Probus. International Journal of Latin and Romance Linguistics 25 (2013), 301-344.
DOI: 10.1515/probus-2013-0012, [The final publication is available at www.degruyter.com]
Object Clitics, Agreement and Dialectal Variation*
Javier Ormazabal1 2 & Juan Romero2 3
Abstract
This article presents an analysis of object clitics in Spanish and some of its
consequences for the theory of agreement and Case. On the empirical side, we present
syntactic, morphological and semantic arguments supporting a mixed approach to object
clitics where 3rd person Direct Object (DO) cliticization constitutes a genuine case of
Determiner movement, but other DO and Indirect Object (IO) clitics are agreement
elements. Once third person object clitics are set aside, the emerging picture is a single
agreement that does not discriminate between DOs and IOs in the syntax. This idea
finds striking support in Basque Leísta Dialect, where there is a 3rd person DO
agreement clitic that behaves in all relevant respects like all other agreement clitics.
Moreover, the consequences of this analysis extend to other properties of the object
relation in Spanish, such as Differential Object Marking (DOM), and dialectal variation
in the clitic field. An interesting observation that arises from this study is that the
agreement nature of 1st and 2nd person clitics and the whole series of IOs is extremely
robust in Spanish and remains invariable across all the dialects analyzed. Variation is
thus restricted to 3rd person DO objects, where in contrast the changes are diverse and
take very different directions, a fact that raises interesting questions related both to the
historical evolution of the clitic system and to the theoretical analysis of Case and
agreement.
KEYWORDS: clitics, agreement, animacy, Case, Differential Object Marking (DOM),
dialectal variation.
1. Introduction
In the syntactic analysis of Romance clitics two have been the main hypotheses within
the generative tradition since Kayne’s (1975) original discussion of the topic. The
movement hypothesis (Kayne 1975, 1991, Rizzi 1986, among others) analyzes clitics as
pronominal elements generated in their argumental position within VP and moved to the
verbal or inflectional projection. The base-generated hypothesis (among others, Strozer
1976, Aoun 1981, Jaeggli 1982) in its modern version argues that Romance clitics are
agreement markers attached to inflectional heads (Borer 1984, Saltarelli 1987, Suñer
1988, Fernández Soriano 1989, Franco 1993, Sportiche 1993, Fontana 1993, Landa
1995, Barbosa 2000, Anderson 2005, among others).
1
University of the Basque Country (UPV/EHU)
Basque Group of Theoretical Linguistics (HiTT)
3
University of Extremadura (UNEX)
2
Both hypotheses share the underlying assumption that object clitics form a
uniform class with respect to the issues at stake. More recently, an alternative, mixed,
analysis has arisen that postulates a different division of work among clitics and
suggests that Romance clitics are not as uniform a class as we would expect even within
the same language (Uriagereka 1988, 1995; Roca 1992, 1996; Torrego 1998, Bleam
2000, Ormazabal and Romero 2002, Schroten 2010, among others). In particular, the
idea is that 3rd person Direct Object (DO) cliticization in Standard Spanish constitutes a
genuine case of Determiner movement, but other DO clitics and Indirect Object (IO)
clitics are of a very different nature, most probably agreement-like elements.
The immediate descriptive goal of this paper is to present syntactic,
morphological and semantic arguments supporting a specific version of this view,
where all the agreement-type clitics constitute a single system, which does not
discriminate between DOs and IOs in the syntax. Once this system is well established,
we explore in more detail the properties of object clitics from that mixed perspective,
and analyze its effects on microdialectal variation. An interesting observation that arises
from this study is that the agreement nature of 1st and 2nd person clitics and the whole
series of IOs is extremely robust in Spanish and remains invariable across all the
dialects analyzed. Variation is thus restricted to 3rd person DO objects, where in contrast
the changes are diverse and take very different directions, a fact that raises interesting
questions related both to the historical evolution of the clitic system and to the
theoretical analysis of Case and agreement.
The paper is organized as follows. In section 2, heavily based on work by Roca (1996),
we argue that syntactic, morphological and semantic differences between 3rd person DO
clitics, lo(s)/la(s), and IO clitic le(s) support the “mixed” analysis. We also extend the
analysis to the entire paradigm, including 1st and 2nd person object clitics; we conclude
that, except for 3rd person DO clitics, all the other clitics in the DO and IO series are
agreement markers.4 In section 3 we analyze the properties of a Spanish dialect spoken
in the Basque Country, which differs from Standard Spanish in one crucial respect: it
4
We leave aside the clitic se in all its uses. However, it has to be noted that there are
two cases where se clearly satisfies an object role. First, when it works as a DO/IO
reflexive. In this function, it behaves as a base generated clitic. The other case involves
spurious se, for which there is no single reason that we know of to suppose that its use is
syntactically relevant.
2
splits 3rd person DO clitics into two groups, the inanimate series lo(s)/la(s) and the
animate one le(s). We show that the animate clitics in this dialect behave in all relevant
respects like all other agreement clitics. This dialect fills up the gap observed in the
paradigm in Standard Spanish corresponding to 3rd person DO agreement, paving the
path for a unification of the DO and IO agreement paradigm. Section 4 analyzes the
distribution of the Differential Object Marker (DOM) a in Spanish. We present a novel
paradigm concerning the interaction of this mechanism and se-constructions that
supports the hypothesis that there is a single Case/agreement system for the internal
arguments of the verb. Our analysis extends naturally to account for unexpected cases of
A-marking
into inanimate arguments previously observed in the literature but never
before accounted for, and cases of clitic doubling in the same inanimate contexts in the
Basque leísta dialect. As we argue, they all involve Raising-to-Object, which makes our
analysis in terms of Case and Agreement better suited to account for these constructions
than morphological analyses based on the animacy specifications of DOM and object
clitic le(s). In section 5 we incorporate the properties of two dialects of Spanish that
followed different paths in the diachronic evolution and, as a result, show mixed
properties in very different ways. On the one hand, we analyze a different leísta dialect
of Spanish, which we call Central Peninsular leísta dialect; we extensively show that
this dialect has developed a determiner-type clitic that is specified for animacy.
Secondly, we discuss the main properties of the clitic system in Rioplatense Spanish, as
described by Zdrojewski (2008). We conclude—minimally modifying his proposal—
that third person DO clitics in this dialect are ambiguous between a determiner-like and
an agreement-like clitic. Section 6 sketches the Case/Agreement system that comes out
from our proposal, and briefly links the result to our previous work on agreement.
2. The Spanish Clitic System Revisited
The goal of this section is to present evidence supporting the idea that object clitics in
Spanish involve two different systems. In particular, we show that 3rd person DO clitics
lo(s)/la(s) contrast with IO clitics, as well as with the rest of DO ones in a number of
morphosyntactic and semantic properties: the availability of clitic doubling, the
encoding of morphological features, their interpretive restrictions, and their behavior in
clitic clusters. Reformulating various proposals in the literature, we argue that these
3
differences support the view according to which 3rd person DO clitics are incorporated
determiners, while the rest of the clitics are better understood as agreement markers.
2.1. Clitic Doubling : differences between 3rd person Direct and Indirect Objects
One of the main properties that have centered the discussion on the nature of clitics is
clitic doubling. Defenders of the movement approach have highlighted the general
absence of clitic doubling in languages like Italian or French. In contrast, those who
argue for the base-generation view tend to focus on its existence in many contexts in
other Romance languages. Both views seem to be partially right concerning a subset of
the data. The case of Standard Spanish is particularly interesting in that respect, given
that it shows a different behavior depending on the type of clitic involved and the
syntactic context where it appears. Consider the example in (1).5
(1)
Les
recomendé
un libro a los estudiantes
3pIO recommended.I
a book A the students
‘I recommended the students a book’
The presence of the full DP los estudiantes (‘the students’) in the argument position
makes that position in principle unavailable as the source of the 3rd person IO clitic, a
fact that seems to suggest that the clitic is base-generated somewhere in the inflectional
head to which it appears attached, triggering an agreement relation with the NP in the
argument position. Yet, if that is correct, the agreement analysis does not extend to 3rd
person DO clitics in Standard Spanish, where the object clitic and the DP do not
coappear in general:
(2)
(*La) vimos
3fsDO saw.we
la casa
the house
5
The following abbreviations are used in the glosses: 1, 2, 3 = first, second, third person
respectively; s = singular; p = plural, f = feminine, m = masculine, a = animate, df = default, IO
= Indirect Object, DO = Direct Object, RFL = reflexive, SUBJ=subjunctive, arb = arbitrary.
Subject agreement in the verb is glossed by the corresponding English pronoun preceded with a
dot, except in the case of the examples with the impersonal construction where a more detailed
information is important for our argument. We also use BLD for Basque leísta dialect and
CPLD for Central Peninsular leísta dialect.
The animacy/Case marker a that appears with animate specific direct objects in Spanish
has been left untranslated and is represented by the small capital A. For reasons that will become
clear in the discussion in section 4, the same notation has been extended to clitic doubled IOs, as
in (1) in the text. Whenever possible, we also translated these constructions with the
corresponding DOC in English. For the same reasons, our analysis does not distinguish between
DO and IO agreement clitics, as will become clear from our argumentation,. However, we keep
the distinction in the glosses to facilitate the comprehension to the reader.
4
‘We saw the house’
The same contrast extends to a large variety of syntactic and semantic configurations, as
widely observed in the literature (see especially Roca 1996 for discussion).
It is not our intention to discuss the nature of Spanish clitic lo(s)/la(s) in detail
here. Various authors have argued that they belong to the category Determiner (see
specially Uriagereka 1988, 1995; Roca 1992, 1996; Torrego 1998, Bleam 2000 and
references therein). The same authors extensively argue that the few contexts where DO
clitic doubling is obligatory (Clitic Left Dislocation [3a], and strong pronouns [3b]) or
possible (universal quantifier todo [3c]), do not constitute counterexamples to this
analysis:
(3)
a.
b.
c.
La casa todavía no *(la)
he
visto
The house still
not 3fsDO have.I seen
‘The house, I haven’t seen it yet’
*(La) vimos a ella
3fsDO saw.we A her
‘We saw HER’
Ayer
(los)
vimos
todos
yesterday
3mpDO saw.we
all
‘We saw them all yesterday’
These authors, as well as many others (see especially Rigau 1988; Suñer 1988; Luján
1984; Fernández Soriano 1993) present different analyses to clitic doubling in (3a-b)
wholly compatible with the determiner hypothesis.6
Furthermore, as we show next, the distributional properties of the quantifier
todo(s) (‘every’) strongly support the determiner nature of 3rd person DO clitics,
minimally contrasting with their IO counterparts. First note that as observed by Roca
(1992, 1996), the universal quantifier in Spanish is the only one that selects a
complement headed by the definite determiner, i.e. that heads a construction of the type
in (4), a contrast illustrated in (5):
(4)
[ Quantifier [Det NP]]
6
Concerning (3a), there is general agreement in that the DP la casa (‘the house’)
is in a left-dislocated position, and that the clitic is the argumental element. With respect
to (3b), Rigau (1988) and Roca (1992, 1996) argue that strong pronouns are not in
argument position, in which case the real argument would be, once again, the clitic. On
the other hand, Uriagereka (1995, 1997) and Bleam (2000) propose a more complex
structure of DP including an internal small clause, associated to the integral relation,
where the pronoun occupies the specifier position of IntP. See references for details and
discussion.
5
(5)
a.
Vimos todos *(los) libros
Saw.we all
the books
‘We saw all the books’
b.
Vimos algunos/bastantes/suficientes/muchos
saw.we some /quite
/enough /a lot
‘We saw some/quite a lot/enough/a lot of books’
(*los) libros
the books
As he notes, this difference correlates with the fact that the universal quantifier is the
only one that doubles.7
(6)
a.
b.
Los
vimos
3mpDO saw.we
‘We saw them all’
todos
all
* Los
vimos algunos/bastantes/suficientes/muchos
3mpDO saw.we some /quite
/enough /a lot
‘We saw them some/quite a lot/enough/a lot’
In this context, there is an interesting contrast between DOs and IOs when the universal
quantifier appears restricted by a strong pronoun. Consider the following minimal pair:
(7)
a.
(Los)
he comprado todos (los= los libros)
3mpDO have.I bought all (3mpDO= the books)
I have bought them all
b.
(*Los) he comprado todos ellos
3mpDO have.I bought all them
I have bought them all
As shown in (7b), the presence of the strong pronoun is not compatible with the
presence of the accusative clitic (see Bosque 2010, p. 320).8 If we turn now to the dative
paradigm we find that the non-doubled version is ungrammatical, but the doubled
version is now perfectly grammatical:
(8)
*(Les) he
pegado pegatinas a todos
3pIO have-I stuck stickers A all-mp
‘I stuck stickers in all of them’
ellos
(les=
a los libros)
them-mp (3mpIO= A the books)
7
As observed by an anonymous reviewer, clitic doubling is not obligatory in this
context. As Olga Fernández Soriano observes to us, the quantifier may appear nondoubled when it refers to an unspecified set of objects:
(i)
Tengo el primero, el segundo,
el tercero,... tengo todos
Have.I the first, the second,
the third,...
Have.I all
‘I have the first one, the second one, the third one,... I have (them) all’
As far as we know, the properties of todos ellos (‘them all’) have not been discussed in
the literature; they are rather peculiar. In particular, this is the only case we are aware of where a
strong pronoun in object position may refer to an inanimate argument. We leave this issue for
further research.
8
6
The contrast between (7b) and (8) shows that 3rd person DO and IO clitics are different
types of morphosyntactic elements, and suggests a determiner-like distribution of
lo(s)/la(s) and an agreement –like analysis of le(s), a proposal supported by other
differences between the two clitic systems, as we argue next.
2.2. The Morphology of 3rd person DO clitics and determiners: gender marking
An important respect in which Spanish DO clitics pair together with determiners is their
morphological feature specification. Both DO clitics lo(s)/la(s) and determiners are
marked for number and gender.
(9)
a.
La
vimos (e.g. the girl)
3fsDO saw.we
‘We saw her’
b.
Lo
vimos (e.g. the boy)
3msDO saw.we
‘We saw him’
In contrast neither IO clitics nor the subject agreement paradigm specify gender as part
of their formal feature cluster. This parallelism is even stronger when we consider the
definite neuter determiner lo. Since in Spanish there are no neuter nouns, this
determiner is only used in combination with an adjectival or, in some restricted
contexts, with a clausal complement:
(10)
Lo
más hermoso de la primavera eres tú
The-neuter
most beautiful of the spring are you
‘The most beautiful thing in the spring is you’
The same form appears in both determiner (10) and clitic position (11).9
(11)
a.
No creo
que vaya
a venir
Not believe-I that is-SUBJUNCTIVE to come
‘I do not believe that he/she is coming’
b.
No lo
creo
Not 3msDO believe
‘I do not believe it’
In fact, abstracting away from the difference in the masculine singular (a historical
accident related to stress), the forms are exactly the same for both the clitic and the
determiner paradigms.10
9
Picallo (2002) gives additional arguments showing the determiner nature of this neuter
clitic; she also argues that this clitic lacks person features.
7
10
Spanish is by no means special in the Romance clitic system. In the entire family
determiners and 3rd person object clitics share their origin and, with some differences, their
morphological shape and general properties. See, among many others, Lapesa (1968) and Bleam
(2000) for Spanish, Uriagereka (1995) for Galician, and Renzi (1997), Giusti (2001) and
references cited there for formal analyses that take into consideration some relevant diachronic
and comparative details. See also footnote 25 for details.
8
Table 1. Determiners and Object Clitics
Masculine
Feminine
Neuter
Singular
Plural
Singular
Plural
Determiner
el
los
la
las
lo
Clitic
lo
los
la
las
lo
From a historical point of view, the fact that the two paradigms overlap is not at
all surprising, since they share the same demonstrative origin. However, the IO forms
also shares the same diachronic origin (see footnote 10 below and references there), and
nevertheless they have dissociated and pattern together with first and second person
clitics. In particular, as examples in (12) show, IO clitics are indifferent to the
distinction between feminine and masculine and are marked morphologically only for
number:
(12)
a.
Le
di
el libro a María/a Juan
3sIO gave the book A María/ A Juan
‘I gave Mary/John the book’
b.
Les
di
el libro
a mis hermanas/ hermanos
3pIO gave the book
A my.pl sisters
/ brothers
‘I gave my sisters/brothers the book’
This suggests that they passed the cliticization stage and became agreement morphemes,
a rather common process in many languages of the world.11
2.3. Restrictions on the range of DPs
The behavior of 3rd person DO and IO clitics not only differs syntactically and
morphologically, but also semantically, and the range of DPs allowed to corefer with
the DO clitic is narrower than the ones allowed by the IO clitic. There is a structural
complication that makes a straightforward comparison of these structures difficult,
11
The diachronic change strong pronoun > clitic > agreement marker—a particular case
of the so-called cline of grammaticality (Hooper & Trauggot 2003)—is a very common process
in the languages of the world, and has been documented in a variety of unrelated language
families including, among many others, Bantu (e.g. Bresnan & Mchombo 1987), Semitic (e.g.
Fashi Fehri 1981, 1988), Celtic (e.g. McCoone 2006), Romance (e.g. Lapesa 1968, Renzi 1997),
Haitian Creole (DeGraff 1994) and Basque (e.g. Gómez & Sáinz 1995). See Hooper & Trauggot
(2003) for a recent typological approach to the phenomenon.
9
because lo(s)/la(s) do not allow doubling of the argument, but an argument may be
constructed that shows the contrast also in this respect. First, note that definite DPs are
allowed in Clitic Left Dislocation (CLLD) constructions in Spanish with both DO and
IO clitic doubling:
(13)
a.
El libro lo
han
The book 3msDO
have-they
‘The book, they sold it’
vendido
sold
b.
A la hija
del
rector le
han
A the daughter of.the rector 3sIO have-they
‘To the rector’s daughter they gave the degree’
dado el título
given the degree
What is interesting is that, when the dislocated element is a negative quantified DP,
there is a clear contrast in that 3rd person IO clitics can double dislocated negative DPs
but DO clitics cannot:
(14)
a.
b.
* Ningún libro lo
han
None book 3msDO have-they
A ninguna
estudiante le
A none
student
3sIO
‘No student they gave the degree’
vendido
sold
han
have-they
dado el título
given the degree
The ungrammaticality of (14a) cannot be attributed to an absolute ban against CLLD
with 3rd person DO clitics, because (13a), with the definite description, is grammatical.
It cannot be due to an impossibility for the negative DP to appear in the left-dislocated
position, because ninguna estudiante (‘no student’) is perfectly ok in that position in
(14b). The natural conclusion to draw from the paradigm is that what fails in this case is
the connection between the dislocated DP and the clitic. This conclusion is totally
coherent with Roca’s (1996) results that the semantic range of the clitic lo(s)/la(s)
coincides with those found with the definite determiner. In any case, the important
observation to draw from the paradigm in (13)-(14) is that 3rd person IO clitics allow a
wider range of DPs as potential antecedents than DO ones do.12
In sum, there are reasons to assume that third person DO and IO clitics in
Standard Spanish differ radically in a set of properties, the former belonging to the
Determiner system. In the next subsections we will show that all first and second person
clitics –in both the IO and the DO paradigms-- pattern together with IO clitics.
12
See next section and, especially, Roca 1996 for a more complete paradigm concerning
the range of DPs. In contrast, the behavior of IO clitics mirrors that of subject agreement.
10
2.4. The other persons: first and second vs. third person DO
1st and 2nd person DO clitics clearly differ from 3rd person DO clitics and converge with
IO clitics in all respects analyzed in the previous subsections: doubling, morphological
features and interpretive restrictions. In particular, they do not show any restriction on
the type of argument they can double. They clearly double definite DPs and quantified
expressions headed by quantifiers other than todos (‘all’):
(15)
a.
Os
han
visto a los niños
2pDO have- they
seen A the children
‘They saw you the children’
(cfr. (2))
b.
Os
vimos a algunos/muchos niños
2pDO saw.we A some / many
children
‘We saw some/many of you children’
(cfr. (5b))
They also double Wh-phrases, minimally contrasting with 3rd person DO clitics:
(16)
a.
¿A quiénes/ cuántos os
han
elegido para el puesto?
A who.pl/how many 2pDO have.they
selected for the position
’Who/how many (of you) did they chose for the position?’
b. * ¿A quiénes/cuántos los
han elegido para el puesto?
A who.pl/how many 3pDO have selected for
the position
’Who/how many did they chose for the position?’
Contextually, it is more difficult to find nondefinite interpretations for 1st and 2nd person
argument than for 3rd person ones; yet, when this difficulty is controlled for 1st and 2nd
DO clitic doubling shows the same lack of restrictions as IO ones and may appear in
contexts where DO clitics would never appear:
(17)
a.
No os
encontraron a nadie /ninguno
Not 2pDO
found.they
A nobody/none
‘They found none of you’
(cfr. (14a))
b.
Me han
dicho que os
han
pillado a unos fumando
1sIO have-they told that 2pDO have-they catch A a-pl smoking
‘I was told that some people (you included) have been caught smoking’
Following the same logic proposed for 3rd person DO and IO clitics, 1st and 2nd person
DO clitics should be treated together with all IO clitics as belonging to the same
paradigm. This conclusion is reinforced by the facts concerning the combination of the
universal quantifier and strong pronouns mentioned in sec. 2.2. As expected, first and
second DO pronouns, contrary to third person ones, demand the presence of the clitic:
11
(18)
a.
* (Nos) han
pillado a todos nosotros
1pDO have.they caught A all us
They have caught us all out
b.
*(Os) han
visto a todos vosotros
2pDO have.they seen A all you
They have seen you all
(cfr. (7b))
Next, we show that this distinction between 3rd person DO clitics, on the one hand, and
the rest of the object clitics, on the other, is also supported by the way these clitics
interact in clitic clusters.
2. 5. The Object Agreement Constraint
The Object Agreement Constraint (OAC) is an extension of the me-lui constraint, a
restriction originally observed by Perlmutter (1971) in Spanish and French but attested
in the grammar of numerous languages consisting in a general gap in the combination of
possible clitic and agreement clusters. In a nutshell, this constraint distinguishes
between 3rd person and 1st/2nd person DO clitics in ditransitive structures. While 3rd
person DO clitics are compatible with the presence of an IO clitic in the same cluster
(19), 1st and 2nd DO clitics can never co-occur with an argumental dative clitic (20).13
(19)
(20)
a.
Pedro te
lo
envía
Pedro 2sIO 3msDO send.he
‘Pedro sends it to you’
b.
Pedro me
lo
envía
Pedro 1sIO 3msDO send.he
‘Pedro sends it/him to me’
a.
* Pedro te
me
envía
Pedro 2sIO 1sDO send.he
‘Pedro sends me to you’
b.
* Pedro te
le
envía
Pedro 2sDO 3sIO send.he
‘Peter sends you to him’
[Spanish]
Bonet (1991) links this constraint to a similar restriction systematically
applying in rich agreement languages, and in the last years there is a considerable
number of works that extensively argue that it is a syntactic restriction related to the
structural properties of ditransitive constructions when both DO and IO arguments enter
13
In languages that make a distinction between animate and inanimate 3rd person objects,
animates pattern together with 1st and 2nd person (Ormazabal & Romero 2007), an issue to
which we will come back in the next sections.
12
into agreement/structural Case relations.14 In Ormazabal and Romero (2007) we also
extend the empirical scope of the phenomenon considerably. The relevant observation
here is that once again all DO-IO clitic combinations are banned except when the DO
clitic is 3rd person lo(s)/la(s), which indicates that the last ones are external to the clitic
agreement system altogether.
2.6. Concluding remarks
From the discussion in this section we can conclude that 3rd person DO clitics in
Spanish show properties and licensing conditions that set them apart from their IO
counterparts, as well as from 1st and 2nd person clitics in both series.
Table 2. Properties of Standard Spanish clitics
IO clitics
DO clitics
st nd
rd
1 ,2
3
1st, 2nd
Doubling
Gender
Range of DPs
OAC violations effects
sg: me, te,
pl.: nos, os
le
les
me, te,
nos, os
yes
no
FULL
yes
yes
no
FULL
yes
yes
no
FULL
yes
3rd pers.
lo (m.), la
(fem.)
los, las
no
yes
NARROW
no
In the following sections we explore in more detail the properties of the clitic
system from this mixed perspective, and analyze its effect on microdialectal variation.
An interesting observation that arises from this study is that the agreement nature of 1st
and 2nd person clitics and the whole series of IOs is extremely robust in Spanish and
remains almost invariable across all the dialects analyzed. Variation is thus restricted to
3rd person DO objects, where in contrast the changes are diverse and take very different
directions. At the same time, although from a diachronic and dialectal perspective the
Since Bonet’s (1991) formulation of the generalization as the Person Case Constraint,
which unifies quite a lot of phenomena in clitic clusters and rich agreement environments, in the
last years an extensive literature has been produced that argues for its syntactic nature related to
Case/agreement. See, among many others, Ormazabal & Romero 1998, 2007, Anagnastopoulou
2002, Béjar & Rezac 2003, 2009, Karimi 2010, Rezac 2010, Preminger 2011, Walkow 2012, as
well as the collection of papers in D’Alessandro et al. 2008 and references therein. Also see
section 6 below. Albizu 1997, although framed in a morphological analysis, already anticipates
many of the syntactic features developed in more recent proposals.
14
13
system could be perceived as being very unstable in this particular area, internally each
dialect shows a high degree of coherence and systematic behavior.
3. Multiple Agreement in Spanish and 3rd Person DO Agreement
In section 2 we have extensively argued that all object clitics in Spanish except 3rd
person DO clitics are agreement markers base-generated in the verbal or inflectional
head; that includes 1st and 2nd DO clitics and the entire IO series. In this section we
concentrate on two aspects of the system as considered so far: the apparent gap left in
the paradigm by 3rd person DO, on the one hand, and the similarities between the DO
and IO agreement. Our main goal is not only to establish that we are dealing with a
genuine, well-behaved, agreement system; but this state of affairs also allows us to go
one step further and propose a single agreement paradigm —and, consequently, a single
agreement relation/position—for DO and IO arguments, a conclusion that has important
theoretical consequences both concerning linguistic change and the theory of
agreement.
If our results so far are correct, it can be argued that languages like Spanish
show a multiple agreement pattern, quite close to the system of well established
multiple agreement languages such as Quechua, Georgian or Basque, as extensively
argued in some previous works (see Franco 1993 and Ormazabal and Romero 1998, and
references there). Table 3 reflects the three-way agreement system we postulate for
Standard Spanish as a starting point.15
Table 3: Spanish Agreement System
DO Agreement clitics
IO Agreement clitics
Subject Agreement
Singular
Plural
Singular
Plural
Singular
Plural
1st person me
nos
me
nos
-o
-mos
2nd person te
os
te
os
-s
-is
3rd person ---
---
le
les
--
-n
15
The subject agreement paradigm is more complex than suggested in the text. The
paradigm in the table corresponds to the present tense of the first conjugation, and what often
appears as the 3rd person agreement marker corresponds to the thematic vowel.
14
From a purely morphological point of view, table 3 reveals an almost complete
similarity between the DO and the IO agreement system, while the subject agreement
seems to go in a different direction. This is confirmed by other properties of the two
groups such as the position each occupies with respect to the verbal root. But the table
also reveals an interesting division in the 3rd person of DO and IO agreements: there is a
gap in the paradigm in that 3rd person DOs are the only arguments that are excluded
from the agreement relation with the verbal system. The lack of agreement with 3rd
person DOs is not in any way an oddity of the Spanish clitic-system; it pairs together in
that respect with many well-established agreement systems where third person, both in
subject and object agreement also lacks a specific marker, a fact known since the
seminal work of Benveniste (1966) that is often represented by coding a
-morpheme.
But the situation is a bit more complex; in fact, that conclusion clearly holds in
Standard Spanish, but not in the leísta dialect spoken in the North of Spain, which, for
the sake of the exposition, we will dub Basque Leísta Dialect (BLD).16 The
characteristic property of this dialect is that a distinction is systematically made in the
3rd person DO system between animate and inanimate objects. When the object is not
marked for animacy, these dialects follow the standard lo(s)/la(s) clitic system; when it
is animate, they make use of the clitic le(s):17
(21)
a.
Lo
vi (*el libro)
3msDO saw.I (the book)
‘I saw it
b.
Le
vi
(al niño/a la niña)
16
The term leísmo covers a wide range of phenomena in the crossroad of Case, animacy,
person and number of internal argument representation. In some leísta dialects, the clitic forms
le (singular) and les (plural) are used for all masculine 3rd person forms of direct and for indirect
objects, and therefore they do not make any distinction between animate/inanimate, or
accusative/dative objects. In other areas the use of clitics lo and le clearly makes a distinction
between inanimate and animate objects. These areas can furthermore be split between those that
allow clitic doubling, and those that do not, and in some other intricate ways. For the ease of
exposition, we first deal with BLD, which allows doubling, and leave the details and the
discussion of Central Peninsular leísta dialects until section 5, where we deal with some of these
intricacies. See also Landa (1995), Fernández-Ordóñez (1999), Bleam (2000) and references
there for an extensive description of these dialectal differences and their theoretical
consequences.
17
In section 4 we show that there are very specific syntactic configurations where the
clitic le may stand for (and double) an inanimate DO argument in BLD, a fact that will turn out
to be crucial evidence in support of our general approach. However, until we present the
relevant evidence, we maintain the standard description on the animacy feature associated to le.
15
3saDO saw.I (the boy/the girl)
‘I saw him/her/the boy/the girl’
Clearly, this cannot be considered just a morphological choice, since its consequences
expand through the entire system. [+animate] DO clitics behave exactly like other
agreement clitics, radically contrasting with the other 3rd person DO clitic lo(s)/la(s),
which keep the same general properties they have in Standard Spanish. First, regarding
its morphological shape, 3rd person animate object le(s) in these dialects are the same as
the 3rd person IO ones, which belong to the agreement paradigm together with the other
agreement clitics me and te. Second, in contrast to Standard Spanish, these dialects do
not discriminate between masculine and feminine DO forms in the animate series.
Third, as (21b) illustrates, DO 3rd person animate le-clitics can be doubled (Franco
1993, Franco and Mejías-Bikandi 1995, Landa 1995), contrasting in that respect with
non-animate ones. Finally contrary to what has been said in the literature (see especially
Bleam 2000), DO [+animate] clitics in BLD do not show the same interpretive
restrictions 3rd person DO clitics lo(s)/la(s) do. Remember from our discussion in
section 2.3 that 3rd person DO clitics co-occur with a narrower range of DPs than the
rest of the clitics in Standard Spanish. In particular, example (14a), repeated in (22),
shows that lo(s)/la(s) cannot refer to a negative quantifier in a CLLD configuration. As
(23) illustrates, 3rd person animate DO clitic le in BLD is not restricted. (23a) shows a
case completely parallel to (22), where the quantified phrase is left dislocated, but the
clitic le(s) is grammatical; in (23b), also grammatical, the clitic doubles the negative
quantifier:
(22) * Ningún libro lo
han
None book 3msDO have-they
(23)
visto en la universidad
seen in the university
a.
A ningún estudiante le
han
visto en la universidad
A none
student
3sIO have-they
seen in the university
‘None of the students they saw at the university’
b.
No le
han
visto a ningún estudiante en la universidad
Not 3sIO
have-they
seen A no
student
in the university
‘They saw no student at the university’
In other words, lo(s)/la(s) clitics in these dialects have the same set of properties as in
Standard Spanish, but le(s) behaves together with IO le(s) and 1st/2nd clitics in general.
There is another interesting respect in which DO clitics le(s) and lo(s)/la(s)
differ in leísta dialects: when combined with other agreement clitics, animate 3rd person
16
DO clitic le(s) shows me-lui (OAC) effects, as illustrated in (24) from Ormazabal and
Romero (1998; 2007):
(24)
a.
Le
llevé
a tu hijo a casa
3saDO brought.I
A your son to home
‘I brought your son home’
b.
Te
(*le) llevé
(a) tu
2sIO 3saDO brought.I A your
‘I brought you your son home’
hijo a casa
son to home
Interestingly, to avoid me-lui effects, BLD makes use of an alternative strategy:
the use of the Determiner-like clitic lo referring to the animate object. It is important to
note that this is possibly the only context where the use of the non-agreement clitic to
refer to animates sounds completely natural for many BLD speakers, not having any
normative connotation.
(25)
Te
lo
llevé
a casa
2sIO 3msDO brought.I to home
‘I brought you it/him (home)’
But the choice of the clitic lo is more than just a morphological strategy to avoid the melui combinations; it comes together with the syntactic consequences associated to this
choice: clitic doubling is again impossible, as illustrated in (26).
(26) * Te
lo
llevé
2sIO 3msDO
brought.I
‘I brought you the child home’
al /el niño a casa
A.the/the child to home
Moreover, following our discussion in section 2, the construction gets restricted
once again to contexts where the interpretation is compatible with the determiner-like
clitic lo. Thus, in a me-lui Repair environment the sentence is ungrammatical when the
clitic corefers with a dislocated indefinite NP (27a), but grammatical if the corefering
element is a dislocated or an implicit definite DP (27b):
(27)
a. *
A unos estudiantes nos los
han
enviado desde la universidad
A some students 1pIO 3pmDO have.they sent
from the university
‘Some students were sent to us from the university’
b.
A tus estudiantes nos los
han
enviado desde la universidad
A your students
1pIO 3pmDO have.they sent
from the university
‘Your students, they were sent to us from the university’
Finally, animate clitic le(s) contrasts with lo(s)/la(s) in that it is necessary with
the strong pronoun headed by a universal quantifier:
(28)
a. *
Los
he
comprado
3mpDO have.I bought
todos ellos [= Standard Spanish (7b)]]
all
them
17
‘I have bought them all’
b.
Les
han
pillado a todos ellos
3pDO have.they caught A all
them
‘They have caught them all’
[see (18a-b)]
In sum, as table 4 illustrates, the properties of the two clitics strongly support our claim
that while le in BLD is an agreement marker lo(s)/la(s) are uniformly determiner-like
clitics in this dialect, even in those restricted contexts where they refer to animate
objects.
18
Table 4. Properties of Basque Spanish clitics
IO clitics
DO clitics
1st, 2nd
3rd animate
1st, 2nd
3rd
Doubling
Gender
Range of DPs
OAC violations
effects
3rd inanimate
sg: me, te, le
pl.: nos, os les
me, te,
nos, os
le
les
yes
no
FULL
yes
no
FULL
yes
no
FULL
yes
no
FULL
lo (m.), la
(fem.)
los, las
no
yes
NARROW
yes
yes
yes
yes
no
In addition to filling up the gap we originally had in the paradigm (cfr. table 3), this
dialect presents an additional point of interest: in BLD the distribution of agreement
clitics comes close to that of the Differential Object Marker (DOM) a in all dialects of
Spanish. This opens a way to treat DOM in a formal and unified analysis together with
agreement clitics. In the next section we explore the interrelation between Case, as
manifested by the marker a, and agreement, as manifested in the clitic paradigm. The
analysis of DOM uncovers new interesting relations between the two systems and, at
the same time, it allows us to set the basis for the analysis of microdialectal differences
we develop in section 5, where we deal with a different leísta dialect and a different
agreement clitic system.
4. Differential Object Marking and Clitics
A property of Spanish distinguishing it from other Romance languages is the presence
of a marker a introducing animate specific DOs. The nature of this element has been the
center of much debate in the Romance literature, and it has often been assimilated to
other cases of Differential Object Marking (DOM). 18 What interests us is the interaction
18
The description in the text is a simplification of the properties that govern the
introduction of the marker a in Spanish. The literature on the topic is huge, and many are the
factors that have been adduced to determine its distribution (see the collection of articles in
Pensado 1995 and in Probus 20.1 and references there, as well as Torrego 1998, Leonetti 2003,
2008, Rodriguez-Mondoñedo 2007 and Zdrojewski 2008, among many others). In this paper we
are only concerned with some syntactic effects resulting from the interaction of a with the clitic
system, and do not have anything particularly insightful to say concerning the relevant features
that govern its presence/absence with DOs.
19
of this animacy marker with dative clitics. On the basis of (i) their agreement behavior
in contexts where se-constructions and datives are combined, and (ii) their presence
with inanimates in raising-to-O contexts, we argue that A-marked DOs maintain a Case
relation with the verbal complex that regular inanimate DOs do not maintain.
4.1. The Marker A and agreement
A clear context where the Differential Object Marker a is possible for all, and
obligatory for most speakers of Spanish is with animate specific objects, as in (29a),
which clearly contrasts with the impossibility of regular inanimate DOs headed by a.
(29)
a.
He
encontrado
Have.I found
‘I found the girl’
*(a) la niña
A-the child-fem
b.
He
encontrado
Have.I found
‘I found the book’
(*a) el libro
A the book
The presence of the animacy marker in these configurations is independent of whether
that object may be doubled by an agreement clitic or not. This is shown by the fact that
a must appear in contexts where no clitic doubling appears (29a), as well as when the
animate specific object is doubled by an agreement clitic: (30a) corresponds to BLD
where animate object clitic doubling is in most cases obligatory (see section 3), and
(30b) is a context of 2nd person DO agreement in all dialects, as discussed in section 2.4:
(30)
a.
Le
he
encontrado
3saDO have.I found
‘I found the girl’
*(a) la niña
A the child.fem
(BLD)
b.
Os
veré *(a) los
que vengáis a la reunión (all dialects)
2pDO willsee.I A the-mpl that come.you to the meeting
‘I will see those of you that come to the meeting’
Interestingly, the only context where a cannot precede animate specific DOs is when
combined with a doubled IO, as in (31b):
(31)
a.
Enviaron
*(a) los enfermos
a la doctora
sent.they
A the sick.people
to the doctor
‘They sent the sick people to the doctor’
b.
Le
enviaron (*a) los enfermos a la doctora
3sIO sent.they A the sick.people A the doctor
‘They sent the doctor the sick people’
20
The contrast in (31a-b) shows that the relation between the verb and the DO changes
completely in the presence of the dative clitic. 19 Although in this case the animate DO
does not show an overt agreement marker in the verb, the situation is very similar to the
OAC discussed above in that the ban on the presence of the marker a in the DO
argument is triggered by the agreement relation of the dative argument with the verb.
This suggests that some Case relation holds also between the verb and the animate
specific DO, the animacy marker being a morphological manifestation of that relation,
as already proposed by some authors in the literature (see, for instance, López 2008).
When there is no dative clitic, specific animate objects receive accusative Case,
morphologically marked by means of the DOM a. As in the case of OAC contexts, this
relation is blocked when an agreeing dative is present. In that case, the animate object
does not receive accusative Case from the verb, and the DOM marker disappears
accordingly (compare (31) with the OAC paradigm in BLD in (24)).
Support for this conclusion comes from the behaviour of animate objects in socalled se-constructions. First note that these constructions share with regular passives the
property that the external argument is dropped and the internal argument raises to subject
position triggering agreement with the verb (see Mendikoetxea 1999 and references
there):
(32)
a.
Se
llevaron
(los) regalos a la doctora
took.3pl
the presents to the doctor
‘(The) presents were sent to the doctor’
SEarb
b. * Se
SEarb
llevó
took.df
(los) regalos a la doctora
the presents to the doctor
Complements headed by the animacy marker are frozen in this construction and unable
to raise to subject position (33a). The only option available in this case is marking the
verb with a default 3rd person singular agreement (33b).
(33)
a. *
Se
SEarb
llevaron
took.3pl
(a) los enfermos
A the sick.people
a la doctora
to the doctor
19
Many authors have described the restriction illustrated in (31) as the impossibility of a
in the presence of an IO headed by a (see Laca 1995, Zdrojewski 2008, Saab & Zdrojewski
2011, and references there), yielding a violation of some kind of filter against double a,
comparable to the double-o of Japanese. However, it is the presence of the IO agreement clitic
le that makes the difference. As (31a) shows, nothing goes wrong when the dative shows up as a
non-agreeing PP. In more general terms, the discussion in the text strongly suggests that the
animacy marker a in Spanish is not a prepositional element and that the so-called Kayne’s
generalization on clitic doubling (Jaeggli 1982, Roberts 2010, among others) may be in fact a
spurious generalization.
21
b.
Se
llevó
a los enfermos
SEarb took.df
A the sick.people
‘The sick people were sent to the doctor’
a la doctora
to the doctor
This suggests that, as we propose, the animate DO is already Case-marked and is inert
for further movement to subject position.
Assuming that, let us come back to the impossibility of a-marking on the
animate specific DO in the presence of a dative clitic in (31). If the presence of the
animacy marker a is in fact the overt indication that a Case relation has been established
between the verbal complex and the argument, the impossibility of DOM with the
animate specific DO in (31b) may be considered to indicate that such a relation is not
possible in the context of an agreeing IO. We expect then that it will be free to raise to
the subject position and agree with Tense (34a), and default agreement should not be
available (34b), a prediction that is borne out:
(34)
a.
Se
SEarb
b.
le
3sIO
llevaron
took.3pl
los enfermos
the sick.people
a la doctora
A the doctora
* Se
le
llevó
los enfermos al doctor
SEarb 3sIO
took.df
the sick-people to.the doctor
‘The sick people were taken to the doctor’
We thus conclude that Case-marking in Spanish tears apart specific animate direct
objects as well as indirect objects from inanimate direct objects. The analysis of DOM
in the presence of dative clitics in Spanish shows that there is a tight relation between
object agreement and DOM, and that a dative agreement-clitic blocks DOM on the
object. This receives a straightforward explanation if in these contexts object
agreement—and, consequently, DOM—is established with the IO, which must also be
a-marked.20
4.2. Inanimates headed by a and doubled by le
Laca (1995), based on Roegiest (1979), and Zdrojewski (2008) discuss quite a few
syntactic configurations where a appears with inanimate DOs virtually in all dialects of
Spanish (also see Campos 1999:1543). Some of Laca’s examples are repeated below:
Note that the prepositional variant may represent different prepositions (en ‘in’, de
‘from’, a ‘to’, etc.), but the DP in the clitic doubling construction can only appear preceded by
A. See section 6 and references there for discussion.
20
22
(35)
a.
Emergiendo sobre una ola, veo al avión caer
envuelto en llamas
Emerging over a wave, see.I A.the plane fall down enveloped in flames
‘Emerging over a wave, I see the plane fall down ablaze’
[Laca 1995, ex. (8b); translation and glosses ours]
b.
Por eso, sólo por eso, prefiero llamar “historia” y no “novela” a esta obra
mía.
For that, only for that, prefer.I call “story” and not “novel” A this work mine
‘For that reason, and only for that reason, I prefer calling that work of
mine “story”, and not “novel”’[Roegiest 1980:145, from Laca 1995;
idem]
c.
La tormenta dejó sin
hojas a los árboles
The storm
left.it without leaves A the trees
‘The storm left the trees leafless’
As Zdrojewski (2008) notes for Rioplatense Spanish, to which we come back in
the next section, Basque Leísta Dialects (BLD) also shows clitic doubling in these
contexts. Consider the following paradigm:
(36)
a.
b.
(37)
* Le
trajeron
3saDO brought
(al
avión) a través de las montañas
(A.the plane) across of the mountains
El avión,
lo
trajeron
a través de las montañas
The plane,
3msDO brought.they across of the mountains
‘The plane, they brought it across the mountains’
Le
vimos al
avión caer
envuelto en llamas
3saDO saw.we A.the plane fall down enveloped in flames
‘We saw the plane fall down ablaze’
As we expect, in regular DO contexts, the clitic le cannot double or corefer with the
inanimate el avión (‘the plane’) ((36a)); instead this argument can appear in a clitic
right-dislocated configuration where the clitic is the determiner-like lo (36b). In
contrast, in the context of a perception verb with an infinitival complement, the
inanimate DO shows up marked with DOM and doubled by the agreement-clitic le (37).
As far as we can see, what all the contexts in (35)-(37) as well as the rest of the
cases discussed by Laca have in common is that they all suit the conditions to be
analyzed as cases of raising-to-object of the A-marked argument from an embedded
position. Thus, not only do they show that the marker A, in general, and the clitic le, in
BLD, are something more than a morphological manifestation of animacy, etc., but they
provide direct evidence that a Case relation is established between the verbal complex
and the embedded argument that is not maintained with regular DOs. In a series of
articles Lasnik (see especially Lasnik and Saito 1991, Lasnik 1995, 1999) argues that
23
ECM-subjects undergo overt object shift in English. Based on Lasnik's arguments,
Bošković (1997, 2002) shows that ECM accusative subjects and regular transitive
objects behave differently in that respect: ECM-arguments are subject to obligatory
object shift while regular objects do not overtly A-move to the object-agreement
position except as an intermediate step of some additional movement to a higher
position (Wh-movement, passive, etc.). Ormazabal and Romero (2002, 2010) and
Boeckx and Hornstein (2003) have independently argued in support of this distinction,
showing that the impossibility of combining ECM and DOCs illustrated in (38) derives
from this difference, as a particular case of the me-lui/OAC effects (see Ormazabal and
Romero 2010 for discussion).
(38)
a.
I showed you the proof
b.
I showed you that the defendants were guilty
c.
I showed the defendants to be guilty
d. * I showed you the defendants to be guilty
Example (38a) shows that the verb show is a Double Object-type verb; as (38b)
illustrates, a clausal complement in DO position is compatible with an accusative
indirect object; and example (38c) shows that the verb belongs to the ECM group as
well. However, as illustrated in (38d), the combination of ECM and DOC is impossible
but, given standard assumptions about the Case of ECM subjects, nothing seems to
prevent it. Instead, as argued in the references above, the ungrammaticality of (38d)
follows without any additional assumption if the obligatory movement of the embedded
subject to the object agreement position is blocked by the presence of the dative
argument in that position.
Given all the above, the discussion of the Spanish cases in (35)-(37) shows that,
as we would expect, the reported asymmetry between regular objects and raising-to-O is
not particular to English, but reflects some deeper property that also comes up in
Spanish in a different shape. The order of the constituents in these constructions
suggests the same conclusion: unlike in regular sentences, the embedded subject of the
small clause appears most naturally in a position between the perception verb and the
embedded infinitive, postverbal subjects being somehow very marginal.
(39)
a.
Le
vimos al
avión estrellarse contra la montaña
3saDO saw.we A-the plane crass
against the mountain
‘We saw the plane crash into the mountain’
24
b. ?? Le
vimos estrellarse
al
avión contra la montaña
3saDO saw.we crash
A-the plane against the mountain
‘We saw the plane crash into the mountain’
In more general terms, the presence of le in the verbal clitic system of BLD almost
mirrors that of a in the nominal DO; in particular, a-marking is a prerequisite for
agreement with le. This suggests that the innovation of the BLD has been to encode
morphologically an agreement relation that was already there in some way encoded in the
argument with the DOM a. The way to achieve this goal was to extend the use of an
agreement marker that was already available in the system for 3rd person IO agreement
in general.
5. Micro-parametric Variation
Our analysis of Standard Spanish and BLD and, in particular, the properties of Spanish
DOM discussed in the previous section gives us the tools to analyze some of the
dialectal variation concerning object clitics in Spanish. It is important to highlight that
the dialectal variation is restricted almost completely to 3rd person DO clitics, a fact that
raises interesting theoretical considerations to which we briefly return in section 6. In
contrast, no variation may be observed in the behavior of 1st and 2nd person DO clitics;
and in the IO system the sole area of variation, much more restricted than in the DO, is
the so-called laísmo, also limited to 3rd person clitics (see Romero 2012 for discussion).
In this section we focus on two dialects that we find particularly difficult to incorporate.
We show that an extension of our analysis in the previous sections explains their
properties fairly naturally.
5.1. A different leísta dialect
5.1.1. A determiner in disguise. The Central Peninsular leísta dialect of Spanish
[henceforth CPLD] coincides with BLD in that it makes a morphological distinction
between animate and inanimate 3rd person masculine DO clitics in a systematic way
(40a-b). They both share with DOM their sensitivity to me-lui and, more generally,
Object Agreement Constraint effects ((41); see Ormazabal and Romero 2007):
(40)
a.
El libro
lo
The book
3msDO
‘I saw the book’
vi
saw.I
25
(41)
b.
Al
niño le
A.the child 3msaDO
‘I saw the child’
a.
Le
envié a la alcaldesa
3msaDO
sent.I to the mayor
‘I sent him to the mayor’
b. * Se
le
3IO 3smaDO
‘I sent him to her’
vi
saw.I
envié
sent.I
However, 3rd person [+animate] clitics le(s) in CPLD present many properties that set
them apart from BLD agreement markers and bring them closer to DO determiners
lo(s)/la(s). Most characteristically, CPLD clitics le(s) do not allow doubling, except in
those contexts discussed in section 2 where the inanimate clitics lo(s) also allows it,
sharply contrasting with BLD animate clitics in that respect.
(42) * Le
vi
al
niño
[CPLD; cfr. BLD in (21b)]
3msaDO
saw.I A.the boy
‘I saw the boy’
Other properties of this dialect also set it apart from BLD and bring the clitic le(s) closer
to determiners than to agreement markers. To begin with, the animate clitic le(s) in
CPLD shows the interpretive restrictions typically observed with lo(s)/la(s) and argued
by Roca (1996) to correspond to the semantics of determiners. Thus, le in these dialects
cannot corefer with an indefinite DP, as the contrast in (43) illustrates.
(43)
a.
* A un amigo le
A a friend
3msaDO
‘I saw a friend yesterday’
b.
vi
ayer
saw.I yesterday
A mi amigo le
vi
ayer
A my friend
3msaDO
saw.I yesterday
‘I saw my friend yestarday’
The clitic cannot double a negative quantifier in CLLD configurations either, also
supporting the Determiner-analysis. Compare the ungrammaticality of (44) with (23a) in
BLD.
(44) * A ningún estudiante le
han
visto en la universidad
A none
student
3sIO have-they seen in the university
‘None of the students they saw at the university’
The parallelism with lo(s)/la(s) also extends to the difficulty to double quantifier
phrases containing a strong pronoun, contrasting both with IO clitics and le(s) in BLD:
(45)
a. *
Los
he
comprado
3mpDO have.I bought
todos ellos [=(7b) all dialects]
all
them
26
‘I have bought them all’
b. ?* Les
han
pillado a todos ellos [CPLD; cfr. BLD (28)]
3mpaDO have.they caught A all them
‘They have caught them all’
c.
Les
he
pegado pegatinas
3pIO have-I stuck stickers
‘I stuck stickers in all of them’
a todos ellos (les= a los libros)
A all-mpl them-mpl
[= (8); all dialects]
In addition, unlike BLD, CPLD shows a gender distinction in the DO system,
discriminating between animate masculine clitic le and feminine la:
(46)
a.
Al
niño
A.the child.m
‘I saw the boy’
le
3msaDO
vi
saw.I
b.
A la niña
A the child-f
‘I saw the girl’
la
3fsDO
vi
saw.I
It is worth mentioning that in all these respects, the behavior of 3rd person animate DO
clitic le(s) also differs from its homophone IO clitic le(s), which patterns like IO clitics
in all the dialects. In particular, the latter can double animate IOs (47a), as well as
inanimate ones (47b).
(47)
a.
Le
recomendé
un libro al estudiante
3IO recommended.I a book A-the student
‘I reccomended the students a book’
b.
Le
apreté
las clavijas
a la guitarra
3sIO squeezed.I
the pegheads A the guitar.FEM
(Lit.)’ I squeezed the guitar the pegheads’
Summarizing, the cluster of properties observed in CPLD suggests that in these dialects
we are dealing with a determiner system similar to the standard one, with the
particularity that it further represents the object’s animacy in the masculine definite
determiner.21 As a determiner head, it behaves together with object clitics lo(s)/la(s)
with respect to doubling and the semantic properties associated to determiners;
similarly, gender distinctions are trivially expected. At the same time, the feature
specifications of these determiners may be expected to play a role in deriving its
differences with lo(s)/la(s), as we show next.
21
This distinction is not uncommon, and is also encoded in other domains of the
pronominal/determiner system such as subject pronouns (masc. él vs. neuter ello).
27
5.1.2. Comparing Animate Determiner le(s) and DOM. As the cliticized determiner
head of an animate DP, le(s) in CPLD shares some properties with the syntactic
configuration of DOM headed by the marker a, and indirectly with agreement clitics
le(s) in BLDs. Specifically, it shares its sensitivity to the me-lui phenomenon, as
illustrated in (41) above. This can be derived if the requirement that animate DOs,
unlike regular inanimate ones, check Case with the verb is independent of whether the
argument appears as a full DP headed by A or as a weak pronoun le(s). If so, we expect
its presence to compete with IOs agreeing with the verbal complex in the same way they
compete in BLD.
In fact, the parallelism with DOM goes beyond OAC facts. It is a well known—
although not very well delimited—fact about DOM in Spanish that, in addition to
animacy, its presence requires that the DP it appears attached to have some kind of
specific interpretation, and the presence of the clitic le(s) is restricted in CPLD in the
same way. Indeed, there is interesting evidence showing that this correlation has a deep
syntactic motivation that goes beyond a pure morphological codification of DOM.
Consider the following contrast, common to all dialects (see Leonetti 2008 for
discussion):
(48)
a.
Pedro busca
Pedro looks.for
(*a) un secretario que hable
inglés
A a secretary that speaks.SUBJUNCTIVE
English
‘Pedro is looking for an English speaking secretary’
b.
Pedro busca
*(a) un secretario que habla
Pedro looks.for
A a secretary that speaks.INDICATIVE
English
‘Pedro is looking for a secretary who speaks English’
inglés
The contrast in (48) builds on the well observed fact that the use of a subjunctive form
in the relative clause (48a) correlates with a non specific reading for the object, while
indicative mood in (48b) correlates with a specific one; in these contexts, despite being
animate, the object in (48a) cannot appear preceded by the preposition a (see references
in footnote 18 for details and discussion). With this contrast in mind, consider now the
following sentences:
(49)
a.
Pedro lo
busca
Pedro 3msDO looks-for.he
‘Pedro looks for someone (who speaks English)’
28
b.
Pedro le
busca
Pedro 3msaDO looks-for.he
‘Pedro looks for him’
If the encoding of animacy in the determiner system in CPLD is syntactically active, the
prediction is that the choice of the clitic will determine the specificity interpretation of
the object just like the absence/presence of a determines the choice of subjunctive or
indicative in the relative clause in these dialects, a prediction that is borne out: sentences
in (49a) and (49b) convey the referential interpretation of (48a) and (48b) respectively.
However, there is an important difference between DOM and agreement le(s) in
BLD on the one hand and the clitic le(s) in CPLD on the other that confirms the
determiner nature of the latter. As we showed in section 4, in addition to animate DOs,
DOM also shows up in contexts of raising-to-Object, even in those cases where the
derived object is inanimate (examples (35)-(37)). In contrast, in CPLD we are dealing
with a determiner le that is specified for the feature [+animate] and, consequently,
cannot be inserted in other contexts where DOM and the BLD agreement marker le(s)
are allowed. In particular, in Raising-to Object environments the use of le in CPLD is
restricted to animate masculine objects (50a) (see Hernanz, 1999).
(50)
a.
(Al niño / *al avión)
le
vimos caer envuelto en llamas
(A.the boy / A.the plane)
3msDO saw.we fall enveloped in flames
‘We saw him/it fall down ablaze’
b.
A este libro mío prefiero llamar-lo
/*le
historia y no novela.
a this book mine prefer.I call-3msDO/3msaDO “story” and not “novel”
“I prefer calling it 'story' and not 'novel'”
c.
La tormenta los
/*les
dejó sin
hojas
the storm
3mpDO/3mpaDO
left.it without leaves
‘The storm left them without leaves’
The different behavior the animate clitic shows in CPLD and BLD also in this respect is
a strong indication of their different nature despite their morphological similarity. All
these differences support our conclusion that it is a determiner in the former and an
agreement marker in the latter.
29
5.1.3. The Object Agreement Constraint (OAC). Consider now me-lui and, more
generally, OAC-effects in (41), repeated in (51).
(51)
a.
b.
Le
envié a la alcaldesa
3msaDO sent.I to the mayor
‘I sent him to the mayor’
* Se
le
3sIO 3msaDO
‘I sent him to her’
envié
sent.I
As observed, CPLD behaves like the BLD in that it precludes le in the presence of a
dative clitic. At a first glance, this fact could constitute a problem for our analysis, given
that in spite of its surface similarity, the clitic le is of a very different nature in the two
dialects: an agreement head in BLD and an incorporated determiner in CPLD. However,
a unified explanation is possible that leans on our observation that in both dialects the
presence of the clitic le correlates partially with the distribution of a-insertion.
Remember that in section 4 we observed that a-insertion is blocked in the presence of a
dative clitic, as shown in example (31), repeated here as (52).
(52)
a.
Enviaron
*(a) los niños a la doctora
sent.they
A
the children to the doctor
‘They sent the children to the doctor’
b.
Le
enviaron
(*a) los niños
3IO sent.they
A the children
‘They sent the doctor the children’
a la doctora
A the doctor
If the animacy marker a is in fact a morphological manifestation of Case, OAC effects
arise in both dialects as a consequence of the failure of the verb to assign Case to both
the DO and the IO in the same configuration.
Interestingly, there is one important respect in which BLD and CPLD differ also
with regard to this phenomenon: in order to avoid OAC effects each dialect resorts to a
different repair strategy, indirectly supporting our claim that le in these dialect is not the
same object. As discussed in section 4, BLD resorts to the determiner clitic lo (53).
However, as the grammaticality judgments in (54) illustrate, this alternative strategy has
an unnatural flavor in CPLD (54a). In fact, in a clear contrast with speakers of BLD,
30
speakers of these dialects strongly prefer not to double the dative in this context (54b),
it being the only context where an IO strong pronoun may appear non-doubled.
(53)
(54)
a.
Se
lo
enviaron
3sIO 3msDO sent.they
‘They sent him to her’
b.
* Le
enviaron a ella
3saDO sent.they to her
‘They sent him to her’
a. ?? Se
lo
enviaron
3IO 3msDO sent.they
‘They sent him to her’
b.
BLD
CPLD
Le
enviaron a ella
3msaDO sent.they to her
‘They sent him to her’
A difference between the two dialects that could account for this contrast is that in
CPLD the repair strategy in (54a) neutralizes the [ animate] distinction within the
determiner clitic system. Consequently, the unnatural character of this strategy in this
dialect can be attributed to the fact that there is a morphological mismatch between the
argument’s actual clitic realization (lo, unspecified for animacy) and its interpretation as
animate. This mismatch does not appear in the BLD, where no issue of determiner clitic
choice arises, but rather there is a shift from a syntactic device involving the agreement
system to a morphological one involving cliticization of a determiner head.22
22
In fact, a similar effect also appears in the Standard dialect, where, somehow
surprisingly, (ic) is also reported as unnatural (Fernández-Ordoñez 1999), despite the
fact that this dialect does not make a morphological difference between animate and
inanimate objects:
(i)
a.
envió a los niños al
médico
sent.he A the children to.the doctor
‘He sent the children to the doctor’
b.
Los
envió al
médico
3mpDO sent.he
to.the doctor
‘He sent them to the doctor’
c. ?? Se
los
envió al
médico
3sgIO 3mpDO sent.he A-the doctor
‘He sent the doctor them’
31
In conclusion, the mixed properties of the clitic le in CPLD derive naturally
from its animate determiner nature.
5.2. An Agreement in disguise
Rioplatense, a mostly urban dialect of Spanish spoken in Buenos Aires and a large area
of Argentina and Uruguay, presents a different challenge for our analysis. The main
particularity of Rioplatense that distinguishes it from Standard Spanish is that 3rd person
DO clitics lo(s)/la(s) allow doubling, although this possibility is restricted to very
specific contexts. The main problem this dialect raises is that lo(s)/la(s) double in many
contexts where they cannot double in Standard Spanish and that the Determiner-analysis
of lo(s)/la(s) in the latter does not seem to be extensible to the former.
But, as Zdrojewski (2008) argues extensively, when considered in more detail
the distribution of doubled and non-doubled clitics correlates with a series of properties
that set them apart in a systematic way. This fact supports an analysis where there are
two sets of (homophone) clitics. To be more precise, if Zdrojewski is correct we are
dealing with two different sets of syntactic objects that happen to share enough features
to converge at PF, when insertion of the clitic applies in the morphological component.
Zdrojewski assumes that the doubling clitic and the non-doubling one are both
determiner phrases that differ in their feature specifications. But, if our analysis in the
previous sections is on the right track, it would be natural to assume that the nondoubling clitic is a determiner head while the doubling one is an agreement. Being
syntactically different, they systematically differ in their behavior with respect to the
properties discussed through the paper, despite having the same morpho-phonological
shape.23 In particular, the agreement marker is restricted to contexts where a [+definite]
This contrast suggests that lo in Standard Spanish is the morphological form of
the two different syntactic objects in (54), in which case the difference between
Standard Spanish and CPLD would reduce to a question of morphological marking in
the determiner system.
23
In the evolution from pronominal clitic systems to agreement systems, it is not
infrequent the existence of languages and dialects where the clitic systems develop paradigms
that are ambiguous. See, for instance, Bresnan & Mchombo’s (1987) battery of tests to postulate
homophone grammatical and anaphoric agreement markers in Chichewa, and Fashi Fehri’s
(1989) similar analysis for Arabic. One of the interests of these proposals, shared with
Zdrojewski’s (2008) study of Rioplatense, is that they all present effective tests to determine
unambiguously whether the clitic behaves as an agreement marker or as a pronominal one in a
32
DP headed by the DOM a is available, and may double this DP. This is, of course, true
with animate objects, as the grammaticality contrast between the indefinite expression
in (55a) and the definite one in (55b) shows. But, more interestingly, the agreement
clitic is also possible doubling inanimate objects introduced by a in raising contexts of
the type discussed at the end of section 4. Thus, although regular inanimates DOs such
as un camión (‘a truck’) in (56a) is impossible and, consequently, the presence of a
doubling clitic is not an option, the clitic is possible in the same restricted contexts
where raising-to-Object has taken place and other dialects allow DOM (56b):
(55)
a. *
Los
planean
echar
a algunos
3DO plan.they
through.away A some
‘They plan to expel some students’
estudiantes
students
b.
Mi perro lo
odia al
cartero
My dog 3DO hate.it A.the postman
‘My dog hates the postman’
[Zdrojewski 2008: 26, (24)-(25); translation and glosses
ours]
(56)
a.
* Lo
vi
a un camión venir a toda velocidad
3DO saw.I A a truck come at all speed
‘I saw a truck coming at high speed’
b.
Lo
vi
a-l
camión venir a toda velocidad
3DO saw.I A-the truck come at all speed
‘I saw the truck coming at high speed’
[Zdrojewski 2008: 25, (21)]
In addition, doubling of the agreement clitic in Rioplatense induces me-lui/OAC effects
exactly in the same contexts agreement clitic le does in leísta dialects, as expected. As
in the case of (31a-b), discussed above, (57a-b) also shows that overt agreement with an
animate DO is banned from a context where dative agreement is also present.24 In
contrast, determiner clitics are allowed in such contexts (58):
(57)
a. *
Juan se
la
presentó
la enfermera a-l doctor
Juan 3IO 3DOf presented.he the nurse
A-the doctor
‘Juan introduced the nurse to the doctor’
b. * Juan se
la
presentó
María
al
doctor
Juan 3IO 3DOf presented.he María
A-the doctor
‘Juan introduced María to the doctor’
given configuration. See also Embrick & Noyer’s (2001) analysis of Swedish determiners,
mentioned by Zdrojewski as the main theoretical model for his proposal.
24
Zdrojewski interprets these facts in slightly different terms, but they may be easily
assimilated to the analysis of the OAC in BLD and the repair strategy discussed in section 3.
33
(58)
Juan se
la
presentó
al
doctor
Juan 3IO 3DOf presented.he A-the doctor
‘Juan introduce her to the doctor
Zdrojewski 2008: 41, (55)-(57)
Moreover, the Determiner clitic is not restricted to definite or to a-marked antecedents
(59), it does not allow doubling (60), and does not induce me-lui/OAC effects, as shown
in (58) above and in (61) below.
(59)
a.
En el tren, Juan vio
(a) una persona que le pareció
conocida,
pero no la
saludó
In the train, Juan saw.he (A) a person that 3sIO appeared.(s)he known-fem,
but not 3DOfem greated.he
‘In the train, Juan saw a person that look familiar to him, but he didn’t say hello to
him/her’
b.
Juan quiere comprar una casa, en pleno centro de Buenos Aires, que tenga patio,
parrilla y pileta.
Yo creo que le va
resultar muy difícil conseguirla
Juan wants.he buy a house, in full center of Buenos Aires, that have.subj patio,
grill and swiming-pool. I think that IO3 is-going-to result very dificult to-get-3DOfem
‘Juan wants to buy a house, in the middle of Buenos Aires, with patio, grill and
swiming-pool. I think that it will be very difficult for him to get something like that’
[Zdrojewski 2008: 26-27, (26),
(28)]
(60) * En el tren, Juan la
vio
a una persona que le pareció
conocida.
In the train, Juan 3sfDOsaw.he A a
person that 3sIO appeared.(s)he known-fem.
(61)
Juan quiere vender un libro que le
se
lo
quiso comprar
regaló
su novia, pero todavía nadie
Juan wants sell a book that 3sIO gave.she his girlfriend, but yet nobody
3sIO 3msDO wanted buy
‘Juan wants to sell a book that his girlfriend gave to him as a present, but nobody
wanted to buy it yet’
If our proposal is on the right track, the specific property of Rioplatense is that it
encodes gender features in the object agreement paradigm. This is the marked case in
the languages of the world, and it differentiates Rioplatense from the peninsular dialects
we have analyzed in the previous sections. But it is shared with other American dialects
such as certain varieties of Chilean and Peruvian Spanish (see Sanchez 2010 and
references there for discussion), and is not uncommon in the historical shift from
pronominal to agreement morphemes in a variety of languages of the world (see, e.g.
references in footnote 23).
34
6. A Single Agreement Paradigm: unifying DO and IO agreement
The dialects discussed in this section show that the feature specifications relevant for
agreement may vary considerably from dialect to dialect (see Ormazabal and Romero
2007 for extensive discussion of these issues in a typologically broader context). Since
the determiner system may also have different feature specifications, the clitic system
interacts in rather complex ways, yielding subtle dialectal differences of the type
discussed in the previous section. These case studies do not exhaust all logical
possibilities, but they present an interesting picture of micro-parametric variation, and
contribute to make the special status of 3rd person DOs in the system more explicit.
In fact, if we abstract away from the morphological complication posed by
Rioplatense, the result of the discussion in the previous sections concerning DO and IO
agreement and Case in Spanish leads us to take a step forward and to unify the two
agreement paradigms in a single one.25
25
From a historical and a morphological point of view, this is nothing particularly new in
what concerns 1st and 2nd person clitics. It is well known that in these cases the two series, the
IO and the DO ones, have the same origin. In the evolution from Latin to Romance, together
with the process of simplification of the declension system, there is a process of specialization
in the distribution of personal pronouns; in the case of Spanish, 1st and 2nd person accusative
pronouns yield the objective clitic paradigm and the dative ones become strong pronouns. Thus
the origin of both Direct and Indirect Object clitics in Spanish is the accusative pronoun of Latin
(i.e. Lat. me (Acc.)> Romance me (DO/IO) , Lat. te > Romance te, Lat. nos > Sp. nos, Lat. vos >
Sp. os, Fr. vous, Port. vos, etc.); on the other hand, the dative pronouns in Latin derive in the
oblique strong pronoun series in Spanish (Lat. mihi (Dat.)> Sp. mí (DO/IO), Lat. tibi > Sp. tí,
etc.). However, all 3rd person pronouns –strong and weak- derive from Latin demonstratives; in
this case, Spanish maintained the Latin Accusative/Dative distinction.
35
Table 5: Spanish Single Agreement Clitic System (DO and IO)
Singular
Plural
1st person
me
nos
2nd person
te
os
le
les
rd
3 person
If this move is on the right track, there is some sense in which the long standing
view in the Spanish clitic literature that treats le of leísta dialects as a dative clitic (e.g.
Torrego 1998) may be observationally correct. However, we have shown that
empirically its agreement properties are nothing exclusive of le in some leísta dialects
but general to the entire set of agreement clitics —in particular to 1st and 2nd person
clitics me, te (sg.), nos, os (pl.)—and the unification of the two le clitics is just a
particular case.
One side of our proposal concerns the elements that show verbal agreement: as
observed, we argue that there is a single agreement system holding for DO and IO
arguments; in its most radical, but coherent, formulation, our claim is that there is a
single agreement relation holding for all internal arguments agreeing with the verbal
complex. Thus, the clitic os in (62) marks exactly the same agreement relation between
v and the argument los que vayáis pronto, independently of whether that argument
corresponds to the DO, as in (62a) or to the IO, as in (62b):
(62)
a.
Os
verán
a los que vayáis pronto
2pDO will.see.they A the that go.you early
‘They will see those of you who arrive early’
b.
Os
darán
el libro
a los que vayáis pronto
2pIO will.give.they the book
A the that go.you early
‘They will give it to those of you who arrive early’
Similarly, the animate DO in the BLD example in (63a) and the IO in (63b) maintain
exactly the same agreement relation with v as the second person argument in the
previous examples:
(63)
a.
Les
verán
a los que vayan pronto
3pDO will.see.they A the that go.they early
‘They will see those people who arrive early’
b.
Les
darán
el libro a los que vayan pronto
2p[IO] will.give.they the book A the that go.they early
‘They will give the book to those people who arrive early’
BLD
36
In fact, we have argued elsewhere that the me-lui constraint and its extensions, the
Object Agreement Constraint, come down to a conflict arising when two arguments
compete for the only available agreement slot. This constraint extends to a larger range
of constructions that typologically have received different names in the literature but
that share many properties and, we claim, the same underlying structural
configuration.26
Obviously, this system only deals with verbal agreement relations; when no
syntactic agreement holds, no agreement clitic appears in the verbal complex, as
expected. There are two cases that interest us particularly here: on the one hand, 3rd
person DOs in the contexts discussed in this paper (i.e., non-specific or inanimate
objects, etc.), and on the other IOs in PP-constructions headed by prepositional heads
such as a (‘to’), en (‘in’, ‘on’), de (‘from’), and maybe others (see Ormazabal and
Romero 2011). Concerning the latter, consider the following paradigm:
(64)
a.
Juan envió un libro a María
John sent.he a book to Mary
‘John sent Mary a book’
b.
Juan le
envió un libro a María
John 3sIO
sent.he a book to Mary
‘John sent Mary a book’
c.
Juan le
envió un libro
John 3sIO
sent.he a book
‘John sent him/her a book’
Some authors have argued that dative clitic doubling is optional and, consequently, does
not behave in that respect like genuine agreement markers –e.g., subject agreement in
Spanish--, which are obligatory (see references in Franco 1993; ch. 2 for discussion).
However, it has been fairly established in the literature (see Uriagereka 1988; Demonte
1995; Romero 1997; Ormazabal and Romero 1998; 2007; Bleam 2000;
Anagnastopoulou 2002; Levin 2008, among others) that the options in (64) correspond
26
Based on our previous analysis of a variety of agreement restrictions across languages (see,
especially, Ormazabal & Romero 1998, 2007), and on a natural reinterpretation of general
conditions on agreement postulated in the literature such as Béjar & Rezac’s (2003) Person
Licensing Condition (see also Preminger 2011), Ormazabal & Romero (2013) argue for a theory
where the triggers for the agreement relation are the properties of the syntactic phrases
undergoing agreement, contrary to current standard assumptions about probe-goal relations. We
propose a detailed analysis of object agreement, and argue that the restrictions generally arise as
a competition between two phrases that for structural reasons are required to enter agreement
when only one appropriate host is available. See references above for details.
37
to two separate constructions that show very different syntactic and semantic properties
depending on whether the dative clitic is present or not. In particular, the Dative Clitic
Construction (64b-c) is subject to the same c-commanding asymmetries discussed in
Barss and Lasnik (1988), and to the same restrictions as English DOCs are. Thus, the
absence of the clitic in (64a) corresponds to a Prepositional Phrase, which never triggers
agreement with v; on the other hand, (64b-c) constitute a particular case of dative shift
and triggers obligatory agreement and Case-checking with the verbal complex,
manifested in Spanish as clitic doubling. From that perspective, (64c) is a case of object
pro-drop, similar to well-known pro-drop cases in rich agreement languages like
Basque. This alternation has been analyzed from very different theoretical perspectives,
but all of them acknowledge the existence of two different structures. In one of them,
the goal argument is governed by a preposition, and in the other it is licensed by object
Case/Agreement.
The second case is related to the internal arguments that do not agree with the
verb. As we have shown, the impossibility of 3rd person agreement for underived
inanimates or non-specific objects is not just a question of morphological marking. The
singular behavior they show with respect to the whole battery of morphological and
syntactic properties discussed through the paper, as well as their particular status across
languages with respect to the me-lui/Object Agreement constraint, clearly supports the
conclusion that these arguments do not compete for the agreement/Case position. This
split regarding the formal licensing of objects is especially clear in the case of se
constructions analyzed in section 4.1.
The main problem for this hypothesis is that substantive uniformity in
Case/agreement relations is lost. It has to be noted, however, that the assumption that
objects constitute a uniform class with respect to Case/agreement relations is based on a
GB modular architecture, where Case and agreement was understood as a filter on NP
licensing; in the Minimalist Program, there seems to be no clear way to encode that
condition. In this paper we have shown that there are strong empirical reasons to
remove this condition from the theory. Crosslinguistic empirical evidence supporting
the idea that only a subset of objects is formally licensed via Case/agreement with the
verb is in fact very compelling.
38
ACKNOWLEDGMENTS
*
The two authors, listed in alphabetical order, are equally responsible for the entire
content of the paper. We are very thankful to the organizers and audiences of the IV
Nantes Workshop in in Linguistics (JEL-2004), the Université de Paris VIII-Saint
Denis-CNRS research group (UMR-7023) and the III Meeting of the European
Research Network in Linguistics at the University of the Basque Country, and at the
joint colloquia series organized by the Psycholinguistics Lab (Elebilab) and the Group
of Theoretical Linguistics (HiTT) at the University of the Basque Country
(UPV/EHU). Very special thanks to Pablo Albizu, Mark Baker, Hamida Demirdache,
Aritz Irurtzun, Julie Ane Legate, Paco Ordóñez, Milan Rezac, Francesc Roca, Andrés
Saab, Esther Torrego, and Myriam Uribe-Etxebarria, as well as to Jairo Nunes, the
journal’s editor, and two anonymous reviewers for very valuable discussion and
comments. Parts of this material were also presented at a course on Clitics and
Agreement taught at the University of Paris 8-Saint Denis and a course on Agreement
Conditions taught at the University of the Basque Country (UPV/EHU); we are very
thankful to the students participating at that seminar.
This work was financially supported in part by the institutions supporting the
research activities of the Group of Theoretical Linguistics (HiTT): the Basque
Government grant numbers GIC07/144-IT-210-07 and IT769-13 (Euskal
Unibertsitate Sistemako Ikerketa-taldeak), the Spanish Government’s Ministry of
Science and Innovation (MICIIN) grant numbers FFI2008-04786 and FFI201129218, and the University of the Basque Country’s (UPV/EHU) UFI-11/14, as well as
the Basque Government’s Ayudas para la consolidación de Grupos de Investigación,
cara a establecer redes de investigación y cooperación 2009 (Universal Grammar and
Linguistic Variation, ref.: HM-2009-1-1) and the Euro-region Aquitania-Euskadi (La
phrase dans la langue basque et les langues voisines : approche comparative, 2012).
39
REFERENCES.
Albizu, Pablo. 1997. The Syntax of Person Agreement. Los Angeles, University of
Southern California dissertation.
Anagnostopoulou, Elena. 2003. The Syntax of Ditransitives: evidence from clitics,
Berlin & New York: Mouton de Gruyter.
Anderson, Stephen R. 2005. Aspects of the Theory of Clitics, Oxford: Oxford University
Press.
Aoun, Joseph. 1981. On Government, Case marking and Clitic Placement. Manuscript,
Massachusetts Institute of Technology.
Barbosa, Pilar. 2000. Clitics. A Window into the Null Subject Property. In Joao Costa
(ed.), Portuguese Syntax. New comparative studies, Oxford: Oxford University
Press, 31-93.
Barss, Andrew & Howard Lasnik. 1986. A Note on Anaphora and Double Objects.
Linguistic Inquiry 17. 347-353.
Béjar, Susana & Milan Rezac. 2003. Person Lincensing and the Derivation of PCC
Effects. In A.T. Pérez-Leroux & Y. Roberge (eds.), Romance Linguistics: theory
and acquisition, Amsterdam: John Benjamins, 49-62.
Béjar, Susana & Milan Rezac. 2009. Cyclic Agree. Linguistic Inquiry 40. 35-73.
Benveniste, Emile. 1966. Problèmes de Linguistique Générale. Paris: Gallimard.
Bleam, Tonia M. 2000. Leísta Spanish and the Syntax of Clitic Doubling. Newark:
University of Delaware dissertation.
Boeckx, Cedric & Norbert Hornstein. 2005. A Gap in the ECM Paradigm, Linguistic
Inquiry 36. 437-441.
Bonet, Eulalia. 1991. Morphology after Syntax: Pronominal Clitics in
Romance.Cambridge (Massachusetts): Massachusetts Institute of Technology
dissertation.
Borer, Hagit. 1984. Parametric Syntax: case studies in Semitic and Romance languages.
Dordrecht: Foris Pub.
Bošković, Željko. 1997. The Syntax of Nonfinite Complementation: an Economy
approach, Cambridge (Massachusetts): MIT Press.
Bošković, Željko. 2002. A-movement and the EPP. Syntax 5. 167-218.
Bošković, Žejko. 2007. On the Locality and Motivation of Move and Agree: An Even
More Minimal Theory. Linguistic Inquiry 38. 589-644.
Bosque, Ignacio (académico ponente). 2010. Nueva Gramática de la lengua española.
Madrid: Espasa Calpe.
Bresnan, J. & M. Mochambo. 1987. Topic, Pronoun, and Agreement in Chichewa.
Language 63. 741-782.
Campos, Héctor. 1999. Transitividad e intrasitividad. In Ignacio Bosque & Violeta
Demonte (eds.), Gramática descriptiva de la lengua española, voume. 2, 15191574, Madrid: Espasa Calpe.
40
D’Alessandro, Roberta, Susan Fisher & Gunar Hrafn Hrafnbjargarson (eds.). 2008.
Agreement Restrictions. Berlin : Mouton de Gruyter.
DeGraff, Michel. 1994. “The morphology-syntax interface in creolization (and
diachrony)”, Studies in the Linguistic Sciences 24. 115-132.
Demonte, Violeta. 1995. Dative Alternation in Spanish. Probus 7. 5-30.
Embrick, David & Rolf Noyer. 2001. “Movement Operations After Syntax”. Linguistic
Inquiry 32. 555-595.
Fassi-Fehri, Abdelkader. 1981. Theorie Lexicale-Fonctionnelle, controle et accord en
arabe moderne. Arabica 28 . 299-332.
Fassi-Fehri, Abdelkader. 1988. Agreement in Arabic, Binding and Coherence. In
Michael Barlow & Charles A. Ferguson (eds.), Agreement in Natural Language :
approaches, theories, descriptions, 107-158, Stanford (CA): CSLI Publications.
Fernández-Ordoñez, Inés. 1999. Leísmo, laísmo y loísmo. In Ignacio Bosque & Violeta
Demonte (eds.), Gramática Descriptiva de la Lengua Española, vol. I, 1317-1398,
Madrid: Espasa Calpe.
Fernández Soriano, Olga. 1989. Rección y ligamiento en español. Aspectos de
parámetro del sujeto nulo. Madrid: Universidad Autónoma de Madrid dissertation.
Fernández Soriano, Olga. 1993. Los pronombres átonos, Madrid: Taurus.
Fontana, Josep M. 1993. Phrase Structure and the Syntax of Clitics in the History of
Spanish. Philadelphia: Univerity of Pennsylvania dissertation.
Franco, Jon. 1993. On Object Agreement in Spanish. Los Angeles: University of
Southern California dissertation.
Franco, Jon & Errapel Mejías-Bikandi. 1995. The Presuppositionality Condition on
Spanish Clitic Doubling”, in M. Authier, Barbara Bullock & Lisa Reed (eds.),
Formal Perspectives on Romance Linguistics, 107-120, Amsterdam: John Benjamins.
Giusti, Giuliana. 2001. The Birth of a Functional Category: from Latin ILLE to the
Romance article and personal pronouns. In Guglielmo Cinque & Gian Paolo Salvi
(eds.) Current Studies in Italian Syntax. Essays offered to Lorenzo Renzi, 157-171,
Amsterdam: Elsevier.
Gómez, Ricardo eta Luis Sáinz. 1995. “On the Origin of the Finite Forms of the Basque
Verb”, in Jose Ignacio Hualde, Joseba Lakarra & Larry Trask (eds.), Towards a
History of the Basque Language, 235-274, John Benjamins, Philadelphia.
Hernanz, M. Lluïsa. 1999. El infinitivo. In Ignacio Bosque & Violeta Demonte (eds.),
Gramática descriptiva de la lengua española, voume. 2, 2197-2356, Madrid:
Espasa Calpe.
Hooper, Paul & Elizabeth C. Trauggot. 2003. Grammaticalization. Cambridge:
Cambridge University Press.
Jaeggli, Osvaldo. 1982. Topics in Romance Syntax. Dordrecht: Foris Pub.
Karimi, Yadgar. 2010. Unaccusative Transitives and the Person-Case Constraint Effects
in Kurdish. Lingua 120. 693-716.
41
Kayne, Richard. 1975. French Syntax. Cambridge (Massachusetts): MIT Press.
Kayne, Richard. 1991. Romance clitics, verb movement, and pro. Linguistic Inquiry 22.
647-686.
Laca, Brenda. 1995. Sobre el uso del acusativo personal en español. In . Carmen
Pensado (ed.), El acusativo Personal, Visor.
Landa, Alazne. 1995. Conditions on Null Objects in Basque Spanish and Their Relation
to Leísmo and Clitic Doubling. Los Angeles: University of Southern California
dissertation.
Lapesa, Rafael. 1968. Sobre los orígenes y evolución del leísmo, laísmo y loísmo. In
Fistschrift Walther von Wartbourg, 523-551,Tübingen: Max Niemeyer.
Lasnik, Howard. 1995. Last Resort and Attract F. In Proceedings of FLSM 6, 62-81,
[reprinted in Lasnik 1999, Minimalist Analysis, 120-145, Blackwell, Pub.
Cambridge (Massachusetts)].
Lasnik, Howard. 1999. Chains of Arguments. In Samuel D. Epstein & Norbert
Hornstein (eds.) Working Minimalism, MIT Press, 189-215, Cambridge
(Massachusetts).
Lasnik Howard & Mamoru Saito. 1991. On the Subject of Infinitives. In L. Dobrin et al
(eds.) Papers from the 27th Regional Meeting of the Chicago Linguistic Society,
324-343, Chicago: Chicago Linguistic Society [reprinted in Lasnik 1999, Minimalist
Analysis, 7-24, Cambridge (Massachusetts): Blackwell].
Leonetti, Manuel. 2003. Specificity and Object Marking: the case of Spanish a”. In K.
von Heusinger & G. A. Kaiser (eds.) Proceedings of the Workshop “Semantic and
Syntactic Aspects of Specificity in Romance Languages, 67-101, Fachbereich
Sprachwissenschaft: Universität Konstanz.
Leonetti, Manuel. 2008. Specificity in Clitic Doubling and in Differential Object
Marking in Spanish. Probus 20. 33-66.
Levin, Beth. 2008. Dative Verbs. A crosslinguistic perspective. Linguisticae
Investigationes 31. 285-312.
López, Luis. 2008. Locality and the Architecture of Syntactic Dependencies, Palgrave
Macmillan.
Luján, Marta. 1984. Null Arguments and Overt Pronouns in Spanish. Ms. University of
Texas.
Mendikoetxea, Amaya. 1999. Construcciones con se: Medias, pasivas e impersonales.
In Ignacio Bosque & Violeta Demonte (eds.), Gramática Descriptiva de la Lengua
Española, volume II, 1631-1722, Madrid: Espasa Calpe.
McCoone, Kim. 2006. The origins and development of the insular Celtic verbal
complex, Maynooth: National University of Ireland.
Montalbetti, Mario. 1984. After Binding. On the interpretation of pronouns. Cambridge
(Massacusetts): Massachusetts Institute of Technology dissertation.
Nevins, Andrew. 2007. The Representation of third person and Its Consequences for
Person-Case Effects. Natural Language & Linguistic Theory 25. 273-313.
42
Ormazabal, Javier & Juan Romero. 1998. On the syntactic nature of the me-lui and the
Person-Case Constraint. Anuario del Seminario Julio de Urquijo XXXII. 415-434.
Ormazabal, Javier & Juan Romero. 2002. Agreement Restrictions. Manuscript,
University of the Basque Country and University of Alcalá de Henares.
Ormazabal, Javier & Juan Romero. 2007. The Object Agreement Constraint, Natural
Language and Linguistic Theory 25. 315-347.
Ormazabal, Javier & Juan Romero. 2010. The Derivation of Dative Alternations. In
Maia Duguine, Susana Huidobro & Nerea Madariaga (eds.) Argument Structure and
Syntactic Relations from a Crosslinguistic Perspective, 203-232 AmsterdamPhiladelphia: John Benjamis.
Ormazabal, Javier & Juan Romero. 2011. Why P. Manuscript University of the Basque
Country (UPV/EHU) & University of Extremadura.
Ormazabal, Javier & Juan Romero. 2013. Choosing Goals. Manuscript University of the
Basque Country (UPV/EHU) & University of Extremadura.
Pensado, Carmen (ed.) 1995. El complemento directo preposicional. Madrid: Visor.
Perlmutter, David. 1971. Deep and Surface Structure Constraints in Syntax. New York:
Rinehart & Winston Inc.
Preminger, Omer. 2011. Agreement as a Fallible Operation. Cambridge
(Massachusetts): Massachusetts Institute of Technology dissertation.
Picallo, Carme. 2002. Abstract Agreement and Clausal Arguments. Syntax 5. 116-147.
Renzi, Lorenzo. 1997. Fissione di lat ILLE nelle lingue romanze. In J. Günter-Holtus &
W. Sweckhard (eds.) Italica et Romanica. Festschrft für Max Pfister zum 65.
Geburtstag, Tübingen: Niemeyer.
Rezac, Milan. 2010. Phi-features and the Modular Architecture of Language,
Dordrecht: Springer.
Rigau, Gemma. 1988. Strong Pronouns. Linguistic Inquiry 19. 503-511.
Rizzi, Luigi. 1986. Null objects in Italian and the Theory of pro. Linguistic Inquiry
17.3. 501-557.
Roberts, Ian. 2010. Agreement and Head Movement. Clitics, Incorporation, and
Defective Goals. Cambridge (Massachusetts): MIT Press.
Roca, Francesc. 1992. On the Licensing of Pronominal Clitics: the properties of object
clitics in Spanish and Catalan. Unpublished Universitat Autònoma de Barcelona.
Roca, Francesc. 1996. Morfemas objetivos y determinantes: los clíticos del español.
Verba 23. 83-119.
Rodriguez-Mondoñedo, Miguel. 2007. The Syntax of Objects: Agree and Differential
Object Marking. Storrs (Connecticut): The University of Connecticut dissertation.
Roegiest, Eugeen. 1979. “A propos de l’accusatif prepositionnel dans quelques langues
romanes”, Vox Romanica 37. 37-54.
Romero, Juan. 1997. Construcciones de doble objeto y gramática universal. Madrid:
Universidad Autónoma de Madrid dissertation.
43
Romero, Juan. 2012. Accusative Datives in Spanish. To appear in R. Etxepare & B.
Fernández (eds.) Variation in Datives: a microcomparative perspective. Oxford,
Oxford University Press.
Saab, Andrés & Pablo Zdrojewski. 2011. Anti-repair effects under ellipsis: diagnosing
(post-) syntactic clitics in Spanish. Paper presented at the Islands in Contemporary
Linguistic Theory workshop, University of the Basque Country, November 2011.
http://ehutb.ehu.es/es/serial/705.html.
Saltarelli, Mario. 1987. The Acquisition of Agreement. In W. Bahner et al (eds.)
Proceedings of the Fourteenth International Congress of Linguistics, 810-815,
Berlin:, Akademie-Verlag.
Sanchez, Liliana. 2010. La aparente opcionalidad del doblado de clíticos en el español
limeño. Cuadernos de la ALFAL (nueva serie) 1. 94-105.
Schroten, Jan. 2010. Clitic Doubling in Spanish. Agreement of the third kind. In Martin
Everaert, Tom Lentz, Hannah De Mulder, Øystein Nilsen & Arjen Zondervan (eds.),
The Linguistic Enterprise. From knowledge of language to knowledge in linguistics,
315-326, Amsterdam/Philadelphia: John Benjamins.
Sportiche, Dominique. 1993. Clitic Constructions. Unpublished manuscript, University
of California at Los Angeles, Los Angeles (California).
Strozer, J. 1976. Clitics in Spanish. Los Angeles (California):, University of California
at Los Angeles dissertation.
Suñer, Margarita. 1988. The Role of Agreement in Clitic-Doubled Constructions,
Natural Language and Linguistic Theory 6. 391-434.
Torrego, Esther. 1998. The Dependencies of Objects. Cambridge (Massachusetts): MIT
Press.
Uriagereka, Juan. 1988. On Government. Storrs (Connecticut): The University of
Connecticut dissertation.
Uriagereka, Juan. 1995. Aspects of the syntax of clitic placement in Western Romance.
Linguistic Inquiry, 26. 79-124.
Uriagereka, Juan. 1997. Formal and Substantive Elegance in the Minimalist Program. In
Chris Wilder, Hans-Martin Gärtner & Manfred Bierwisch (eds.) The Role of
Economy Principles in Linguistic Theory, 170-204, Berlin: Akademie Verlag.
Walkow, Martin. 2012. Goals, Big and Small. Amherst (Massachusetts): The University
of Massachusetts, Amherst, dissertation. [Open Access:
http://scholarworks.umass.edu/open_access_dissertations]
Zdrojewski, Pablo. 2008. Por quién doblan los clíticos. Univ. Nacional Comahue
Master Thesis.
44
Javier Ormazabal
University of the Basque Country (UPV/EHU)
Basque Group of Theoretical Linguistics (HiTT)
Department of Linguistics & Basque Studies
University of the Basque Country (UPV/EHU)
Unibertsitateen Ibilbidea 5
01006 Vitoria-Gasteiz, Spain
[email protected]
Juan Romero
University of Extremadura (UNEX)
Basque Group of Theoretical Linguistics (HiTT)
Departament of Hispanic Philology
University of Extremadura (UNEX)
Avenida de la Universidad s/n
10071 Cáceres, Spain
[email protected]
45