Cuervo 2014

Available online at www.sciencedirect.
com
ScienceDirect
Lingua 141 (2014) 48--70
www.elsevier.com/locate/lingua
Alternating unaccusatives and the distribution of roots

María Cristina Cuervo *
Department of Spanish & Portuguese, University of Toronto, Canada
Received 21 July 2012; received in revised form 28 November 2013; accepted 3 December 2013
Available online 14 January 2014
Abstract
This paper argues for two structural types of unaccusative constructions which systematically differ in semantic and morphosyntactic
properties. Evidence is presented from alternating Spanish unaccusative verbs which have a reflexive se-variant and a se-less variant,
such as caer(se) ‘fall’, salir(se) ‘come out/off’, morir(se) ‘die’, ir(se) ‘go/leave’.
The analysis derives the contrasts between the SE and the SE-less variants from their syntactic event structure, and reveals a parallel with
SE-anticausatives and non-alternating SE-less unaccusatives, respectively. The SE-less variant is argued to be the poster case of unaccusative
constructions in which the root expresses a manner of change/motion and the sole argument is introduced as a complement. Contrastingly,
the SE-variant corresponds to a reflexive inchoative configuration comprising two sub-events: an unspecified event of change embedding a
resulting state, lexicalized by the root. The argument DP is licensed as the specifier/subject of the stative verb, and is also the undergoer of the
change. The structural decomposition of SE-unaccusatives into a change and a state captures the dual role of the argument DP, the presence
of the reflexive clitic and the dyadic nature of the predicate without reference to an external argument or a transitive counterpart.
These results provide new arguments against derivational approaches to the causative--inchoative alternation, extending the
empirical and theoretical support for the view that the meaning of verbs is formed within a syntactic-event structure, crucially depending on
the particular combination of the root with verbal-functional elements.
© 2013 Elsevier B.V. All rights reserved.
Keywords: Unaccusative; Anticausative; Argument structure alternation; Reflexive; Roots; Spanish se
1. Introduction
Unaccusative verbs have been central in the study of language, providing a window into the formal and semantic
determinants of argument structure. Their investigation has illuminated the theory of the lexicon, syntactic theory and the
relationship between the two. The idea that there are two types of intransitive verbs and that the only argument in
unaccusatives is an internal argument is now generally assumed in current formal linguistics (since Perlmutter’s
Unaccusative Hypothesis, 1978). Particular approaches have focused on the sole argument being a complement of the verb
or on the lack of external argument as the defining characteristics of unaccusatives, as opposed to unergatives. Dynamic
unaccusatives have been characterized as denoting changes (of state or location), while unergatives are taken to express
activities performed by an agent or instrument. As for other argument structure phenomena, part of the questions and
controversies have centred around whether unaccusativity is ultimately determined by semantic or syntactic properties.
In terms of the empirical base, change of state predicates such as (intransitive) break, freeze and melt have figured
prominently as the prototypical unaccusatives.1 In turn, the alternation between a transitive and an intransitive use of
* Tel.: ++1 416 813 4051.

E-mail address: [email protected].
1
Two-argument unaccusative psych predicates, such as importar ‘matter’ and gustar ‘like’ are left aside.
0024-3841/$ -- see front matter © 2013 Elsevier B.V. All rights reserved.
http://dx.doi.org/10.1016/j.lingua.2013.12.001
M.C. Cuervo / Lingua 141 (2014) 48--70 49
these verbs (notably without any overt morphology in English) has made the study of transitive causatives a necessary
component of theories of unaccusativity. As a consequence, most theories consider that the meaning and syntactic
components of unaccusatives are a subpart of those of transitives, and that the variants are---lexically or syntactically---
derivationally related (Chierchia, 2004; Hale and Keyser, 1993, 2002; Levin and Rappaport Hovav, 1995; Reinhart, 2002;
and others). In many languages, however, there exist unaccusative verbs which do not have a corresponding transitive
use, typically verbs expressing happenings, changes, and existentials (e.g., arrive, die, fall, be missing). This has
generated accounts of this gap as lexical accidents (Chierchia, 2004; see also Alexiadou, 2010), but otherwise these
verbs have been largely ignored in many syntactic discussions of unaccusativity (some exceptions being Masullo, 1992;
Moro, 1997). A systematic, structural comparison between alternating and non-alternating unaccusatives is still pending.
In Spanish, two classes of unaccusative verbs can be distinguished according to the presence or absence of reflexive
morphology in the form of a clitic. Verbs without clitic (SE-less), such as llegar ‘arrive’, venir ‘come’, crecer ‘grow’, nacer ‘be
born’, rodar ‘roll’, suceder ‘happen’, faltar ‘lack’, brotar ‘sprout’, usually do not have transitive variants (with the exception of
some degree achievements such as aumentar ‘increase’, mejorar ‘improve’, engordar ‘get fat(ter), etc.). On the other
hand, verbs with SE, such as hundirse ‘sink’, romperse ‘break’, derretirse ‘melt’, abrirse ‘open’, correspond to the
unaccusative inchoative variant of transitives.2
Interestingly, there is a group of verbs that exhibit two variants, a SE and a SE-less variant, which are both intransitive
(and have no transitive variant), as illustrated in (1). This class includes caer(se) ‘fall’, morir(se) ‘die’, ir(se) ‘go’, salir(se)
‘come out/off’, bajar(se) ‘get down’, subir(se) ‘get up’, escapar(se) ‘escape’, venir(se) ‘come’, quedar(se) ‘stay/remain’,
resbalar(se) ‘slip’. A few verbs that participate in this unaccusative alternation also have a transitive variant, such as
quemar(se) ‘burn’, trepar(se) ‘climb’, volver ‘turn/return’.
(1) a. Cayeron tres hojas V-DP

fell.PL three leaves
‘Three leaves fell’
b. Se cayeron tres vasos SE-V-DP
SE fell.PL three glasses
‘Three glasses fell (down)’
In neither variant is the argument interpreted as an agent or instrument performing an activity, which suggests this DP is
not an external argument. Word order in out-of-the-blue-contexts is V--S in both variants, which contrasts with normal S--V
order for unergative predicates. The clitic in the pronominal variant agrees in person and number with the nominative
subject. Additionally, these verbs can participate in absolute participial clauses (2), which distinguish derived subjects of
unaccusatives (3a) from subjects of unergatives and transitives (3b--c) (Campos, 1999).
(2) a. Una vez caídos 30mm, paró de llover. from caer ‘fall’
One time fallen 30mm, stopped of rain.
‘Once 30mm had fallen, it stopped raining’
b. Salidos los clavos, puede comenzar a lijar la pared. from salirse ‘come off’
Come-off the nails, can start to sand the wall.
‘Once the nails are off (the wall), you can start to sand the wall’
(3) a. Llegados los invitados, empezó la fiesta.

Arrived the guests, began the party
‘Once the guests arrived, the party began’
b. *Trabajados los empleados, el jefe se fue.
Worked the employees, the boss left
Intended: ‘Once the employees had finished working, the boss left’
c. *Entregados los estudiantes las composiciones, el semestre va a terminar.
Turned-in the students the compositions, the semester will end
Intended: ‘Once the students have turned in the books, the semester will end’
(Armstrong, 2011:6)
2
Exceptions are verbs of posture change that either have no transitive variant (arrodillarse ‘kneel down’, acodarse ‘lean one’s elbows on’) or, as
a reviewer notes, limit their transitivity to objects denoting body parts (agacharse ‘crouch’, agachar la cabeza/las orejas/*a alguien ‘bow one’s
head/ ears/ *somebody’).
50 M.C. Cuervo / Lingua 141 (2014) 48--70
Based on these data, I will assume throughout that both variants in (1) are unaccusative. In section 3 this issue is further
discussed in the context of the distribution of bare nouns.
Although the analysis of the diverse SE constructions in Spanish has figured prominently in formal linguistics
(Armstrong, 2011; Basilico, 2010; de Miguel and Fernández-Lagunilla, 2000; Mendikoetxea, 1999, 2012; Zagona, 1996;
Zubizarreta, 1987, among others), the alternation illustrated in (1) has not received a lot of attention nor has it been
analysed as such. Several questions arise with respect to this alternation, both at the descriptive and theoretical levels.
Descriptively, the alternation must be studied in order to determine the properties that the two alternants have in common
and the characteristics that distinguish them. In particular, is there a systematic contrast between the variants? Given that
the alternation does not involve an obvious change in the number or type of arguments, is this a true argument structure
alternation, on a par with the causative/inchoative alternation, or the transitive/anti-passive alternations? On the basis of a
detailed description of morphosyntactic and interpretative properties, and if systematic differences are found, we can ask
about the source and the nature of the contrast, including the contribution of SE. In turn, the analysis of the alternation
makes it possible to consider whether the contrast is an idiosyncratic property of a reduced set of verbs or if, on the
contrary, the variants correspond each to productive configurations. The answers to these questions can not only
illuminate this less studied domain of Spanish, but also contribute more generally to our understanding of unaccusativity, a
phenomenon at the core of the theory of the lexicon--syntax interface. Furthermore, the study of the distribution of verbal
roots in diverse configurations is crucial for a comprehensive theory of argument structure.
This paper is organized as follows. In Section 2, several semantic and morphosyntactic properties of the variants are
presented. The descriptive work evidences that there exist systematic interpretative and formal contrasts between the
variants which suggest a true argument structure alternation. The proposed analysis of the alternation is developed in
Section 3. The central idea is that the two variants correspond to two different base syntactic structures which contrast in
terms of the number of sub-events in the structure, the syntactic licensing and interpretation of the argument, the
combination of the root with functional heads within the verbal projection, and the morphological spell-out of the functional
heads. Differences in interpretation, adverbial modification and aspectual properties derive from the syntactic analysis.
Previous approaches to unaccusatives---inchoatives in particular---are discussed as they illuminate or fail to account for
the SE and the SE-less configurations, and for the alternation between the two. In Section 4, the proposal is further tested in
some of its predictions with respect to idiomatic uses, and the availability and interpretation of dative arguments. Section 5
presents the conclusions.
2. Properties of the variants
As discussed above, the variants in the alternation present three characteristics of unaccusative constructions in
Spanish: word order in out-of-the-blue-contexts is V--S, the postverbal argument triggers verbal agreement, and participial
absolute constructions are possible. Another morphosyntactic property that has been claimed to distinguish Spanish
unaccusatives from other types of predicates is that post-verbal subjects of unaccusatives can be bare nominals without
particular intonation or any type of modification (4a). As noted by Masullo (1992), however, bare nouns are disallowed as
subjects of inchoatives, even in post-verbal position (5).
(4) a. Llegaron invitados.

Arrived guests
‘Guests arrived’
b. *Trabajaron empleados.
Worked employees
‘Employees worked’
(5) *Se derritió manteca. (ok under an impersonal reading)

‘Butter melted’ (Masullo, 1992:272)
The contrast between the simple SE-less unaccusatives in (4a) and SE-anticausatives (5) is reproduced by the variants in
the unaccusative alternation. This is a systematic difference which holds for all the alternating verbs, illustrated in (6) with
caer ‘fall’.
(6) a. Cayeron hojas SE-less

‘(Some) leaves fell’
b’. *Se cayeron hojas SE
‘(Some) leaves fell down’
M.C. Cuervo / Lingua 141 (2014) 48--70 51
Besides this striking syntactic contrast, the variants differ in several semantic and aspectual properties. While some
arguments may be compatible as subjects of both variants (as leaves for fall in (6)), many arguments are only acceptable
for one or the other. The sentences in (7a) show that caer can combine with arguments expressing individuals that fall
‘naturally’, as water falling as rain, leaves falling in autumn. Since glasses do not fall naturally, they are incompatible with
SE-less caer; but they can fall down, as expressed with caerse in (7a’). In parallel fashion, weeds come out (salir) but they
do not come off (salirse) (7b); the opposite holds for nails (7b’).
(7) a. Ayer (*se) cayeron 30 milímetros (de lluvia) a’. *(Se) cayeron tres vasos.
‘Thirty millimetres (of rain) fell yesterday’ ‘Three glasses fell yesterday’
b. Ayer (*se) salieron tres yuyos. b’. *(Se) salieron tres clavos.
‘Three weeds come out yesterday’ ‘Three nails came off yesterday’
Dative arguments can, in principle, appear in both variants, but, as in the case of subjects, there are restrictions in their
compatibility. The locative/recipient dative in (8a) is only compatible with the SE-less variant, while the affected dative in
(8b) requires the SE-variant.
(8) a. Joaquín (*se) le fue con problemas a la jefa

‘Joaquín went to his boss with problems’
b. *(Se) le murió el helecho a Joaquín
‘The fern died on Joaquín’
The variants also differ aspectually. The SE-less variant of some verbs is atelic while for other verbs it is telic, as evidenced
by their compatibility with durative or frame modifiers durante and en, respectively. Caer, like crecer ‘grow’, is atelic
in the absence of elements that bound or measure out the event (9a--b). Morir ‘die’, in contrast, is telic as is llegar ‘arrive’,
(9c--d).3
(9) Atelic
a. El avión cayó durante tres minutos/ *en tres minutos antes de estrellarse.
‘The plane fell for three minutes/ *in three minutes before crashing’
b. Los yuyos crecieron durante tres días/ *en tres días antes de que Juan los arrancara.
‘The weeds grew for three days/ *in three days before Juan pulled them out.’
Telic
c. Los soldados murieron *durante 10 minutos/ en 10 minutos.
‘The soldiers died *for 10 minutes/ in ten minutes’
d. Los soldados llegaron *durante 10 minutos/ en 10 minutos.
‘The soldiers arrived *for 10 minutes/ in ten minutes’
This variability in telicity found among SE-less unaccusatives is not found in the SE-variant of alternating unaccusatives
(10a--b), just as it is not found in SE-inchoatives which alternate with transitive variants (10c)4: all these SE-constructions
are robustly telic.
(10) a. El avión se cayó *durante tres minutos /en tres minutos (# antes de estrellarse).
‘The plane fell down *for three minutes /in three minutes (before crashing)’
b. Los soldados se murieron *durante 10 minutos/ en 10 minutos.
‘The soldiers died *for 10 minutes/ in ten minutes’
c. Los yuyos se secaron *durante tres días /en tres días y Juan los arrancó.
‘The weeds dried up *for three days/ in three days and Juan pulled them out.’
3
For acceptability, I consider here only the reading by which each individual is said to fall, grow, die or arrive during/in a certain period (many
individual events). The readings of durante scoping over the individuals as a group (one collective event) are irrelevant here.
4
As an anonymous reviewer notes, however, alternating degree achievement enfriar(se) ‘cool (off)’ seems to allow an atelic interpretation in its
SE-variant. The same holds for degree achievements which have a SE-less unaccusative variant (aumentar ‘increase’, engordar ‘get fat(ter)’).
Determining whether the current analysis can directly be extended to cover SE and SE-less unaccusative variants of degree achievements requires
further work.
52 M.C. Cuervo / Lingua 141 (2014) 48--70
3. Analysis
The unaccusative alternation between a SE and a SE-less variant applies to a relatively small set of verbs. This
alternation, however, displays an array of systematic morphosyntactic and semantic contrasts reminiscent of other, well-
studied argument structure alternations such as the causative, the dative and the antipassive alternations. These
properties point to different conclusions: while the systematic contrasts point towards a productive, structural alternation,
the restriction to a small set of verbs suggests a lexical, idiosyncratic operation. I develop below an analysis of the
alternation as a true argument structure alternation between two kinds of structurally different unaccusatives. In particular,
I propose that the variants differ in their components and the way the verbal projection is built, involving a different
interpretation of the root (Manner or Result; see Mateu and Acedo-Matellán, 2012). None of the variants is postulated as
basic, the other as derived. If the assumption that the variants are derivationally related via a rule is abandoned, the
apparent lack of productivity of the alternation can be understood in a new light. Although the set of roots that are
compatible with both variants is relatively small, each of the two constructions is independently attested and productive in
the language.5
The main claim is that the SE-less variant corresponds to one class of unaccusative verbs: the non-alternating simple
verbs of change or motion (‘‘true unaccusatives’’ in Masullo, 1992; Mendikoetxea, 2012) such as llegar ‘arrive’, occurrir
‘happen’, crecer ‘grow’, etc. In this configuration, the argument is interpreted as the undergoer of the change. This broad
notion of change is similar to Rappaport Hovav and Levin’s (2010:28), although it includes what they consider non-scalar
change as well as multiple-point scalar changes, as in verbs of directed motion, e.g., fall, ascend, etc. The SE-variant, on
the other hand, corresponds to a change of state, just as the inchoative SE-variant of verbs that participate in the causative
alternation, such as abrir ‘open’, romper ‘break’, derretir ‘melt’, quemar ‘burn’. In this configuration, the argument has a
dual role of undergoer of a change and a subject or holder of a resulting state. The structure of each variant appears below.
As represented in (11a), the verb in the SE-less variant is composed of a dynamic non-agentive verbalizing head vGO and
the root (Cuervo, 2010a; Ramchand, 2008). The structure of this simple, mono-eventive variant is similar to the lower
structure proposed for transitive activity verbs, in which the root merges as an adjunct to the (phonologically null) light verb
v, specifying a Manner (as Mateu and Acedo-Matellán, 2012; Marantz, 2005, among others); in other words, the root
names and specifies the event (the activity for transitives, the change or process for unaccusatives).6 This composed verb
takes the argument DP as its complement. The complement position of v+Root is the position for the semantic licensing of
‘non-canonical’ objects in transitives (Harley, in press; Levin, 1999; Cuervo, 2010a); here this type of licensing is taken to
also hold for simple unaccusatives. In this configuration without an external argument or event that initiates the change
externally, the argument DP names an entity capable of undergoing the change ‘naturally’, as an autonomous process (a
notion similar to ‘internally caused’ in Levin and Rappaport Hovav, 1995) without entailing a final state. Thus, the argument
of SE-less caer ‘fall’ can be rain, snow, leaves, planes (and metaphorically, prices, quality, etc.), as in (12), but typically not
glasses, chairs or food (illustrated in (7) above, and in (13)).
5
In other words, these constructions are not ‘‘lexically determined or semantically frozen’’, as Zubizarreta and Oh (2007:152) characterize
unproductive compounding in Spanish.
6
I assume Cuervo’s (2003) distinction between the two dynamic verbalizing heads vDO for activities with an external argument, and vGO for
changes (including motion) without external argument. The contrast between activities and changes could be captured by distributing the
licensing of the external argument between Voice and v more directly, maintaining that there is only one kind of dynamic event (processes). See
Labelle and Doron (2010) for such a proposal.
M.C. Cuervo / Lingua 141 (2014) 48--70 53
(12) a. Este añ o (*se) cayó la calidad/ el precio de la ropa.

This year SE fell the quality/ the price of the clothes
‘The quality/ price of clothes fell this year’.
b. Esta mañ ana (*se) salieron pimpollos / noticias de Lisboa.
This morning SE went-out rosebuds/ news from Lisboa
‘This morning rosebuds/ news from Lisboa came out’.
The SE-variant in (11b) consists of two sub-events, each represented by a light verb v. The lower subevent represents
the resulting state: the root combines with a stative light verb vBE as its complement (see Basilico, 2010; Embick,
2004; Folli and Harley, 2005; Hale and Keyser, 2002; Labelle and Doron, 2010; Mateu and Acedo-Matellán, 2012,
among others, for somewhat diverse implementations of this structure within a similar spirit). Unlike in the SE-less
variant, here the root names the result: the new state of the argument DP. The composed verb---a phrase---licenses
the DP as its specifier. This type of licensing (semantic and syntactic licensing in Levin, 1999) determines that the DP
has a predictable, structural interpretation as the subject or holder of the state. The sentence in (11b), however, does
not express just a state (estar salido ‘be off’) but a change to a state (‘become off’). This interpretation arises in (11b)
from the embedding of the state under a dynamic light verb vGO which introduces the change. The meaning of BECOME
is syntactically composed of two sub-events (Cuervo, in press). Given that the root merges as the complement of vBE,
it cannot also head-adjoin to the dynamic vGO adding a manner component (see Mateu and Acedo-Matellán, 2012 for
relevant discussion).
Although the change remains unspecified, it is understood that the argument DP undergoes this change until it reaches
the resulting state. Thus, entities that would not fall as per an autonomous process, such as glasses, chairs and food,
can nevertheless be compatible with the SE-variant of caer. Similar facts obtain for salirse ‘come off’, as opposed to
salir ‘come out’.
(13) a. Recién *(se) cayeron tres vasos/ sillas/ galletitas.

just SE fell three glasses/ chairs/ cookies
‘Three glasses/ chairs /cookies have just fallen’
b. Esta mañ ana *(se) salieron tres tornillos/ hojas/ las tapas.
This morning SE went-out three screws/ pages/ the lids
‘Three screws/ pages / the lids came off this morning’
The dual role of the argument as undergoer and holder and, more generally, that the argument is interpreted as a
participant in two sub-events is expressed morphosyntactically by the presence of the reflexive clitic. Unlike in many
previous approaches to anticausatives (Basilico, 2010; Cuervo, 2003; Labelle and Doron, 2010; Schäfer, 2008, among
others), the clitic se in (11b) is not the spell-out of the verbal head per se, that is, se is not the spell-out of a feature (such as
‘non-agentive’) of a verbal functional head v or Voice. The clitic is a pronominal clitic, a D element, which spells-out---as
agreement on vGO---the w-features of the holder of the state, the DP in the specifier of vBE. The reflexive clitic thus satisfies
vGO’s requirement of an argument and gives rise to the interpretation of the DP as both the holder of the state and the
undergoer of the unspecified change. The status of se is further discussed in Section 3.3.
To sum up, the unaccusative alternation is based in the participation of a root in two alternative non-agentive
unaccusative configurations, which contrast in:
the licensing position of the argument (complement vs. specifier)

the number of (sub)events (one vs. two)
the combination of the root with a verbalizing head (adjunct vs. complement)
the morphological spell-out of the v that introduces the dynamic event (Ø vs. reflexive clitic)
the aspectual properties of the configuration (variable telicity vs. necessarily telic)
The SE-less variant of the alternation corresponds to the structure of mono-eventive ‘true’ unaccusatives, while
the SE-variant shares the structure of inchoatives, which alternate with a transitive causative counterpart (anticausatives).
3.1. Evidence for licensing position of argument: restrictions on bare NPs
In Spanish, subjects cannot be bare NPs. This restriction includes pre and postverbal external arguments, pre-verbal
subjects of unaccusatives, and subjects of small clauses. Objects of transitive (non-psych) verbs and post-verbal subjects
of unaccusatives can be bare NPs.
54 M.C. Cuervo / Lingua 141 (2014) 48--70
The restriction also applies to the postverbal nominative DP in psych-verbs and, with some exceptions, to objects of
transitive causatives. In order to capture these two cases as specifiers acting as inner subjects, Cuervo (2003, 2010a),
reformulates Suñer’s (1982) generalization on bare nominals and proposes the following constraint:
(14) The Bare Noun Phrase Constraint

‘‘An unmodified common noun cannot be the subject of a predicate under conditions of normal stress
and intonation’’
This generalization based on a notion of subject that covers internal subjects, together with the analysis in (11) predicts
that there should be a contrast in the availability of bare NPs as postverbal subject of the variants. Specifically, for SE-less
variants---analysed as identical to the non-alternating simple unaccusative predicates such as llegar ‘arrive’---a bare NP as
postverbal subject should be perfectly acceptable. This is so because the argument is licensed as a complement and, as a
consequence, does not fall under the restriction. Bare nominals as postverbal subjects of the inchoative SE-variant, on the
contrary, should be ungrammatical since the arguments are licensed as inner subjects, the specifier of the lower vP. This
is exactly what we find, as noted by Masullo (1992), Fernández Soriano (1999), and illustrated in (6) and below.
(15) Simple unaccusatives (=(11a))

a. No sé dónde cayó (el) agua
‘I don’t know where (the) water fell’
b. No sé dónde murieron (los) hombres
‘I don’t know where (the) men died’
(16) Inchoatives (=(11b))

a. No sé dónde se cayó *(el) agua
‘I don’t know where the water fell’
b. No sé dónde se murieron *(los) hombres
‘I don’t know where the men died’
3.2. Evidence for bi-eventive structure: ambiguous scope of casi ‘almost’ and otra vez ‘again’
In bi-eventive structures, it is predicted that certain adverbials can modify one or the other event; no ambiguity is
expected for mono-eventive structures. The adverbials casi, por poco ‘almost’, which can modify both dynamic and stative
eventualities, should be ambiguous in the SE-variant between scoping over the whole event (vGOP) or just the lower state
(vBEP); there should be no such scopal ambiguity for mono-eventive unaccusatives. These predictions hold, as illustrated
in (17) and (18).
(17) Simple unaccusatives: 1 event

Casi salen tres yuyos
‘There almost appeared three weeds’
1. They almost emerged, but nothing actually did; scope = vGOP
(18) Inchoatives: 2 events

Casi se salen dos clavos
almost SE come-off two nails
1. The nails almost started to come out, but did not; scope = vGOP
2. The nails did start to come out, but did not come off completely (there was no attainment of a final state);
scope = vBEP
The lack of ambiguity of casi for SE-less salir and the ambiguity with salirse fall directly from the event structures in (11). von
Stechow (1995) argues that again is sensitive to the number of events in a syntactic configuration. Specifically, he notes
the ambiguity between a restitutive reading (narrow scope over the result sub-event) and a repetitive reading (wide scope
over the whole event) in bi-eventive causatives as opposed to mono-eventive activities. The same reasoning applies to
the contrast among unaccusative verbs developed in (11): again should be ambiguous in the bi-eventive SE-variant but
unambiguous in the SE-less variant.
M.C. Cuervo / Lingua 141 (2014) 48--70 55
(19) Simple unaccusatives: 1 event

Bajaron otra vez los bonos hoy.
‘Bonds dropped again today’
1. Repetitive reading: the bonds dropped yesterday, and they dropped again today.
(20) Inchoatives: 2 events

Subimos al omnibus, pero nos bajamos otra vez.
‘We got on the bus, but we got off again’.
1. Repetitive reading: we got off the bus once, then we got on, and we got off again.
2. Restitutive: we were off the bus, we got on, and then were off the bus again.
As predicted, with the se-less variant bajar, otra vez ‘again’ can only refer to the repetition of the event vGO of going down
(in this case, getting off the bus). With the inchoative variant, in contrast, otra vez can mean that the whole event took place
again but, crucially, it can also mean that the argument DP was restituted to an initial state of being down (off the bus in
(20)); otra vez takes narrow scope over the final state vBEP.
In sum, adverbial ambiguity (or lack thereof) supports the proposed contrast between the variants in terms of the
number of events (and their corresponding verbal heads).
3.3. Evidence for argument-related SE
Although most literature on Spanish unaccusatives presents only cases with SE, the argument DP is actually not
restricted to 3rd person. If the argument DP is not 3rd person, the clitic is the reflexive corresponding to the person and
number of the DP. That is, unlike impersonal SE, the SE in unaccusatives is paradigmatic (‘anaphoric’ in Mendikoetxea,
2012). This is the case for the SE-variant of both alternating unaccusatives (21a--b) and causative--anticausatives (22a--b).
(21) SE variant
a. Yo nunca me caí al agua.
‘I never CL.1.SG fell to.the water
‘I never fell into the water’
?
b. Por qué te fuiste de ese trabajo?
for what CL.2.SG went.2.SG of that job?
‘Why did you leave that job?’
(22) ‘Regular’ inchoatives

a. Vos nunca te hundiste en la arena/ derretiste con ese calor.
‘You never CL.2.SG sank.2.SG in the sand/ melted with that heat
‘You never sank in the sand/melted in that heat’
b. Después de la pileta nos secamos al sol.
After of the pool CL.1.PL dried.1.PL at.the sun
‘After coming out of the pool, we dried in the sun’
If the argument DP participates in both sub-events, as the undergoer of the unspecified change and the holder of the state, the
presence of reflexive morphology falls straightforwardly. The variation (agreement) in SE supports the idea that the clitic is a
thematic clitic and accounts for the w-features in a more natural way than proposals that analyse inchoative SE as the spell-out
of a verbal feature of the v or Voice head, or as a telicity marker. Some of these accounts are briefly discussed in Section 3.7.
In terms of their meaning, Cuervo (in press) presents the following characterization for change of state predicates such
as derretirse ‘intr. melt’. The formalization in (23) states that there is an event of change GO in which there is an Undergoer
participant and that there is a state BE of being melted of the same participant. Building on Rothstein (2001), a rule of
semantic composition ensures the event is understood as the culmination of a change that is included in the state.
(23) derretirse ‘melt’: lxlels [GO (e) & Undergoer (e, x) & BE (s) & Holder (s, x) melted (s)]
Similarly, the DP in the SE-variant of alternating unaccusatives associates with two roles, one for each sub-event, as in (24).
(24) caerse ‘fall’: lxlels [GO (e) & Undergoer (e, x) & BE (s) & Holder (s, x) fallen (s)]
Assuming a system such as Adger’s (2003) in which selectional requirements of verbal heads are expressed as the head
having an uninterpretable nominal feature uN, vGO has a uN feature that must be satisfied. In simple unaccusatives, the
56 M.C. Cuervo / Lingua 141 (2014) 48--70
complement DP satisfies this requirement via an agreement relation with the DP argument. In the case of SE-
unaccusatives there is a uN selectional feature for each verbal head; the uN feature of the lower vBE is checked first
when the DP is merged, and then the uN feature of vGO is satisfied via agreement. This feature checking may be
understood as valuing of the uN in vGO with w-features once the lower verb (vBE+Root) raises and incorporates in vGO.
Under this analysis, SE-unaccusatives are cases of reflexivization as argued for in Chierchia (2004), Koontz-Garboden
(2009), Reinhart and Siloni (2005), among others. The crucial difference is that here ‘reflexivization’ is not taken to be a
lexical operation of reduction of the external argument; rather, it is a syntactic phenomenon of structure building and
feature checking. Given that SE-unaccusatives are a sort of dyadic structure without the syntax or the semantics of an
external argument (cf. Masullo, 1992; Mendikoetxea, 2012), this view allows for the construction of SE-unaccusatives with
roots that do not have transitive-causative uses.7
Another piece of evidence for the clitic in these SE-unaccusatives being thematic is that it patterns with other thematic
reflexives---true reflexives and reciprocals, and anticausatives---in the contexts of causative verbs such as hacer ‘make’
and dejar ‘let’. When embedded under these causative light verbs, reflexive verbs and inchoatives can (apparently
optionally) appear with or without SE.
(25) a. Lo dejaron afeitar(se) Reflexive

him let.3.PL shave-SE
‘They let him shave’
b. Nos hicieron dar(nos) la mano. Reciprocal
us made.3.PL give-SE the hand
‘They made us shake hands’
c. Lo dejaron quemar(se) Anticausative
it let.3.PL burn-SE
‘They let it burn’
d. Los hicieron caer(se) SE-variant
them made.3.PL fall-SE
‘They made them fall’
In contrast, obligatorily pronominal verbs such as quejarse ‘complain’, atreverse ‘dare’, abstenerse ‘abstain’, arrepentirse
‘regret’, darse cuenta ‘realize’, as well as antipassive lamentarse ‘lament’ and olvidarse ‘forget’, for which the reflexive
clitic has no argumental interpretation, must keep the reflexive morphology in the same context.8
(26) a. *Lo dejaron quejar

him let.3.PL complain
‘They let him complain’
b. *Nos hicieron atrever a cruzar el río
us made.3.PL dare to cross the river
‘They made us dare cross the river’
7
Given an appropriate human subject, many anticausatives are in fact ambiguous between an inchoative interpretation and a true (agentive)
reflexive reading (as discussed in Koontz-Garboden, 2009). Interestingly, cases can be found in which the SE-variant of an alternating
unaccusative is used as agentive too (i). These uses, together with the transitive use of alternating unaccusatives in certain dialects and even
cases of transitive use outside those dialects as in (iia,b) illustrate the flexibility of roots to appear in non-typical environments with clear, structural
meanings.
? ?
(i) Tropezó? Se cayó a propósito? Que cada uno juzgue.
‘Did she trip? Did she SE-fall on purpose? Let each one decide’
www.tvzapping.org/programas/esta-pasando
(ii) a. . . . por si alguien especula que lo cayeron para borrar huellas. Argentina
‘In case somebody speculates that they ‘‘fell’’ it (the plane) in order to erase the clues’
http://www.aviacionargentina.net/foros/aviacion-naval-o-coan.23/8416-accidente-electra
b. Se queda, o lo quedaron. Mexico
‘He is staying (in his position), or they ‘‘stayed’’ him’.
http://www.debate.com.mx/eldebate/noticias/default.asp?IdArt=11579142&IdCat=6115
8
As noted by Armstrong (2011), there is some controversy regarding the (un)acceptability of omission of the clitic in some examples. I have
provided examples in which the embedded subject is pronominalized, which sharpens the contrast between (25) and (26). Google searches
confirm these judgments: for cases like (25), the SE-less variants are much more numerous; for (26), the SE-less variants are either much fewer
than the SE-variant or are not attested at all. See Armstrong (2011) for discussion of previous work.
M.C. Cuervo / Lingua 141 (2014) 48--70 57
Although an analysis of these facts is beyond the scope of the present work, the presence or absence of the clitic in
embedded unaccusatives (e.g. (25c--d)) is not a question of simple variability (cf. Armstrong, 2011; Kempchinsky, 2004),
but rather each option corresponds to a structural difference in the complement of the causative verb. In particular, the
SE-less variant (e.g., hacer caer, ‘make fall’) may correspond to the causative verb embedding a mono-eventive vGOP
(the simple unaccusative meaning of change; a process with no external initiator) while the SE-variant (hacer caerse,
‘make SE-fall’) is the bi-eventive inchoative structure with the change of state meaning. The causative vP layer (with hacer
‘make’ or dejar ‘let’), by adding an event and the external initiator, allows the combination of SE-less unaccusatives with
arguments that would usually require the SE-inchoative structure (27).
(27) a. *(Se) cayeron tres vasos/ sillas/ galletitas

SE fell.3.PL three glasses/ chairs/ cookies
‘Three glasses/ chairs /cookies have just fallen’
b. Dejaron/ hicieron caer tres vasos/ sillas/ galletitas.
let.3.PL made.3.PL fall three glasses/ chairs/ cookies
‘They let/made three glasses/ chairs /cookies fall’
When the argument DP would, in principle, be an adequate argument for either variant, the presence or absence of
SE correlates with the expected difference in meaning between a change (or motion) and a change of state (or location).
This can be seen both with regular inchoatives (28) and with alternating unaccusatives (29).9
(28) a. Al tronco, lo dejaron quemar (durante la noche) Change10

the log, it let.3.PL burn
‘The log, they let it burn’
b. Al tronco, lo dejaron quemarse Change of state
the log, it let.3.PL burn-SE
‘The log, they let it burn out’
(29) a. Al payaso, lo dejaron bajar (al sótano) Motion

the clown, him let.3.PL descend to.the basement
‘The clown, they let him go down (to the cellar)’
b. Al payaso, lo dejaron bajarse (*al sótano) Change of location
the clown, him let.3.PL descend-SE to.the basement
‘The clown, they let him get-off’
In contrast to inchoatives, if the SE in obligatorily pronominal verbs as those in (26) is not thematic, it is natural that its
presence would not be affected by changes in argument structure, such as the embedding under a causative verbal
layer. Interestingly, a subgroup of obligatorily pronominal verbs has been analysed as different from the pseudo-
reflexive quejarse ‘complain’, atreverse ‘dare’ class. Armstrong (2011) presents evidence that acatarrarse ‘get a cold’,
acalambrarse ‘cramp up’, afiebrarse ‘get a fever’, ensimisimarse ‘get lost in thought’---and I would add arrodillarse ‘kneel
down’ and acodarse ‘lean on one’s elbows’---are obligatorily inchoative; that is, they are verbs like quemarse ‘burn’ and
derretirse ‘melt’ but which do not have a transitive causative variant, presumably due to the oddity of the meaning the
transitive structure would have (#to kneel somebody down, #to cramp somebody up). This subgroup of verbs allows the
omission of the clitic just like reflexives and regular inchoatives, supporting the analysis sketched above. Compare (26)
with (30).
(30) a. Lo dejaron arrodillar(se) ante la estatua.

him let.3.PL kneel-SE at the statue
‘They let him kneel at the statue’
b. Me hizo acalambrar(me)
me made.3.SG cramp up.
‘It made me cramp up’
9
As an anonymous reviewer points out, it is predicted that there would be a three-way ambiguity of otra vez ‘again’ in the case of SE-
unaccusatives embedded under a causative verb, and a two-way ambiguity in SE-less unaccusatives embedded under a causative verb. This is
indeed the case. The addition of causative hacer to (19) and (20) adds a repetitive reading of the whole event to the one and two readings,
respectively, obtained in each case without hacer. The same also holds if otra vez is added to (29a--b).
58 M.C. Cuervo / Lingua 141 (2014) 48--70
To sum up, the paradigmatic nature of SE, the contrast in possible arguments for one or the other of the variants, the
possibility of ambiguity with true reflexives, the behaviour under causative light verbs and, more indirectly, the evidence for
a higher dynamic sub-event all suggest that SE in alternating unaccusatives is an argument-related reflexive. Never-
theless, claiming that the clitic is a thematic reflexive does not imply that it has all the properties of ‘true’ reflexives. In
particular, as opposed to true reflexives in transitives (31), the reflexive in SE-unaccusatives (32) cannot be replaced by a
referential DP nor can the argument DP be doubled by the strong reflexive ‘a sí mismo’.
(31) a. Agustín se afeitó Reflexive

Agustín SE shaved
‘Agustín shaved’
b. Agustín afeitó a Joaquín
Agustín shaved Joaquín-ACC
‘Agustín shaved Joaquín’
c. Agustín se afeitó a sí mismo
Agustín SE shaved himself-ACC
‘Agustín shaved himself’
(32) a. Agustín se cayó SE-unaccusative

Agustín SE fell
‘Agustín fell down’
b. *Agustín cayó a Joaquín
Agustín fell Joaquín-ACC
*‘Agustín fell Joaquín’
c. *Agustín se cayó a sí mismo
Agustín SE fell himself-ACC
*‘Agustín fell down himself’
Given these facts, reflexive unaccusatives correspond to ‘‘inherent reflexives’’ in Schäfer (2012a), characterized by being
always reflexive.11 Thus, reflexive unaccusatives are not unique in consisting of a verb that takes two arguments that must
refer to the same entity. The crucial difference with transitive reflexives is that their reflexivity is the expression of an
argument having a role in each sub-event of a complex verb, not two roles in a ‘standard’ transitive verbal configuration.
That the undergoer of the change and the holder of the state are necessarily co-referential is reminiscent of reflexive
resultative constructions in English such as drink oneself silly, sing oneself hoarse, in which the same argument
participates in two events, as agent of the activity and holder of the result state. This view of reflexive resultatives and the
contrast with true reflexives is argued for in Everaert and Dimitriadis (2013) in terms of argument bundling. The roles of
undergoer of a change and holder of a state have been independently proposed in previous work, Koontz-Garboden
(2009) and Ramchand (2008) being particularly relevant.
3.4. Distribution of roots
A central claim of this analysis is that the variants in the unaccusative alternation involve a different structural
interpretation of the root. In the SE-less variant, the root is interpreted as Manner (in an abstract, structural sense) by virtue
of its combining as an event modifier of the unaccusative dynamic vGO head; the complex head v+√ then takes an
argument as complement. In the SE-variant, the root combines with a stative vBE head as its complement, forming a
predicate which takes the argument DP as its specifier/subject. This proposal builds on a distinction that has been
developed for transitive verbs and has been the focus of recent research. This kind of contrast is represented below in
terms of the Manner or Result interpretation that one and the same root can have when merged differently in two different
structures according to Mateu and Acedo-Matellán (2012).
10
This meaning of the unaccusative SE-less variant of quemar, not common in every variety of Spanish, is equivalent to another verb arder, a
non-alternating unaccusative as English burn in The fire is burning. This same use is also attested in Greek, the verb appearing, interestingly,
without Nonactive morphology typical of intransitives alternating with transitive counterparts (Alexiadou and Anagnostopoulou, 2004:124). For
this environment, the authors state that ‘‘the single argument does not undergo a change of state’’.
11
Schäfer distinguishes between inherently reflexive (‘be ashamed’ in German and Dutch), naturally reflexive (shave, wash) and naturally
disjoint predicates (hate, accuse).
M.C. Cuervo / Lingua 141 (2014) 48--70 59
(33) a. [vP [v √BREAK v] [SC [DP he] [into the room]]] (He broke into the room)
b. [vP v [SC [DP the glass] [√BREAK]]] (The glass broke)
Mateu and Acedo-Matellán (2012:211)
c. [vP [v √CAER vGO] [DP/NP nieve]] Manner: Cayó nieve ‘Snow fell’
d. [vP vGO [vP [DP un libro] [vP vBE √CAER]]] Result: Se cayó un libro ‘A book fell down’
The root √BREAK can alternatively merge as a Manner modifier in (33a), or it can merge as the predicate in a small clause
structure embedded under (unmodified) v in inchoative break (33b). This idea that a root can merge in different
configurations and, in particular, that the root can name a Manner or a Result can also be found in, among others,
Acedo-Matellán (2013), Cuervo (2003), Embick (2004), Harley (2005), Labelle and Doron (2010), Marantz (2005), Mateu
(2002), Ramchand (2008), Rappaport Hovav and Levin (2010).12 Manner, in this structural sense, is a notion broader than
the semantic notion attributed to roots, as in Rappaport Hovav and Levin’s (2010) characterization of manner as non-
scalar change. This expanded notion of Manner as an adjunct specifying a v head, as in (33c), is concomitant with the
more restricted concept of Result as a root forming a stative predicate embedded under a dynamic event.13 The difference
in interpretation of the root which derives from this analysis is evidenced by the contrasting compatibilities of subject
arguments discussed in Section 2, and illustrated in (12)--(13). The difference in the structural position of the root and the
verbal meaning built on this basis is also manifested by the contrasting interpretation of manner and quantity adverbials
such as mucho ‘a lot’, bien ‘well’ and rápido ‘quickly’.
3.4.1. Mucho
The adverb mucho varies in interpretation according to the type of unaccusative variant. In the SE-less variant of caer
‘fall’ (34a) it normally expresses the distance, measures the (atelic) event of falling in its extension. In the SE-variant this
interpretation is not accessible, and mucho is typically interpreted iteratively, expressing that the whole event occurred
many times (34b), or it acts as a degree modifier of the result (34c).
(34) a. La bolsa no cayó mucho
The stock market not fell much
‘‘The stock market did not fall too much.’’
b. Para ser tan torpe, no se cayó mucho
to be so clumsy not SE fell much
‘‘Considering (s)he is so clumsy, (s)he did not fall too many times’’
c. Por suerte, los clavos no se salieron mucho
for luck the nails not SE came.out much
‘Luckily, the nails did not come out a lot’
The extension interpretation obtained in the SE-less variant (34a) is typical of mucho with unergative manner verbs: mucho is
interpreted as for a long time with activities (35a), and for a long distance with manner of motion verbs (35b). With causatives
and anticausatives, in contrast, we get a degree modification of the result (35c,d), or an iterative interpretation (35e).
(35) a. Tomás trabajó/ jugó/ bailó mucho Activities

Tomás worked/ played/ danced a lot/for a long time
b. Tomás corrió/ caminó mucho Manner of motion
Tomás ran/ walked a lot/for a long distance
c. El sol derritió mucho la manteca Causative
‘The sun melted the butter a lot (=well melted)’
d. La manteca se derritió mucho Anticausative
‘The butter melted a lot (=got well melted)’
e. Los Peugeots se rompen mucho Anticausative
‘Peugeots break a lot (=all the time)’
12
I do not subscribe, however, to analysing the contrast in terms of conflation versus incorporation processes (cf. Hale & Keyser, 1993, 2002;
Haugen, 2009; Mateu and Acedo-Matellán, 2012), since I assume that in both cases an acategorial root merges directly with a v head to form a
verb. I remain agnostic with respect to the formation of verbs via head movement of nominals or prepositions.
13
Note that some dynamic telic verbs which are usually assumed to lexically involve a change of state or express a result (transitive find,
unaccusative arrive) do not necessarily involve a Result phrase (an embedded stative vP lexicallized by the root) under the current analysis, the
root merging as an adjunct of the dynamic v. This broad, more abstract notion of Manner also covers what have been described as verbs
lexicalizing paths and incremental themes.
60 M.C. Cuervo / Lingua 141 (2014) 48--70
3.4.2. Bien
The adverb bien ‘well’ has a salient manner meaning, but can also function as a degree modifier (Armstrong, 2011).
The manner interpretation is perfectly compatible with the SE-less variant of alternating unaccusatives (36a,b), as it is with
activity verbs, (36c). In the SE-variant, in contrast, bien is either infelicitous (37a), or is interpreted as a degree modifier of
the resulting state (37b), the reading available with anticausatives (37c).
(36) a. La paracaidista principiante cayó bien. SE-less unacc

The skydiver beginner fell well
‘‘The beginner skydiver descended well.’’
b. La torta salió bien. SE-less unacc
The cake came-out well
‘‘The cake came out fine’’
c. Tomás trabajó/ bailó bien. Unergative
‘Tomás worked/ danced well’
(37) a. ??La paracaidista principiante se cayó bien. SE-unacc

The skydiver beginner SE fell well
‘‘The beginner skydiver fell down well.’’
b. Los clavos no se salieron bien. SE-unacc
the nails not SE came.out well
‘The nails did not come properly/completely off’
c. El chocolate se derritió bien. Anticausative
the chocolate SE melted well
‘The chocolate melted completely (well melted)’
3.4.3. Rápido
The adverb rápido ‘quickly, fast’ can alternatively be interpreted as ‘at high speed’ (manner) or ‘within a short period of
time’ (temporal frame). The manner interpretation is available for both transitive or unergative activity/manner verbs and
SE-less (atelic) unaccusatives (38); the frame is the salient reading for causatives, anticausatives and the SE-variant: the
result state is achieved within a short time (39).
(38) a. Tomás manejó rápido la camioneta Transitive

‘Tomás drove the truck fast’
b. Tomás trabajó/ caminó rápido Unergative
‘Tomás worked/ walked fast’
c. La pelota cayó rápido por la escalera SE-less unacc
‘The ball fell down the stairs fast’
(39) a. El sol derritió la manteca rápido Causative

‘The sun melted the butter quickly (=in a short time)’
b. El Peugeot se rompió rápido, cuando era nuevo todavía Anticausative
‘The Peugeot broke quickly, when it was still new’
c. Pero no se cayó rápido, pasaron 20 días más o menos. SE-unacc
‘It (the wart) did not fall off quickly, 20 days went by more or less’
To sum up, these interpretative facts show that a manner component is not available for modification in the SE-variant of
alternating unaccusatives, but is available when the same verb appears in the SE-less variant. This follows from the
structural analysis in (11): the manner is specified in simple unaccusatives and is therefore a possible target of adverbial
modification. In the SE-inchoative variant, just as in SE-anticausatives and causatives, the manner is unspecified and
cannot be targeted; modifiers can either modify the result or the whole complex event.
3.5. Telicity
We saw in Section 2, in examples (9) and (10), that while the event expressed by the SE-variant is always telic, the event
expressed by the SE-less variant can be telic or atelic depending on the particular root (caer ‘fall’, bajar ‘descend’, subir
M.C. Cuervo / Lingua 141 (2014) 48--70 61
‘ascend’, and morir ‘die’, respectively).14 This variation in telicity reproduces exactly, as expected, the one found among
simple unaccusative verbs such as atelic crecer ‘grow’, rodar ‘roll’ on the one hand, and telic llegar ‘arrive’, nacer ‘be born’,
on the other. Again, this is also found among agentive manner verbs such as dance and hammer (atelic) and kill (telic).
The contrast in telicity among transitives gets neutralized when the verbs appear in a telic construction, such as
resultatives, in which a manner root embeds a result state (the event expressed being a complex causative).
(40) Manner: atelic or telic

a. Daniel hammered the metal for two hours/ *in two hours Atelic root
b. Daniel killed the cockroach *for two seconds/ in two seconds Telic root
Resultatives: only telic
a’. Daniel hammered the metal flat *for two hours/ in two hours Atelic root
b’. Daniel killed the fly well dead *for two seconds/ in two seconds Telic root
The neutralization of telicity contrasts among alternating unaccusatives functions the same way: the SE-variant, by virtue
of having an underlying inchoative, complex structure (a dynamic event vP embedding a result vBEP) is necessarily (i.e.,
structurally) telic. In terms of the alternation, the change of a verb such as caer ‘fall’ from an atelic simple unaccusative to a
telic change of state caerse parallels the alternation between the simple manner use of hammer in (40a) and its use in the
resultative (complex causative) construction (40a’).
The contrast also emerges in the entailments of the two constructions. An atelic root in the SE-less variant does not entail a
final result (41). In contrast, denying the attainment of the final result in the SE-variant generates unacceptability (42).
(41) SE-less atelic

a. El avión cayó durante tres minutos. . . pero no se estrelló.
‘The plane fell for three minutes. . . but it did not crash.’
b. Los soldados subieron por la escalera pero no llegaron ni al primer piso.
‘The soldiers went up the stairs. . . but they did not even reach the second floor.’
c. Los chicos fueron al supermercado. . . pero ya están de vuelta.
‘The kids went to the supermarket. . . but they are already back.’
(42) SE-variant telic

a. El avión se cayó. . . #pero no se estrelló.
‘The plane fell down. . . but it did not crash.’
b. Los soldados se subieron al techo. . . #pero se resbalaron y nunca llegaron.
‘The soldiers climbed up to the roof. . .#but they slid off and never reached it.’
c. Los chicos se fueron al supermercado. . . # pero ya están de vuelta.
‘The kids are gone to the supermarket. . . #but they are already back.’
Leaving technical details aside, what is clear is that the telicity depends on the aspectual properties of the root when there
is only one dynamic event the root lexicalizes---as is the case in the SE-less variant---while the SE-variant is structurally telic
because it contains a final state, named by the root.
3.6. Summary
Table 1 summarizes the central properties of each construction, as discussed in the previous sections.
3.7. Previous approaches
The analysis of the semantic and morphosyntactic properties of the two variants developed in the previous section is
informed by many previous studies. In turn, the facts of this alternation are relevant for issues discussed in previous work, in
particular those concerning anti-causatives, types of unaccusative structures, the distribution of roots in various
configurations, the role of unaccusative SE, and the relationship between variants in argument structure alternations. Many
aspects of the current analysis have been proposed before for other, related phenomena. I focus below on the specific
challenges raised by the unaccusative alternation and how some of the well articulated previous approaches still fail to
account for them.
14
The telicity of the SE-configuration can nevertheless be altered by aspectual operators external to the vP such as progressive or habitual.
62 M.C. Cuervo / Lingua 141 (2014) 48--70
Table 1
Properties of the variants in the unaccusative alternation.
# of participants/ Type of Subject licensed as Interpretation of Interpretation of Telicity

# of events eventuality argument root
SE-less: Simple 1/1 Change Complement of Undergoer Adjunct/Manner Telic or

unaccusatives the root = object atelic
SE-variant: 1/2 Change Specifier of lower Undergoer & Complement/ Telic
Inchoatives to a state vP = subject Holder Result
The work that goes closest to dealing with the phenomena presented here in terms of an alternation is Masullo (1992).
He observes that a group of movement verbs, such as ir ‘go’, venir ‘come’, volver ‘return’, and bajar ‘descend’, alternate
between a ‘reflexive’ and a ‘non-reflexive’ construction. He argues that in the SE variant the clitic stands for an incorporated
source argument. He contrasts this structure with anticausatives, such as romperse ‘break, intr.’ and derretirse ‘melt, intr.’,
for which he proposes an analysis of SE as the incorporated external argument. The derivation for the sentence Juan se
fue de casa ‘Juan left home’ appears in (43).
(43) DS: [IP [VP fue [NP Juan] [PP se[SOURCE]i] de casai]]
SS: [IP Juanj [VP fue-sei tj ti de casaj]]
PF: Juan se fue de casa Masullo (1992:246)
The claim that these movement verbs imply a source argument prevents the approach from accounting for the SE variant of
a verb like subirse ‘go up’ (which, if anything, implies a goal), caerse ‘fall down’ (which can imply a source or a goal),
quedarse ‘stay’ (which could imply a location), or morirse ‘die’ (which does not seem to imply either a source or a goal, or a
location argument).15 Thus, the analysis effectively divorces these verbs from other verbs of change of location and, more
explicitly, from verbs of change of state (anticausatives). As a result, this analysis of the alternation cannot explain in a
straightforward way the set of properties shared by alternating unaccusatives and anticausatives. Lastly, there are no
hints whatsoever of why such an initial configuration (DS, Deep Structure in (43)) would exist. Mendikoetxea (2012) and
Zagona (1996), among others, present irse ‘leave’ and bajarse ‘get down/off’ as cases of aspectual SE, again to be
distinguished from unaccusative--inchoative SE constructions.
Beyond an account of the particular semantic and morphosyntactic properties of each of the variants, the challenge for
an analysis of the alternation is to derive the systematic contrasts between the SE and SE-less variants, on the one hand,
and their parallel with SE-anticausatives and non-alternating SE-less unaccusatives, respectively, on the other. This
crucially includes an explicit analysis of the role of the reflexive. Since this reflexive has been mostly analysed for
anticausatives, previous accounts are organized first in terms of their analysis of SE.
As was observed for anticausatives and alternating unaccusatives above, the presence of SE correlates with telicity,
specifically, SE-constructions are telic constructions. This correlation is also found in other constructions in Spanish---as in
Pedro se comió la pizza ‘‘Pedro ate up the pizza’---and it has been proposed that SE is a verbal operator or aspectual
marker (de Miguel and Fernández-Lagunilla, 2000; Nishida, 1994; Zagona, 1996; Zubizarreta, 1987, among others).
Although this view incorporates the idea that arguments are related to sub-events, if the role of SE is to add or signal telicity,
it is still not clear how other contrasts between the variants are derived, in particular, the change in meaning and the
difference in the arguments that can appear with one or the other of the variants.
An influential approach to reflexive SE in inchoatives is that it is the expression of the suppression of the external
argument, a process that has been claimed to take place either in the lexicon or in the syntax, by lexical binding, real
reflexivization, bundling, incorporation or case absorption (Chierchia, 2004; Koontz-Garboden, 2009; Horvath and Siloni,
2013; Levin and Rappaport Hovav, 1995; Masullo, 1992; Reinhart and Siloni, 2005, i.a.). Given the evidence that the SE-
variant of alternating unaccusatives has the structure of inchoatives/anticausatives, the reduction-based approach to
anticausative SE cannot be correct: there is no sense in which the SE in caerse can signal the suppression of the external
argument of the non-reflexive variant caer, which is an unaccusative lacking a transitive variant. More generally, the
existence of the unaccusative alternation constitutes a critical challenge for theories which claim that there is a
derivational relation between the variants of the causative alternation. Such theories leave the unaccusative alternation in
need of an independent explanation.
15
As an anonymous reviewer points out, another reason for not treating se as a source argument comes from other Romance languages, such
as Catalan, Aragonese and French, in which these verbs appear with both se and a separate source argument expressed by a source/partitive
clitic en/ne (e.g., Catalan Joan se n’ha anat, ‘Joan has left’; Joan se n’ha tornat, ‘Joan came back’).
M.C. Cuervo / Lingua 141 (2014) 48--70 63
A related kind of approach is that SE, although not literally indicating the suppression of the external argument of the
transitive variant, is a marker of its absence, or the spell-out of a non-agentive Voice or causative v head (Alexiadou, 2010;
Armstrong, 2011; Basilico, 2010; Embick, 1998; Labelle and Doron, 2010; Nishida, 1994; Schäfer, 2008). However, these
analyses do not directly account for the differences between the variants in terms of possible subjects and complexity of
the structure. The reflexive and non-reflexive anticausative variants of causative verbs in French discussed by Labelle
and Doron (2010) exhibit a very strong parallel to the alternation in Spanish. These authors also develop an account that
derives the difference in the verbal meaning of each construction as arising from different underlying structures, which
contrast in aspects shared by my approach, such as the insertion position of the root, the interpretation of the subject, the
focus on a change or a result. For them, however, the reflexive variant is a less complex configuration, just the expression
of a result without a representation of a dynamic non-agentive event. Without this higher dynamic sub-event, several
properties of the Spanish alternation are left unaccounted for (e.g., adverbial ambiguity, dual role of the argument, different
licensing position of the argument). Additionally, if the clitic is the spell-out of a verbal head per se, that is, it is not related to
selectional nominal or w-features, there is no reason for SE to be sensitive not only to the absence of external argument but
also to the features of the subject argument; thus, the paradigmatic nature of unaccusative SE seems to be an accidental
morphological fact.
Some authors have analysed SE as a verbal head v of a different type: one marking a special subject or meaning (Folli
and Harley, 2005; Basilico, 2010; Armstrong, 2011). In Folli and Harley (2005), for instance, SE can appear in Italian with
activity verbs of consumption such as mangiare ‘eat’, in which case the subject can be an unusual (inanimate) unselected
subject, as in English eat versus eat away in The sea ate away at the beach. Although there are several important
differences between the analyses in the structure, the aspectual properties and the nature of the clitic, my proposal shares
with these works the idea that in the structure with SE the root does not modify the dynamic event but forms a verb by
combining with a different type of verbal head. Thus, the subject is not interpreted as the agent of a manner verb (vDO+
adjunct root) but as a different type of argument---causer in Folli & Harley, effector in Armstrong. In a parallel fashion, the
derived subject of SE-unaccusatives is not interpreted as the undergoer of a manner verb of change (vGO+ adjunct root)
since the root does not specify the change but combines with stative vBE to form the (result) state. However, the structures
discussed by these authors, also attested in Spanish, are of a different nature; crucially they do not involve a resulting
state and therefore, do not express a structurally telic change of state.
Finally, the analysis of the SE-variant crucially involves two verbal layers, which sets this analysis apart from other syntactic
approaches to the argument structure of inchoatives which propose a single verbal head for anticausatives/unaccusatives,
usually a specialized head vBECOME or RESULT (Alexiadou and Anagnostopoulou, 2004; Harley, 2008; Marantz, 1997, i.a.).
These analyses cannot easily cover the existence of the simple unaccusative structure assigned here to both alternating and
non-alternating SE-less unaccusatives such as caer ‘fall’ and llegar ‘arrive’. In order to account for these configurations
another type or flavour of v head would have to be proposed. This is the case even in proposals that argue for more than one
structural kind of unaccusative, such as Alexiadou (2010), Alexiadou and Anagnostopoulou (2004). Besides Cuervo’s
(2003) inventory of three types of v heads assumed here, Ramchand’s (2008) syntax-based theory of argument structure
(which also argues for three types of verbal heads) is the approach that can most naturally express the alternation studied
here by finely articulating each variant. The details of her analysis of inchoatives such as melt (consisting of a unique verbal
head of change, process) however, are not directly applicable to the unaccusative alternation.
4. Additional evidence
A series of characteristics of the unaccusative variants further argue for a structure-based analysis of the alternation in
the lines of the proposal in Section 3. Data involve formation of idiomatic expressions and light verb constructions, and the
interaction of the unaccusative alternation with dative non-core arguments.
4.1. Idiomatic and light verb uses
Alternating unaccusatives participate in many idiomatic expressions; several are used as light verbs. Full idiomatic
expressions exist with either variant (44). Most idioms involving these verbs are formed on the basis of one variant, the
other variant lacking the idiomatic reading, (45).16
16
The idiomatic uses that are compatible with either variant---with no obvious difference in meaning---all involve morir ‘die’: morir(se) de hambre
‘be starving’/de risa ‘laugh out loud’/ de miedo ‘be very scared’. Morirse seems to be, in some varieties of Spanish, lexicalized in the pronominal
form, the SE-less variant signalling a more formal register or affectation (therefore, as for an anonymous reviewer and myself, the above informal
expressions are only acceptable with morirse). Importantly, however, in constructions that are only compatible with the SE-less variant, the
contrast is preserved and morir is used (see (49)).
64 M.C. Cuervo / Lingua 141 (2014) 48--70
(44) Idioms with alternating verbs

a. SE-less variant: caer (en la cuenta) ‘realize’, caer (de sorpresa) ‘drop by’, salir de juerga ‘go out partying’,
b. SE-variant: salirse de las casillas ‘get very angry’, irse de mambo ‘exaggerate’, venirse encima ‘approach
too close/stalk’, bajarse del caballo ‘stop being presumptuous’
(45) SE-lessvariant SE-variant

a. Cayeron muchos vecinos a’. Se cayeron muchos vecinos
‘Many neighbours fell’ ‘Many neighbours fell down’
‘Many neighbours dropped by’ *‘Many neighbours dropped by’
b. Joaquín salió de las casillas b’. Joaquín se salió de las casillas
‘Joaquín came out of the slots’ ‘Joaquín got free from the slots’
*‘Joaquín got very angry’ ‘Joaquín got very angry’
Interestingly, light verb constructions, in contrast, only select the SE-less variant.17
(46) a. caer: caer bien ‘be liked’; caer mal ‘be disliked’; caer enfermo ‘get sick’; caer muerto ‘drop dead’.
b. salir: salir bien ‘turn out fine’; salir mal ‘come out wrong’.
c. ir: ir bien ‘go well’; ir mal ‘go badly’
d. quedar: quedar bien ‘save face‘; quedar mal ‘loose face’; quedar muerto ‘get exhausted’
The verb caer is particularly interesting in its idiomatic and light verb possibilities. In terms of its translation into English, the
pronominal variant caerse seems closer to English fall down, while caer closer to fall. In both languages, the pure
unaccusative of change can be used as a light verb of change and combine with a state expressed by a separate predicate
(47a), and it can have metaphorical uses (47b). Caerse and fall down cannot substitute for caer and fall (48). These facts
highlight how two languages using different elements (clitics or particles) arrive at the same level of structural complexity,
and the same syntactic and semantic behaviour obtains.
(47) SE-less: Verb of change (vGO)

a. Emilio cayó enfermo/ dormido/ muerto
Emilio fell sick/ asleep/ dead
b. Veinte soldados cayeron durante la batalla
Twenty soldiers fell (died) in the battle
(48) SE-variant: Verb of change of state (vGO+vBE)

a. Emilio se cayó *enfermo/ dormido/ *muerto
Emilio fell down sick/ asleep/ dead
Se cayó dormido = he fell while asleep; no light verb meaning: cannot mean =(47a)
b. Veinte soldados se cayeron durante la batalla
Twenty soldiers fell down during the battle
No idiomatic meaning: cannot mean =(47b)
The general restriction on co-occurrence of an adjective (expressing a resulting state) and SE (or particle in English)
follows from the analysis: in the SE variant, the root lexicalizes and forms the result; no other (result) predicate can be
added without generating ungrammaticality (48a); the change of state meaning thus obtained is incompatible with the root
acting as a light verb in vGO. In the SE-less variant, in contrast, caer combines with the only v head available, dynamic vGO
(as its spell-out in its light verb use) while the adjective expresses the end state (47a).18
17
An exception to this generalization are inchoative volverse ‘turn/get’ and ponerse ‘put/get’ + adjective, as in volverse loco ‘get crazy’ and
ponerse nervioso ‘get nervous’. These light verbs are special, however, in that as light verbs they alternate between a SE-less transitive-causative
and the SE-inchoative variants.
18
A parallel case is found in the predicate doubling (found in some varieties) of bajar ‘go down’, salir ‘go out’ and subir ‘go up’, which is not
acceptable in the SE-variant.
(i) Recién bajó abajo./ Recién salió afuera./ Recién subió arriba.
‘He just went down (down)’ /He just went out (outside)’ /‘He just went up (upstairs)’
M.C. Cuervo / Lingua 141 (2014) 48--70 65
The alternants of morir ‘die’ exhibit a similar behaviour when combined with a participle that names at the same time the
way of dying and the end result.19 Here, again, the predicate can modify the root when it is merged as an adjunct of vGO,
but it cannot modify the null vGO or the root merged as complement of vBE in the inchoative SE configuration.
(49) a. La rata murió electrocutada/ ahogada SE-less variant

the rat died electrocuted/ drowned
b. *La rata se murió electrocutada /ahogada SE-variant
the rat SE died electrocuted/ drowned
The contrasts discussed above point towards a structural difference between the variants, here expressed, in part, in
terms of the different combination of the root with a verbal head (Result versus Manner, complement versus head
adjunction), with consequences for the compositional meaning of the vP and the compatibility with a predicative adjective.
4.2. Interaction with dative arguments
As noted in Section 2, dative arguments are compatible with both variants of Spanish alternating unaccusatives,
although they are subject to restrictions, as illustrated in (8), repeated below.

Integrating Pylkkänen’s (2008) analysis of double-object constructions as Low Applicatives and Cuervo’s (2003) analysis of
dative arguments in causative constructions as a third kind of applicative, Affected Applicative, interesting predictions arise
with respect to the availability and interpretation of dative arguments with the different types of unaccusative predicates.
Cuervo takes Pylkkänen’s structural and semantic definition of low applicatives as a relation between two individuals
merged as complement of the root, and proposes the analyses in (51) for Spanish double-objects.
(51) Spanish low applicatives (Cuervo, 2003)
As one of the diagnostics to distinguish low applicatives from verb-modifying high applicatives, Pylkkänen proposes there
is a transitivity restriction on the availability of low applicatives.
(52) Pylkkänen’s transitivity restriction on low applicatives

Only high applicative heads should be able to combine with unergatives. Since a low applicative head denotes
a relation between the direct and indirect object, it cannot appear in a structure that lacks a direct object.
In order to understand the restriction, we must have an explicit idea of how ‘direct object’ is to be interpreted: whether it is to
cover any internal argument, or the complement of a transitive verb, or the complement of any type of transitive or
intransitive verb. Additionally, it is important to look for a structural rationale for the restriction on unergatives. If the
relevant notion of ‘direct object’ is taken to be structural rather than surface---an internal argument licensed as a
19
As an anonymous reviewer notes, morir in (49) is not acting as a light verb but as a main verb, which is modified by the participle. The
difference with (47) is evidenced by the contrast between Cómo murió la rata?---Electrocutada. ‘How did the rat died?---Electrocuted.’ vs. Cómo
? ?
cayó Emilio?---# Enfermo. ‘How did Emilio fall?---#Ill.’ Note, additionally, that sentence (49b) would be grammatical under the odd reading that the
rat was in a state of being electrocuted when it died.
66 M.C. Cuervo / Lingua 141 (2014) 48--70
complement---then the restriction on (52) can be interpreted in terms of the following two statements on the structural
distribution of low applicatives.
(53) Structural restrictions on low applicatives

A. Low applicatives are only compatible with a complement of the verb. A low applicative head cannot
take as its complement an argument DP licensed as a specifier (subject).
B. Low applicatives are structurally compatible with the complement of a verb in both transitive and
unaccusative configurations.
The restriction in A amounts to the systematic unavailability of low applicative arguments with deep subjects in general
(external arguments), and also with surface objects which are licensed as inner subjects in specifier position, as has been
proposed for subjects of small clauses and objects of causatives. It is exactly the analysis of the object of causatives as an
inner subject that forms the basis of Cuervo’s (2003, 2010a) proposal of datives with causatives as affected applicatives, a
type of applicative necessarily distinct from low (and high). The systematic incompatibility of low applicatives with
specifiers as stated in A also derives Borer and Grodzinsky (1986) generalization on possessor datives as related to DPs
in properly governed positions (complements), never to subjects. What is of central importance here are the predictions
that arise from the two distributional statements in (53) in interaction with the syntactic structures of alternating
unaccusatives in (11), repeated below.
The complementary predictions for the availability and interpretation of dative arguments with SE-less and SE-variants are
as follows:
(55) Two complementary predictions:

a) A dative argument as low applicative is possible with the SE-less variant of alternating unaccusatives, since
the (nominative) argument DP is licensed as a complement (of the head vGO+Root).
b) A dative argument as low applicative is impossible with the SE-variant of alternating unaccusatives, since
the (nominative) argument DP is licensed as a specifier (inner subject) of the vP. If a dative argument is
possible in this configuration, then the applicative must be another kind of applicative (affected, as with
causatives, or high).
4.2.1. Datives with the SE-less variant

The structure of the SE-less variant allows for a low applicative dative. In this configuration, a dative DP should, in
principle, be possible and receive one of the three possession-related meanings that the double-object construction can
have: recipient, source or (static) possessor (Cuervo, 2003). The sentences in (56) show that this is the case. The low
applicative structure of (56a) appears in (57). The structure of these sentences is parallel to the structure of a double-
object construction in a transitive configuration: the dative DP is the specifier of a LowAppl that takes the ‘direct object’ as
its complement; in turn, the LowApplP merges as the complement of the combined head vGO+Root.
(56) SE-less variant

a. A Gabi le cayeron visitas de Londres Recipient
Gabi.DAT CL.DAT fell.PL guests of London
‘Gabi got guests from London’
b. A todos los árboles les salieron hongos Source
all the trees.DAT CL.DAT came.out.PL fungi
‘All the trees got fungi’
M.C. Cuervo / Lingua 141 (2014) 48--70 67
These data confirm that the prediction in (55a) holds. Let’s turn now to datives in the SE-variant.
4.2.2. Datives with the SE-variant

Dative arguments are indeed possible with the SE-variant of alternating verbs, as illustrated in (50b), and (58) below.
(58) SE-variant
a. A Carolina se le salieron dos clavos (de la pared)
Carolina.DAT SE CL.DAT came-out.PL two nails from the wall
‘Two nails came out (from the wall) on Carolina’
b. A Gabi se le cayó la biblioteca (encima de la comida)
Gabi.DAT SE CL.DAT fell.PL the bookcase on.top of the food
‘The bookcase fell down (on the food) on Gabi’
In contrast to (56), the dative arguments in (58) are not interpreted as related to another individual (the nominative DP) either
as recipients, possessors or sources; in other words, they are not interpreted as low applicatives. Their interpretation is that of
an individual affected by the change of state of the nominative DP: Affected Applicatives (Cuervo, 2003). In the bi-eventive
configuration, the dative DP (or the ApplP) cannot merge as complement of the root or v+Root because the root is the
complement, and the DPNom is a specifier, as in (59). The complement of the applicative is the whole stative vP; that is, the
dative gets or has the state: the applied dative is affected by the nails being out. This confirms the prediction stated in (55b).
This difference in structure and meaning of applicatives is responsible for the contrasts presented in (60), (= (49)). The
dative a la jefa in (60a) is interpreted (at a certain level of abstraction) as the recipient of the undergoer/theme, Joaquín.
This interpretation does not arise in the SE-variant irse ‘leave’. In (60b), the dative in the SE-variant is interpreted as affected
by the death of the fern, whether Joaquín was its possessor or not.

The semantic contrasts are also evidenced in the possibilities of interpretation of a location that is questioned by dónde
‘where’. In (61a), the location where the coffee fell can only be interpreted as part of Gabi, that is, the dative is interpreted
as a location/recipient. In the context of the SE-variant, in contrast, there is no such restriction, and the question for where
the coffee fell can be answered with any location.
? ?
(61) a. Dónde le cayó el café a Gabi? b. Dónde se le cayó el café a Gabi?
Where fell the coffee Gabi.DAT Where SE fell the coffee Gabi.DAT
On the arm/ #Onto Juan/ #In the kitchen On the arm/ Onto Juan/ In the kitchen
68 M.C. Cuervo / Lingua 141 (2014) 48--70
In Cuervo (2010b) it is shown that these same contrasts in interpretation hold of the inchoative variant of transitive
causatives (quemarse ‘burn’, abrirse ‘open’) versus datives with simple dynamic or existential unaccusative verbs such as
llegar ‘arrive’, faltar ‘lack’, escapar ‘escape’, etc. This further demonstrates that the reflexive and non-reflexive structures
of alternating verbs correspond exactly to the structures of alternating anticausatives and non-alternating simple
unaccusatives, respectively.
Finally, a dative argument in the reflexive variant of alternating unaccusatives can also be interpreted as the accidental
causer of the change of state. These datives with anticausatives are analysed by Cuervo (2003) as a high applicative that
merges above vGOP, that is, above the whole complex event (see also Fernández Soriano and Mendikoetxea, 2012;
Kalluli, 2006; Kim, 2011; Schäfer, 2008 for relevant work on other languages). The SE-less variant, just as non-alternating
simple unaccusatives, does not exhibit this ambiguity. This is shown in (62) by the (in)compatibility of ‘‘accidental
responsibility’’ modifiers such as por/de torpe ‘for being clumsy’; sin querer ‘without wanting’.
(62) a. A Tomás le cayeron hojas *por torpe / * sin querer

Tomás.DAT CL.DAT fell.PL leaves for clumsy / without want
‘Leaves fell on Tomás (*because he is clumsy/*without him wanting to)’
b. A Tomás se le cayeron los libros por torpe/ sin querer
Tomás.DAT SE CL.DAT fell.PL the books for clumsy / without want
‘Tomás dropped the books (because he is clumsy/ without him wanting to).’
The availability of accidental causers arises only with true change of state verbs---what are usually termed ‘‘externally
caused’’ verbs---as opposed to SE-less degree achievements or ‘‘internally caused change of state verbs’’, such as
florecer ‘blossom’ and engordar ‘get fat(ter)’, and simple non-alternating unaccusatives such as llegar ‘arrive’. Fernández
Soriano and Mendikoetxea (2012) derive the lack of the accidental causer reading from the lack of a v CAUSE component in
the structure of internally caused change of state verbs, the head that takes a pro subject and allows, in SE-anticausatives,
for a high applicative to be interpreted as the (non-volitional) initiator of the causing event. In the current analysis there is
no particular CAUSE component in the syntax of either type of unaccusative constructions. The accidental causer reading
derives from structural properties of the SE-configuration: there is an unspecified dynamic event vGO that embeds a state
vP. The cause interpretation arises from this type of bi-eventive configuration, which is absent in the simple unaccusatives
(see Cuervo, in press; Marantz, 2005 for discussion).20 Although the proper analysis of these facts falls beyond the scope
of this paper, what is crucial is that in the SE-construction there is no specification of the dynamic event that brings about
the new state of the object (the result). In fact, the general notion of ‘causer’ that emerges is simply the interpretation of an
external argument of non lexicalized dynamic events which embed a state; that is, the external argument of a v head to
which no manner root attaches, be it a potentially agentive head vDO (causatives) or a non-agentive vGO (inchoatives).21
This idea is also expressed in Schäfer (2012b), and constitutes a syntactic reformulation of the lexical characterization of
alternating causatives as those verbs that do not specify the causing event (e.g., Levin and Rappaport Hovav, 1995).
5. Conclusions
This paper discussed several formal and semantic properties of the two unaccusative variants of a relatively small
group of verbs in Spanish. In spite of both variants being unaccusative and apparently not varying in the number or type of
argument, the alternation has been shown to indeed be an argument structure alternation on a par with other well studied
ones. The arguments presented show that the reflexive and non-reflexive variants correspond to distinct underlying
20
It is important to recall that the dative argument in the SE-variant can always have the affected applicative reading which arises if the underlying
structure is that in (59). In turn, simple SE-less unaccusatives are, in principle, structurally compatible not only with low applicatives but also with a
high applicative merging above vGOP. In this case, the dative argument would not be interpreted (necessarily) as a possessor (recipient or source)
of the nominative DP but as a non-agentive, non-volitional argument indirectly involved in (getting or having) the event of change or motion of the
theme/undergoer specified by the root which adjoins to vGO (see Kim, 2012 for relevant discussion). This is the most salient reading in sentences
such as (i) (see Schäfer, 2008, contra Fernández Soriano and Mendikoetxea, 2012, who deny this type of reading and seem to assume that only
low applicatives are possible in this case, although with an ‘‘affectedness’’ reading).
(i) a. A Tomás no le crecen las plantas.

Tomás.DAT not CL.DAT grow.PL the plants
‘Tomás cannot get plants to grow’ (i.e., Tomás does not have a green thumb)
b. Perdón. Me hirvió el agua.
sorry. me.DAT boiled.SG the water.
‘Sorry. I accidentally let the water boil’.
21
Under this notion of causer, the agents of resultative constructions, as [John] in John hammered the metal flat, are not considered causers.
M.C. Cuervo / Lingua 141 (2014) 48--70 69
syntactic structures from which their main morphosyntactic and semantic properties derive. The structures of the SE-less
and SE-variants correspond to---in fact are---the structure of simple non-alternating unaccusatives and SE-inchoatives/
anticausatives, respectively. Thus, what initially appears to be a semi-productive, idiosyncratic alternation, attested only
with a small group of verbs, turns out to be, when examined with a focus on the syntactic structures rather than on types of
lexical verbs, just another alternation between two independently very productive constructions. The compatibility of the
set of roots with the two structures is based, in part, on their possibility of expressing both a type of change, process or
motion (Manner), or a state (Result). The structural decomposition of SE-unaccusatives into a change and a state allows us
to capture the dual role of the argument DP and the dyadic nature of the predicate (a complex verbal phrase) without
reference to an external argument or a transitive counterpart.
The phenomena discussed here constitute a strong challenge to two influential types of analyses of unaccusatives. On
the one hand, they challenge proposals that unify the structure of unaccusative verbs that participate (anticausative/
inchoatives) and do not participate (simple unaccusatives) in the causative alternation. On the other hand, the systematic
alternation itself---independently of whether the analysis developed here is the correct one---goes against the view that the
clitic SE signals that a valency reduction operation has taken place. In particular, the data discussed here challenge the
idea that anticausatives are derived from transitive causatives via one such operation (be it decausativization or
reflexivization); instead, the data strongly suggest that the ‘‘common base’’ approach to alternations is on the right track,
as proposed by Alexiadou et al. (2006), Cuervo (2003), Piñón (2001), among others.
The idea that verbs are formed in the syntax directly, and that the position of insertion of a root is particularly relevant for
the interpretation of the verbal projection and the licensing of arguments receives further support from this work. A
particular contribution has been to show that, just as there are roots that can function as agentive manners in the
construction of activity verbs, there are roots that can function as unaccusative manners in the construction of verbs of
change or motion.
Acknowledgments
I thank participants in the Workshop on verb meaning, event semantics and argument structure (Universitat Autonòma
de Barcelona, December 2010) and the Workshop of Approaches to the Lexicon (Roots III) (The Hebrew University of
Jerusalem, June 2011), and two anonymous reviewers for their insightful and useful comments on earlier versions of this
work.
References
Acedo-Matellán, V., 2013. Merging roots in bare phrase structure and the conflation/incorporation distinction. In: Keine, S., Sloggett, S. (Eds.),
Proceedings of the 42nd Annual Meeting of the North East Linguistic Society, Volume 1. GLSA Publications, Amherst, pp. 1--14.
Adger, D., 2003. Core Syntax. A Minimalist Approach. Oxford University Press, Oxford.
Alexiadou, A., 2010. On the morphosyntax of (anti)causative verbs. In: Rappaport Hovav, M., Doron, E., Sichel, I. (Eds.), Lexical Semantics,
Syntax and Event Structure. Oxford University Press, Oxford, pp. 177--203.
Alexiadou, A., Anagnostopoulou, A., 2004. Voice morphology in the causative--inchoative alternation: evidence for a non-unified structural
analysis of unaccusatives. In: Alexiadou, A., Anagnostopoulou, E., Everaert, M. (Eds.), The Unaccusativity Puzzle. Oxford University Press,
Oxford, pp. 114--136.
Alexiadou, A., Anagnostopoulou, E., Schäfer, F., 2006. The properties of anticausatives cross-linguistically. In: Frascarelli, M. (Ed.), Phases of
Interpretation. Mouton de Gruyter, Berlin, pp. 187--211.
Armstrong, G., 2011. Two Classes of Transitive Verbs: Evidence from Spanish. Doctoral dissertation, Georgetown University.
Basilico, D., 2010. The se clitic and its relationship to paths. Probus 22, 271--302.
Borer, H., Grodzinsky, Y., 1986. Syntactic cliticization and lexical cliticization: The case of Hebrew dative clitics. In: Borer, H. (Ed.), Syntax and
semantics 19: The syntax of pronominal clitics. Academic Press, Orlando, pp. 175--217.
Campos, H., 1999. Transitividad e intransitividad. In: Bosque, I., Demonte, V. (Eds.), Gramática descriptiva de la lengua española, vol. 2. Espasa-
Calpe, Madrid, pp. 1519--1574.
Chierchia, G., 2004. A semantics for unaccusatives and its syntactic consequences. In: Alexiadou, A., Anagnostopoulou, E., Everaert, M. (Eds.),
The Unaccusativity Puzzle. Oxford University Press, Oxford, pp. 22--59.
Cuervo, M.C., 2003. Datives at Large. Doctoral dissertation, MIT.
Cuervo, M.C., 2010a. Against ditransitivity. Probus 22, 151--180.
Cuervo, M.C., 2010b. Some dative subjects are born, some are made. In: Borgonovo, C., Español-Echevarría, M., Prévost, P. (Eds.), Selected
Proceedings of the Hispanic Linguistic Symposium 2008. Cascadilla Press, Somerville, MA, pp. 26--37.
Cuervo, M.C., in press. Causation without a CAUSE. Syntax.
de Miguel, E., Fernández-Lagunilla, M., 2000. El operador aspectual ‘se’. Revista española de lingüística 30, 13--44.
Embick, D., 1998. Voice System and the Syntax/Morphology Interface. MIT Working Papers in Linguistics, pp. 41--72.
Embick, D., 2004. On the structure of resultatives participles in English. Linguistic Inquiry 35, 355--392.
Everaert, M., Dimitriadis, A., 2013. On Reflexive Resultatives, Handout downloadable at http://www.hum.uu.nl/medewerkers/o.m.matushansky/
SPIFF/me&ad.pdf.
70 M.C. Cuervo / Lingua 141 (2014) 48--70
Fernández Soriano, O., 1999. Two types of impersonal sentences in Spanish: locative and dative subjects. Syntax 2 (2), 101--140.
Fernández Soriano, O., Mendikoetxea, A., 2012. Non selected datives in anticausative constructions. Archivio Glottologico Ita XCVI, 87--128.
Folli, R., Harley, H., 2005. Flavours of v: consuming results in Italian & English. In: Kempchinsky, P., Slabakova, R. (Eds.), Aspectual Enquiries.
Springer, Dordrecht, pp. 95--120.
Hale, K., Keyser, J., 1993. On argument structure and the lexical representation of syntactic relations. In: Hale, K., Keyser, J. (Eds.), The View from
Building 20. MIT Press, Cambridge, MA, pp. 53--109.
Hale, K., Keyser, J., 2002. Prolegomenon to a Theory of Argument Structure. MIT Press, Cambridge, MA.
Harley, H., 2005. How do verbs get their names? Denominal verbs, manner incorporation and the ontology of verb roots in English. In: Erteschik-
Shir, N., Rapoport, T. (Eds.), The Syntax of Aspect. Deriving Thematic and Aspectual Interpretation. Oxford University Press, Oxford, pp. 42--
64.
Harley, H., 2008. On the causative construction. In: Miyagawa, S., Saito, M. (Eds.), Handbook of Japanese Linguistics. Oxford University Press,
Oxford, pp. 20--53.
Harley, H., in press. On the identity of roots. Theoretical Linguistics.
Haugen, J.D., 2009. Hyponymous objects and Late Insertion. Lingua 119, 242--262.
Horvath, J., Siloni, T., 2013. Anticausatives have no cause(r): a rejoinder to Beavers and Koontz-Garboden. Lingua 131, 217--230.
Kalluli, D., 2006. Unaccusatives with dative causers and experiencers: a unified account. In: Hole, D., Meinunger, A., Abraham, W. (Eds.), Dative
and Other Cases. Benjamins, Amsterdam, pp. 271--301.
Kempchinsky, P., 2004. Romance SE as an aspectual element. In: Auger, J., et al. (Eds.), Contemporary Approaches to Romance Linguistics.
Benjamins, Amsterdam, pp. 239--256.
Kim, K., 2011. High applicatives in Korean causatives and passives. Lingua 121, 487--510.
Kim, K., 2012. External argument-introducing heads: voice and appl. In: Cuervo, M.C., Roberge, Y. (Eds.), The End of Argument Structure?
Emerald, Bingley, UK, pp. 131--154.
Koontz-Garboden, A., 2009. Anticausativization. Natural Language & Linguistic Theory 27, 77--138.
Labelle, M., Doron, E., 2010. Anticausative derivations (and other valency alternations) in French. Probus 22, 303--316.
Levin, B., 1999. Objecthood: an event structure perspective. In: Proceedings of CLS 35, volume 1: The Main Session. Chicago Linguistic Society,
University of Chicago, Chicago, IL, pp. 223--247.
Levin, B., Rappaport Hovav, M., 1995. Unaccusativity. MIT Press, Cambridge, MA.
Marantz, A., 1997. No escape from syntax: don’t try morphological analysis in the privacy of your own lexicon. In: U. Penn Working Papers in
Linguistics, vol. 4.2. pp. 201--225.
Marantz, A., 2005. Objects Out of the Lexicon! Argument Structure in the Syntax. Ms., MIT.
Masullo, P., 1992. Incorporation and Case Theory in Spanish. A Crosslinguistic Perspective. Doctoral dissertation, University of Washington,
Seattle.
Mateu, J., 2002. Argument Structure. Relational Construal at the Syntax--Semantics Interface. Doctoral dissertation, UAB, Bellaterra, retrieved
from http://hdl.handle.net/10803/4828.
Mateu, J., Acedo-Matellán, V., 2012. The manner/result complementarity revisited: a syntactic approach. In: Cuervo, M.C., Roberge, Y. (Eds.),
The End of Argument Structure? Emerald, Bingley, UK, pp. 209--228.
Mendikoetxea, A., 1999. Construcciones con se: medias, pasivas e impersonales. In: Bosque, I., Demonte, V. (Eds.), Gramática Descriptiva de la
Lengua Española. Espasa-Calpe, Madrid, pp. 1575--1629.
Mendikoetxea, A., 2012. Passives and se constructions. In: Hualde, J.I., Olarrea, A., O-Rouke, E. (Eds.), Handbook of Hispanic Linguistics. Wiley-
Blackwell, Oxford, pp. 485--510.
Moro, A., 1997. Raising of Predicates: Predicative Noun Phrases and the Theory of Clause Structure. Cambridge University Press, Cambridge.
Nishida, C., 1994. The Spanish reflexive clitic se as an aspectual class marker. Linguistics 32 (3), 425--458.
Perlmutter, D., 1978. Impersonal passives and the unaccusative hypothesis. In: Proceedings of the Fourth Annual Meeting of the Berkeley
Linguistic Society. University of California, Berkeley, pp. 157--189.
Piñón, C., 2001. A finer look at the causative--inchoative alternation. In: SALT XI. Cornell Linguistics Circle, Ithaca, NY, pp. 346--364.
Pylkkänen, L., 2008. Introducing Arguments. MIT Press, Cambridge, MA.
Ramchand, G.C., 2008. Verb Meaning and the Lexicon. Cambridge University Press, Cambridge.
Rappaport Hovav, M., Levin, B., 2010. Reflections on Manner/Result Complementarity. In: Rappaport Hovav, M., Doron, E., Sichel, I. (Eds.),
Syntax, Lexical Semantics and Event Structure. Oxford University Press, New York, NY, pp. 21--38.
Reinhart, T., 2002. The theta system: an overview. Theoretical Linguistics 28, 229--290.
Reinhart, T., Siloni, T., 2005. The lexicon--syntax parameter: reflexivization and other arity operations. Linguistic Inquiry 36, 389--436.
Rothstein, S., 2001. Predicates and their Subjects. Kluwer Academic Publishers, Dordrecht.
Schäfer, F., 2008. The Syntax of (Anti-)causatives. External Arguments in Change-of-State Contexts. Benjamins, Amsterdam.
Schäfer, F., 2012a. The passive of reflexive verbs and its implications for theories of binding and case. Journal of Comparative Germanic
Linguistics 15.3, 213--268.
Schäfer, F., 2012b. Two types of external argument licensing -- the case of causers. Studia Linguistica 66, 128--180.
Suñer, M., 1982. The Syntax and Semantics of Presentational Sentence-Types in Spanish. Georgetown University Press, Washington, DC.
von Stechow, A., 1995. Lexical decomposition in syntax. In: Egli, U., Pause, P.E., Schwarze, C., von Stechow, A., Wienold, G. (Eds.), Lexical
Knowledge in the Organization of Language. Benjamins, Amsterdam, pp. 81--117.
Zagona, K., 1996. Compositionality of aspect: evidence from Spanish aspectual se. In: Parodi, C., Quicoli, C., Saltarelli, M., Zubizarreta, M.L.
(Eds.), Aspects of Romance Linguistics. Georgetown University Press, Washington, DC, pp. 475--488.
Zubizarreta, M.L., 1987. Levels of Representation in the Lexicon and in the Syntax. Foris, Dordrecht, The Netherlands.
Zubizarreta, M.L., Oh, E., 2007. On the syntactic composition of manner and motion. MIT Press, Cambridge, MA.

Cuervo 2014

Uploaded by

Document Informationclick to expand document information

Copyright:

Available Formats

Cuervo 2014

Uploaded by

Document Information

Original Title

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Cuervo 2014

Uploaded by

Copyright:

Available Formats

Available online at www.sciencedirect.

Alternating unaccusatives and the distribution of roots

Keywords: Unaccusative; Anticausative; Argument structure alternation; Reflexive; Roots; Spanish se

* Tel.: ++1 416 813 4051.

(1) a. Cayeron tres hojas V-DP

(3) a. Llegados los invitados, empezó la fiesta.

2. Properties of the variants

(4) a. Llegaron invitados.

(5) *Se derritió manteca. (ok under an impersonal reading)

(6) a. Cayeron hojas SE-less

(8) a. Joaquín (*se) le fue con problemas a la jefa

(12) a. Este añ o (*se) cayó la calidad/ el precio de la ropa.

(13) a. Recién *(se) cayeron tres vasos/ sillas/ galletitas.

 the licensing position of the argument (complement vs. specifier)

3.1. Evidence for licensing position of argument: restrictions on bare NPs

(14) The Bare Noun Phrase Constraint

(15) Simple unaccusatives (=(11a))

(16) Inchoatives (=(11b))

(17) Simple unaccusatives: 1 event

(18) Inchoatives: 2 events

(19) Simple unaccusatives: 1 event

(20) Inchoatives: 2 events

3.3. Evidence for argument-related SE

(22) ‘Regular’ inchoatives

(25) a. Lo dejaron afeitar(se) Reflexive

(26) a. *Lo dejaron quejar

(27) a. *(Se) cayeron tres vasos/ sillas/ galletitas

(28) a. Al tronco, lo dejaron quemar (durante la noche) Change10

(29) a. Al payaso, lo dejaron bajar (al sótano) Motion

(30) a. Lo dejaron arrodillar(se) ante la estatua.

(31) a. Agustín se afeitó Reflexive

(32) a. Agustín se cayó SE-unaccusative

3.4. Distribution of roots

(35) a. Tomás trabajó/ jugó/ bailó mucho Activities

(36) a. La paracaidista principiante cayó bien. SE-less unacc

(37) a. ??La paracaidista principiante se cayó bien. SE-unacc

(38) a. Tomás manejó rápido la camioneta Transitive

(39) a. El sol derritió la manteca rápido Causative

(40) Manner: atelic or telic

(41) SE-less atelic

(42) SE-variant telic

3.7. Previous approaches

# of participants/ Type of Subject licensed as Interpretation of Interpretation of Telicity

SE-less: Simple 1/1 Change Complement of Undergoer Adjunct/Manner Telic or

4.1. Idiomatic and light verb uses

(44) Idioms with alternating verbs

(45) SE-lessvariant SE-variant

(47) SE-less: Verb of change (vGO)

(48) SE-variant: Verb of change of state (vGO+vBE)

(49) a. La rata murió electrocutada/ ahogada SE-less variant

4.2. Interaction with dative arguments

(50) a. Joaquín (*se) le fue con problemas a la jefa

(51) Spanish low applicatives (Cuervo, 2003)

(52) Pylkkänen’s transitivity restriction on low applicatives

(53) Structural restrictions on low applicatives

(55) Two complementary predictions:

4.2.1. Datives with the SE-less variant

(56) SE-less variant

the licensing position of the argument (complement vs. specifier)

(62) a. A Tomás le cayeron hojas por torpe / sin querer