics in theories of clitics
The SE clitic paradigm is a member of the more general category of object clitics in Spanish (and
other Romance languages). Recent literature on object clitics has highlighted the difficulty in
determining whether these are moved determiners or reflexes of agreement (Kramer 2014;
Ormazabal & Romero 2013, and references therein). In principle, any object clitic in Spanish could
be a determiner in argument position that moves to a functional head as in (23a) or the reflex of an
agreement relation established between a functional head and a DP/pro in argument position.
(23)
a. [VoiceP [DP …] Voice [VP V [DP clitic]]]
Determiner movement
b. [VoiceP [DP …] Voice[uϕ] [VP V [DP pro]]]
Agreement
AGREE
ϕ
Ormazabal & Romero (2013) have claimed that both of these approaches are correct for subsets
of clitics in different varieties of Spanish. They provide evidence showing that 1 st and 2nd person
direct and indirect objects as well as 3rd person indirect objects are agreement clitics (= Agr-clitics)
that are generated in a configuration like that of the simplified (23b). This is a stable characteristic
of all varieties of Spanish. On the other hand, third person direct object clitics exhibit much more
variation and can be analyzed either as determiner elements (D-clitics) as in (23a) or as Agr-clitics
(24b) in different varieties.
12
Reflexive object clitics have generally been studied independently of non-reflexive ones.
Indeed, Ormazabal & Romero (2013) do not mention reflexive clitics in their study but suggest
that they are likely Agr-clitics. The main difficulty associated with reflexive clitics is that, in
addition to distinguishing between the determiner or agreement status of clitics in general, there is
also a relation of construal between two theta positions to contend with. For non-reflexive clitics,
it is uncontroversial that there is a syntactically active internal theta role in configurations like
(24), which makes it simpler to focus on the morphosyntactic status of the clitic and how it is
related to that internal position. Now, consider the reflexive constructions in (24).
(24)
a. Juan se
critica
Juan SE.3s criticizes
‘Juan criticizes himself’
b. Te
compraste un carro
SE.2s you bought a car
‘You bought yourself a car
In these examples a theta role in an internal position is discharged to the same argument that
occupies an external thematic position. One approach to reflexives contends that they contain
syntactically active internal thematic positions just like constructions with non-reflexive
predicates. The main difference is that this position is filled with a referentially defective element
that must be linked to another argument (subject to locality), an anaphor in P&P parlance. On this
approach the SE clitic paradigm can either be treated as a referentially defective determiner that
moves (23a) or as the reflex of agreement between a functional head and a bound internal thematic
position (23b) occupied by some kind of deficient pronoun. Different proposals along these lines
include Mendikoetxea (1997, 2012) and Teomiro (2010), for whom the internal theta position is
occupied by PRO and SE is agreement, and Torrego (1995), who claims the internal thematic
position may be occupied either by SE (and a double) or a bound pronoun in which case SE is a
verbal operator. The general idea represented in such approaches is shown in (25).
(25)
a. [VoiceP [DP …] Voice [VP V [DP
SE + double ]]]
b. [VoiceP [DP …] Voice[uϕ] [VP V [DP PRO/pro]]]
AGREE
Determiner movement
Agreement
SE
A second family of approaches to reflexive constructions, however, does not treat the SE
clitic as a determiner or the reflex of agreement. In many early analyses of SE clitics in Romance,
it was proposed that SE attaches to a verb in the lexicon, absorbs accusative case and eliminates a
theta role from the verb’s lexical entry (see Grimshaw 1990; →ehrli 1986; among others). Even
though this idea has been modified in current approaches, the unique status of SE remains: it is
neither a D-clitic nor an Agr-clitic. For instance, Reinhart & Siloni (2005) argue that Romance
reflexive constructions are built from a verb that contains a syntactically active internal thematic
position, but this position is simply not filled by an argument in syntax. Instead of causing a
derivational crash, the SE clitic adjoins directly to the verb in order to absorb its accusative case
feature and the unlinked internal theta role is bundled to the role discharged to the external
13
argument at LF. In this sense, reflexive verbs behave like unergative intransitives since they lack
a syntactic internal argument. Pujalte & Saab (2012) argue for a similar idea, though using different
technical machinery. They claim that reflexives, along with all other SE constructions, arise when
an external-argument selecting head fails to discharge a categorial feature, labeled a D-feature (see
Schäfer 2008 for additional information about categorial D-features). An unchecked D-feature is
interpreted by the PF-component of the grammar and repaired through the insertion of SE. On the
LF side of the grammar, SE constructions receive various interpretations depending on structural
and conceptual factors – a reflexive interpretation is one of these. In both approaches, the SE clitic
is neither a D-clitic nor an Agr-clitic in the sense outlined in Kramer (2014) and Ormazabal &
Romero (2013). The approaches described above are schematized in (26).
(26)
a. [VoiceP [DP …] Voiceθ-1 [VP SE-Vθ-2 ]]
LF: θ-1 (AGENT) = θ-2 (THEME)
(Reinhart & Siloni 2005)
b. [VoiceP Voice[D] [VP V [DP … ]]
PF: Insert SE
LF: AGENT = THEME
(Pujalte & Saab 2012)
A further complicating factor that arises in the study of reflexive clitics, as is obvious based on the
discussion up to now, is that they commonly appear in many constructions that do not appear to
involve reflexive interpretations such as impersonals, passives, middles, anticausatives and, most
importantly, pronominal verbs (see Mendikoextea 2012 for an excellent review of these in
Spanish). The relevant question for the purposes of this paper is if examples like (27), often labeled
intrinsic or inherent reflexives, 11 are more appropriately analyzed as D-clitics or Agr-clitics in the
sense discussed in Kramer (2014) and Ormazabal & Romero (2013) or as something different as
in Reinhart & Siloni (2005) or Pujalte & Saab (2012).
(27)
a. La sopa se
enfrió
The soup SE.3s cooled
‘The soup cooled’
b. Ana se
desmayó
Ana SE.3s fainted
‘Ana fainted’
c. Guillermo se
jacta de su riqueza
Guillermo SE.3s boasts about his riches
‘Guillermo boasts about his riches’
Some researchers have proposed that anticausatives like (27a) have two syntactically active
theta positions but that one of them is filled with a special type of PRO. For Mendikoextea (1997,
11
I am using the term intrinsic reflexive as in Otero (1999) and Reinhart & Reuland (1993) to refer to a verb that must
be marked by a “weak” (reflexive or SE-anaphor) reflexive element in languages like Spanish and Dutch but can never
take “strong” reflexive elements (SELF-anaphor) or a non-reflexive DP argument in these languages. These include
schaamt (shame) in Dutch and portar-se bien/mal (behave well/poorly) in Spanish. The same types of verbs are labeled
inherent reflexives in Alexiadou & Schäfer (2014). This type of reflexivity is different than extrinsic reflexivity, which
applies to verbs that can take strong reflexive elements or non-reflexive DPs in argument positions. Extrinsic
reflexivity may be further divided into naturally reflexive verbs such as wash and naturally disjoint verbs such as hate
(see Alexiadou & Schäfer 2014 for an overview).
14
2012), (27a) involves an internal position that is filled with a DP and an external position is filled
by a PRO. The internal argument raises and controls this PRO resulting in a co-reference between
the external and internal theta positions. The specific interpretation that arises in such cases is a
kind of reflexivization discussed at length in Chierchia (2004) and Koontz-Garboden (2009) called
stative reflexivation. 12 In an example like (27a), the interpretation (on this account) is that an
internal property of the soup causes it to cool down. Mendikoextea (2012) suggests that a similar
analysis could be extended to cases like (27b) as these are also unaccusative, but it is unclear how
cases like (27c) would fit into such an analysis.
Another set of approaches posits that this entire family of constructions is syntactically
transitive but semantically intransitive. The basic idea is that intrinsically reflexive SE
constructions have essentially one argument position in their semantic interpretation but, for
formal reasons, they project a second, semantically inert, syntactic argument position that must be
filled by an expletive. For instance, Teomiro (2010) claims that Spanish contains a PRO that is
basically a null SE anaphor and may merge in either an external or internal position. When PRO
merges, it does not get a theta role, but simply occupies a syntactic position that is required for
purely formal reasons. He claims that certain verbs like those in (27) can undergo specific types
of lexical processes in which a theta role is eliminated or bundled (see Reinhart & Siloni 2005) but
that, formally, a position still must be filled in syntax by an expletive element and this is why SE
appears with them. Schäfer’s (2008) proposal for marked anticausatives like (27a) is similar. He
argues extensively that such constructions have an expletive Voice head that requires the presence
of a particular type of Voice morphology or a referentially defective D head in many languages.
Thus, on this analysis, the Voice head does not introduce a theta role (so, there is no stative
reflexive interpretation) but it does act as the trigger for SE insertion.
Finally, both Lidz (2001) and Pujalte & Saab (2012) argue that all SE constructions,
including those in (27), are reflections of a particular type of configuration that triggers SE insertion
in a morphological component of the grammar. For Lidz (2001), a verbal reflexive morpheme is
inserted to reflect a mismatch between linking of thematic and aspectual positions in a presyntactic lexicon. Thus, it lacks any kind of specific grammatical function (i.e. agreement, caseabsorption) or semantic content. In Pujalte & Saab, SE is inserted at PF whenever a syntactic
configuration contains an external-argument-introducing head, mainly v, with no external
argument. Such a derivation converges if there is an appropriate interpretation that can be assigned
to the structure based on compositional and conceptual principles operative at LF and in the
encyclopedia. Again, there is no specific grammatical function attached to SE in this theory, as it
takes on a default character that is inserted to reflect a particular type of syntactic structure. In
these theories, there is no sense in which SE is either a D-clitic that moves from an argument
position to a functional projection or a reflex of agreement with an internal theta position.
In this section, current theories of object clitics have been discussed. Object clitics are
treated either as determiners that move from an internal theta position to a functional projection or
as reflexes of agreement between an internal theta position and a functional projection. Reflexive
clitics have often been investigated independently both because of the construal between subject
and object that they mark, and due to the variety of contexts in which they appear. Taking up the
issue of PVs once again, the main task at hand is to develop an analysis that can account for the
data presented in section 2 regarding unaccusativity and integrate the presence of SE in such a way
12
Cuervo (2014a) proposes an analysis in which SE acts as the spell-out of a reflexive relation between an UNDERGOER
and HOLDER for verbs such as caer-se (fall-se), subir-se (ascend-se) and bajar-se (descend-se).
15
that respects this previous work on the status of clitics more generally. This is the objective of the
next sections.
3.2 The treatment of SE in PVs
The main evidence against treating SE in PVs as either a D-clitic or an Agr-clitic in the sense
described above is that SE can never be doubled by a strong anaphor as in (28), and apart from
certain idiosyncratic transitive uses described in section 2, no non-reflexive objects are permitted
in PVs.
(28)
a. *Pedro se arrodilló a sí mismo
Pedro SE.3s knelt A himself
Intended: ‘Pedro knelt himself down’
b. *Juan se desquitó a sí mismo de su rival
Juan SE.3s got even A himself of his opponent
Intended: ‘Juan got himself even with his opponent’
Data like (28) provide evidence against a syntactically active internal theta position in PVs that
could potentially be a merge site for SE or be filled with an anaphor that would trigger the presence
of SE through agreement. The absence of such a position follows from the proposal presented in
section 2. For unaccusative PVs like arrodillar-se (28a) this is because the sole argument of the
verb is an internal argument and thus this position cannot be simultaneously occupied by a strong
anaphor. For unergative PVs like desquitar-se (28b), the root itself can only license a PP and not
a DP argument. These structures are repeated in (29) for convenience.
(29)
Pronominal Verb Structures13
a. Unaccusative (arrodillar-se)
b. Unergative (desquitar-se)
VoiceP
Voice
VoiceP
vP
v
DP
Juan Voice
SC
DP
Pedro
Voice´
a-√rodilla
vP
v
√desquit-
√P
PP
de su rival
13
As mentioned in footnote 5 when these structures were first introduced, I am making a distinction between prefixes
with clearly locative meanings like a-, and those like des-, which do not. I have opted not to separate des- from the
root in (29b). If des- were separated from the root in this example, it would most likely not be a P. One possible
treatment of it would be as a type of light verb that must combine with certain bound roots or fully formed verbal
structures. I leave this as an open research question.
16
The proposals that characterize SE in terms distinct from a D-clitic or an Agr-clitic such as Lidz
(2001), Schäfer (2008) and Pujalte & Saab (2012) thus seem to be the most promising ways of
integrating the presence of SE with the argument structures that have been proposed for PVs. I
argue that SE is a reflex of a deficiency in Voice, but the type of deficiency is different in each of
the structures in (29). I will discuss each of these in turn in sections 4 and 5.
4. Unaccusative PVs: expletive Voice
It was argued in section 2 that unaccusative PVs have argument structure properties identical to
those of marked anticausatives, so an analysis that captures these shared properties, in addition to
why SE appears in these constructions, is desirable. Fortunately, Schäfer’s (2008) analysis of
anticausatives provides a straightforward way of accounting for unaccusative PVs like arrodillarse. This analysis has been worked out in detail for languages like German, which use a SE anaphor
to mark both reflexives (30a) and anticausatives (30b).
(30)
a. Hans mag sich
John likes self
‘John likes himself’
b. weil
sich die Tur offnet
because self the door opens
‘because the door opens’
One of the main goals in Schäfer (2008) is to attempt to derive why it is that reflexive morphology
is used to mark anticausatives across languages, rather than simply stipulate that the presence of
reflexive morphology reflects an anticausative interpretation with some change of state verbs. He
sets out to do this within a minimalist framework that is based on the following assumptions. First,
theta roles are claimed to be purely configurational. An argument (an element with a categorial D
feature that is referential) is assigned a theta role at the C-I interface that is unique to its syntactic
position and based on the event described by the verb/root. Argument positions are complements
of V/root and spec of Voice. Second, syntactic binding is an AGREE relation (see Reuland 2011 for
further details) between a fully referential DP and a syntactic variable that has a D feature but is
completely unspecified for phi-features and is thus non-referential (see Burzio 1991 for a proposal
that SE lacks phi-features). When a c-commanding, fully-referential DP enters into an AGREE
relation with the variable, the latter becomes referential and may be assigned a theta role at the CI interface if it is in an argument position. Finally, Schäfer (2008) assumes a dependent
case/agreement system like that of Marantz (1991). In a nutshell, Voice does not act as a probe
that values accusative case. There is one probe, T, which enters into an AGREE relation with the
closest active goal. This AGREE relation is visible to PF and may trigger spell out of phi-features
on the verb or spell out of the goal DP with a particular case morphology if available in the
language. If there is an additional DP in the same domain, it receives accusative case at PF.
The structure of sentence (30a) above is shown in (31). The argument positions, spec Voice
and complement V, respectively, contain a fully referential DP and a variable. Syntactic binding
through AGREE values the unvalued phi features of the variable in object position. This gives the
variable referential content and allows it to receive a theta role at LF. It also determines how it will
be spelled out at PF (= sich). T is a probe that contains unvalued phi-features, but notice that it is
not endowed with any specific case feature. It probes its c-command domain and enters an AGREE
relation with Hans, which values its phi-features and determines the nature of verbal agreement at
17
PF. This AGREE relation is also legible at PF in that it determines the case of Hans (= nominative),
which also indirectly triggers accusative case on the variable.
(31)
AGREE
NOM
[TP Tu:P,N,G [VoiceP [DP Hans[D; P,N,G]] Voice{D, θ} [VP mag [DP u:P,N,G]]]]
AGREE
Binding
sich (gets dependent case = ACC)
The key ingredient to the proposal is that in order for a variable to be referential (and receive a
theta role), it must be bound, where binding is AGREE with a c-commanding DP. Now, turning to
reflexive-marked anticausatives, Schäfer (2008) argues that they have the same elements as a
reflexive-marked transitive verb, but the arguments occupy the opposite positions: the variable is
in spec Voice and the fully referential DP is in an internal argument (here, the complement of the
V/Root). So (30b) has a structure like (32).
(32)
AGREE
NOM
No phi-features
[TP Tu:P,N,G [VoiceP [D u:P,N,G] Voice{D, Ø} [VP offnet [DP die Tur[D; P,N,G]]]]]
sich (gets dependent case = ACC)
In this configuration, there is no binding since the only referential DP in the structure does not ccommand the variable. However, there is an AGREE relation that is initiated by T. T probes its ccommand domain and finds D, the variable, but it has no valued phi-features. It probes further
down and agrees with the internal argument, which values the phi-features on both T and D (the
variable). Since the variable does not receive nominative case, it will be assigned the dependent
case, accusative, at PF and be spelled out as sich. The main insight offered in this proposal is that
the variable in this configuration cannot be referential since it is not bound. This means that no
theta role can be assigned to the position it occupies. The logical conclusion that Schäfer (2008)
derives from this is that Voice in anticausatives is an expletive Voice with no theta role. This is
represented in (32) simply as Voice{D, Ø} as opposed to Voice{D, θ}. The next question concerns the
distribution of this expletive Voice – why does it appear with the verbal roots that it appears with?
Following earlier work by Haspelmath (1993) and Levin & Rappaport Hovav (1995),
Schäfer (2008) claims that the distribution of expletive is tied to certain groups of roots that
describe changes of state with an unspecified cause. Roots are organized in each language
according to whether they can describe non-spontaneous (agentive, externally cause) changes such
as murder that require transitive syntax, or purely spontaneous changes such as blossom that
require intransitive syntax. Those that fall in between can potentially appear in either a transitive
or intransitive syntactic environment. These are typically called ‘cause-unspecified’ verbs since
they can be conceptualized as describing either a non-spontaneous externally-caused event such
as John opened the door or as a spontaneous event such as the door opened. The in-between status
of these roots is reflected in marked anti-causatives, which have the formal, configurational
properties of an externally caused event (a Voice head) but the semantic properties of one that is
internally caused (a single argument undergoes a change of state).
18
Two questions arise in adapting this analysis to Spanish marked anticausatives and
unaccusatives PVs. The first is a technical one: does the reflexive clitic have the same status as the
SE anaphor in German? Schäfer (2008) suggests that reflexive clitics in Romance could either be
treated in the same way as SE anaphors with the extra requirement that they must move to a
functional head or they can be treated as the spell-out of the expletive Voice head. I will follow
Pujalte & Saab (2012) in answering this. These authors have proposed independently that SE is
inserted in contexts exactly like those identified in Schäfer (2008) and offer a post-syntactic
morphological mechanism to explain how this is done. Adapting this idea to Schäfer’s system, the
sentence in (33a) would have the representation in (33b).
(33)
a. Se
abrió la puerta
the door
‘The door opened’
SE.3s opened
AGREE
NOM
b. [TP Tu:P,N,G [VoiceP Voice{D, Ø} [vP v [SC [DP la puerta [D; P,N,G]] [√ abr-] ] ] ]
Thus, the difference between German and Spanish is that there is no SE anaphor in the latter so no
D element is present in the specifier of expletive Voice. This unchecked categorial feature must
be eliminated prior to PF. Pujalte & Saab (2012) propose that once the cycle above Voice is sent
to the interfaces, there are number of morphological operations that apply to the head adjunction
structure at the T node. This is shown in (34).
(34)
T
CLuφ
T
T
Voice
…
AGRφ
T
Voice[D]
The first is that an extra node is added for agreement morphology. The agree relation that is
establish between the T and the internal argument requires an extra node to host its exponent. In
this case it will be [3], [sing] and [past]. At this particular cycle, the [D] feature on Voice must be
eliminated and this is what clitic insertion does. What is inserted at T is essentially a D-element
with a set of unvalued phi-features (a morphological version of Schäfer’s SE anaphor). The actual
form of the clitic is determined by the AGR node. The CL has unvalued phi features that are valued
through an AGREE relation with AGR. In this system, the explanation behind the SE clitic
‘paradigm’ being the one that is inserted in such contexts is related to their default status in the
grammar – they are variables that do not have valued phi-features.
From a purely technical point of view, there is no difference between unaccusative PVs
and marked anticausatives. The remaining task is to account for their most notable difference:
unaccusative PVs can only appear in syntactic configurations with expletive Voice while marked
19
anticausatives can also be transitive. Given the relevance of externally caused non-spontaneous
changes versus internally caused spontaneous ones in determining whether particular roots have
marked anticausatives, we might expect to find some semantic unity in the kinds of roots that
appear solely as unaccusative PVs, which could be linked to work on the lexical semantics of the
causative alternation more generally. Consider once again the types of events described by
unaccusative PVs, repeated from above.
(35)
acalambrar-se ‘to cramp up’, acalenturar-se ‘to become feverish’, acodar-se ‘to lean on
one’s elbows’, acurrucar-se ‘to curl up’, afiebrar-se ‘to get a fever’, (a)gangrenar-se ‘to
become gangrenous’, agolpar-se ‘to crowd together’, apoderar-se (de) ‘to take over’,
arrellenar-se ‘to sprawl out’, arrepentir-se (de) ‘to repent’, arrodillar-se ‘to kneel’,
desmayar-se ‘to faint/pass out’, enamorar-se (de) ‘to fall in love (with)’, enfurruñar-se ‘to
get angry’, ensimismar-se ‘to become engrossed/lose oneself (in something)’, privar-se
(de) ‘to be/become deprived of’, repantingar-se/repanchingar-se ‘to sprawl out’
Recall that changes of state or configuration that are conceptualized as externally caused or nonspontaneous require a transitive syntax, those that are internally caused or spontaneous require an
intransitive syntax while those that are cause-unspecified may appear in both types of constructions
and these typically in the causative alternation with marked anticausatives. As noted in Haspelmath
(1993) and Schäfer (2008) the set of roots in a given language (or perhaps even in varieties of the
same language) may be conceptualized differently and thus we expect to see differences in terms
of which roots appear in which syntactic configurations. Taking this into consideration, I propose
that unaccusative PVs describe changes that are conceptualized in terms of a causal chain in which
an external causer initiates something that leads to an internally caused change. For example,
acalambrar-se (to cramp up) is a change that is caused indirectly by some external stimulus that
leads to a process of change that is internal to the subject. It is not something that is directly caused
by some external source. Thus, it usually makes no sense to use these in transitive configurations.
That said, certain contexts may permit the presence of an external causer that acts directly to bring
about the change described by one of these roots, accounting for the idiosyncratic transitive uses
that were described in section 2, repeated in (36).
(36)
Ven
gatito, te
voy a acurrucar en mis brazos
Come here cat, 2s.ACC I will curl up in my arms
‘Come here cat, I’m going to curl you up in my arms’
Summarizing, in this section, I have defended the hypothesis introduced in section 2 that
unaccusative PVs are identical to marked anticausatives with respect to their morphosyntax and
used proposals by Schäfer (2008) and Pujalte & Saab (2012) in order to account for why and how
SE comes to mark these verbs. Moreover, I have suggested that the in-between status with respect
to how these verb roots are conceptualized as causal chains in which an indirect external cause
leads to a direct internally caused change is a plausible reason for why unaccusative PVs typically
only appear in this particular syntactic construction. Note that all of the properties of unaccusatives
PVs discussed in section 2, in addition to why and how SE comes to mark them have been
accounted for in a way consistent with current approaches to argument structure and treatment of
reflexive clitics.
20
5. Unergative PVs: failed agreement
Turning now to unergative PVs, there is comparatively little work to draw on in order to integrate
SE into the structure proposed for unergative verbs. Clearly, using the notion of expletive Voice
introduced above would not be desirable since unergative PVs appear to have an external argument
that gets a theta role by virtue of being in the spec of Voice. What we have in the case of unergative
PVs seems to be the following: a Voice head that licenses an external argument like a transitive
verb, but fails to assign structural case to an internal argument, unlike a transitive verb. This
particular property has led Bogard (2006) and Masullo (1992a,b) to label a subset of the unergative
PVs as antipassives in the sense that SE appears to mark the absence of accusative case and requires
any internal arguments to appear with oblique case. I will first consider treating SE as a type of
antipassive morpheme in unergative PVs, ultimately rejecting it in favor of a different type of
analysis in which SE appears as a result of a failed agreement relation between Voice and a DP in
its c-command domain. First, let us consider the antipassive analysis.
The antipassive is a type of morpheme that demotes the internal argument of a transitive
verb, yielding an intransitive verb (see Polinsky 2013 for a general overview). There are two
principal ways in which the newly created transitive verb may surface syntactically. It may have
an internal argument that is marked as oblique (with a P or specific type of case form such as
instrumental) or it may have an implicit argument. The following examples from Akatek, a Mayan
language of the Q’anjobalan branch spoken in Guatemala (Zavala 1997), serve to illustrate the first
type of demotion.
(37)
Akatek Antipassives (Zavala 1997, 456 as cited in Coon to appear, 11)
a. X-s-nooch-toj
no’ txitam ixim aan
ASP-A3sg-eat.biting-DIR CLF pig
CLF corncob
‘The pig ate the corncob’
b. X-nooch-wa
no’ txitam y-iin
ixim aan
ASP-eat.biting-ANTIP CLF pig
A3sg-LOC CLF corncob
‘The pig was eating on/at the corncob’
Observe that when the verb nooch (eat by biting) contains the antipassive morpheme –wa, there is
no longer ergative agreement with the subject (glossed as A3sg here) and the object is introduced
by a locative element called a relation noun in Mayan linguistics that is similar to a preposition.
This example looks superficially like (38) since in the presence of SE, the internal argument of the
verb olvidar is marked as oblique.
(38)
a. Juan olvidó la tarea
Juan forgot the homework
‘Juan forgot the homework’
b. Juan se olvidó de la tarea
Juan SE.3s forgot of the homework
‘Juan forgot about the homework’
There is an important difference to note though. In (37b), the demotion of the argument correlates
with low affectedness and atelicity. This is often brought out by translating antipassive
constructions with the English preposition at, which marks the conative alternation (Levin 1993:
41) as in I ate the cake in an hour (I finished it) versus I ate at the cake for hours but never finished
21
it. While this correlation is not absolute (see Polinsky 2013 for details), it does not seem to exist
at all in the Spanish examples that exhibit the relevant pattern. In fact, Sanz (2000) has argued that
in (38b), SE is a marker of telicity rather than atelicity. The conclusion that I draw from this brief
comparison is that the formal properties of some antipassives and some unergative PVs are similar
but that the set of interpretative effects typically induced by an antipassive morpheme do not obtain
in the Spanish examples.
The second type of demotion involves an internal argument that is implicit and is illustrated
below with an example from the Baffin Island dialect of Inuit (Spreng 2005).
(39)
Inuit Antipassives (Spreng 2005, as cited in Basilico 2012, 81)
a. Kapi-jara
Stab-PART.1sg/3sg
‘I stabbed him’
b. Kapi-si-juq
stab-ANTIP-PART.3sg
‘He is stabbing/poking someone/something’
In this pair of examples, the antipassive morpheme –si attaches to the verb and no internal
argument is present in the syntax as exhibited by the lack of object agreement on the verb. In (39b),
only subject agreement appears. The internal argument is implicit in such cases. Certain types of
unergative PVs show a similar pattern, illustrated in (40).
(40)
a. La bailarina contonea sus caderas al
ritmo de la música
The dancer sways her hips to.the rhythm of the music
‘The dance sways her hips to the rhythm of the music’
b. La bailarina se contonea al ritmo de la música
The dancer SE.3s sways to.the rhythm of the music
‘The dancer sways to the rhythm of the music’
In (40b), it is implicit that something is swaying (hips, etc.), but this is simply left unexpressed in
the presence of SE. As in (37), the antipassive in (39) correlates with a particular aspectual change.
(39a) is interpreted as a completed event – there is a fatal stabbing that results in death whereas in
(39b), only an activity reading is possible. This distinction does not arise in (40) as both are atelic
activities.
It is clear based on the discussion above that antipassives in ergative languages of the
Mayan and Inuit families are productive and plausibly constitute argument structure alternation in
which an internal argument of a transitive verb receives a special grammatical marking in the
antipassive that often correlates with an aspectual effect. The cited examples from Spanish seem
to cluster around a handful of verbs and tend to exhibit notable idiosyncratic changes in meaning
from transitive to intransitive. Moreover, most unergative PVs in Spanish do not present the types
of alterations described above. Instead, they are obligatorily intransitive and marked with SE. Given
these differences, it would not be desirable to treat unergative PVs as antipassives that are derived
from a transitive counterpart. A further complication concerns the phi-features on SE: while it
would signal demotion of an internal argument on the antipassive analysis, it ultimately agrees in
person with the external argument. It is not clear how treating SE as an antipassive morpheme could
capture this characteristic.
22
An interesting alternative to the antipassive approach can be found in Preminger (2011).
This is based on an analysis of unergative constructions in Basque, shown in (41).
(41)
a. Jon-ek danzatu d-Ø-u-Ø
John-ERG dance.PRT 3.ABS-sg.ABS-AUX-3sg.ERG
‘John danced’
b. Dantza-n egin d-Ø-u-te
dance-LOC do 3.ABS-sg.ABS-AUX-3pl.ERG
‘They danced repeatedly’
(Preminger 2011, 163)
Preminger (2011) notes that a curious characteristic of the unergative constructions in (41) is that
they exhibit 3rd person singular absolutive agreement on the auxiliary, which is indicated in bold.
He shows that examples like (41) lack a syntactically active implicit object, so it is somewhat
mysterious why the agreement reflex appears here. The solution he proposes is that in Basque
unergative constructions, there is an active probe that does not find a goal. In (41a) this is because
there is no internal argument at all and in (41b), the so-called repetitive unergative construction,
the internal argument is marked with a locative or adverbial suffix, which arguably blocks a probe
from reaching a DP inside it. Thus, the operation AGREE fails in the narrow syntax in each case.
However, he argues independently that failure to establish an agreement relation does not
necessarily lead to ungrammaticality, which is different from the derivational crash that failed
agreement would provoke in the model outlined in Chomsky (2000, 2001). Many languages have
repair strategies in which a default form of agreement appears, mainly 3rd person singular, appears
in contexts where an active probe fails to find a goal. The 3rd person singular absolutive is precisely
one such default form.
While unergative PVs are different in important ways from Basque unergative
constructions, I would like to propose that an adaptation of the main insight offered in Preminger’s
(2011) analysis provides a way of linking the presence of SE to Voice and thus connecting
unergative PVs to unaccusative PVs. As described above, the operation AGREE is taken to be an
obligatory, but fallible, operation for Preminger (2011). If there is an active probe with unvalued
phi-features, it must attempt to find a goal, but a default mechanism kicks in if it fails to do so. In
order to apply this idea to unergative PVs, I must first claim that there is an active Voice probe in
unergative PVs that fails to establish an AGREE relation with an active goal. Second, this failure is
interpreted at PF as a trigger for an independently established default mechanism that checks a D
feature or values phi-features on Voice, mainly SE-insertion as proposed in Pujalte & Saab
(2012).14
Turning to concrete examples now, failed agreement followed by SE insertion can be found
in two main contexts. If there is no internal argument, Voice cannot establish an AGREE relation
with a goal and thus it will have unvalued phi-features that are passed on to the interfaces. I take
SE-insertion to be a general repair mechanism for unvalued D and phi-feature in Voice. This is
shown in (42).
14
It is important to note that the appearance of SE as a reflex of failed agreement does not arise in the other types of
case studies discussed in Preminger (2011). This could be due to the fact that there is no probe which fails to find a
goal in such environments in Spanish or due to the fact that such instances involve clitic doubling and not agreement.
23
(42)
a. La bailarina se contonea
The dancer SE.3s swayed
‘The dancer sways’
AGREE
NOM
NO GOAL
b. [TP Tuϕ [VoiceP [DP la bailarina[D, ϕ]] Voice{D, θ}, uϕ [vP v [√P conton-] ] ] ]
c.
T
CLuφ
T
T
Voice
…
AGRφ
T
Voice [uϕ]
The second context in which failed agreement followed by SE-insertion appears, is when there is
an oblique internal argument that is licensed as a PP by the root as in (43). I take such licensing to
be an idiosyncratic property of the root.
(43)
a. Juan se olvidó de la tarea
Juan SE.3s forgot of the homework
‘Juan forgot about the homework’
AGREE
NOM
P blocks AGREE
b. [TP Tuϕ [VoiceP [DP Juan [D, ϕ]] Voice{D, θ}, uϕ [vP v [√P olvid- [PP de [DP la tarea[D, ϕ] ] ] ]
This analysis fulfills the basic desiderata outlined at the beginning of the paper in that the
shared formal properties of PVs are derived in essentially the same way and at the same time the
different argument structure properties they exhibit are accounted for. That said, there still exists
the question of why Voice is an active probe to begin with. Why would a root that does not license
a direct internal argument require a Voice head that needs a goal? This question seems all the more
important when taking into consideration the fact that most unergative verbs do not require this
Voice head. These include pure unergatives such as caminar (walk), bailar (dance) and trabjar
(work) and unergatives that take prepositional complements such as cuidar de (take care of), soñar
con (dream about) and confiar en (confide/trust in). Note that this question is similar to the one
that arose in the last section: why would an unaccusative verb that is perfectly happy without Voice
require an expletive Voice? In that section, I suggested that this may have a lexical semantic
explanation tied to the notions of external/internal causation and (non)spontaneity (Haspelmath
1993; Schäfer 2008) that have been studied independently in analyses of the causative alternation
and anticausatives. In the case of unergative PVs, the answer does not seem to be as clear, at least
24
from a synchronic lexical semantic perspective. However, there is reason to believe that a
diachronic explanation for why such a peculiar set of verbs requires this Voice head may be on the
right track.
The structure in (41b) is, formally speaking, a transitive syntactic structure, but there is a
mismatch between what this structure requires to converge and what the idiosyncratic semantics
of the root provides. The idea that unergative PVs are formally transitive finds support in the
historical record. While in modern Spanish verbs such as quejar-se (complain) and jactar-se
(boast/praises) require SE and cannot be used transitively, they appeared as transitives in earlier
states of the language as shown in (43).
(43)
a. Véo-te, señora, por una parte
quejar el dolor …
I see-2s, madam, on the one hand complain the pain …
‘I see you, madam, on the one hand complain about your pain …’
[Source: La celestina, Author: Fernando de Rojas, Year: 1499-1502]
b. Que no
jacto
valor de mis pasados (Campos & Kempchinsky 1991: 174)
That NEG I boast/praise worth of my ancestors
‘That I don’t boast/praise the worth of my ancestors’
[Source: La cueva de Salamanca, Author: Juan Ruiz de Alarcón, Year: 1608-1620]
My claim is that unergative PVs can be considered cases in which there is a fossilized connection
between a particular root and a transitive Voice head, even though the root has lost the ability to
semantically license a direct internal argument. In modern Spanish, the root √quej(ar)
idiosyncratically licenses a PP argument, not a DP, but there persists a fossilized link between
transitive Voice and this root. This fossilized link is what triggers SE insertion, as shown in (44).
This is based in part on the proposal in Harley & Noyer (2000) that roots may contain certain
licensing features that determine the contexts in which they may appear. I claim that these are only
necessary in cases where roots have very strict licensing features that would have to be learned
individually (like PVs). In other cases, the conceptual meaning of the root will interact with the
structure in which it appears, ultimately determining if the resulting verb makes sense or not. As
can be observed in (44b-c), the particular syntactic licensing conditions imposed on this root and
others like it create a situation in which a probe cannot establish an AGREE relation with a goal. As
shown in (44c), the PP constitutes a locality domain which blocks an AGREE relation between
Voice and the DP, which triggers SE-insertion at PF.
(44)
a. √quej(ar) : PP, [vDO]-[Voice{D,θ}]
b. David se
queja
de su dolor
David SE.3s complains about the pain
‘David complains about his pain’
AGREE
NOM
P blocks AGREE
b. [TP Tuϕ [VoiceP [DP David[D, ϕ]] Voice{D, θ}, uϕ [vP v [√P quej- [PP de [DP su dolor[D, ϕ] ] ] ]
25
To conclude this section, I would like to make an observation regarding a potential
advantage offered by this analysis with respect to idiosyncrasy and meaning change. In the analysis
outlined above, the root does the semantic work in licensing the internal argument, so we expect
to see certain idiosyncrasies regarding whether an internal argument can be licensed at all, and for
those that can license an internal argument, we expect to see variation in category (i.e. – DPs, and
all types of PPs) and interpretation. These expectations are all borne out in the data. Another
important result is that unergative PVs may be built from the same root as certain transitive
structures, but they are not derived from them. This provides a simple way of accounting for
changes in meaning between those verbs that appear to exhibit an alternation, but in reality do not.
Thus, despedirse de Juan (say goodbye to Juan) is not derived from despedir a Juan (fire Juan)
through argument demotion. Rather, the two structures are built independently with the same root.
The root in each of the independent structures is interpreted differently, those specific
interpretations being triggered by local idiosyncratic complements of the root, PP versus DP,
respectively (see Harley 2014 for detailed treatment of root allomorphy and allosemy, conditioned
by local contexts).
To sum up this section, I have proposed that unergative PVs arise due to the interaction of
particular selectional and semantic properties of a set of roots in the language. These roots either
license a PP argument or have implicit internal arguments but select a transitive Voice. The Voice
cannot establish an AGREE relation with DP in its c-command domain and this triggers SE insertion
at PF.
6. Conclusion
In the shift from Principles & Parameters towards Minimalism and Distributed
Morphology, the role of the lexicon has changed dramatically. With respect to transitivity
alternations, one view currently held in this framework is that these are not actually alternations in
which one entry, transitive or intransitive, is more basic and serves as the input to a lexical rule
that derives the other (Schäfer 2008). Instead, these alternations represent different syntactic
constructions in which certain roots may appear. Thus, the number of verbal argument structures
is limited by the same principles that are operative in sentential syntax, (Hale & Keyser 1993,
2002; Marantz 2013) and the meaning of a root (and perhaps other idiosyncratic historical
properties and licensing features) is what ultimately determines how many types of configurations
it may appear in. Given these changes, the traditional conception of a pronominal verb that is stored
as a transitive verb and then undergoes an obligatory lexical operation that is marked by SE is
simply untenable. If the traditional view of PVs is correct, they would constitute a serious problem
for these particular views of the lexicon. In this paper I have demonstrated that both the form and
meaning of PVs can be adequately handled in the MP/DM approach to argument structure without
positing an obligatory lexical rule. I hope to have shown that the division of labor between roots,
argument licensing heads and formal features across different components (syntax and the
interfaces) provides a better solution to the puzzle posed by pronominal verbs than one based on
obligatory lexical operations.
References
Alexiadou, Artemis. 2014. Roots don’t take complements. Theoretical Linguistics 40: 287-297.
Alexiadou, Artemis & Florian Schäfer. 2014. Towards a Non-uniform Analysis of Naturally
Reflexive Verbs. In R.E. Santana-LaBarge (ed.) Proceedings of WCCFL 31, 1-10. Amherst,
Somerville, MA: Cascadilla Proceedings Project.
26
Aranovich, Raúl. 2000. Split intransitivity and reflexives in Spanish. Probus 12: 165-186.
Armstrong, Grant. To appear. Spanish participios activos are adjectival antipassives. The
Linguistic Review.
Basilico, David. 2012. The antipassive and its relation to scalar structure. In M.C. Cuervo & Y.
Roberge (eds.) The End of Argument Structure?, 75-103. Bingley, UK: Emerald Press.
Bogard, Sergio. 2006. El clítico se. Valores y evolución. In C. Company (dir.) Sintaxis histórica
de la lengua española. Primera parte: La frase verbal. V.2, 755-870. Mexico City:
UNAM/Fondo de cultura económica.
Bowers, John. 2010. Arguments as Relations. Cambridge, MA: MIT Press.
Burzio, Luigi. 1991. The morphological basis of anaphora. Journal of Linguistics 2: 81-105.
Campos, Héctor & Paula Kempchinsky. 1991. Case Absorption, Theta Structure and Pronominal
Verbs. In D. Wanner & D. Kibbee (eds.) New Analyses in Romance Linguistic, 171-185.
Amsterdam/Philadelphia: Benjamins.
Cano Aguilar, Rafael. 1999. Los complementos de régimen verbal. In I. Bosque & V. Demonte
(eds.) Gramática descriptiva de la lengua española, Volumen 2, 1807-1854. Madrid: EspasaCalpe.
Chierchia, Gennaro. 2004. A Semantics for Unaccusatives and its Syntactic Consequences. In A.
Alexiadou, E. Anagnostopoulou & M. Everaert (eds.) The Unaccusativity Puzzle: explorations
of the syntax-lexicon interface, 22-69, Oxford, UK: Oxford University Press.
Chomsky, Noam. 2000. Minimalist inquiries: The framework. In R. Martin, D. Michaels, and J.
Uriagereka (eds) Step by step: Essays on minimalist syntax in honor of Howard Lasnik, 89–
156. Cambridge, MA: MIT Press.
Chomsky, Noam. 2001. Derivation by phase. In M. Kenstowicz (ed) Ken Hale: A life in language.
Cambridge, MA: MIT Press.
Coon, Jessica. To appear. Mayan Morphosyntax. Language and Linguistic Compass.
Cuervo, María Cristina. 2003. Datives at large. PhD Dissertation, MIT.
Cuervo, María Cristina. 2014a. Alternating unaccusatives and distribution of roots. Lingua 141:
48-70.
Cuervo, María Cristina. 2014b. Arguments for a root. Theoretical Linguistics 40: 375-387.
Dobrovie-Sorin, Carmen. 2006. The se anaphor and its role in argument realization. In M. Everaert
& H. van Riemsdijk (eds.) The Blackwell Companion to Syntax, v.4, 118-179. Oxford, UK:
Blackwell.
Embick, David. 2004. On the Structure of Resultative Participles in English. Linguistic Inquiry 35:
355-392.
Folli, Raffaella. 2002. Construction Telicity in English and Italian. PhD. Dissertation, Oxford
University.
Grimshaw, Jane. 1990. Argument Structure. Cambridge, MA: MIT Press.
Hale, Ken & Keyser, Samuel J. 1993. ‘On Argument Structure and the Lexical Expression of
Syntactic Relations.’ In K. Hale & S.J. Keyser (eds.) The view from Building 20: Essays in
linguistics in honor of Sylvain Bromberger, 53-109. Cambridge, MA: MIT Press.
Hale, Ken. & Keyser, Samuel J. 2002. Prolegomenon to a Theory of Argument Structure.
Cambridge, MA: MIT Press.
Harley, Heidi. 1995. Subjects, Events and Licensing. PhD. Dissertation, MIT.
Harley, Heidi. 2014. On the identity of roots. Theoretical Linguistics 40: 225-276.
27
Harley, Heidi & Rolf Noyer. 2000. Formal versus Encyclopedic Properties of Vocabulary:
Evidence from Nominalisations. In B. Peeters (ed.) The Lexicon-Encyclopedia Interface, 349374. Oxford, UK: Elsevier
Haspelmath, Martin. 1993. More on the typology of inchoative/causative verb alternations. In B.
Comrie & M. Polinsky (eds.) Causatives and Transitivity, 87-120. Amsterdam: John
Benjamins.
Kayne, Richard. 2009. Antisymmetry and the lexicon. Linguistic Variation Yearbook 8: 1-31.
Koontz-Garboden, Andrew. 2009. Anticausativization. Natural Language and Linguistic Theory
27: 77-138.
Kratzer, Angelika. 1996. Severing the External Argument from its Verb. In J. Rooryck & L. Zaring
(eds.) Phrase Structure and the Lexicon, 109-137. Dordrecht: Kluwer.
Kratzer, Angelika. 2000. Building Statives. In L.J. Conathan et. al. (eds.) Proceedings of the 26th
Annual Meeting of the Berkeley Linguistics Society, 385-399. Berkeley, CA: UC Berkeley
Linguistics Society.
Levin, Beth. 1993. English Verb Classes and Alternations. Chicago: University of Chicago Press.
Levin, Beth. 1999. Objecthood: An Event Structure Perspective. In Proceedings of the Chicago
Linguistics Society 35, volume 1: The Main Session, 223-247.
Levin, Beth. & Malka Rappaport Hovav. 1995. Unaccusativity: At the Syntax Lexical Semantics
Interface. Cambridge, MA: MIT Press.
Lidz, Jeff. 1996. Dimensions of Reflexivity. PhD Dissertation, University of Delaware.
Lidz, Jeff. 2001. The Argument Structure of Verbal Reflexives. Natural Language and Linguistic
Theory 19: 311-353.
MacDonald, Jonathan E. 2008. The Syntax of Inner Aspect. A Minimalist Approach.
Amsterdam/Philadelphia: John Benjamins.
Marantz, Alec. 1991. Case and licensing. In G. Westphal, B. Ao & H-K. Chae (eds.) Proceedings
of the 8th Eastern States Conference on Linguistics (ESCOL 8), 234-253. Ithaca, NY: CLC
Publications.
Marantz, Alec. 2013. Verbal argument structure: Events and participants. Lingua 130: 152-168.
Masullo, Pascual José. 1992a. Incorporation and Case Theory in Spanish: A Cross Linguistic
Perspective. PhD Dissertation, University of Washington.
Masullo, Pascual José. 1992b. Antipassive constructions in Spanish. In P. Hirschbühler & K.
Koerner (eds.) Romance Languages and Modern Linguistic Theory, 175-194.
Amsterdam/Philadelphia: Benjamins.
Mateu, Jaume. 2002. Argument Structure: Relational Construal at the Syntax-Semantics Interface.
PhD Dissertation, Universitat Autónoma de Barcelona.
Mateu, Jaume. 2012. Structure of the Verb Phrase. In J.I. Hualde et. al. (eds.) The Handbook of
Hispanic Linguistics, pp. 333-353. Oxford: Blackwell.
McIntyre, Andrew. 2013. Adjectival passives and adjectival participles in English. In A. Alexiadou
& F. Schäfer (eds.) Non-canonical Passives, 21-41. Amsterdam: Benjamins.
Mendikoetxea, Amaya. 1997. Clitics as AGR, and PRO in Finite clauses. In N. Català & M.
Bargalló (eds.) Proceedings of the 4th Colloquium on Generative Grammar, 84-121.
Tarragona: Servicio de Publicaciones de la Universitat Rovira I Virgili.
Mendikoetxea, Amaya. 1999. Construcciones inacusativas y pasivas. In I. Bosque & V. Demonte
Gramática descriptiva de la lengua española, Volumen 2, 1575-1629. Madrid: Espasa-Calpe.
Mendikoetxea, Amaya. 2012. Passives and se constructions. In J.I. Hualde et. al. (eds.) The
Handbook of Hispanic Linguistics, 477-502. Oxford, UK: Blackwell.
28
Ormazabal, Javier & Juan Romero. 2013. Object clitics, agreement and dialectal variation. Probus
25: 301-344.
Otero, Carlos P. 1999. Pronombres reflexivos y recíprocos. In I. Bosque & V. Demonte (eds.)
Gramática descriptiva de la lengua española, Volumen 1, 1429-1517. Madrid: Espasa-Calpe.
Polinsky, Maria. 2013. Antipassive constructions. In M. Dryer & M. Haspelmath, (eds.) The World
Atlas of Language Structures Online. Leipzig: Max Plank Institute for Evolutionary
Anthropology (Available online at http://wals.info/chapter/108)
Preminger, Omer. 2011. Agreement as a fallible operation. PhD Dissertation, MIT.
Pujalte, Mercedes & Andrés Saab. 2012. Syncretism as PF-repair: The case of se-insertation in
Spanish. In M.C. Cuervo & Y. Roberge (eds.) The End of Argument Structure?, 229-260.
Bingley, UK: Emerald Press.
Ramchand, Gillian. 2008. Verb Meaning and the Lexicon. A First Phase Syntax. Cambridge, UK:
Cambridge University Press.
Real Academia Española. 2009. Nueva gramática de la lengua española. Madrid: Espasa-Calpe.
Reinhart, Tanya & Eric Reuland. 1993. Reflexivity. Linguistic Inquiry 24: 657-720
Reinhart, Tanya. & Tal Siloni. 2005. The Lexicon-Syntax Parameter: Reflexivization and Other
Arity Operations. Linguistic Inquiry 36: 389-436
Reuland, Eric. 2011. Anaphora and Language Design. Cambridge, MA: MIT Press.
Sánchez López, Cristina. 2002. Las construcciones con se. Estado de la cuestión. In C. Sánchez
López (ed) Las construcciones con se, 13-142. Madrid: Visor.
Sanz, Monserrat. 2000. Events and Predication: A new approach to syntactic processing in
English and Spanish. Amsterdam/Philadelphia: Benjamins.
Schäfer, Florian. 2008. The Syntax of (Anti-)Causatives. Amsterdam/Philadelphia: John
Benjamins.
Spreng, Bettina. 2005. Events in Inuktitut: Voice alternations and viewpoint aspect. In R.L.
Edwards, P.J. Midtylyng, C.L. Sprague & K.G. Stensrud (eds.) Proceedings from the main
session of the 41st meeting of the Chicago Linguistics Society, 473-489. Chicago, IL: Chicago
Linguistics Society.
Suñer, Margarita. 1982. The Syntax and Semantics of Presentational Sentence Types in Spanish.
Washington, DC: Georgetown University Press.
Teomiro, Ismael. 2010. Anaphors at the interfaces. PhD Dissertation, Universidad Autónoma de
Madrid (UAM).
Torrego, Esther. 1989. Unergative–unaccusative Alternations in Spanish. MIT Working Papers in
Linguistics 10: 253–269.
Torrego, Esther. 1995. From Argumental to Non-Argumental Pronouns: Spanish Doubled
Reflexives. Probus 7: 221-241.
Varela, Soledad. 2002. Active or Subjective Adjectival-Participles in Spanish. In J.K. Lee, K.L.
Geeslin & J.C. Clements (eds.) Structure, Meaning and Acquisition in Spanish: Papers from
the 4th Hispanic Linguistics Symposium, 304-316. Somerville, MA: Cascadilla Press.
Wehrli, Eric. 1986. On Some Properties of the French Clitic Se. In H. Borer (ed.), The Syntax of
Pronominal Clitics, Syntax and Semantics 19. New York: Academic Press.
Zavala, Roberto. 1997. Functional analysis of Akatek voice constructions. International Journal
of American Linguistics 63: 439-474.
29
Grant Armstrong