Professional Documents
Culture Documents
1. Introduction
Arguments with the function of subject or object of predication relate to a
functional projection through the operation Agree, which applies at the
computational component (Chomsky 1998, 1999). Agree consists of feature
matching between two syntactic objects within a given structural space. For
abstract agreement to take place, each of these objects must be endowed with
a set of u-features of the types person, number, and gender (henceforth PNG
features). In the Minimalist Program (Chomsky 1995), functional categories
such as TP and vP are assumed to be headed by a noninterpretable u set of
features. These are checked against the corresponding formal features of the
argument subject and object, respectively.
In Spanish, nominal categories are endowed with Case and grammatical
number, person, and gender. Structural Case on a nominal is realized as
nominative or accusative. Number is expressed as [plural] and gender as
[feminine]. For the purposes of this discussion, I assume that person is
expressed with a subset of the possible combinations of the values [I, II].1
The structural Case assigned to a nominal makes this argument active and
renders possible the operation Agree. The noninterpretable features of a
functional projection and its related argument are deleted under Agree: the
complete u set in TP/vP and structural Case in DP. In Spanish, the
morphophonological effect of the operation is overt subject-verb agreement.
The object is assumed to undergo covert agreement. Thus, given a sentence
like (1), the configuration (2) abstractly represents the mechanisms that are
taking place at the computational component (irrelevant details are omitted).
* This paper has gone through several stages of elaboration. A previous version appeared in
Picallo 2001. For comments and suggestions, I would like to thank I. Bosque, V. Demonte, J.
Elordi, A. Fontich, J. Quer, G. Rigau, the participants of the 2000 LEHIA Workshop (EHUVitoria), and the anonymous reviewers for Syntax. Research was supported by the grants
BFF2000-0403-C02-02 (Ministerio de Ciencia y Tecnologa) and 2001SGR00150,
2000XT00032 (Generalitat de Catalunya).
1
First person can be identified as [+I, II], second person as [I, +II], and third person as
[I, II].
Blackwell Publishers Ltd, 2002. Published by Blackwell Publishers, 108 Cowley Road, Oxford OX4 1JF, UK and
350 Main Street, Malden, MA 02148, USA
(2)
117
Las parejas
bailaban el tango.
the couples.FEM.PL danced.3PL the tango.MASC.SG
The couples were dancing the tango.
TP
DP[+P,+N
+G, NOM]
[T[P,
N, (G)]] vP
[P, N, (G)]]
V DP[+P,+N,
+G, ACC]
a.
b.
c.
118
M. Carme Picallo
(5)
a.
b.
119
Chomsky 1999, that Case-u checking (i.e., the operation Agree) is a syntactic
process that applies independently of movement.
2. The Constituent Structure of Nominalized Clauses
Nominalized clauses (DP-CPs) are structures selected by factive predicates.
In Spanish, they conform to sequences of the types El que IP (lit.: The that
IP), as in (6a), or to sequences consisting of the determiner followed by an
infinitive clause, as in (6b).
(6)
a.
b.
The structure (7) accounts for the obligatory absence of the preposition de
of introducing the clause, which is required with all CP complements of
nouns in Spanish.3 Note that the preposition should appear even if the noun is
phonologically null:
2
See Demonte 1977 (p. 123), Plann 1981, and Iatridou and Embick 1997, who, among many
others, have claimed that nominalized clauses are complexes with a phonologically null nominal
head.
3
The contrast between (i-a), a complex nominal, and (i-b), a nominalized clause, shows that
the preposition introducing CP is obligatory in the first case and ungrammatical in the second
case:
(i)
120
M. Carme Picallo
(8)
a.
b.
Considero
varios hechos independientemente. El [e] de que
the of that
considered.3SG several facts independently
hubieran apoyado tal
propuesta era el mas conspicuo.
had.3PL supported such-a proposal was the most conspicuous
S/he considered several facts independently. The (fact) that they
had supported such a proposal was the most conspicuous one.
Su sugerencia de que deba dimitir no fue aceptada pero
POSS suggestion of that had.3SG to-resign not was accepted but
la [e] de que poda continuar en el cargo fue aprobada
the of that had.3SG to-continue in the post was approved
inmediatamente.
immediately
His/her/their suggestion that s/he had to resign was not accepted,
but the (suggestion) that s/he may continue in the post was
immediately approved.
a.
b.
Lo de ir
a Mallorca este verano no nos convence.
the of to-go to Mallorca this summer not us convince
The (idea/proposal) of going to Mallorca this summer does not
convince us.
Lo de que se
tenga que pagar un impuesto adicional
the of that people have that to-pay a tax
additional
provocara un unanime rechazo.
will-cause an unanimous revolt
The (idea/proposal) that people have to pay an additional tax will
cause a unanimous revolt.
121
which the exchange is taking place. Note that in (9) the determiner is realized
as the neuter lo and not as the masculine form el, unlike the nominalized
clause I am discussing, which corresponds to the structure (7). I would like to
suggest that the so-called neuter Determiner lo in (9) can be analyzed as the
complex [l+o] formed by l, standing for D and o, corresponding to the default
morphological realization of the functional projection Agr (or the projections
[Number, Gender]) immediately dominating the NP with a null head.4
As shown in the following discussion, the fact that nominalized clauses
conform to the agreement patterns of argument CPs, and not that of complex
nominal constructions, with a phonologically realized or a phonologically
null N head, supports the analysis represented in (7).
3. Clausal Arguments and the u-Featureless Hypothesis
Based on the preliminary data provided in (3)(5), let us initially assume that
DP-CPs and argument CPs do not have any specification for u-features and
are not endowed with Case. With respect to this particular aspect of their
formal content, this amounts to considering that argument CPs or DP-CPs are
similar to categories like adverbs, for example, which have no specification
for Case or PNG features. According to this idea, let us examine examples
like (10) where the subject is a DP-CP and the object a CP.
(10) [El que hayas
suspendido este examen] demuestra [que
this exam
shows
that
the that have.2SG failed
estudiaste muy poco]
studied.2SG very little
That you have failed this exam shows that you studied very little.
In these cases, the derivation should crash if T and v enter the numeration
as u-complete functional categories. The abstract hypothetical representation
provided in (11) shows that the noninterpretable u set of the functional
elements cannot relate by agreement with the arguments, which are inert for
this operation because they are Caseless and u-featureless under the present
hypothesis.
4
I am following the lines of the analysis proposed by Bernstein (1993) for Spanish nominals
with a null head and an indefinite determiner, as in this example:
(i)
Busco
uno rojo.
look.1SG.-for a.AGR red.MASC
I am looking for a red one.
The indefinite determiner overtly appears with the morpheme realizing the Agr ([Gender,
Number]) functional projection (i.e., un+o) and contrasts in this respect with the indefinite form
un introducing a phonologically realized N head as in (ii):
(ii)
Busco
un
paquete
rojo.
look.1SG.-for a.AGR parcel.MASC red.MASC
I am looking for a red parcel.
122
M. Carme Picallo
N, (G) ]
] [vP v
[P, N, (G)]
V CP ]]
5
A reviewer suggests the alternative view that Select could be considered a two-step process,
as suggested by Giorgi and Pianesi (1997:9). The first step would consist of selection of items
from the lexicon with only their inherent features. Noninherent features would be added in a
second step as the lexical item is used in the actual derivation. The two-step process would
account for the fact that a same lexical item can have different instantiations in the same
numeration, which are produced by different choices of noninherent features (as in, e.g., el perro
mira los perros the dog is seeing the dogs). I believe that the cases under consideration are
slightly different than having several instantiations of a Noun or a Determiner in a given
numeration. Note that T or v are assumed to enter the numeration with an unvalued set of ufeatures in all cases, unlike N or (arguably) D. The noninterpretable PNG content of T/v is valued
under agreement with an argument. As applied to these functional categories, the two-step
process of Select would have to consist of adding u-features to T or to v as required by the
categorial choice of their related goal, otherwise the derivation would crash. As I see it, the twostep process alternative also implies a look-ahead cost incorporated into the derivation. It also
appears to violate the inclusiveness condition (see Chomsky 1998, 1999) because features have to
be added in the course of the computation.
123
b.
c.
(13) a.
b.
The Spanish expression sus orejas, for example, corresponds to his/her/its/their ears.
I adopt Kaynes (1998) proposal that the morphemes m-/n-, t-/v-, and s- of the Romance
possessives are the morphological realization of specific features for first, second, and third
person, respectively. The m-/n-, t-/v-, and s- forms correspond to the Spanish mi-/nuestr- my/
our[fem, pl], tu-/vuestr- your[fem, pl], and su- his/her/its/their[fem].
7
124
M. Carme Picallo
b.
(15) a.
b.
(16) a.
b.
I suggest that the ungrammaticality of (13b), (15b), and (16b) must be due
to a mismatch in person feature specification between the anaphoric
8
125
CPs and DP-CPs can, however, be the antecedents of pronouns other than
possessives, such as lo, ello, or the null subject pro:
(18) a.
(19) a.
10
Recall that the Spanish possessive su/sus is inherently numberless. The corresponding
morpheme is assigned in concord with the number of its selecting N. Therefore, the possessive
can anaphorically be linked to a singular or a plural antecedent indistinctively. No number
mismatch can take place, as shown in the following examples:
(i)
126
M. Carme Picallo
11
Note that examples in (17) become grammatical if the s- possessive is substituted by the socalled possessive determiner:
(i)
127
4.2 Gender
Grammatical gender is assumed to be a noninterpretable feature that allows
us to assess possible relations, anaphoric or otherwise, in a given
construction. As is known, Spanish nouns lexically belong to the masculine
or to the feminine class ([fem]). Determiners, adjectives, pronouns, and
participial forms are also suffixed for gender morphemes in concord with the
noun they are syntactically related to.
There are several items of the DP-types, such as the demonstrative forms
esto/eso/aquello this/that and some l-pronouns (like the strong form ello it
and the accusative clitic lo it) that have traditionally been classified as
neuter. Although neuter is used as a descriptive label for these elements, I
would like to claim that there is no third neuter gender, complementary to
masculine and feminine, in Spanish. On the contrary, I suggest that that the
Spanish neuter corresponds to the [Gender] specification of a class of
syntactic objects, which include argument CPs and DP-CPs, as well as the socalled neuter demonstratives and pronouns. In this section, this suggestion is
supported by the distribution of some interrogative elements as well as by the
characteristics of constructions involving VP-ellipsis.
Consider first the distribution of the interrogative cual which. I. Bosque
(p.c.) has pointed out that this form is only compatible with [fem] nominals,
demonstratives or pronouns as shown in (21).
(i)
As the discussion proceeds, I intend to show that CPs are not u-featureless. Thus, a referential pro
anaphoric with a CP argument agrees in PNG content with its antecedent and can be in contexts
where verbal agreement appears with default morphology.
Blackwell Publishers Ltd, 2002
128
M. Carme Picallo
(21) a.
b.
c.
d.
Cual escoges,
la pluma o la cartera de piel?
which choose.2SG the pen.FEM or the bag.FEM of leather?
Which one do you choose, the fountain pen or the leather bag?
Cual lamentas mas, el hecho
de PRO haberte
to-have
which regret.2SG more the fact.MASC of
callado
o el hecho
de que pro se haya enfadado?
has been annoyed
remained-silent or the fact.MASC of that
Which (fact) do you regret more, the fact of having remained
silent or the fact that s/he has been annoyed?
En cual confias mas, en el o en ella?
in which trust.2SG more in him or in her
Who do you trust more, him or her?
Cual prefieres, este
o aquel?
which prefer.2SG this.MASC or that.MASC
Which one do you prefer, this one or that one?
d.
e.
f.
129
b.
c.
d.
Additional evidence in favor of the hypothesis that CPs and DP-CPs are
specified as [Gender] is provided by VP-ellipsis data. VP-deletion is only
possible in Spanish in the context of a modal verb. In these constructions, an
anaphor with a sloppy reading is possible in examples of the types (24),
involving a clausal antecedent in (24a) and a neuter demonstrative in (24b).
The deleted string is crossed out within parentheses:
14
The following examples are taken from Seco et al. (1999, I:1340):
(i)
130
M. Carme Picallo
(24) a.
b.
Ellipsis cannot apply in pairs of examples of the types (25a,b) and (25a,c).
In these cases, the first member of the pair is a complex nominal headed by
hecho fact, as shown in (25a), whereas the second member of the pair has
been constructed with a DP-CP, as in (25b), or with an argument CP as in
(25c).
(25) a.
b.
c.
[El hecho
de que Juan no me salude] debo
the fact.MASC of that Juan not me greet must.1SG
lamentarlo. . .
regret-it.MASC
*. . .pero [el que Mara no me haya hablado] no puedo.
but the that Maria not me has spoken not can.1SG
*. . .pero [que Mara no me haya hablado] no puedo.
but that Maria not me has spoken not can.1SG
b.
c.
[El hecho
de que Juan no me salude] debo
the fact.MASC of that Juan not me greet must.1SG
lamentarlo. . .
regret-it.MASC
. . .pero [el que Mara no me haya hablado] no puedo
but the that Maria not me has spoken not can.1SG
lamentarlo.
regret-it.NEUT
. . .pero [que Mara no me haya hablado] no puedo
but that Maria not me has spoken not can.1SG
lamentarlo.
regret-it.NEUT
Note that the anaphoric pronoun corresponds to the clitic form lo in every
case. The pronoun is phonologically null in (24b) and (25b,c) and
phonologically overt in (26). In (24a,b), both the antecedents (a CP and a
demonstrative) and the respective anaphoric pronouns (overt and covert lo)
are neuter elementsthat is, they are [Gender] according to my
Blackwell Publishers Ltd, 2002
131
b.
c.
Si el obispo anunciara
[su dimision],
el
if the bishop would-announce POSS resignation.FEM the
presidente debera
aceptarla. . .
president would-have-to accept-it.FEM
If the bishop would announce his resignation, the president would
have to accept it. . .
. . .pero si el Papa anunciara
[su abdicacion],
el
but if the Pope would.3SG-announce POSS abdication.FEM the
presidente no podra (aceptarla).
president not could (accept-it.fem)
. . .but if the Pope would announce his abdication, the president
would not be able to.
?(??). . .pero si el Papa anunciara
[su
but if the Pope would.3SG-announce POSS
procesamiento],
el presidente no podra (aceptarlo).
impeachment.MASC the president not could (accept-it.masc)
In conclusion, I have shown that argument CPs and DP-CPs behave like
the so-called neuter categories. I have suggested that, formally speaking,
15
Speakers differ with respect to the degree of (un)acceptability that can be attributed to the
pair (27a) and (27c). They agree, however, in considering ungrammatical the pairs (25a) and
(25b), and (25a) and (25c).
16
The deleted elements are also represented as crossed out within parentheses in these
examples.
132
M. Carme Picallo
there is no third gender (the neuter) in Spanish, but the label neuter is best
thought of as corresponding to the expression of [Gender] specification of
some syntactic objects. These include a types of DP-like elements, such as
some demonstratives and a subclass of l-pronouns, as well as clausal
arguments and nominalized clauses.
4.3 Number
Consider the coordinated complex nominals in (28)(30). The head noun in
the second member of the conjunct is usually deleted under identity and is
represented as crossed out. Note that the preposition de of introducing the
CP complement of N remains (see section 2).17 Example (28) shows that
coordination of nominal structures triggers plurality on the verb. In (29), the
coordinated nominals are the antecedents or a pronoun, which must also be in
the plural. Finally, (30) shows that coordinated nominals are able to link a
floating quantifier.
(28) [El hecho de que hubiera desaparecido] junto
con [el hecho
the fact of that had.3SG disappeared together with the fact
de que no tuviera una coartada] lo hicieron sospechoso ante
him made.3PL suspect
to
of that not had.3SG an alibi
la polica.
the police
The fact that he had disappeared together with the (fact) that he didnt
have an alibi made him a police suspect.
(29) Pedro sugirio [la hipotesis de que las oraciones tienen Caso]i
Pedro suggested the hypothesis of that the sentences have Case
y [la hipotesis de que no tienen rasgos u]j, pero nosotros no
not
and the hypothesis of that not have features u but we
lasi+j
asumiremos.
them.FEM assume
Pedro suggested the hypothesis that sentences have Case and the
(hypothesis) that they do not have u-features, but we will not assume
them.
17
All types of nominals with subordinated clauses, arguments as in (i), or adjuncts as in the
examples in the text, behave alike in this respect.
(i)
b.
133
The verbal inflection, the ability to link a floating quantifier, and the plural
features of the anaphoric pronoun indicate that complex nominals are assigned
a [+Number] feature specification. Coordination of these categories allows
them to be assigned grammatical plural. If we now compare the preceding
examples with the following ones with coordinated DP-CPs or coordinated
argument CPs, we can see that they always require the verb in the singular, as
in (31).18 They are also unable to link a floating quantifier, as shown in (32),
or to antecede an anaphoric pronoun in the plural, as shown in (33).
(31) a.
b.
18
Example (31b) is the Spanish counterpart of the coordinated clausal subject constructions
discussed by McCloskey (1991). He shows that preverbal coordinated CPs that express mutually
excluding propositions appear to trigger plurality on the verb in English. The examples in (31)
(33) show that singular inflection on the verb is required in Spanish, independently of whether the
coordinated construction is propositional or factive and also independently of whether it
expresses mutually excluding states of affairs.
134
M. Carme Picallo
(32) a.
b.
(33) Juan lamenta [el que estuvieras en America]i y [el que nunca
Juan regrets the that were.2SG in America and the that never
conocieras a Luisa]j pero yo no loi+j/*losi+j lamento en absoluto.
Luisa but I not it/*them
regret at all
met.2SG
Juan regrets (the fact) that you were in America and (the fact) that you
never met Luisa, but I do not regret it at all.
Coordinated DP-CPs and coordinated CPs can be interpreted as a plurality
of facts or of states of affairs. This interpretation, however, does not surface
in the verbal inflection or result in the ability of these types of coordinated
arguments to link floating quantifiers or antecede anaphoric pronouns in the
plural. In the previous section, I discussed how DP-CPs and argument CPs
behave like so-called neuter items, such as the pronouns ello and lo it and
the demonstratives esto/eso/aquello this/that, and attributed their behavior
to the fact that all are [Gender] items. Neuters also happen to behave like
CPs and DP-CPs with respect to the formal specification for number. They
lack plural morphology,19 require the verb in the singular under coordination,
and are unable to link floating quantifiers as well. The following examples
show the contrasts between neuter demonstratives and pronouns and their
corresponding masculine or feminine counterparts:
19
The demonstrative plurals estos these and esos/aquellos those are masculine, as well as
the pronouns ellos they and los them.
b.
(35) a.
b.
c.
135
Me gustan este
y aquel,
pero no los
to-me like.3PL this.MASC and that.MASC but not them
probare.
will-taste.1SG
I like this one and that one, but I will not taste them.
Me gusta /*-n esto
y aquello, pero no lo/*los
to-me likes/*like this.NEUT and that.NEUT but not it/*them
probare
will-taste.1SG
I like this and that, but I will not taste it/*them.
Esta
y aquella me
parecen ambas/las dos/cada
both/the two/each
this.FEM and that.FEM to-me seem
una exquisita(s).
one exquisite
This and that seem to me both/the two/each one exquisite.
*Esto
y aquello me
parece(n) ambos/los dos/cada
both/the two/each
this.NEUT and that.NEUT to-me seem
uno exquisito(s).
one exquisite
*Ello y lo
otro son ambos/los dos/cada uno
it and the.NEUT other are both/the two/each one
asombrosos.
amazing
It has been observed that there is a correlation between gender and number
feature specification in Spanish, a language that has no plural neuters (see
Ambadiang 1999:4901 and references cited there). Categories that are
grammatically feminine or masculine, namely [+Gender] items, are also
grammatically [+Number], either singular or plural. It appears, however, that
the specification [Gender] prevents the specification [+Number] from being
possible. Suppose there is a correlation (or a hierarchy) between gender and
number, such that all [Gender] syntactic objects are also [Number]. I
cannot offer a hypothesis that can accounts for why an implicational relation
between gender and number specification should exist in Spanish. I can
merely say that the data points out in that direction.20 As far as such
20
A counterexample against the implicational relation that appears to exist between gender
and number is provided by dative clitic pronouns in standard Spanish, which are not inflected for
gender, as shown in (i-a,b), but are inflected for number, as in (i-c):
(i)
a. (A Fernando) le
dije
que esto no me interesaba.
to Fernando l.DAT.SG told.1SG that this not me interested
(Fernando) I told him that this didnt interest me.
b. (A Margarita) le
dije
que esto no me interesaba.
to Margarita l.DAT.SG told.1SG that this not me interested
(Margarita) I told her that this didnt interest me.
136
M. Carme Picallo
correlation seems to exist, it explains why argument CPs and DP-CPs behave
like neuter items and cannot trigger any phenomena related to plurality.21
c. (A Fernando y a Margarita) les
dije
que esto no me interesaba.
to Fernando and to Margarita l.DAT.PL told.1SG that this not me interested
(Fernando and Margarita) I told them that this didnt interest me.
I believe, however, that the counterexample is only apparent. The dative clitic le/les shows
some characteristics similar to those of the interrogative cual which (see section 4.2). I suggest
that le/les has also [+Gender] content but its realization as [fem] is not morphologically
expressed in standard Spanish. Some indication that this may be so is provided by the
phenomenon of lasmo, present in some nonstandard varieties. In the lasta dialects, dative clitics
are overtly inflected for [fem], like accusative clitics, according to the gender of their intended
referent:
(ii)
a. (A Margarita) la
dije
que esto no me interesaba.
to Margarita l.DAT.FEM.SG told.1SG that this not me interested
(Margarita) I told her that this didnt interest me.
b. (A Emma y a Isabel) las
dije
que esto no me interesaba.
to Emma and to Isabel) l.DAT.FEM.PL told.1SG that this not me interested
(Emma and Isabel) I told them that this didnt interest me.
It must also be pointed out that the plural morpheme of a dative clitic doubling an indirect
object is frequently dropped in Peninsular Spanish. The example (iii-a), where the plural mark /-s/
does not appear, corresponds to the normative (iii-b):
(iii) a. Le
voy
a contar esto a mis hijos.
l.DAT.SG am-going to tell
this to my children
b. Les
voy
a contar esto a mis hijos.
this to my children
l.DAT.PL am-going to tell
I am going to tell this to my children.
The dative paradigm shows that this clitic can be morphologically underspecified for gender as
well as number. For discussion on number underspecification, see Fernandez Soriano
(1999:1259) and references cited there.
21
There is one exception in this paradigm. Bello (1847:sect. 829) points out that coordinated
infinitives with the grammatical function of subject of a symmetric predicate can trigger plurality
on the verb, as shown in (i), taken from Bello. I. Bosque (p.c.) points out that the same exception
applies to the so-called neuter demonstratives, as shown in (ii).
(i)
(ii)
Symmetric predicates take collective arguments that necessarily denote sets of entities. The
phenomenon observed in (i) and (ii) appears to be similar to the one found in constructions with
committee-type NPs, which can trigger plural agreement with the verb (see, more recently, Den
Dikken 2001). It must be pointed out, however, that coordinated tensed clauses are unacceptable
with the verb in the plural in the same contexts:
(iii) a. Que
that
b. Que
that
holgazanees
y que aprendas es/??son incompatible/??-s.
you-laze-around and that you-learn is/??are incompatible.SG/??PL
digas esto y que actues de esta forma es/*son contradictorio/*-s.
you-say this and that you-act of that way is/*are contradictory.SG/*PL
I have no insightful explanation to offer for the contrasts between (i), (ii), and (iii). I suggest
that the acceptable cases might be attributed to the DP-like status of demonstratives and to the
quasinominal characteristics of Spanish infinitives, which are in this respect somewhat similar to
English gerund forms (see Hernanz 1999).
Blackwell Publishers Ltd, 2002
137
138
M. Carme Picallo
b.
139
(G)]
. . . CP[P,N,G]
140
M. Carme Picallo
b.
c.
24
2001.
Blackwell Publishers Ltd, 2002
141
25
142
M. Carme Picallo
b.
[Etxean
daudela] iruditu
zait.
house.INES.SG AUX.3PL.C seem.PERF AUX.3SG.ABS, 1SG.DAT
It has seemed to me that they are at home.
[Gaixorik zaudela] ahaztu
zait.
AUX.2SG.C forget.PERF AUX.3SG.ABS, 1SG.DAT
sick
I have forgotten that you are sick. (That you are sick forgotten
is to me.)
Recall once more that argument CPs have been claimed to have the
specification [P, N, G]. This characteristic is expressed on the verbal
inflection by resorting to the default third-person singular morphology. I
suggest that a similar strategy is used in Euskara for the Case feature.
Suppose that argument CPs are endowed with the specification [Case], as I
suggest. The related auxiliary must therefore exhibit absolutive, which is the
unmarked, default, or elsewhere Case in this language.27 Any other
morphological expression of Case is impossible because it manifests a form
of the [+Case] specification, a property of nominals or nominal-like syntactic
objects. Under this idea, consider the following transitive constructions
showing the incompatibility between ergative assignment and subject tensed
clauses. The ungrammatical ergative morphology on the main auxiliary is
shown in boldface in (43).
26
With the exception of (41), the Euskara examples in this section were provided by J. Elordi,
whom I thank for his patient and careful discussion of many data. For a detailed discussion of the
structure of verbal inflection in Euskara, see Laka 1996.
27
Following Schutze (2001), I assume that Universal Grammar may include a notion of
default Case. He suggests that default Case surfaces in nominal elements that do not receive a
Case specification by syntactic means. I propose to extend the notion to include arguments that
cannot be associated with a [+Case] feature.
b.
143
*[Etor zaizeten]
[arduragabekeria handia] erakusten
come AUX.2SG.ABS(Subj) irresponsibility big.ABS prove.IMPERF
du
AUX.3SG.ABS/3SG.ERG
That you have come has proved/shown a big irresponsibility.
*[Jon
etorri
dela]
[arazo
asko]
Jon.ABS come.PERF AUX.3SG.ABS.C problem many.ABS
eragingo ditu.
cause.FUT AUX.3PL.ABS/3SG.ERG
That Jon has come will cause many problems.
b.
[Zuk
Jon
gonbidatzeak]
[arduragabekeria handia]
you.ERG Jon.ABS invite.NOM.ERG.SG irresponsibility big.ABS
erakusten
du.
show.IMPERF AUX.3SG.ABS/3SG.ERG
You inviting Jon has shown/proved a big irresponsibility.
[Mozioa
onartzeak]
[arazo
asko]
motion.DET.ABS approve.NOM.ERG.SG problem many.ABS
eragingo ditu.
cause.FUT AUX.3PL.ABS/3SG.ERG
Approving the motion will cause many problems.
144
M. Carme Picallo
(45) a.
b.
[Zuk
Jon
gonbidatu izanak]
[arduragabekeria
you.ERG Jon.ABS invite.PERF AUX.INF.ERG.SG irresponsibility
handia] erakusten
du.
big.ABS show.IMPERF AUX.3SG.ABS/3SG.ERG
You to have invited Jon has shown/proved a big irresponsibility.
[Mozioa
onartu
izanak]
[arazo
motion.DET.ABS approve.PERF AUX.INF.ERG.SG problem
asko]
eragingo ditu.
many.ABS cause.FUT AUX.3PL.ABS/3SG.ERG
You to have approved the motion will cause many problems.
Finally, consider (46), which has an infinitive subject in the ergative and a
complement tensed clause. The main auxiliary du appears with default
absolutive, for the CP object, and with ergative agreement with the infinitive
(nominal-like) subject. Both appear in boldface in the glosses:
(46) [Zuk
gainditu izanak]
[ikasi
you.ERG pass.PERF AUX.INF.ERG.SG learn.PERF
duzula]
erakusten
du.
AUX.3SG.ABS/2SG.ERG.C show.IMPERF AUX.3SG.ABS, 3SG.ERG
You to have passed has shown/proved that you have studied.
I believe these Euskara examples offer support to the hypothesis that
argument CPs are not Caseless but have, in fact, specification for Case. In
structures like (41), the presence of a Case feature, with the value [Case],
allows ergative assignment to the NP subject. The concrete specification of
this noninterpretable feature has its morphological reflex on the verbal
auxiliary. The [Case] value assigned to argument CPs is expressed by
resorting to the less marked inflection of the Case paradigm (absolutive), the
same way that the interpretable [P, N, G] content of the CP argument is
expressed with the less marked inflection of the u paradigm [III, Sg].
Independently of this interpretation of the Euskara data, another
consideration can be brought up in favor of the proposal that argument
CPs have Case and u content. Assume that [Case] and the u-features [P,
G, N] are properties of Comp. Such a formal content is phonologically
realized in Spanish either as que (tensed CPs), el que (DP-CPs), or a
phonologically null head (infinitive CPs). As suggested by an anonymous
reviewer, the proposal that clausal arguments have a complete formal feature
content in Comp can account for the possibility or the impossibility of
embedded subjects to be the goal of a functional projection in the
superordinate structure. A T or a v in the main clause is able to enter into
an agreement relation with an embedded subject only if a CP is not
selectedthat is, in raising and ECM constructions as in the abstract
representations (47a,b), respectively. If a CP is present, as in (47c), the
agreement relation between the main T/v and the embedded subject is
Blackwell Publishers Ltd, 2002
145
blocked because Comp c-commands the subject and Comp has Case and u
content.28
(47) a.
b.
[TP [T[P,
N, (G)]
[TP DP [vP [v
SEEM [TP
[P, N, (G)]
DP[+P,
+N, +G][+Case]
BELIEVE [TP
to VP]]]]
DP[+P,
+N, +G][+Case]
to VP]]]]
c. [TP XP [vP [v [P, N, (G)] V [CP C[P, N, G][Case] [TP DP[+P, +N, +G][+Case] to VP]]]]]
Summarizing, the proposal that argument CPs are endowed with Case and
PNG features as part of their lexical makeup leads us toward a simpler
system. The need for ad hoc stipulations based on category types is
eliminated because all arguments are alike at the relevant level of abstraction.
In this paper, I have included nonnominal categories within the general
process of abstract agreement. The agreement mechanism proposed in the
minimalist framework can easily be imagined as a species of grammatical
synapse between elements having an interpretable u set of features with a
combination of different values (for nominal and for nonnominal arguments)
and elements endowed with u-receptors (functional categories). The synaptic
contact between syntactic objects in the lexical and in the functional phases
of the clause is possible by the noninterpretable Case feature of the arguments
that activates the process.
References
ABNEY, S. 1987. The English NP in its sentential aspect. Ph.D. dissertation, MIT,
Cambridge, Mass.
AMBADIANG, T. 1999. La flexion nominal: Genero y Numero. In Gramatica
descriptiva de la lengua espanola, ed. I. Bosque & V. Demonte, 48434913.
Madrid: Espasa Calpe.
ANDREWS, A. D. 1982. The representation of Case in Modern Icelandic. In The
mental representation of grammatical relations, ed. J. Bresnan, 427503.
Cambridge, Mass.: MIT Press.
BELLO, A. 1847. Gramatica de la lengua castellana. Madrid: EDAF Ediciones
(1980).
BERNSTEIN, J. 1993. Topics in the syntax of nominal structure across Romance.
Ph.D. dissertation, CUNY, New York.
BRUCART, J. M. 1998. El artculo neutro lo in El foro del espanol en el Centro
Virtual Cervantes. Available at: cvc.cervantes.es/foros.
28
The Phase Impenetrability Condition (PIC) proposed in Chomsky 1999 also rules out the
impossible agreement relation represented in (47c). The PIC states that the domain of a head X
(X, v, or Comp) is not accessible to operations outside its projection XP. Any operation can only
apply to X and to its specifiers or adjuncts.
146
M. Carme Picallo
147
M. Carme Picallo
Universitat Auto`noma de Barcelona
Filologia Catalana, Edifici B
08193 Bellaterra (Barcelona)
Spain
carme.picallo@uab.es