The Syntax of Spanish

KAREN ZAGONA

CAMBRIDGE UNIVERSITY PRESS

Clear and well-organized, this textbook is an introduction to Spanish syntax,

which assumes no prior knowledge of current theory. Beginning with a
descriptive overview of the major characteristics of the grammar, it goes on
to describe facts about Spanish, such as its word order, notions of subject,
direct object, auxiliary verb and so on. The book combines traditional
grammatical description with perspectives gained from recent research in the
Principles and Parameters framework. It also presents useful theoretical
notions such as semantic roles, Case and Predication.
Accessibly written, the book gives just enough background so as to allow
the reader to understand the lines of investigation that have been pursued in
accounting for such issues as clause structure and constituent order. It will be
of use to students who are interested in grammar, Spanish, or in some of the
basic results of modern, formal linguistic theory.
is Associate Professor of Linguistics and adjunct Associate
Professor of Spanish at the University of Washington. She is author of Verb
Phrase Syntax: A Parametric Study of English and Spanish (1988), editor of
Grammatical Theory and Romance Languages (1995), and co-editor, with
Ivonne Bordelois and Heles Contreras, of Generative Studies in Spanish
Syntax (1986).


General editors:
J. Bresnan, D. Lightfoot, I. Robertson, N. V. Smith, N. Vincent
Responding to the increasing interest in comparative syntax, the goal of the
Cambridge Syntax Guides is to make available to all linguists major ndings,
both descriptive and theoretical, which have emerged from the study of particular languages. The series is not committed to working in any particular
framework, but rather seeks to make language-specic research available to
theoreticians and practitioners of all persuasions.
Written by leading gures in the eld, these guides will each include an
overview of the grammatical structures of the language concerned. For the
descriptivist, the books will provide an accessible introduction to the
methods and results of the theoretical literature; for the theoretician, they
will show how constructions that have achieved theoretical notoriety t into
the structure of the language as a whole; for everyone, they will promote
cross-theoretical and cross-linguistic comparison with respect to a welldened body of data.

The Syntax of Spanish

KAREN ZAGONA

PUBLISHED BY CAMBRIDGE UNIVERSITY PRESS (VIRTUAL PUBLISHING)

FOR AND ON BEHALF OF THE PRESS SYNDICATE OF THE UNIVERSITY OF
CAMBRIDGE
The Pitt Building, Trumpington Street, Cambridge CB2 IRP
40 West 20th Street, New York, NY 10011-4211, USA
477 Williamstown Road, Port Melbourne, VIC 3207, Australia
http://www.cambridge.org
Karen Zagona 2002
This edition Karen Zagona 2003
First published in printed format 2002

A catalogue record for the original printed book is available

from the British Library and from the Library of Congress
Original ISBN 0 521 57177 4 hardback
Original ISBN 0 521 57684 9 paperback

ISBN 0 511 01581 X virtual (netLibrary Edition)

For Heles

This page intentionally left blank

Contents
Preface
ix
Symbols and abbreviations
1

Overview of the grammar

1.1
1.2
1.3
1.4
1.5
1.6
1.7

73

Introduction
73
The distribution of argument NPs
73
Predicative NPs
82
The constituents of NP
85
Determiners of argument NPs
94
Pre- and post-determiners
102
Speciers of predicative NPs
107
Constituent order within NP
109
Summary
116

The Verb Phrase

3.1
3.2
3.3
3.4
3.5
3.6
3.7
3.8

Introduction
1
General characteristics of the syntax
The subject constituent
26
The predicate constituent
33
Main clauses
48
Subordinate clauses
54
Syntactic dialects
68

The Noun Phrase

2.1
2.2
2.3
2.4
2.5
2.6
2.7
2.8
2.9

xi

118

Introduction
118
The distribution of VP
118
The external argument of VP
126
Complements of V: prepositional complements vs. adjuncts
Complements: direct object DPs
135
Indirect objects
141
Complements of unaccusative verbs
152
Summary
156

131

vii

viii

Contents

VP-related functional categories

4.1
4.2
4.3
4.4
4.5
4.6

160

Subjects, topics, and declarative constituent order

5.1
5.2
5.3
5.4
5.5
5.6
5.7

158

Introduction
158
VP-adverbs and the verb/tense relation
Auxiliary verbs, tense and aspect
170
Clitics
184
Negation
194
Summary
200

A-movement and Xo movement through COMP

6.1
6.2
6.3
6.4
6.5
6.6

202

Introduction
202
The problem of free subject order
204
Discourse roles: Focus and Topic
208
[Topic] movement to the specier of IP
213
Dislocated Topics
220
Subject order and the NS parameter
229
Summary
239

Introduction
241
Wh-movement
241
(Contrastive) Focus
248
Other A-movements
254
Head movement to (and through) COMP
Summary
271

References
273
Index
281

241

263

Preface
This textbook is intended to present a broad view of Spanish syntax,
one which takes into account the results of recent research, but which does not
focus on theoretical discussion, nor assume familiarity with current theory. In
order to describe insights based on recent research, it is of course necessary to
introduce enough theoretical machinery so that the approaches that have been
explored are understandable. Earlier discussions, especially Chapters 2 and 3,
are framed within the assumptions of the Principles and Parameters framework as developed in Chomsky (1981, 1986). Chapters 4 and 5 introduce some
basic elements of the Minimalist framework of Chomsky (1993, 1995). That
discussion is largely informal, and rather than providing a comprehensive
introduction to the theory, it is intended to give just enough background to
allow the reader to understand the lines of investigation that have been
pursued in accounting for such issues as clause structure and constituent
order.
Chapter 1 presents a descriptive overview of the grammar, combining many
generalizations of a traditional nature with some generalizations that arise
within generative grammar. This description is intended to include both those
generalizations that would be of particular interest to students of Spanish linguistics, and information of a broader nature for readers who are not Spanish
specialists. Chapter 2 focuses on the Noun Phrase (NP). In the course of the
discussion, basic theoretical mechanisms of the Principles and Parameters
framework, such as Theta-role assignment, Case assignment and Predication
are introduced, in order to account for the external distribution of NP. In
examining the internal structure of NP, we introduce the DP-hypothesis, an
important development which has a role in accounting for determiners, and
for NP-internal constituent order. Chapter 3 discusses the Verb Phrase (VP)
from a Principles and Parameters perspective. We begin with the external distribution of the phrase, focusing on Predication and the relationship between
the Verb Phrase and Tense. In considering phrase-internal constituent relations, we return to Theta-role assignment, and introduce the distinction
between external arguments and internal arguments, which, together with
ix

Preface

Case features, determines the mapping of the grammatical subject and complements. The properties of these constituents are summarized, including how
they dier from adjuncts.
Chapters 4 and 5 are concerned with the functional categories associated
with VP. In Chapter 4, we examine the distribution of VP- and IP-adverbs,
auxiliary verbs, clitics and negation. In each of these areas, we introduce
certain empirical generalizations, then consider how the postulation of functional categories may account for them. Chapter 5 is devoted to the issue of
the position of the clausal subject in declaratives, and, more generally, to the
exible order of constituents that is possible in Spanish declaratives. We will
see how the idea that movement is not optional has aected the analysis of
constituent order in a exible constituent order language such as Spanish.
We will also consider the hypothesis, developed in many recent studies, that
the information content of constituents (reected in such notions as
Topic and Focus) is central to the analysis of declarative constituent
order. We will summarize recent analyses, and nish with an overview of the
null subject parameter. Finally, Chapter 6 discusses a variety of constructions whose standard analysis involves the uppermost part of the clause the
Complementizer Phrase and whose derivation involves movement to a nonargument position such as the Specier of the Complementizer Phrase. This
chapter is primarily descriptive, as it discusses the constructions in Spanish
which seem to have the properties of this type of movement.

Symbols and abbreviations

*
?
*()
(* )
Vo
Po
Ao
No
Do
Co
VP
PP
AP
NP
DP
CP

#
m.
f.
neu.
1st.
2nd.
3rd.
sg.
pl.
Re.
CL

ungrammatical sequence
not fully grammatical
ungrammatical in the absence of the parenthesized material
ungrammatical in the presence of the parenthesized material
verb
preposition
adjective
noun
determiner
complementizer (subordinating conjunction)
Verb Phrase
Prepositional Phrase
Adjective Phrase
Noun Phrase
Determiner Phrase (a Noun Phrase introduced by a determiner,
e.g., [the red car] is a DP)
Complementizer Phrase (subordinate clause)
an orthographic symbol which accompanies ? to mark
interrogatives
plus/minus: either value for a given feature
intonational juncture (pause)
masculine
feminine
neuter
rst person
second person
third person
singular
plural
Reexive
clitic (unstressed pronoun form)
xi

xii

Symbols and abbreviations

Det
IO
DO
inc.
Nom.
Gen.
Dat.
Acc.
Obl.
PA
inf.
fut.
cond.
pr.
pa.
imp.
I
pret.
ind.
subj.
pas.
prt.
pprt.

determiner
indirect object
direct object
inchoative
Nominative
Genitive
Dative
Accusative
Oblique
Personal (accusative) a
innitive
future
conditional
present tense
past tense
past imperfect indicative
imperative
past preterite indicative
indicative mood
subjunctive mood
passive voice
present participle
past participle

Overview of the grammar

1.1

Introduction

1.1.1

The extent of Modern Spanish

Modern Spanish is spoken by just under 300 million people worldwide, and is thus one of the three or four most widely spoken languages, after
Mandarin Chinese, English and possibly Hindi.1 Spanish is the primary or
ocial language in numerous countries, including Spain and its dependencies,
Equatorial Guinea, eighteen countries of Central and South America, and the
US protectorate of Puerto Rico.2 Spanish is robust as a rst or second language in many areas of the southwestern United States, as well as in other agricultural areas of the US, and urban areas such as Miami and New York.
According to the 1990 census, about 17.3 million people over the age of ve
speak Spanish at home in the US.
Many countries in which Spanish is the ocial or primary language are linguistically diverse, with bilingualism a common, but not universal, phenomenon. In the north of Spain, primary languages include Basque, Catalan and
Galician.3 In Latin America, many indigenous languages are used alongside
Spanish. In Bolivia, for example, at least half the population speaks either
Aymara or Quechua natively, and it is estimated that 40% of these speakers
1

Mandarin has well over 700 million speakers, English over 400 million. Estimates for
Spanish speakers range from 266 million (Bright 1992) to 290 million (Green 1992),
and estimates for Hindi range from 182 million (Bright 1992) to 290 million (Dcsy
1986).
Spanish is the ocial language of most countries of Latin America. In Peru, both
Spanish and Quechua are ocial languages. In Bolivia, Spanish, Quechua and
Aymara are all ocial languages.
Although Spanish is the ocial language of Equatorial Guinea, it is estimated that
only 45% of the population speaks Spanish (Kurian 1992:600).
Galician or Gallego is considered more closely related to Portuguese than to Spanish.
Catalan is more closely related to Occitan than to Spanish. Basque is a linguistic
isolate.
1

The syntax of Spanish

do not speak Spanish (Grimes 1988:8587; Kurian 1992:184). In Paraguay,

Guaran is spoken by over 3 million speakers, with a majority of rural speakers being monolingual (Grimes 1988:125). Relatively large populations of
speakers of indigenous languages are also found in Peru (Ayacucho Quechua
and Cuzco Quechua), Guatemala (Mayan languages) and Ecuador
(Quichua). Many other indigenous languages are spoken, by populations
numbering from dozens of speakers to tens of thousands. Relatively small
populations speak Creole languages in Honduras, Nicaragua, Costa Rica, the
Dominican Republic and Panama.4 English is growing as a second language
in some parts of the Caribbean, such as the Dominican Republic and Puerto
Rico, in northern Mexico, and in urban areas elsewhere in Latin America.
Dialects of Modern Spanish on the Iberian peninsula include Castilian, the
northern dialect families of Navarro-Aragonese, Leonese and Asturian, and
the southern, Andaluz dialects.5,6 Ladino or Judeo-espaol is a dialect of
Spanish spoken by Sephardic Jews expelled from Spain at the end of the fteenth century. It is a fossil dialect in that it retains characteristics of the pronunciation of that time. In Latin America, the problem of dening dialect
boundaries is a complex one.7 The grammar is dierentiated along phonological, morphological, syntactic and lexical lines, but the degree of variation
makes classifying discrete dialect boundaries extremely dicult. Latin
America is more conveniently described in terms of dialect areas which are
associated loosely with general linguistic patterns. These include such areas as
the River Plate region of Uruguay and Argentina, the Andean highlands, and
the Caribbean. Section 1.7 below summarizes general patterns of syntactic
variation in these areas.

1.1.2

The spread of the Castilian dialect

Although Spanish is spoken over an extremely broad geographical

expanse, it is nevertheless relatively uniform syntactically. This is due in part
4

5
6

In Belize, 2540% of the population is Spanish-speaking, and most of the population speaks an English-based Creole (Kriol). The ocial language of Belize is
English. Statistics on the occurrence of Creoles are based on Grimes (1988) and
Kurian (1992).
For detailed discussion of Iberian dialects see Alvar (1996), Otero (1971).
Among Andaluz dialects, which are characterized by weakening of word-nal -s,
there are areas in which nal -s appears to be disappearing. This (eventually) may
have syntactic consequences with respect to the richness of features for number
and person, since -s distinguishes plurality in nominals and distinguishes 2nd person
in verbal paradigms.
For detailed discussion of the problem of classication of Latin American dialects
see Lipski (1994).

Overview of the grammar

to the early political unication of Spain, and to the spread of the Castilian
dialect throughout the unied area. This unication was a consequence of the
drive to re-conquer the peninsula after its occupation by the Moors in the
early eighth century. The area from which the reconquest was launched was
Castilla la Vieja (Old Castille). In the course of the centuries-long battle
against the Moors, the Castilian dialect spread throughout much of modern
Spain. Castilian thereby coexisted with other Spanish dialects that had
evolved in various areas, and largely replaced them over the course of time.
Most of Iberia had been Romanized during the period of the expansion of
spoken Latin.8 With the decline of Rome, the peninsula was invaded by successive waves of Germanic tribes, and eventually came under the control of
Visigothic kingdoms during the fth to eighth centuries. This period marks a
transition during which spoken Latin was initially similar enough to the
written form of Classical Latin to remain viable for administrative purposes.9
Meanwhile the increasing political weakness of the Visigothic kingdoms and
the beginnings of feudalism accelerated the growth of local Romance varieties. This was especially characteristic of northern and northwestern Iberia,
where Romanization was never extensive, urbanization was minimal, and
Romance coexisted with Basque, and perhaps other indigenous languages.
With the Moorish conquest, Iberia was for a time severed from the rest of
Europe, where emerging monasteries provided a linguistic and cultural
counterweight to feudal isolation. Throughout much of Iberia, Mozrabe10
became the standard form of Romance. The mountainous north, however,
which the Moors never successfully settled, retained its dialect diversity
(Alatorre 1989:108). As Moorish control of the peninsula receded, the north
and northwest became Christian strongholds with renewed ties to the rest of
Europe. Santiago de Compostela was an important destination for Christians
from throughout Europe, and monasteries and cathedrals emerged. At the
18

19

10

Although spoken Latin was in use and undergoing evolution from much earlier
times, the period of its great geographic expansion might be taken to begin around
100 BC, when Latin replaced Oscan as the ocial language of central Italy, to AD
200, when the empire reached its broadest expanse. Although Romanization of the
Hispanic peninsula began earlier with the Second Punic War, the legionnaires (and
colonizers) of this period were perhaps not predominantly Latin speakers. Lapesa
(1981:94101) notes that signicant numbers may have been speakers of the
OscanUmbrian subfamily of Italic, which was spoken in southern regions of Italy.
The question of whether speakers considered their spoken and written languages to
be one and the same has been debated in recent studies. For discussion and references see Wright (1991).
The term Mozrabe refers either to Christians who lived in Moorish-controlled
Spain, or to the variety of Spanish spoken by Christians (and non-Christians). See
Galms de Fuentes (1996).

The syntax of Spanish

southern periphery of Asturias (the then kingdom of Oviedo), a relatively

unpopulated area known previously as Bardulia (Alvar 1994:81) had been
newly settled and fortied with castiellas against Moslem raids. By the ninth
century the area was known as the place of the castles, or Castille.
According to Lloyd (1987:177), Castille was populated by settlers from dierent areas, who abandoned peculiar features of pronunciation associated with
their origins. Castille was also an area where Basque was spoken, and some
features of Spanish, such as initial f>h have been attributed to Basque inuence.
Over the subsequent centuries, Castille became a dominant power in the
north, and was the center from which the reconquest of the peninsula was
launched. Although Castilian was not a prestige dialect, it gradually spread
southward and became dominant as Spain was politically unied and
Christianized.11 The religious zealotry which followed the reconquest included
linguistic purication, as Arabic books were burned in Granada, and the use
of Arabic (and even Arabic borrowings) was increasingly condemned
throughout the sixteenth century. Between 1609 and 1614, as many as 300,000
moriscos (non-assimilated or partially assimilated Moors and their descendants) were expelled from Spain.
The form of the language that took root in Latin America was aected by
a number of unifying inuences. One of these was the social climate of conformity including linguistic conformity which held sway in Spain at the
time of colonization. This tendency was made concrete policy with respect to
colonization, as the monarchs prohibited emigration of Jews and Moors to
the new world (Snchez-Albornoz 1984:15). Another factor that minimized
diversity during the era of colonization was the relatively short time frame
during which much of the settlement occurred. Immigration was most extensive before 1650, and dropped o sharply by the 1700s.12 Colonization also
coincided with the introduction of the printing press, the rst of which was
brought to Mexico City by the 1530s (Alatorre 1989:138). Subsequent inuences, such as ongoing commerce with Spain, the independence movements,
bilingualism and the growth of mass media, have resulted in a rich range of
11

12

Lloyd (1987:179180) suggests that reverse prestige may have enhanced the spread
of Castilian, given the role of Castille in the liberation of the peninsula from the
Moors. An additional factor in the spread of Castilian was migration. An economic
breakdown in the north triggered signicant migration from northern Castille to the
south during the sixteenth century, which reinforced the spread of Castilian.
Snchez-Albornoz (1984:1516) estimates that from 200,000 to 243,000 people
immigrated during the sixteenth century, and an almost equal number during the
rst half of the seventeenth century. The extent of immigration is small overall,
compared with immigration to the United States from other countries.

Overview of the grammar

phonological and morphological variations in the grammar, but less variation

in the syntax.

1.1.3

The evolution of Spanish syntax

The evolution of spoken Latin into proto-Romance was characterized from early on by simplication of inectional paradigms for nouns, adjectives and verbs, and emergence or broader use of periphrastic constructions
which fullled some of the same grammatical functions. The nominal case
paradigms were reduced to a Nominative/Accusative distinction, and prepositions emerged as markers of other cases. Denite and indenite articles
evolved (from Latin demonstrative ille that and the cardinal unum one,
respectively). Periphrastic comparative forms of adjectives replaced synthetic
forms. In the verbal paradigms, simplication of Classical inections included
the loss of the future tense, of synthetic passives, and of diverse non-nite
forms. Many of these changes were incipient or well underway in spoken
Latin, and some were accelerated as a result of phonological changes such as
loss of many word-nal consonants and loss of distinctive vowel quantity. The
most stable inectional features were person, number and masculine/feminine
gender markers, and the [] inection for verbs.
The break-up of proto-Romance into the early dierentiated Romance
languages is generally dated from the point at which written Latin was no
longer comprehensible to the Romance speaker, roughly between the fth and
ninth centuries.13 Characteristics of early Spanish are deduced from documents dating from the eleventh century. Grammatical changes during this
period continued those trends described above: inectional simplication and
grammaticalization of functional and quasi-functional morphemes; in many
instances these changes were common across languages. For example, nouns
lost their Nominative/Accusative distinction. In western varieties of
Romance, accusative plural -s was reanalyzed as a plural marker. Object pronouns were de-stressed and became clitics. Verbal auxiliaries evolved in passives, compound perfect, future and conditional tenses. The clitic se (Latin
3rd.sg./pl. Re.) was grammaticalized, rst as a detransitive (anti-causative)
morpheme, then as a marker of middles, and (in Spanish) as a marker of
passive voice (Hanssen 1945:230231).
13

Because classical Latin was used as a written form under the Visigothic administrations, it is more dicult to date the transition from proto-Romance to Romance in
the Iberian peninsula than elsewhere. In France, by contrast, translations began
to occur in 813 (cf. note 8; see also Palmer 1954:178179). Only in the eleventh
century did Carolingian writing replace the Visigothic system (Lapesa 1981:169).

The syntax of Spanish

One syntactic innovation from this period is the emergence in Spanish of

the personal a, a marker of specic, human direct objects. Personal a
occurred most consistently at rst with proper names and pronouns, less consistently with common nouns (Lapesa 1981:213). Torrego (1998:42; citing
Lapesa 1968) mentions an additional factor which governed the distribution
of personal a around the thirteenth and fourteenth centuries. A appeared with
the complements of verbs that denote an action that aects an individual
physically or psychologically. Only later did it occur with non-aected animate
direct objects.
The constituent order of Old Spanish diers from that of Modern Spanish
in several respects. In Old Spanish, only phrases headed by closed class items
(such as articles, complementizers and prepositions) were clearly head-initial.
Lexical, or open class, heads of phrases (nouns, adjectives and verbs)
allowed both complementhead and headcomplement order. The basic order
of the verb and its objects is analyzed as having switched from OV to VO order
(Otero 1975; Saltarelli 1994). It is interesting to note that auxiliarymain verb
complexes gradually evolved from verbauxiliary to auxiliaryverb (Rivero
1993; Lapesa 1981:217; Hanssen 1945:249, 251). The constituents of clauses
also patterned dierently in Old Spanish. Fontana (1993) argues that Old
Spanish is a V2 (verb second) language, not of the German type (which exhibits second-position verbs in main clauses only), but of the Icelandic type: with
verbs occupying second position in subordinate clauses also. Fontana terms
this symmetric V2.
Another dierence between Old Spanish and Modern Spanish concerns the
behavior and the placement of pronominal clitics. Modern Spanish clitics
attach only to verbs, and either precede or follow the verb according to
whether the verb is nite or non-nite. Old Spanish pronominal clitics occupied second position in the clause, and were phonologically dependent on the
preceding constituent whether that constituent was a verb or not. This is
shown by the fact that they could not occur clause-initially following a pause.14
In this respect, the pronominal clitics behaved like other atonic elements,
including non not, conjunctions and some auxiliaries. Auxiliaries mostly lost
this restriction during the period of Old Spanish (cf. Hanssen 1945:251252).
Old Spanish displayed auxiliary switch, similar to that of Modern French
and Italian (Vincent 1982). Auxiliary ser to be alternated with aver to
have in the compound perfect tenses. In these tenses, ser was generally used
with unaccusatives and reexive (anticausative) intransitives, and aver with
14

For detailed discussion of the syntax of Old Spanish clitics see Rivero (1986, 1991),
Wanner (1987), Fontana (1993).

Overview of the grammar

transitives (Lapesa 1981:212; Hanssen 1945:230233). The compound perfect

tense also displayed past participle agreement with the object. However, both
auxiliary switch and past participle agreement were inconsistent.15

1.2

General characteristics of the syntax

Many characteristics of Spanish syntax are typical of the IndoEuropean family, including the relative richness of verbal morphology compared with nominal morphology, and the overt movement of interrogative
phrases and of noun phrases (e.g., in passives). Other characteristics are prevalent within the Romance family. These include head-initial constituent order,
pronominal clitics, negative concord, rich agreement morphology and null
subject phenomena. Two characteristics of Spanish which are relatively isolated within Romance include the so-called personal a which precedes
animate direct objects under certain conditions,16 and clitic doubling of
indirect objects (and dialectally, direct objects). This section summarizes features of Spanish syntax which place the language typologically, and which
provide an introduction for subsequent discussion.

1.2.1

Constituent order

Modern Spanish is a head-initial language. As shown in (1), the construction of a phrasal head, or Xo with a complement, gives the order: headcomplement. Thus, nouns, adjectives, verbs and prepositions precede their
complements. Examples are in (2):
(1)

X
(2)

15

16

YP

a. construyeron un puente
built
a bridge
(they) built a bridge

[Vo NP]

Lapesa (1981:212) notes the inconsistent usage of ser and of past participle agreement, and notes that contradictory uses due to foreign inuences were not uncommon.
Lapesa (1981:94101) observes that the use of personal a is one of several grammatical features which Iberian dialects share with Sicilian and other southern Italian
varieties.

The syntax of Spanish

[Po NP]

b. con un martillo
with a hammer
c. estudiante de fsica
student
of physics
d. leal a los ideales
loyal to the ideals

[No PP]
[Ao PP]

Functional categories also precede the lexical categories which they govern,
for example determiners precede noun phrases, and complementizers precede
clauses. Auxiliary verbs, which might be considered functional or quasifunctional items, also precede the main verb of the clause:
(3)

a. Habamos hablado del

problema.
had
spoken of+the problem
(We) had spoken about the problem.
b. *Hablado habamos del problema.

The order of adjuncts, or optional modifying phrases, relative to the head

varies according to several factors. All of the positions in (4) are possible with
normal (unbroken) intonation:
(4)

[ (adjunct) head (adjunct) complement (adjunct) ]

Structurally complex adjuncts typically follow the head and its complements.
Several factors condition the availability of pre-head adjuncts, including
structural and lexical properties of the adjunct as well as the category of the
head. Adjunct order is discussed in relation to the Noun Phrase (Chapter 2),
the Verb Phrase (Chapter 4) and the clause (Chapter 5). The order of subjects
is addressed below (1.3.) and in Chapter 5.

1.2.2

Case

Spanish has a Nominative/Accusative case system. Case is not manifested morphologically on lexical nouns or determiners; only personal pronouns and some relative pronouns retain vestiges of Latin case distinctions.
The strong (i.e., tonic, or stressed) personal pronouns display morphologically
distinct forms to the extent shown in (5), illustrated with the 1st person singular form:
(5)

17

a. Nominative:
b. Objective:
c. Genitive:

yo
m
m(o/ a (s))
(m./f.(pl.))

I
me
my17

The strong forms of possessive pronouns agree in number and gender with the
modied noun.

Overview of the grammar

Objective Case in (5) is the form common to objects of prepositions. The weak
pronouns (Section 1.2.4) may have dierent form and distribution depending on
whether the object is direct or indirect. These dierences lead to subclasses of
Objective: (a) Accusative (direct object of Vo), (b) Dative (indirect object of Vo)
and (c) Oblique (object of Po). The following discussion will briey summarize
the contexts for Nominative, Genitive and the three subcases of Objective case.
Nominative is the case of subjects of nite clauses, both indicative and subjunctive; of predicative NPs linked to the clausal subject; and of subjects of
participial and innitival adjunct clauses. The example in (6) illustrates that
pronominal subjects of both indicative and subjunctive clauses appear in
Nominative form:
(6)

Insisto
yo en que lo hagas
t.
Insist-pr.ind.1st.sg. I on that it do-pr.subj.2nd.sg. you
I insist that you do it.

Predicative NPs with Nominative form are shown in (7):

(7)

a. El campen eres t.
The champion is you(Nom.).
b. Lo que encontraron era yo.
What (they) found was I(Nom.).

In (7), the verb agrees in person and number with the predicative pronoun (cf.
English It is/*am I).
Adjunct clauses with Nominative subjects are shown in (8):
(8)

a. [Llegada ella]
empez la esta.
arrived-f. she(Nom.) began the party
(With) her arrived, the party began.
b. [Habiendo llegado
ella],
empez
la esta.
have-prt. arrive-pprt. she(Nom.) begin-pret. the party
With her having arrived, the party began.
c. [Al
cantarlo t],
empez la esta
upon+the sing-inf+it you(Nom.) began the party
Upon your singing it, the party began.
d. [De ganar ellos]
los
felicitaremos.
of win-inf. they (Nom.) CL(DO) congratulate-fut.1st.pl.
If they win, we will congratulate them.

In the above constructions, the participle or innitive must precede the

subject, but some dialect variation occurs (see 1.7). The participial clause in
(8a) shows number and gender agreement with the subject; the participial
clause (8b) and innitives (8c), (8d) are non-agreeing forms.18
18

Rigau (1992) shows that constructions like (8c), which appear to be nominalized,
are in fact clausal.

10

The syntax of Spanish

Genitive is the case assumed by the subject of a noun phrase, and is marked
either by the preposition de with a non-pronominal, as in (9), or by the
Genitive form of a pronominal, as in (10). Genitive pronominals have both
weak (pre-nominal) and strong (post-nominal) forms, illustrated in (10a) and
(10b) respectively:
(9)

el retrato de Josena
the portrait of J.
Josenas portrait

(10)

a. mis
libros
my-pl. book-m.pl.
my books
b. los
libros
mos
the-m.pl. book-m.pl. my-m.pl.
my books

In (9), the de-phrase is ambiguous between possessor, agent, and subject of the
portrait. This illustrates that Genitives are not necessarily possessors, and also
that de is not exclusively Genitive. The examples in (10) illustrate that Genitive
pronominals agree in number (and gender) with the possessed noun. In contrast with Italian, determiners do not co-occur with a pre-nominal possessive
(*los mos libros the my books) in most dialects of Spanish. In contrast with
English, double genitives of the form a book of his (*un libro de suyo) do
not occur. Post-nominal genitives show either de, as in (9), or genitive
morphology, as in (10b).
Relative pronouns display a distinguishable Genitive form, although interrogatives do not. This is illustrated by the contrast between the relative
pronoun in (11a) and the interrogatives in (11b, c):19
19

Interrogative forms do not show case distinctions in general. Qu what, and quin
who, for example, serve as both Nominative and Accusative arguments:
(i)

(ii)

a. Qu
pesa
7 kilos?
What(Nom.) weigh-pr.3rd.sg. 7 kilos
What weighs 7 kilos?
b. Qu
dijo Susana?
what(Acc.) said Susana?
What did Susana say?
a. Quin trabaja aqu?
Who works here?
b. (A) quin
buscan?
PA who(Acc.) look-for
Who are they looking for?

The case of non-Nominative interrogatives is marked by prepositions, including personal a, as in (iib).

Overview of the grammar

(11)

11

a. la persona cuyo coche se

vender
the person whose car
CL(pas.) sell-fut.
the person whose car will be sold
b. *Cuyo coche se
vender?
Whose car
CL(pas.) sell-fut.
Whose car will be sold?
c. Se vender el coche de quin?
pas. sell-fut. the car
of who
(Lit.) Will be sold [the car of who]?
Whose car will be sold?

As shown in (11b), the genitive pronoun cuyo is not possible as an interrogative form. In (11c), the interrogative phrase de quin, which remains in-situ
not moved to the beginning of the clause is marked as Genitive by de rather
than by the form of the pronoun.
Genitives do not occur as the subject of nominalized clauses corresponding
to English gerunds. Nominative subjects are possible instead:
(12)

a. [El hacer eso t]

sera
buena idea.
the do-inf. that you(Nom.) be-cond.3rd.sg. good idea
For you to do that would be a good idea.
b. *[Su
hacer eso ] sera
buena idea.
your(Gen.) do-inf. that be-cond.3rd.sg. good idea
Your doing that would be a good idea.

The three types of Objective case are distinguished on the basis of whether
or not they co-occur with clitic pronouns, and, if so, the form which the clitic
takes. Oblique case occurs as the complement of most prepositions,20 both in
prepositional phrases which are adjuncts and those which are prepositional

20

The prepositions hasta even, como like/as and entre between govern
Nominative:
(i)

a. Todos bailaron
en la esta, hasta yo/*m.
all
dance-pa.3rd.pl. at the party, even I/*me
Everyone danced at the party, even I (did).
b. Nadie baila
como yo/*m.
nobody dance-pr.3rd.sg. like I/*me
Nobody dances like I (do).
c. Entre
t y
yo/* m,
between you and I/*me

Depending on its environment, the preposition a can mark Accusative, Dative or

Oblique (al medioda at noon). Likewise, de can mark Genitive or Oblique: un
amigo de Madrid a friend from Madrid.

12

The syntax of Spanish

complements of verbs.21 Oblique complements of a verb do not admit weak

(clitic) forms of pronouns, as shown in (13):22
(13)

a. Hablaron
[de Juan/l].
spoke-3rd.pl. of Juan/him
They talked about Juan/him.
b. *Le
hablaron.
CL(3rd.sg.) spoke-3rd.pl.
Lit.: (They) him-spoke
They talked about him.

Non-oblique objects of verbs do accept (or require) clitic forms.

Accusatives require a clitic when the object is anaphoric or pronominal. For
example, compare the reexive objects in (14a), (14b):
(14)

a. *(Me)
vi
a m misma.
*CL(1st.sg.Acc.) saw-1st.sg. PA my self
I saw *(CL) myself.
b. (*Me)
habl de
m misma.
CL(1st.sg.Acc.) spoke about my self
I talked *(CL) about myself.

As shown in (14a), a reexive direct object requires a clitic double. Oblique

reexives, as in (14b), disallow a clitic double.
An additional characteristic of Accusative case is that Accusative phrases
21

22

As noted above, Oblique, Accusative and Dative strong pronouns all have the same
morphological form. One exception is the forms occurring with the preposition con
with: conmigo, with me; contigo, with you; consigo with him/her/you
(formal). These forms consist of con+pronoun+go; -go is the residue of Latin cum
with.
A possible instance of oblique clitic-doubling with certain motion verbs is shown in
(ic):
(i)

a. Mara fue al
parque.
M.
went to+the park.
Maria went to the park.
b. Mara se
fue.
M.
CL(3rd.sg.) went
Maria went away.
c. Mara se
fue al
parque.
M.
CL(3rd.sg.) went to+the park
Maria went o to the park.

Both PPs and clitics can express the Goal of the verb. In (ic), both are present. Other
verbs which behave similarly are escaparse to escape and fugarse to ee (which
allow both Goal and Source PPs). Unlike standard cases of clitic doubling, the clitic
in (ic) agrees in person and number with the subject of the clause, not with the NP
in the prepositional phrase.

Overview of the grammar

13

are, under certain conditions, marked by a particular morpheme, often

referred to as Personal a (PA). Personal a is, supercially, a preposition,
identical to the preposition a which marks Dative case.23 Personal a occurs primarily when the direct object is [+ ] and [+], as illustrated in
(15) and (16). Compare (15a), with a [+ ] object, with (15b), with an
inanimate object. The contrast between specic and non-specic objects is
shown in (16):
(15)

a. En el mercado vi
*(a) los vecinos.
at the market saw-1st.sg. PA the neighbors
At the market (I) saw the neighbors.
b. En el escritorio vi
(*a) los papeles
on the desk
saw-1st.sg. PA the papers
On the desk (I) saw the papers.

(16)

a. (Yo) busco
a una secretaria.
(I) look for PA a
secretary
I am looking for a (specic) secretary.
b. (Yo) busco
una secretaria.
(I) look for a
secretary
I am looking for a (non-specic) secretary.

The contrast between (16a) and (16b) concerns whether a specic individual
is sought (16a), or whether anyone who happens to be a secretary is sought
(16b). Personal a may be used also with non-human animate direct objects, if
the object is interpreted as specic and individual (as with pets, for example),
or is in some manner personied.24 Personal a also occurs with inanimate
direct objects (from Hanssen 1945:296):
(17)

a. El adjetivo modica al (=a+el) sustantivo.

the adjective modies PA the
noun
The adjective modies the noun.
b. Y a eso llamaban
libertad?
and PA that call-pa.3rd.pl. liberty
And they called that liberty?

In (17a), both subject and object are inanimate; in (17b) the verb llamar call
selects a nominal small clause complement in eect a double Accusative,
23

24

Torrego (1998) argues that the choice of morpheme is not arbitrary. She observes
that languages as diverse as Spanish and Hindi mark Accusatives using a morpheme
that otherwise marks Dative Case.
Hanssen (1945:296) gives the examples Llam a la muerte and Llam la muerte
S/He called out to death, which dier only in the presence or absence of personal
a. The dierence in interpretation might be described in terms of whether one calls
out to an abstraction (perhaps an event), or personies the abstraction.

14

The syntax of Spanish

where both subject and predicate nominal of the small clause are inanimate.25
Zubizarreta (1994) notes that the distribution of personal a in cases such as
these suggests that a is not so much a marker of [+ ] as it is a direct
object marker in constructions in which two arguments are animate or two
arguments are inanimate. That is, when animacy dierences do not independently do so, a identies a unique Accusative argument. It is otherwise dicult to account for examples such as (17).
Turning now to Datives, these indirect objects of verbs occur in the context
of the preposition a, and may co-occur with a clitic double, even if the argument is non-pronominal, non-anaphoric.
(18)

Juan le
mand un paquete a
Jos.
Juan CL(Dat.) sent
a package to(Dat.) Jos
Juan sent a package to Jos.

The preposition a in (18) marks Dative case; its presence is not contingent on
any particular features of the argument, such as animacy (cf.: Le mand el formulario al departamento I sent the form to the department). The dative clitic
(le) in (18) is often characterized as required. However, Demonte (1995)
observes that there are conditions which favor omission of the clitic. In particular, the clitic is, for many speakers, omissible if the transfer expressed by
the verb is not asserted to have been completed.26

1.2.3

Inectional morphology

The major types of axal inections in Spanish, and the types of elements which can exhibit them are summarized in (19):
(19)

25

26

a. NUMBER, GENDER:
amigo
amiga
amigos
amigas
friend(m.sg.)
friend(f.sg.)
friend(m.pl.)
friend(f.pl)
(nouns, demonstratives, denite and indenite determiners,
quantiers, personal pronouns [strong and clitic], interrogative and
relative pronouns, reexive/reciprocals, adjectives, passive participles,
absolutive past participles)

Personal a is optional for some speakers in cases like (17a). This may be due to aspectual characteristics of the verb modicar modify. Torrego (1998:17 .) notes that
predicates may dier in whether or not they require personal a on the basis of their
aspectual properties.
The contrast in interpretation associated with the presence versus absence of the
Dative clitic is similar to the contrast between the English Dative shifted construction, I sent him the package, versus the non-shifted construction, I sent the package
to him. The former sentence disfavors a reading in which the transfer is not completed, while the latter is compatible with this reading.

Overview of the grammar

15

b. PERSON:
yo
t
l/ella
I(1st.sg.)
you(2nd.sg.)
he/she(3rd.sg.)
(personal pronouns [strong and weak], reexive/reciprocals, nite verbs)
c. CONJUGATION CLASS:
I
II
III
cantar
temer
escribir
sing-inf.
fear-inf.
write-inf.
(nite and non-nite verbs)
d. PAST, PRESENT, FUTURE:
canto
cant
cantar
sing(pr.1st.sg.) sing(pa.1st.sg.)
sing(fut.1st.sg.)
(nite verbs)
e. PRETERITE/IMPERFECT TENSE:
cant
cantaba
sing(pret.1st.sg.) sing(imp.1st.sg.)
(nite verbs)
f. PERFECTIVE AND PROGRESSIVE ASPECT:
ha cantado
est cantando
has(pr.) sing(pprt.)
is(pr.) sing(prt.)
has sung
is singing
(non-nite verbs)
g. MOOD:
cantas
cantes
sing(pr.ind.2nd.sg.)
sing(pr.subj.2nd.sg)
(nite verbs)
h. VOICE:
fue cantado
be(pret.3rd.sg.) sing(pas.)
was sung
(participles)

As (19) suggests, many elements display number and gender agreement. Verbs
display the broadest array of inections. Other quasi-inectional axes
include a diminutive sux and an intensier for adjectives (residue of Latin
superlative suxes). Neither adverbs, prepositions nor conjunctions display
axal inection or contextually induced allomorphs.
1.2.4

Clitics
The term clitic refers to elements which are syntactically independent words or phrasal constituents, but which are phonologically dependent.
Phonological dependence typically implies that the clitic undergoes phonological word-formation so that it joins a constituent which bears stress. For
example, English contracted auxiliaries cliticize to a preceding constituent
(e.g., Shell leave). Phonological and syntactic conditions of cliticization vary

16

The syntax of Spanish

from language to language, as do the inventories and properties of particular

clitics.27 Spanish, and Romance in general, developed a robust system of
clitics, derived from Latin demonstrative ille and from strong pronouns and
reexives. The inventory of Spanish clitics is shown in (20)(22), organized
according to their form when they correspond to subject, object and indirect
object arguments:
(20)

Nominative:
se 3rd. sg. one

(21)

Accusative:
me
1st.sg.
te
2nd.sg
lo
3rd.sg m.
la
3rd.sg.f.
se
3rd. sg./pl. re.

nos
(os)
los
las

1st.pl.
2nd.pl.28
3rd.pl.m.
3rd.pl.f.

Dative:
me
te
le
se

nos
(os)
les

1st.pl.
2nd.pl.
3rd.pl.

(22)

1st.sg.
2nd.sg.
3rd.sg.
3rd.sg./pl.29

Spanish clitics are sometimes referred to as pronominal clitics. However, they

are neither uniformly pronominal or anaphoric, nor necessarily related to
verbal arguments. These same clitics may represent non-arguments (e.g., benefactives), and have other grammatical functions, including formation of
middles and passives, and marking lexical aspect.
A simplied summary of the clitic template is given in (23), based on
Perlmutter (1971):
(23)

[se] [2nd.] [1st.] [3rd.(dat.)] [3rd.(Acc.)]

a. For sequences of non-reexive 3rd person clitics, Dative precedes
Accusative;
b. Non-3rd person clitics precede 3rd person;
second person precedes rst person; i.e.: II I III;
c. Se precedes other clitics;
d. Sequences of phonetically identical clitics are excluded.

The rst position clitic se may be the subject clitic, a reexive 3rd person direct
or indirect object, or an inherent clitic (not corresponding to an argument).30
27

28
29

30

See Zwicky (1977) for an overview of clitic types. For detailed analysis of English
auxiliary clitics see Kaisse (1983).
Second person plural inections and clitics are restricted to peninsular dialects.
Dative se is both a reexive/reciprocal clitic and an allomorph of (pronominal) le.
Se replaces le if a third person Accusative clitic follows (e.g., le+lo selo).
A thorough description of clitic sequences and functions is found in Strozer (1976).
See also Bonet (1991, 1995).

Overview of the grammar

17

The only true subject clitic in Spanish is impersonal se one.31

Impersonal se clauses show 3rd person singular verb forms. Subject se does
not double an overt subject:
(24)

a. *Uno/l, se
trabaja
demasiado all.
one/he, CL(Nom.) work-3rd.sg. too much there
One, one works a lot there.
b. *El hombre, se
piensa
demasiado.
the man,
CL(Nom.) think-3rd.sg. too much
Man, one thinks too much.

Turning to Accusative and Dative clitics, 1st and 2nd person forms are identical in the two cases.32 In some dialects, forms from one case encroach partially or wholly on the functions of the other.33 Elsewhere, the syntactic
conditions governing the appearance of direct and indirect object clitics
remain distinct. As was shown in the previous section, a clitic is required for a
pronominal or anaphoric direct object; clitics co-occur with indirect objects
even when the object is non-pronominal, non-anaphoric.
Spanish does not have clitics corresponding to Oblique (prepositional)
arguments, including locatives. Expressions corresponding to French locative/directional y are all, ah, all there, which are strong, non-clitic forms.
Modern Spanish clitics are always immediately adjacent to a verb, and never
occur in construction with other grammatical categories. Clitics follow positive imperatives, innitives and gerunds, as shown in (25)(26). Notice that the
orthographic conventions show enclitics as part of the verb, while proclitics
(those preceding the verb) are orthographically separated:
(25)

31

32

33

a. Hazlo
ahora.
Do-I.+CL(Acc.) now
Do it now!
b. Intent
mandrmelo.
try-pa.3rd.sg. send-inf.+CL(Dat.)+CL(Acc.)
(S/he) tried to send it to me.
c. Estaba
cantndolo.
be(imp.3rd.sg.) sing-prt.+CL(Acc.)
(S/he) was singing it.

The overt forms of personal pronouns are strong (stressed) forms (see 1.7 on
Dominican Spanish).
The 1st and 2nd person clitics derive from Latin pronouns, with normal phonological changes producing merger of Accusative and Dative forms. Non-reexive 3rd
person forms derive from demonstrative ille, which had the -o/-a inectional ending
in the Accusative and -e in the Dative.
These are known as Lesmo (dative le/les are used also for masculine human
Accusatives), Lasmo, Losmo (replacement of 3rd person Datives by
Accusative forms).

18

The syntax of Spanish

(26)

a. *Lo haz ahora. (=25a)

b. *Intent me lo mandar. (=25b)
c. *Estaba lo cantando. (=25c)

Clitics precede negated imperatives (27) and other nite verbs (28):
(27)

No lo
escriba
ahora.
not CL(Acc.) write-I. now
Dont write it now!

(28)

a. Mara lo
escribi
ayer.
M.
CL(Acc.) write-pa.3rd.sg. yesterday
Maria wrote it yesterday.
b. *Mara escribilo ayer.

In progressives, clitics may either precede the auxiliary or follow the participle, as shown in (29); clitics cannot follow past or passive participles, as shown
in (30b), (31b):
(29)

a. Juan lo
estaba preparando.
J.
CL(Acc.) was
prepare-prt.
Juan was preparing it.
b. Juan estaba preparndolo. (=29a)

(30)

a. Mara ya
lo
haba
preparado.
M.
already CL(Acc.) have-pa.3rd.sg. prepare-pprt.
Maria had already prepared it.
b. *Mara ya haba prepardolo. (=30a)

(31)

a. La carta ya
te
fue mandada.
the letter already CL(Dat.) was send-pprt.
The letter was already sent to you.
b. *La carta ya fue manddate. (=31a)

Spanish shares with Italian the phenomenon of Restructuring, or clitic

climbing, in which clitics related to a subordinate innitive appear in construction with a semi-auxiliary matrix verb. Both (32a) and (32b) are grammatical:
(32)

a. Susana quiere
verte.
S.
want-pr.3rd.sg. see-inf.+CL(Acc.)
Susana wants to see you.
b. Susana te quiere ver. (=32a)

In addition, Spanish restructuring also includes verb-participle sequences:

(33)

a. Mara segua
cantndolo.
M.
continue-pa.3rd.sg. sing-prt.+CL(Acc.)
Maria kept on singing it.
b. Mara lo segua cantando. (=33a)

Overview of the grammar

19

Clitics are an ongoing topic of investigation in Spanish syntax. Among the

issues debated are such fundamental matters as the position in which clitics
are generated, the manner in which they are grammatically linked to an argument position, and their syntactic features. General issues and proposals are
introduced in Chapter 4. The phenomenon of restructuring has also been controversial with respect to the structure of the innitive or participle. These
issues will be considered in Chapter 6.

1.2.5

WH- and NP-movement

Interrogative phrases appear in clause-initial position in both direct

and indirect questions. Compare the position of the direct object in the declarative in (34a), and the corresponding interrogative in (34b), (34c):
(34)

a. Juan ley ese libro.

J.
read that book
b. Qu libro ley Juan?
which book read J.
Which book did Juan read?
c. Mara no sabe [qu
libro ley Juan].
M.
not knows which book read J.
Maria doesnt know which book Juan read.

Multiple interrogatives are possible, and require one interrogative constituent to appear in clause-initial position, while the rest remain in situ:34
(35)

a. A
quin le
mand qu
libro?
to(dat.) whom CL(Dat.) sent
which book
To whom did (s/he) send which book?
b. Qu libro le
mand a
quin?
what book CL(Dat.) sent to(Dat.) whom
What book did (s/he) send to whom?

Processes such as passivization and subject-to-subject raising also show

derived positions for NPs. However, since subjects have a degree of freedom
of order relative to other elements, the eects of these processes are not always
transparent. For example, consider the passives in (36):
34

More than one interrogative constituent may move if a second clause-initial

landing-site is available:
(i)

Qu libro no sabe Juan [quin compr]?

what book not know J.
who bought
What book doesnt Juan know who bought?

The grammaticality of this type of extraction out of an indirect question depends

on the grammatical function of the constituents extracted. See Torrego (1984) for
detailed discussion.

20

The syntax of Spanish

(36)

a. El artculo fue publicado.

the article was published
b. Fue publicado el artculo
was published the article

The order in (36a) shows that an object may move to pre-verbal subject position. In (36b), it appears that no movement has taken place. However, the
phrase el artculo may occupy a structurally higher position than direct object
a position available for subjects even when a direct object is present, as in (37):
(37)

Analiza las preposiciones el artculo.

analyzes the prepositions the article
The article analyzes prepositions.

Example (37) shows that post-verbal subjects can occupy a position other
than direct object position, since the latter is occupied by the phrase las preposiciones the prepositions. Whatever position is available for the subject el
artculo the article in (37) should therefore be available in principle also in
(36b). In Chapter 3 (Section 3.3) and in Chapter 5 this issue is examined
further.

1.2.6

Determiners

Determiners and demonstratives agree in number and gender (masculine or feminine) with nouns. Forms of the indenite and denite determiners are shown in (38) and (39):35
(38)

un(o) m.sg.36
una f.sg.

unos m.pl.
unas f.pl.

(39)

el m.sg.
la f.sg.37

los, m.pl.
las, f.pl.

35

36

Demonstratives, which also agree in number and gender with a noun, distinguish
three degrees of proximity to the speaker: este libro this book; ese libro that
book; aquel libro that (distant) book.
The nal vowel in uno is elided before an overt noun. Compare (i) and (ii):
(i)

(ii)

37

Dame
give-I.2nd.sg.+CL(Dat.)
Give me a pencil.
Dame
give-I.2nd.sg.+CL(Dat.)
Give me one.

un
lpiz.
a(m.sg.) pencil(m.sg.)
uno.
a(m.sg.)

For feminine nouns beginning with stressed a, such as agua water, la is replaced
by el: el agua, the water, las aguas the waters. Feminine el and la both derive
from ela (<illa).

Overview of the grammar

21

The neuter determiner lo occurs in DPs with no overt head noun; lo is followed
by an adjective or relative clause:38
(40)

a. Lo
importante de esa pelcula es el dilogo.
the(neu.) important of that lm
is the dialogue
What is important in that lm is the dialogue.
b. Lo
que me
interesa es el dilogo.
the(neu.) that CL(Dat.) interests is the dialogue
What interests me is the dialogue.

Non-overt nouns or noun phrases also occur with other determiners and
demonstratives. The pronoun corresponding to English one is always silent
(e.g., el otro the other one).
DPs may lack an overt determiner under several circumstances. Predicative
DPs normally lack an overt determiner unless the DP is modied:
(41)

a. Susana es doctora.
S.
is doctor
Susana is a doctor.
b. Susana es una doctora excelente.
S.
is a
doctor-f. excellent
Susana is an excellent doctor.

Referential DPs also occur without overt determiners under certain conditions. Bosque (1980) notes that singular DPs occur without determiners in
negative contexts:
(42)

a. Ernesto no lee libro sin

ilustraciones.
E.
not read book without illustrations
Ernesto doesnt read (any) book without illustrations.
b. *Ernesto lee libro sin
ilustraciones.
E.
read book without illustrations
Ernesto reads any book without illustrations.

Negative contexts include both the presence of negative no not, and other
negative elements which allow negative polarity items (see 1.4.).
Bare (determinerless) plural DPs are generally impossible before the verb,
and generally possible in post-verbal positions. This is illustrated by the contrast between (43a) and (43b):
38

Lo also replaces exclamative qu how. Compare (i) and (ii):

(i)

(ii)

Qu bien escribe Caterina!

how well writes C.
How well Caterina writes!
Lo
bien que escribe Caterina!
the(neu.) well that writes C. (=(i))

22

The syntax of Spanish

(43)

a. Llegaron estudiantes.
(Suer 1982)
arrived students
(Some) students arrived.
b. *Estudiantes llegaron.
students
arrived (=43a)

However, pre-verbal bare plurals are possible if the DP is conjoined, contrastively focused, or a topic (left dislocated):
(44)

(45)
(46)

Viejos y

nios

escuchaban con atencin sus

palabras.
(Bello (1847) 1971:231)
old-pl. and children listened
with attention 3rd.pl.(Gen.) words
Old people and children listened attentively to his/her words.
Estudiantes llegaron (y
no profesores).
students
arrived (and not teachers)
Estudiantes, no creo que falten.
Students
not think that lack
Students, (I) dont think are lacking.

Example (45) is grammatical with main sentential stress on estudiantes, indicating that it is contrastively focused (see Chapter 6, Section 6.2). In (46), estudiantes is not strongly stressed, and is separated from the following constituent
by pause intonation.
Post-verbal bare plurals occur in complement positions, including as direct
object of transitive verbs and as subjects of unaccusative verbs (e.g., llegar
arrive in (43a)). Post-verbal subjects of ordinary intransitive verbs (unergative verbs) cannot normally be bare plurals. However, as noted in Torrego
(1989), they become grammatical if locative inversion occurs:
(47)

a. *Juegan
nios.
play-pr.3rd.pl. children
Children are playing.
b. En este parque juegan
nios.
in this park play-pr.3rd.pl children
In this park children play.

Lois (1986) observes that bare plurals may generally appear as subjects in nonnite clauses, except in the case of agreeing participles:
(48)

De llegar
estudiantes, habr que dar clase.
of arrive-inf. students
have that give class
If students arrive, one has to have class.

(49)

*Comprado
caf, nos
fuimos a casa.
bought-pprt.m.sg. coee CL(1st.pl.) went to home
With coee bought, we went home.

Overview of the grammar

23

In all the preceding ungrammatical examples with bare plural and bare mass
nouns, the sentences become grammatical with the addition of an overt determiner, either denite or indenite.
Items which are traditionally analyzed as determiners include quantifying elements such as todos all,pocos few,muchos many.Some recent studies have
proposed that such items should be dierentiated from determiners, both on the
basis of their logical form and on the basis of core syntactic properties such as
coocurrence with determiners (but not with each other), modication and movement possibilities. This area of investigation is complicated by the fact that
quantifying elements do not behave uniformly as a class in many respects. Some
syntactic generalizations concerning quantiers will be reviewed in Chapter 2.

1.2.7

Negative concord
Sentences in (50a) and (50b) are synonymous:

(50)

a. Nadie sali.
nobody left
Nobody left.
b. No sali nadie.
not left anybody (=50a)

The pattern in (50) generalizes to other negated constituents: either a negated

constituent or no precedes the verb:
(51)

a. Mara no canta nunca.

M.
not sings ever
Maria never sings.
b. Mara nunca canta.
M.
never sings (=61a)

(52)

a. *Nadie no canta nunca.

nobody not sings ever
Nobody ever sings.
b. *Nunca no canta nadie.
never not sings anybody

The phenomenon illustrated above has been termed Negative Concord,

which conveys that, in (51a) for example, there is only one negation, rather
than two independent negative elements. The supercial appearance of two
negative elements is analyzed as resulting from concord, or agreement.39
39

A basic issue in the analysis of Negative Concord is whether no or nadie is the true
negative element. Both positions have been adopted. See Bosque (1980), Zanuttini
(1990).

24

The syntax of Spanish

Romance languages are generally Negative Concord languages. Characteristic

of Negative Concord is the absence of lexical alternations between such pairs
as nobody/anybody, both of which are nadie in Spanish.
There is a broader class of items which, although not intrinsically negative
in meaning, nevertheless require a negative context to be grammatical. These
include adverbs like en absoluto at all, todava yet, as well as various other
categories of items and many idioms. These items are referred to as Negative
Polarity Items (NPIs), and are illustrated below:
(53)

Juan *(no) ha llegado todava.

J.
not has arrived yet
Juan hasnt arrived yet.

(54)

Ese nio *(no) come en absoluto.

that child not eats at all
That child doesnt eat at all.

(55)

Susana *(no) ha movido un dedo por l.

S.
not has lifted a nger for him
Susana hasnt lifted a nger for him.

Negative no is not the only element which can trigger the occurrence of NPIs.
Others include certain interrogative contexts and certain classes of verbs (of
lacking, absence, doubt, opposition), prepositions, conjunctions, comparatives
and quantiers.40 The items discussed above with respect to Negative Concord
(nadie, nunca, etc.) can also cooccur with some of these triggers, but not all.

1.2.8

Null subjects

The subject pronouns are displayed in (56). As shown in (57), pronominal subjects may be overt or covert:
(56)
1st:

Singular:
yo
I

2nd:

t (vos)

3rd:

40

you
l(m.)
he
ella(f.)
she
Usted (Ud.)
you(formal)

Plural:
nosotros(m.)
nosotras(f.)
we
vosotros(m.)
vosotras(f.)
you all
ellos(m.)
they
ellas(f.)
they
Ustedes (Uds.)
you(formal)

A comprehensive summary of NPIs and NPI triggers is given in Bosque (1980).

Overview of the grammar

(57)

25

a. Cantaron
ellos.
sing-pa.3rd.pl. they(Nom.)
They sang.
b. Cantaron.
sing-pa.3rd.pl.

The overt subject pronouns in (56) are always strongly stressed. It is generally
assumed that the richness of morphological agreement allows the content of
the pronoun to be recovered, making the overt pronoun unnecessary. Under
some conditions, subject pronouns cannot be overt. One instance is sentences
whose subjects are non-referring:
(58)

(*Ello) es obvio que (*ello) llovi.

(It) is obvious that (it) rained.

Pronouns corresponding to pleonastic (non-referring) it and to the quasipleonastic subject of atmospheric verbs are always covert. Existential sentences have no overt form corresponding to there:
(59)

(*All) hay un unicornio en el jardn.

(There) is a unicorn in the garden.

Personal subject pronouns are not strictly optional. For example, in a discourse in which Juan is the topic, subsequent references to Juan use the covert
pronoun, not overt l, except for contrastive focus. In the following sequence,
where a contrastive focus interpretation (shown by HE in the gloss) is
impossible, l, is ungrammatical:
(60)

Vi a Juan en el mercado. (*l) me salud, y (*l) dijo que (l) pensaba que
iba a llover.
I saw Juan at the market. *HE greeted me and *HE said that HE thought
that it was going to rain.

Subject pronouns are overt only in contexts of contrastive focus or switching of reference. Consequently, (57a) and (57b) above are not strictly synonymous. The question arises as to whether the distinction between the two
interpretations is represented in sentence-grammar or only in discoursegrammar.
Following work by Jaeggli (1982) and Rizzi (1982), it has been argued that
the option of null subjects is one of a cluster of phenomena which, although
supercially unconnected, can be explained in terms of a single feature of the
grammar. Properties observed in null-subject languages are summarized in
(61):41
41

Null-subject languages allow the use of null resumptive pronouns, which may be
taken as a subcase of (61a).

26

The syntax of Spanish

(61)

a. Phonetically null pronominal subjects

b. Non-overt pleonastic pronouns
c. Free postposing of subjects
d. Absence of COMPtrace eects
e. Long extraction of subjects

(61a) and (61b) have been illustrated above. (61c) refers to the grammaticality
of predicatesubject order in declaratives (Juan cant / Cant Juan Juan
sang). The absence of COMPtrace eects is illustrated by (62):
(62)

Quin crees
que vendr?
who
think-2nd.sg. that come-fut.
Who do you believe that will come?

In (62), the interrogative subject of the subordinate clause is compatible with

an overt complementizer que. The corresponding sentence in English is impossible with the complementizer (hence the name COMPtrace eects: the
trace of a moved subject cannot cooccur with an overt complementizer).42
Long extraction of the subject is shown in (63):
(63)

Quin no sabes qu escribi?

who
not know what write-pa.3rd.sg.
Who dont you know what (they) wrote?

In (63), the complement of saber know is an interrogative clause which has

two interrogative constituents: qu, which is in initial position in the subordinate clause, and quin, which is interpreted as the subject of the subordinate
clause. Quin has been long extracted to clause-initial position of the main
clause. The ungrammaticality of the corresponding English sentence shows
that non-null-subject languages disallow long extraction of the subject.
Analyses of the null-subject parameter have accounted for the clustering of
these properties in terms of the richness of verbal inection for subject features, which (stated informally) provides a stronger governor for the subject
of nite clauses than is otherwise possible. Government of the subject by a
strong governor makes possible phonetically null subjects, including null
pronominals and traces of moved subjects.

1.3

The subject constituent

The preceding discussion summarized the characteristics of null subjects. We turn our attention now to a description of overt subjects: their order
relative to other constituents, their occurrence in non-nite clauses, and
42

Perlmutter (1971) rst observed the correlation between COMPtrace eects and
null subjects.

Overview of the grammar

27

subjectverb agreement. As the discussion in 1.3.1 and 1.3.2 will show, the
position of the subject is relatively free in nite declarative clauses, while in
non-declaratives and in non-nite clauses, the subject is more restricted in its
distribution. Section 1.5.1 below will discuss constituent order in declaratives
from the perspective of information structure.

1.3.1

Order

In nite declarative sentences, the subject may either precede or

follow the predicate, as shown in (64a) and (64b). V-S-O order is also possible,
as in (64c):43
(64)

a. Escribi la carta mi hermana.

wrote
the letter my sister
My sister wrote the letter.
b. Mi hermana escribi la carta. (=64a)
c. ?Escribi mi hermana la carta. (=64a)

V-S-O sequences in nite declaratives may be less natural than S-V-O and VO-S orders. The naturalness of V-S-O sentences may vary from speaker to
speaker, and may depend also on lexical properties of the sentence. For
example, (66) is more natural than (65); both are V-S-O:
(65)

?Pint el artista retratos terribles.

painted the artist portraits terrible
The artist painted terrible portraits.

(66)

Sufri el paciente dolores terribles.

suered the patient pains terrible
The patient suered terrible pains.

The mixed results for V-S-O sentences dier from both pre-verbal and postpredicate subjects, which are uniformly grammatical in nite declaratives.
Another type of clause that has freedom of subjectpredicate order is the
small clause. These constituents contain a predicative phrase and a constituent that is the semantic subject of the predicate. Unlike full clauses, small
clauses may lack a verb. In the examples in (67) and (68), the small clause consists of the bracketed sequence. The subjects of small clauses may precede or
follow their predicate:
(67)

43

a. Eligieron
[presidente a Juan].
elected-3rd.pl. [president PA J.
(They) elected Juan president.
b Eligieron [a Juan presidente].
(=67a)

Judgments seem to vary as to whether V-S-O order is fully acceptable in declaratives.

28

The syntax of Spanish

(68)

a. Consideran
[vlida la prueba].
consider-pr.3rd.pl. valid the proof
(They) consider the proof valid.
b. Consideran [la prueba vlida].
(=68a)

Non-nite clauses typically allow only post-verbal subjects, as is illustrated

in (69)(72). (See 1.7.5 for discussion of dialectal variation.)
(69)

a. [ Llegada
ella ]
empez la esta.
arrive-pprt.f.sg. she(Nom.) began the party
b. *[ella
llegada ]
empez la esta.
she(Nom.) arrive-pprt.f.sg. began the party
(With) her arrived, the party began.

(70)

a. [Habiendo llegado
have-prt. arrive-pprt.
b. *[Ella
habiendo
she(Nom.) have-prt.

(71)

a. [Al
cantarlo
t]
upon+the sing-inf.+CL(Acc.) you(Nom.)
Upon your singing it, the party began.
b. *[Al
t
cantarlo]
upon+the you(Nom.) sing-inf.+CL(Acc.)

(72)

ella],
empez la esta.
she(Nom.) began the party
llegado],
empez la esta.
arrive-pprt. began the party.
empez la esta.
began the party
empez la esta.
began the party

a. [De ganar ellos]

los
felicitaremos.
of win-inf. they(Nom.) CL(Acc.) congratulate-fut.1st.pl.
If they win, we will congratulate them.
b. *[De ellos
ganar] los
felicitaremos.
of they(Nom.) win-inf. CL(Acc.) congratulate-fut.1st.pl.

Non-declarative nite clauses generally require a post-verbal subject. This is

illustrated below for imperatives (73), constituent questions (74) and exclamatives (75). (See 1.7.5 for discussion of dialectal variation.)
(73)

a. Hazlo
t.
do(I)+CL(Acc.) you(nom.)
You do it!
b. *T hazlo.44

(74)

a. Qu ley Juan?
what read J.
What did Juan read?
b. *Qu Juan ley?

44

Although imperatives do not freely allow pre-verbal subjects, as in (73b), it is possible to add a topicalized subject to an imperative:
(i)

T, hazlo!
You, do it!

In (i), the overt subject is separated intonationally from the imperative.

Overview of the grammar

(75)

29

a. Qu alto est Julio!

how tall is
J.
How tall Julio is!
b. *Qu alto Julio est!

Relative clauses pattern with declaratives, rather than interrogatives: their subjects may be pre-verbal:
(76)

a. el libro [que Juan ley ]

the book that J.
read
b. el libro [que ley Juan] (=76a)

In some cases, a post-verbal subject is required if a non-subject constituent is

preposed. This will be illustrated in Chapter 6. Finite subordinate clauses
pattern with main clauses with respect to subject position: the subject may be
pre-verbal or post-verbal, except in embedded interrogatives and exclamatives,
and clauses with preposed constituents.

1.3.2

Distribution in non-nite clauses

Nominative subjects are possible in non-nite adjunct clauses, as illustrated above in (69)(72). Nominative NPs also occur as arguments of the prepositions hasta even and entre between (see note 20). With the exception of
causative and perception constructions discussed below, non-nite argument
clauses do not allow non-Nominative subjects. For example, gerunds do not
admit Genitive subjects (*su partiendo your leaving); there is no prepositional complementizer corresponding to English for (e.g., for Mary to leave);
and exceptional case in complements of believe predicates is excluded:45
(77)

*Juan cree
Mara ser
inteligente.
J.
believes M.
be-inf. intelligent
Juan believes Mara to be intelligent.

In innitival causative constructions under hacer make or dejar let, the

innitival subject is an Accusative or Dative object of the causative verb:
(78)

45

a. Susana le
hizo leer la carta a
Jos.
S.
CL(Dat.) made read the letter Dat. J.
Susana made Jos read the letter.
b. Susana har
caminar a la ocina a Jos.
S.
make-fut. walk
to the oce PA J.
Susana will make Jos walk to the oce.

Unlike French, an interrogative is also excluded as the subject of a complement of

creer believe:
(i)

*Quin cree
Juan ser
inteligente?
who believes J.
be-inf. intelligent
Who does Juan believe to be intelligent?

30

The syntax of Spanish

In (78a), the innitival subject (Jos) is Dative, and the object of the innitive
is Accusative. In (78b), the subject of the innitive (Jos) is Accusative. Two
factors determine the Case of the innitival subject. The transitivity of the
innitive is one factor. If the innitive has an Accusative object, the subject is
necessarily Dative. That is, the causative complex cannot contain two
Accusative arguments. A second factor is concerned with the contrast
between direct and indirect causation. Intransitives such as (78b) allow the
innitival subject to be either Accusative (direct causation) or Dative (indirect
causation).
Innitival complements of perception verbs pattern with causative complements in the respects mentioned above. Gerundive and nite clause complements of perception predicates also have subjects that are grammatical objects
of the perception verb:
(79)

a. Juan la
vio cruzando la calle.
J.
CL(f.sg.Acc.) saw cross-prt. the street
Juan saw her crossing the street.
b. Juan vio a Mara que cruzaba
la calle.
J.
saw PA M.
that cross-pa.3rd.sg. the street
Juan saw that Maria crossed the street.

The subjects of small clauses in argument positions take objective case:

(80)

a. La
nombraron
presidenta.
CL(f.sg.Acc.) name-pa.3rd.pl. president
They named her president.
b. Los
consideran
inteligentes.
CL(m.pl.Acc.) consider-pr.3rd.pl. intelligent
They consider them intelligent.

In these examples, the subject of the small clause takes the case appropriate to
the complement of the verb.
Adjunct innitival clauses may have Nominative subjects:
(81)

a. [Al
cantarlo
t]
empez la esta.
upon+the sing-inf.+CL you(Nom.) began the party
Upon your singing it, the party began.
b. [De ganar ellos]
los
felicitaremos.
of win-inf. they(Nom.) CL(Acc.) congratulate-fut.1st.pl.
If they win, we will congratulate them.

1.3.3

Agreement

Most person/number suxes are unambiguous. This is illustrated for

a rst conjugation verb cantar to sing in the present tense, where only 1st
and 3rd person singular of the subjunctive are homophonous:

Overview of the grammar

(82)

cantar
1st.sg.
2nd.sg.
3rd.sg.
1st.pl.
2nd.pl.
3rd.pl.

Indicative:
canto
cantas
canta
cantamos
cantis
cantan

31

Subjunctive:
cante
cantes
cante
cantemos
cantis
canten

Finite clauses generally display person and number agreement between the
clausal subject and the nite verb. Sentences with personal pronominal subjects show the same subjectverb agreement whether the pronoun is overt or
covert. However, some sentences with covert subjects display invariant 3rd
person singular agreement on the verb. One such case is constructions with the
subject clitic se:
(83)

a. Se
trabaja
mucho en este curso.
CL(nom) work-pr.3rd.sg. a lot in this course
b. *Se trabajan
mucho en este curso.
CL work-pr.3rd.pl. a lot
in this course
One works a lot in this course.

Constructions with non-referential subjects (corresponding to English it,

there), also have invariant 3rd singular verb forms:
(84)

a. Parece
que los libros han llegado.
seems-pr.3rd.sg. that the books have arrived
It seems that the books have arrived.
b. *Parecen
que los libros han llegado.
seems-3rd.pl. that the books have arrived

(85)

a. Es
obvio que los libros han llegado.
be-pr.3rd.sg. obvious that the books have arrived
It is obvious that the books have arrived.
b. *Son
obvio(s) que los libros han llegado.
be-pr.3rd.(pl.) obvious that the books have arrived
It is obvious that the books have arrived.

In some instances, invariant verb forms do alternate with agreeing forms.

This is illustrated for existential constructions in (86), and atmospheric verbs
in (87):
(86)

a. Haba
tres libros en la mesa.
be-pr.3rd.sg. three books on the table
There were three books on the table.
b. ?Haban
tres libros en la mesa.
be-pr.3rd.pl. three books on the table

(87)

a. Llueve
monedas del
cielo.
rain-pr.3rd.sg. coins
from+the sky
Its raining coins from heaven.

(Hurtado 1989a)

32

The syntax of Spanish

b. ?Llueven
monedas del
cielo. (=76a)
rain-pr.3rd.pl. coins
from+the sky

Rightward agreement in existentials is fairly common, although it is considered substandard.

Copular sentences with an identicational interpretation require agreement
with the predicative element:
(88)

a. El culpable [soy
yo].
the culprit
be-pr.1st.sg. I(Nom.)
The culprit is me.
b. *El culpable es
yo.
the culprit be-pr.3rd.sg. I(Nom.)

Passives generally display agreement between the nite verb and derived
subject. This is shown for passives composed of ser be +participle and
se+verb:
(89)

a. Esos libros fueron

vendidos.
those books be-pa.3rd.pl. sell-pprt.m.pl.
Those books were sold.
b. *Esos libros fue
vendido(s).
those books be-pa.3rd.sg. sell-pprt.m(pl.)

(90)

a. Esos libros se
vendieron.
those books CL(pas) sell-pa.3rd.pl
b. *Esos libros se
vendi.
those books CL(pas) sell-pa.3rd.sg.

The following pair has received various analyses:

(91)

a. Se
vende
ores.
CL(pas) sell-pr.3rd.sg. owers
Flowers sold/for sale.
b. Se
venden
ores.
CL(pas) sell-pr.3rd.pl. owers

The fact that se may be a subject clitic, a passive morpheme, or an anti-causative (middle) morpheme leads to various possible analyses of (91b). A central
issue that arises with respect to its analysis is that the verb agrees with ores,
suggesting that this phrase is the grammatical subject. However, based on the
fact that it is a bare plural (see 1.2.6), it must be in complement position.
Example (91b) becomes ungrammatical if ores precedes the verb.

Overview of the grammar

1.4

The predicate constituent

1.4.1

Tense and aspect

33

Verbal inections express (a) tense (Past, Present, Future), (b) perfectivity (or telicity) of past events (Imperfect, Preterite Past) and (c) mood
(Subjunctive, Indicative). Each of these is illustrated below. We then summarize periphrastic forms related to tense, aspect, modality and voice.
Indicative tenses include Future, Conditional, Present and two simple past
tenses, Preterite and Imperfect Past. The Future tense is ambiguous between
temporal future reference and a present modal of probability. This is illustrated by the pairs in (92) and (93):
(92)

a. Dnde van
a estar a las dos?
where go-pr.3rd.pl. to be-inf. at the two
Where will they be at two oclock?
b. Estarn
en casa a las dos.
be-fut.3rd.pl. at home at the two
They will be at home at two oclock.

(93)

a. Dnde est Mara?

where is M.
Wheres Maria?
b. Estar
trabajando, sin
duda.
be-fut.3rd.sg. work-prt. without doubt
(She) must be working, no doubt.

Notice in (92a) that temporal future reference is expressed by the periphrastic

ir a+innitive. This is the standard future construction, especially for nearfuture reference. In (93), the future tense does not locate an event in the future.
It asserts a probability or likelihood in the present. On this reading, the future
is most similar to non-indicative tenses, which generally do not assert an
events occurrence.
The conditional tense is sometimes referred to as a past of the future, and
indeed historically the Future and Conditional tenses evolved from Present
and Imperfect forms of auxiliary haber based on its modal reading have to
V. The Conditional is still a grammatically [+PAST] counterpart of the
modal present of probability:
(94)

Estaran
trabajando en ese momento.
be-cond.3rd.pl. work-prt. at that moment
They must have been working at that moment.

The Conditional also behaves like a grammatical [+PAST] with respect to

sequence-of-tense:

34

The syntax of Spanish

(95)

a. Te
prestaran
el coche si lo
necesitaras.
CL(Dat.) lend-cond.3rd.pl. the car
if CL(Acc.) need-pa.subj.2nd.sg.
They would lend you the car if you needed it.
b. *Te
prestaran
el coche si lo
necesitas.
CL(Dat.) lend-cond.3rd.pl. the car
if CL(Acc.) need-pr.ind.2nd.pl.
They would lend you the car if you need it.

In (95b), the Conditional is incompatible with the present tense if-clause. In

(95a), the Conditional is compatible with the past tense if-clause.46 To refer
to past time in counterfactual Conditionals, the compound perfect is
required:
(96)

Te habran
prestado el coche si lo
hubieras
necesitado.
CL have-cond. lend.pprt. the car
if CL(Acc.) have-pa.subj.2nd.sg.
need-pprt
They would have lent you the car if you had needed it.

(97)

*Te habran
prestado el coche si lo
necesitaras.
CL have-cond. lend-pprt. the car if CL(Acc.) need-pa.subj.2nd.sg.

The contrast between (96) and (97) suggests that the Conditional is essentially
modal or atemporal. Although the Conditional in (94) appears to order an
event in the past, this reading is not possible without the adverb unlike the
adverbs in (92) and (93). Thus, while the Future tense is both temporal and
modal, the Conditional may be exclusively modal, with only a formal feature
for [+PAST] (as opposed to a semantic feature).
The Present Indicative potentially has present and future readings. Present
readings for non-stative predicates are ambiguous between habitual (frequentative) and present-moment:
(98)

Mara estudia
geografa en la biblioteca.
M.
study-pr.3rd.sg. geography at the library
Maria studies geography at the library.
Maria is studying geography at the library.

Future readings are possible with non-statives such as (99a), but not with statives such as (99b):
46

Notice that the grammatical feature [+PAST] does not necessarily correspond to
semantic anteriority, as illustrated in (i):
(i)

Te prestaran el coche maana si lo necesitaras hoy.

CL lend-cond. the car tomorrow if CL need-pa.subj. today
They would lend you the car tomorrow if you needed it today.

For discussion of construal of Past tense, see Iatridou (2000).

Overview of the grammar

(99)

35

a. *Salen
(esta noche).
*leave-pr.3rd.pl. this night
*Theyre leaving/going out (tonight).
b. *Mara sabe
la leccin a las tres.
*M.
know-pr.3rd.sg. the lesson at the three
*Mara knows the lesson at three oclock.

Indicative Past tenses refer to events and states which precede the
moment of speech. Spanish distinguishes further between Imperfect and
Preterite Past. The distinction is similar to the distinction described by
Vendler (1967) with respect to delimited and non-delimited events. Delimited
events have an inherent endpoint, while non-delimited events (and states) do
not. For example, to walk is not inherently delimited, but to walk to the store
is, since once the goal is reached, the event necessarily ends. The Preterite
and Imperfect Past tenses make an analogous distinction, but not with
respect to events and states themselves; rather these tenses assert whether or
not the interval during which an event or state occurs is delimited. This
factor interacts with properties of the predicate, including adverbs, to
provide inferences about the delimitedness of the event. To illustrate this
point, let us rst consider the Imperfect, which is often glossed as was Ving, used to V:
(100)

a. Susana tocaba
la auta.
S.
play-imp. the ute
b. Susana used to play the ute.
c. Susana was playing the ute.

The interpretation in (100b) is frequentative or habitual: there is a past interval of non-specic duration, during which the activity of playing the ute is
frequent or habitual. On the reading in (100c) there is a single event of playing
the ute, which corresponds in duration to the interval of non-specic duration. What the two readings have in common is the occurrence of some
event(s) during a non-delimited interval. An event may be neither frequentative within the interval nor correspond with the interval, as in (101):
(101)

Eran
las cinco.
be-imp.3rd.pl. the ve
It was ve oclock.

In (101) the event is the occurrence of ve oclock, which is punctual and

delimited. The Imperfect is shown here not to dene the duration of the event
(or its delimitedness) but instead to refer to the non-delimitedness of the interval during which the event occurs. The Imperfect may have the eect of canceling the delimitedness of an event. Consider the two readings in (102):

36

The syntax of Spanish

(102)

a. (Yo) caminaba
a la tienda.
I walk-imp.1st.sg. to the store
b. I was walking to the store.
c. I used to walk to the store.

The predicate walk to the store is aspectually delimited. However, neither

reading in (102) is delimited. On the single event reading in (102b), the goal is
understood not to be reached. On the habitual reading in (102c), the plurality
of occurrences produces a non-delimited interpretation.47 Thus, in this example
(unlike (101)), this potentially delimited event is non-delimited in the Imperfect.
Consider now the Preterite Past. By contrast, a delimited event such as (102)
is understood to be completed:
(103)

(Yo) camin
a la tienda.
I
walk-pret.1st.sg. to the store

In the Preterite Past, the past interval itself is delimited. The endpoint of the
interval corresponds with the endpoint of delimited events in the predicate.
The Preterite Past typically adds a delimited reading to predicates which do
not have them inherently. This is illustrated for states and activity predicates
in (104):
(104)

a. Cant.
sing-pret.1st.sg.
I sang.
b. Fuiste
amable.
be-pret.2nd.sg. kind
You were kind.
c. Dibuj
crculos.
draw-pret.1st.sg. circles
I drew circles.

The predicates in (104) do not express inherently delimited events or states.

However in the Preterite Past, they have a delimited interpretation, by virtue
of the delimitedness of the past interval during which they occur. The interval
has a denite endpoint, and is followed by a subsequent interval during which
47

The conversion of a (potentially) delimited event to a non-delimited one by pluralizing it, as in the frequentitive reading (102c), is analogous to the conversion of
an Accomplishment to a non-delimited reading by pluralizing its complement:
(i)

a. dibujar un crculo (Accomplishment)

draw a circle
b. dibujar crculos (Activity)
draw circles

In (ib), the plural has the eect of iterating events. The iterated sequence is nondelimited.

Overview of the grammar

37

the event does not hold. For (104a), for example, if I sang (pret.), it is understood that I stopped singing and that some interval of non-singing is subsequent to the event of singing. By contrast, the Imperfect Past lacks this
subsequent interval of non-singing. The Present Perfect lacks it as well, a point
to which we return below.
Periphrastic tenses include a future construction consisting of ir a go to
plus innitive; tener que have to plus innitive; and perfective and progressive constructions which are quite close to English counterparts in both form
and interpretation. The perfect tenses are constructed with auxiliary haber
have followed by a past participle. There is no change of auxiliary as is
found in French, Catalan and Italian, nor does participle show agreement:
(105)

a. Los nios han

lavado
la ventana.
the kids have-pr.3rd.pl. wash-pprt. the window-f.sg.
The kids have washed the window.
b. Los nios la
han
lavado.
the kids CL(f.sg.Acc.) have-pr.3rd.pl. wash-pprt.(m.sg.)
The kids have washed it.
c. *Los nios la
han
lavada.
the kids CL(f.sg.Acc.) have-pr.3rd.pl. wash-pprt.f.sg.

Unlike French and Italian, the Present Perfect does not serve as a simple
past tense in main clauses:48
(106)

*Los nios la han

lavado
ayer.
the kids CL have-pr.3rd.pl. wash-pprt yesterday
The kids have washed it yesterday.

The standard approach to the interpretation of compound tenses is that they

do not temporally locate the event relative to the time of the utterance.
Instead, a reference point is located relative to the time of the utterance, and
the event is in turn located relative to the reference point. In (105), the reference point is the present; the event of washing is anterior to the reference
point. Positing the reference point allows the the present relevance of the event
to be captured. The three points are dierentiated in the Past Perfect (e.g., Los
nios la haban lavado The children had washed it).
48

There is some variation with respect to the acceptability of past adverbs with the
present perfect tense, at least among some peninsular speakers. For example, (i) is
acceptable for some speakers, in contrast to ayer yesterday, which is less so:
(i)

Los nios la han lavado esta maana.

the kids CL have washed this morning
The kids have washed it this morning.

It appears that for these speakers, a proximate/distant gradation may come into play.

38

The syntax of Spanish

The periphrastic progressive consists of auxiliary estar be followed by a

progressive participle:
(107)

a. Estoy
leyendo El poema del Cid.
be-pr.1st.sg. read-prt. El poema del Cid.
b. Estaba
leyendo El poema del Cid.
be-pa.1st.sg. read-prt. El poema del Cid.

(108)

Mara est
estudiando para dentista.
M.
be-pr.3rd.sg. study-prt. for dentist
Mara is studying to be a dentist.

The progressive refers to intervals of activities. As in English, it is not normally

available with predicates that lack activity as part of the event structure:
(109)

a. *La caja est

conteniendo papeles.
the box be-pr.3rd.sg. contain-prt. papers
The box is containing papers.
b. *Est
pareciendo llover.
be-pr.3rd.sg. seem-prt. rain-inf.
Its seeming to rain.

1.4.2

Mood

Turning to expression of mood, we focus here on subjunctive

morphology, which includes simple Present and Past (Imperfect) tenses, as
well as subjunctive forms of the compound tenses and other periphrastic
verbal constructions:
(110)

Marta lamenta ...

M.
regrets ...
a. que yo cante.
that I sing-pr.subj.1st.sg.
that I sing.
b. que yo haya
cantado.
that I have-pr.subj.1st.sg. sing-pprt.
that I have sung.
c. que yo cantara.
that I sing-pa.subj.1st.sg.
that I sang.
d. que hubiera
cantado.
that have-pa.subj.1st.sg. sing-pprt
that I had sung
e. que estuviera
cantando.
that be-pa.subj.1st.sg. sing-prt.
that I was singing.
f. que hubiera
estado cantando.
that have-pa.subj.1st.sg. be-pprt. sing-prt.
that I had been singing.

Overview of the grammar

39

Mood is sometimes characterized as an expression of the speakers attitude toward the factuality of an event or state expressed by a clause. In
Spanish, indicative mood generally occurs in contexts which assert an
events occurrence (hence, factuality). Subjunctive occurs in contexts in
which the events occurrence is in doubt or unknown, and also in contexts
such as (110), in which the events occurrence or factuality is presupposed
rather than asserted. However, optional alternations between subjunctive
and indicative as a means of reecting speakers degree of certainty as to
factuality are available only in ungoverned clauses, such as main clauses
expressing possibility:
(111)

a. Quizs ya
hayas
comido.
maybe already have-pr.subj.2nd.sg. eat-pprt.
Perhaps you have already eaten.
b. Quizs ya
has
comido.
maybe already have-pr.ind.2nd.sg. eat-pprt.

In (111), both sentences contain quizs, therefore express some degree of

doubt or uncertainty as to the factuality of the proposition. The greater degree
of certainty is associated with the indicative in (111b).
Although alternations of the type shown in (111) are possible in main
clauses, they generally are not possible in subordinate clauses. There, the selection of subjunctive or indicative mood is determined by properties of the subordinate clause and its context. For example, adjunct clauses introduced by
cuando when require the subjunctive if the subordinate predicate is interpreted as subsequent to the moment of speech:49
49

Some adjuncts of place and time show alternations of subjunctive/indicative dependent on the specicity of the place or time indicated:
(i)

(ii)

a. Lloraba

cuando estaba

triste.
(Campos 1993:151)
cry-imp.ind.3rd.sg. when be-imp.ind.3rd.sg. sad
S/he cried when s/he was sad.
b. Llorar
cuando est
triste.
cry-fut.3rd.sg. when be-pr.subj.3rd.sg. sad
S/hell cry when(ever) s/hes sad.
a. Estudiar
donde estudia
su novio.
study-fut.3rd.sg. where study-pr.ind.3rd.sg. her boyfriend
Shell study where her boyfriend is studying.
b. Estudiar
donde estudie
su novio.
study-fut.3rd.sg. where study-pr.subj.3rd.sg. her boyfriend
Shell study where(ever) her boyfriend studies.

These alternations appear to be analogous to contrasts in relative clauses, where subjunctive occurs if the relative clause antecedent is nonspecic. The examples in (i)
and (ii) may in fact be relatives with a covert time or place antecedent.

40

The syntax of Spanish

(112)

a. Cenamos
cuando lleguen.
dine-pr.1st.pl. when arrive-pr.subj.3rd.pl.
Well eat dinner when they arrive.
b. Cenamos
cuando llegaron.
dine-pa.1st.pl. when arrive-pa.ind.3rd.pl.
We ate dinner when they arrived.

In argument clauses the mood of the subordinate clause is generally dictated

by the lexical item which selects the clause, and mood alternations are generally impossible. Relative clauses are indicative, unless the antecedent has particular referential characteristics, such as a negative existential (nadie
nobody, nada nothing, etc.) or a non-specic antecedent. That is to say,
the appearance of subjunctive mood is conditioned by syntactic features of
the context in which a clause occurs (see 1.6.2 on relative clauses). Theoretical
issues related to mood include (a) the analysis of the temporal value of clauses
whose verb appears in the subjunctive, and (b) the grammatical feature(s)
which trigger the occurrence of the subjunctive.

1.4.3

Voice

Non-active voice morphology in Spanish includes passive and

middle/passive constructions. Passives are formed with ser be followed by a
passive participle:
(113)

Los argumentos
fueron
rechazados
(por el juez).
the arguments-m.pl. be-pa.3rd.pl. reject-pprt.m.pl. by the judge
The arguments were rejected by the judge.

In (113), the nite verb agrees in person and number with the derived
subject, the participle agrees in number and gender with the derived subject,
and the agent may be expressed in a prepositional phrase headed by por
by. The derived subject may remain in post-verbal position. Both bare
NPs and denite subjects appear between the participle and a locative
complement:50
(114)

50

a. Fue encontrado oro en el Mar del

Norte.
(Bosque 1996:30)
was found
gold in the sea of+the north
Gold was found in the North Sea.
b. Fue encontrado el oro en el Mar del
Norte. (=114a)
was found
the gold in the sea of+the north

There are some restrictions on the occurrence of bare NPs in the post-verbal position of passives. See Bosque (1996:30), Contreras (1996:147).

Overview of the grammar

41

Passivization is restricted to those arguments which correspond to

Accusatives in the active voice. Neither indirect nor oblique complements can
be passivized.51
Middle/passive constructions are formed with the clitic se with an active
verb form. The clitic which appears in this construction is homophonous with
reexive clitics. Consequently, three readings of the clitic in sentences like
(115) are possible (a) transitive reexive, (b) passive voice and (c) intransitive middle voice:
(115)

El coche se movi.
the car
CL move-pa.3rd.sg.
a. The car moved itself.
b. The car was moved.
c. The car moved.

In the middle/passive construction, the verb is active in form. It agrees in

person and number with the derived subject, and the clitic agrees in person
and number with the derived subject. For most speakers, optional agent
phrases headed by por are impossible, both on passive and middle readings.
The middle and passive constructions are supercially identical, but the two
can be dierentiated on the basis of their argument structure and associated
temporal properties. Passives, but not middles, have an implicit agent. This is
shown by the fact that with agent-oriented adverbs and purpose clauses, only
the passive interpretation is grammatical:
(116)

a. El coche se movi voluntariamente.

the car
CL moved voluntarily
The car was moved voluntarily.
(not: *The car moved voluntarily.)
b. El coche se movi para evitar un accidente.
the car
CL moved to
avoid an accident
The car was moved to avoid an accident.
(not: *The car moved to avoid an accident.)

A related temporal property of passives, which also distinguishes them from

middles, is the marginal status of simple present tense with denite subjects.
Thus, the examples in (116) are ungrammatical in the present tense:
(117)

51

a. ?*El coche se mueve voluntariamente.

the car
CL moves voluntarily
The car is moved voluntarily.

In journalese, the indirect object of preguntar ask is sometimes passivized:

(i)

El representante fue preguntado si rmara el acuerdo.

The representative was asked whether he would sign the agreement.

This does not generalize to other indirect objects.

42

The syntax of Spanish

b. ?*El coche se mueve para evitar un accidente.
the car
CL moves to
avoid an accident
The car is moved to avoid an accident.

An aspectual dierence between passives and middles concerns the delimitedness of the underlying transitive predicate. Passive clitics may combine with
transitives without regard to whether the transitive is a perfective (delimited)
or an inchoative (nondelimited) predicate. The middle clitic appears to
combine only with perfectives. Thus, tocar to ring, an inchoative transitive
(118a), takes only passive se (118b), not middle se (118c):
(118)

a. El cura toca la campana.

the priest rings the bell
b. La campana se toc (voluntariamente).
the bell
CL rang voluntarily
The bell was rung (voluntarily).
c. *La campana se toca.
the bell
CL rings
The bell rings.

There are several issues to be addressed in analyzing the middle and passive
constructions. One is whether or not the non-active sentences are related to
corresponding active sentences by way of syntactic or (derivational) lexical
processes. A second issue is the nature of the relationship between the transitivity of the predicates, voice, and temporal properties such as those noted
above.

1.4.4

Modal verbs

The verbs poder may; be able and deber must; should are
modal in meaning, but pattern syntactically with main verbs: they take a
full range of tense, mood and person/number inections; their position relative to negation and auxiliaries is identical to that of main verbs. On epistemic
readings, these verbs do not occur in innitives or in the compound perfect
tense:
(119)

a. Mara parece deber

terminar el proyecto.
M.
seems should-inf. nish-inf. the project
(=It seems that she should nish the project.)
(*It seems that she must nish the project.)
b. Mara parece poder
terminar el proyecto.
M.
seems may-inf. nish-inf. the project
(=It seems that she is allowed/able to nish the project.)
(*It seems possible for her to nish the project.)

(120)

a. Juan piensa poder

terminar el proyecto.
J.
thinks may-inf. nish-inf. the project

Overview of the grammar

43

(=Juan thinks he is allowed/able to nish the project.)

(*Juan thinks it is possible hell nish the project.)
b. Juan piensa deber
terminar el proyecto
J.
thinks should-inf. nish
the project
(=Juan thinks he is obliged to nish the project.)
(*Juan thinks it is necessary that he nish the project.)
(121)

a. Juan lamenta poder terminar el proyecto.

J.
regrets may-inf. nish
the project
(=Juan regrets he is allowed/able to nish the project.)
(*Juan regrets it is possible hell nish the project.)
b. Juan lamenta deber terminar el proyecto.
J.
regrets should nish-inf. the project
(=Juan regrets he should nish the project.)
(*Juan regrets it is necessary he nish the project.)

(122)

a. Mara puede haber

terminado el proyecto.
M.
may have-inf. nish-pprt. the project
(=It is possible M. has nished the project.)
(=M. has been able/allowed to nish the project.)
b. Mara ha podido
terminar el proyecto.
M.
has may-pprt. nish-inf. the project
(=M. has been allowed/able to nish the project.)
(*It has been possible that M. nished the project.)

(123)

a. Juan debe haber

terminado el proyecto.
J.
should have-inf. nish-pprt. the project
(=It is necessary that J. has nished it.)
(=J. should have nished it.)
b. Juan ha debido
terminar el proyecto.
J.
has should-pprt. nish-inf. the project
(=J. should have nished it.)
(*It has been necessary that J. nish it.)

These verbs pattern also with restructuring verbs (see 1.2.4), so that they form
a verbal complex with a following innitive, and occur in construction with
the clitics associated with a following innitive.

1.4.5

Negation
Negation is marked in the basic case by no not:

(124)

Juan no cant.
J.
not sing-pa.3rd.sg.
Juan didnt sing.

Non-negation (armation) is unmarked except in contrastive contexts.

Contrastive contexts such as (125) and (126) show positive polarity marked
for a constituent by s yes, sino but rather:

44

The syntax of Spanish

(125)

Juan s lleg
anoche.
J.
yes arrive-pa.3rd.sg. last night
Juan DID arrive last night.

(126)

Mara se
lo
regal
no a Pedro sino
a
Juan.
M.
CL(Dat.) CL(Acc.) give-pa.3rd.sg. not to P.
but (rather) to
J.
Maria gave it not to Pedro but to Juan.

Contrastive polarity is also possible with s or no in a topic structure with

clausal ellipsis:
(127)

Juana y
Susana llegaron,
pero Jos, no.
J.
and S.
arrive-pa.3rd.pl. but J.,
not
Juana and Susana arrived, but Jos didnt.

(128)

Pedro no comi
pulpo, pero salmn, s.
P.
not eat-pa.3rd.sg. octopus but salmon, yes
Pedro didnt eat octopus, but salmon, yes.

In the second conjunct of (127) and (128), a constituent has been topicalized
(left dislocated), and is followed by the focused polarity marker s or no; the
remainder of the second conjunct is ellipted.
No may be ambiguous in scope:
(129)

a. No com la zanahoria cruda.

not ate the carrot
raw
b. It is not the case that I ate the carrot raw (the raw carrot).
c. I ate the carrot, but it wasnt raw.

On the reading in (129c), no has scope over the secondary predicate, while in
(129b) no has scope over both the primary and secondary predicates, i.e., over
the entire clause. Clausal negation both main and subordinate is distinguished in most cases by the presence of pre-verbal no. Negation of nonclausal constituents is more varied, both with respect to the form and position
of negation. Let us take the two types in turn.
Clausal negation readings are possible when the morpheme no immediately
precedes the verb inected for niteness. Only clitics intervene between no and
the verb:
(130)

a. Mara no ley ese captulo.

M.
not read that chapter
Maria didnt read that chapter.
b. Mara no lo
ley.
M.
not CL(Acc.) read
Maria didnt read it.

Overview of the grammar

45

c. Mara no lo
haba ledo.
M.
not CL(Acc.) had read-pprt.
Maria hadnt read it.

In each of the preceding examples, no precedes the verb inected for tense
(niteness); in (130b) and (130c), a clitic intervenes between no and the verb.
The following examples show that no also precedes non-nite verbs and the
head of nominalized clauses:
(131)

a. Juan prometi ayer

[no intervenir].
J.
promised yesterday not meddle-inf.
Juan promised yesterday not to meddle.
b. Pablo quera [no fracasar].
P.
wanted not fail-inf.
Pablo wanted not to fail.

(132)

a. [La no intervencin] es admirable.

the not meddling
is admirable
Not meddling is admirable.
b. [El no soar] es perder mucho.
the not dream is lose-inf. much
Not to dream is to miss a great deal.

Although adverbs may appear between no and the verb, this order (no-adverbverb) produces constituent negation for the adverb:
(133)

Susana no siempre canta tangos (a veces canta boleros).

S.
not always sings tangos (at times sings boleros)
Susana doesnt always sing tangos, she sometimes sings boleros.52

(134)

Pablo no slo ley ese peridico (adems ley dos libros).

P.
not only read that paper
(also
read two books)
Pablo didnt only read that paper, (he) also read two books.

(135)

?Juan no inmediatamente movi el coche.

J.
not immediately
moved the car
Juan didnt immediately move the car.

The fully grammatical sequences are those whose adverbs are quantier-like
in interpretation.

52

Notice that constituent negation of siempre always is logically equivalent to

propositional negation in this case: Its not the case that (Susana always sings). A
similar eect is observable in (134) with slo only. This could be construed to
imply that the sequence no-adverb-verb is structurally ambiguous between clausal
and adverbial negation. If clausal negation were generally possible with this order,
other adverbs should produce the same ambiguity. Example (135) shows that this is
not generally true.

46

The syntax of Spanish

Subordinate clauses are not necessarily within the scope of negation of a

main clause. Neither (136) nor (137) negates the subordinate clause:
(136)

Juan no dijo que Pedro era inteligente.

J.
not said that P.
was intelligent
Juan didnt say that Pedro was intelligent.

(137)

Susana no lamenta que Pedro sea

inteligente.
S.
not regrets that P.
is-subj. intelligent
Susana doesnt regret that Pedro is intelligent.

In these cases, the events of saying and regretting are negated independently
of the polarity of the events of the complement clauses. However in Spanish
as in other languages, there are specic classes of predicates which seem to
share negation between subordinate and governing clauses (Bosque 1980).
The following pairs are considered as more or less synonymous:
(138)

a. Juan cree
que Jos no ha
llegado.
J.
believes that J.
not has(ind.) arrived
Juan believes that Jos hasnt arrived.
b. Juan no cree
que Jos haya
llegado.
J.
not believes that J.
has(subj.) arrived
Juan doesnt believe that Jos has arrived.

As noted in Bosque (1980), the class of predicates which exhibit this phenomenon is not language-particular. It includes predicates of opinion, expectation,
intention and volition, and of perceptive approximation (seem, be likely)
although individual lexical items may vary somewhat across languages. The
phenomenon has been analyzed transformationally as involving Negraising from the subordinate to the higher clause. As Bosques discussion
indicates however, a movement analysis is problematic in various respects. The
issue remains as to why these particular clausal arguments are interpreted as
within the scope of the governing negation, while other complement clauses
are not.
Multiple negation is possible, both in Neg-raising constructions and in
simple sentences, although the latter is somewhat marginal, and is restricted
to sentences with auxiliary-main verb sequences.53 Multiple negation produces positive polarity:

53

Spanish does not generally admit counterparts to English double negation supported by adverbs such as really, exactly:
(i)

?*No exactamente no llueve.

not exactly
not rains
Its not exactly not raining.

Overview of the grammar

(139)

Juan no crea
que Pedro no llegara.
J.
not believed that P.
not arrived-subj.
Juan didnt believe that Pedro didnt arrive.

(140)

Susana no estaba no estudiando la leccin.

S.
not was
not studying
the lesson
Susana wasnt not studying the lesson.

1.4.6

Copular sentences: Ser and Estar

47

Both ser and estar are copular verbs meaning to be. The category
of complements possible with estar, PP and AP, is a subset of those which are
possible with ser:
(141)

Complements of ser:
a. ___ CP (clefts):
Lo que piensa es [que debe practicar].
What she thinks is [that she should practice].
b. ___ DP (equational sentences):
El siete es [un nmero impar].
Seven is [an odd number].
c. ___ NP (predicate nominals):
Mara es [doctora].
Maria is (a) [doctor].
d. ___ AP:
Susana es [alta].
Susana is [tall].
e. ___ PP:
La reunin es [a las ocho].
The meeting is at eight oclock.

(142)

Complements of estar:
a. __ PP
El lpiz est [en la mesa].
The pencil is on the table.
b. __ AP
Susana est [alta].
Susana is tall.

In many cases, the partition between ser and estar corresponds with the distinction between individual-level predicates and stage-level predicates.
The former ascribe permanent qualities or properties, and the latter ascribe
transient properties. Analyzing estar as the copula for stage-level predicates
would account both for the narrower categorial selection of estar, for its use
with locative PPs, and for the interpretation of adjectives which occur with it.
For example the adjective alta tall in (141d) is understood as an individual

48

The syntax of Spanish

characteristic of Susana; the adjective alta in (142b) is not interpreted as an

individual-level property, but as a changed (recent) property, or a subjective
attribution.
However, Snchez and Camacho (1995) point out problems with this
description of the partition. For example, ser is used with predicates of temporary condition such as expression of the time of day (e.g. (141e)) and the
season. They explore the hypothesis that the basic distinction between the two
is aspectual. The descriptive generalization which emerges is that estar is an
inchoative predicate, while ser has no aspectual properties, and is transparent
to those of its complements.

1.5

Main clauses

Main clauses may have dierent grammatical properties according to

whether they function as declaratives, interrogatives (questions), imperatives
(commands) or exclamatives. This section presents a brief overview of grammatical features that are particular to each of these types.

1.5.1

Declaratives

The constituents of declarative clauses have been described above. As

noted previously, pronominal subjects are normally silent, and overt subjects
may either precede the predicate or follow it. The order of constituents in
declaratives is not xed according to grammatical function (subjectpredicate
order), but may instead be determined in large part by the information load
that constituents have in the discourse. To see this, consider rst that unmarked
order of constituents in declaratives may follow either of the patterns in (143):
(143)

a. Subject Predicate
b. (XP) Predicate Subject

In (143b), (XP) is an optional constituent other than the subject, such as the
modier in (144b):
(144)

a. Susana ley el diario esta maana.

S.
read the paper this morning
b. Esta maana ley el diario Susana.
this morning read the paper S.

The fact that both orders are equally acceptable leads to several questions.
First among these is whether, descriptively, Spanish should be analyzed as
having a basic S-V-O order from which other orders are derived.

Overview of the grammar

49

It has been argued in pre-generative as well as recent generative studies that

an important concept that is relevant for determining constituent order in
declarative sentences is the distinction between presupposed and asserted information (cf. Chapter 5, Section 5.2). Presupposed information is shared, or
known, by the speaker and hearer. Asserted information is not. Other terms
for this distinction are old (or given) versus new information, Theme versus
Rheme, Topic versus Focus (or Comment). The mechanisms which derive
surface order appear to be sensitive to the information content of clausal constituents. For example, the XP which precedes the predicate in (143b) may be
a presupposed constituent (a Topic), as illustrated in (145a), or a Focus constituent, as in (145b):
(145)

a. [Susana]TOPIC [ley el diario

S.
read the paper
b. [Esta maana]FOCUS [ley el
this morning
read the

esta maana]FOCUS.
this morning
diario Susana]TOPIC.
paper S.

Sentence (145a) illustrates a situation in which Susana is the Topic of the discourse, and the predicate contributes new information about her. Sentence
(145b) illustrates a situation in which the speaker and hearer know that Susana
has read the paper; the newly asserted information is that it was this morning
that she read it. The intonation pattern of the sentence distinguishes between
the two meanings. In both sentences, the Focus constituent contains the intonational peak (or main sentence stress). In (145a), this falls within the predicate; in (145b) it falls within the XP esta maana.
The preceding discussion implies that the syntactic component of the
grammar can be sensitive to features corresponding to information content,
rather than to features related to subjecthood, as would be true of xed
word order languages like English or French. The freedom of position of the
subject (see Section 1.3.1) may follow from the fact that the subject constituent may be either presupposed or asserted in most sentences, allowing derivations like those in (145a) or (145b). This generalization is supported by the fact
that certain predicates have subjects that are not freely interpreted as presupposed, and the surface position of the subject is more restricted. (See Hatcher
1956, Contreras 1978, Suer 1982 and references cited.) Presentational predicates are those which, as part of their meaning, introduce the existence or presence of the subject into the discourse. These predicates are not expected to
have derivations corresponding to (145a). This is shown in a neutral context
(146a), where no constituent is presupposed:
(146)

a. Qu pas?
What happened?

50

The syntax of Spanish

b. Empez la resistencia.
began the resistance
The resistance began.
c. ??La resistencia empez.
d. Sali
el sol.
came out the sun
The sun came out.
e. ??El sol sali.

(presentational)

(presentational)

In (146b), the order V-S is the unmarked one for the presentational interpretation of the predicate. The entire sentence is understood as Focus (asserted).
In (146c), the subject appears to the left of the verb, and the subject is analyzed as either the Topic, as in (145a), or a Focus constituent, with the predicate presupposed. Neither of these interpretations is compatible with the
presentational interpretation, which asserts both the subject and predicate.
Similarly, (146d) is presentational, but (146e) is not. The disfavored orders are
perfectly grammatical in other contexts, where the predicate can be interpreted
non-presentationally.
In Chapter 5, the derivation of pre-verbal constituents in declaratives will
be discussed in detail.

1.5.2

Interrogatives

Direct questions include yes/no questions and constituent questions.

Yes/no questions may be marked only by intonation, or by intonation and
constituent order:
(147)

a. Est Mara en casa?

is
M.
at home
b. Mara est en casa?
M.
is
at home?

The intonation of both sentences is distinct from declaratives, in that the tone
remains level or rises after the main sentential stress, rather than falling, as in
declaratives.
Constituent questions, discussed briey above in 1.2.5, normally display
overt movement of an interrogative constituent (Wh-phrase) to clause-initial
position. The subject must be post-verbal, unless the Wh-phrase is an adjunct
(modier) of a certain type. If the Wh-phrase is a modier, the subject may be
pre-verbal:
(148)

a. Qu ley Juan?
what read J.
What did Juan read?
b. *Qu Juan ley?

Overview of the grammar

(149)

a. Por qu Juan dice eso?

why
J.
says that?
Why does Juan say that?
b. Por qu dice Juan eso?
why
says J.
that

(150)

a. En qu medida la constitucin ha contribuido a eso?

in what way
the constitution has contributed to that
In what way has the constitution contributed to that?
b. En qu medida ha contribuido a eso la constitucin?
in what way
has contributed to that the constitution

51

Multiple interrogatives are fully grammatical; one interrogative constituent

moves to clause-initial position, others remain in-situ.54
One issue that has received attention in recent research concerns the derived
position of the verb in interrogatives. Suer (1994) has observed that the verb
is not in the same position as in English, because other phrases can intervene
between the Wh-phrase and the verb:
(151)

A quin jams ofenderas

t con tus acciones?
PA whom never oend-cond. you with your actions
Who(m) would you never oend with your actions?

(152)

Qu

idioma

todava estudia Pepita en su tiempo libre?

(Suer 1994:345)
which language still
studies P.
in her time
free
Which language does Pepita still study in her free time?

Since even the subject can precede the verb in examples like (149) and (150)
above, the verb appears to be at some distance from the preposed Wh-phrase:
(153)

a. Wh-phrase (adverb) Verb . . .

b. Wh-phrase (subject) Verb . . .

Wh-phrases have the same pattern of clitic-doubling as do other nonpronominal phrases. That is, Wh-phrases corresponding to indirect objects
have a clitic double (see 1.2.4.), but those corresponding to direct objects do
not. (See 1.7 for discussion of dialect variation.)
54

Interrogative constituents also remain in-situ with a non-echo interpretation if the

Wh-phrase is contained within an island a syntactic constituent from which
movement is impossible Arnaiz (1996). One type of island is the adjunct clause:
(i).

a. Juan llegara
[antes de saludar a quin]?
J.
arrive-cond. before of greet-inf. PA who?
Juan would arrive before greeting whom?
b. *A quin llegara
Juan [antes de saludar]?
PA who arrive-cond. J.
before of greet-inf.
Who would Juan arrive before greeting?

52

The syntax of Spanish

(154)

a. Qu (*lo)
ley Juan?
what CL(Acc.) read J.
b. A quin (*lo)
vio Juan?
PA whom CL(Acc.) saw J.
Whom did Juan see?

Preposition stranding is never possible. Interrogatives within Prepositional

Phrases move as a phrase:
(155)

a. [Con quin] habl Susana?

with whom spoke S.
With whom did Susana speak?
b. *Quin habl Susana con?
whom spoke S.
with
Whom did Susana speak with?

Chapter 6 discusses interrogatives, focusing primarily on the derived position of the Wh-phrase, the subject, and the verb.

1.5.3

Imperatives

Imperatives may be any person/number except rst person singular.

Positive and negative imperatives have distinct forms and syntax. Negative
commands are identical to Present tense subjunctive forms, and show the same
person/number endings as other verb forms:
(156)

Cantar sing Negative Imperative forms (=pr. subjunctive):

a.
1st.pl.
No cant-emos
b. 2nd.sg.
No cant-es
2nd.pl.
No cant-is
c. 3rd.sg.
No cant-e
3rd.pl.
No cant-en

The order of negation and clitics is the same as for other nite verb forms, with
clitics preceding the verb:
(157)

No se
lo
mande.
not CL(Dat.) CL(Acc.) send-I.
Dont send it to him/her.

Positive imperatives have unique forms in the second person, while the rst
and third person forms are identical to Present subjunctive forms:
(158) Cantar sing Positive Imperative forms:
a.
1st.sg.
b. 2nd.sg.
cant-a
2nd.pl.
c. 3rd.sg.
cant-e
3rd.pl.

cant-emos
cant-ad
cant-en

In the 2nd person singular, the person ending -s is absent, and rather than a
subjunctive form, the desinence is identical to the third person singular

Overview of the grammar

53

Present indicative.55 In the second person plural, the -d is unique to the imperative. Positive imperatives precede clitics:
(159)

a. Cntalo!
sing+CL(Acc.)
Sing it!
b. Mndeselo!
send+CL(Dat.)+CL(Acc.)
Send it to him/her!

Imperatives exclude stative predicates and auxiliaries. Subject pronouns may

be overt, but cannot precede the verb, unless separated by a pause.
Desideratives are expressions of wish or hope, formed with the Present or
Past (counterfactual) subjunctive. The clause is introduced either by the complementizer que or by the expression ojal (etymologically derived from an
expression meaning may Allah will it). This phrase is optionally followed by
que:
(160)

a. (Ojal) que tengan

buen viaje.
that have-pr.subj.3rd.pl. good trip
May they have a good trip.
b. (Ojal) que estuvieran
aqu.
that be-pa.subj.3rd.pl. here
I wish that they were here.

1.5.4

Exclamatives

Exclamatives include the constituents shown in (161), where the

initial XP is the focus of exclamation:
(161)

XP (que) V (Subj.)

Que is the complementizer, and its distribution depends on the nature of the
focus XP. The verb precedes the subject in exclamatives.
Adjectival and adverbial phrases in the XP position may be specied by
qu/cun how or by lo, the neuter determiner/pronominal, with degree interpretation:
(162)

55

a. Qu/cun lista (que) es Miriam!

how
smart (that) is M.
How smart Miriam is!

For rst and second conjugation verbs, the indicative (third person singular) form is
identical to the innitive with its conjugation class or theme-vowel, minus the nal
consonant -r. Third conjugation verbs dier in this vowel, and show that the imperative is formed on the indicative. For example, escribir to write, escribe he/she
writes, escrib-e write!

54

The syntax of Spanish

b. Lo lista * (que) es Miriam!
neut. smart that is M. (=162a)

(163)

a. Qu/cuan rpido (que) escribe Susana!

how
quickly (that) writes S.
How quickly Susana writes!
b. Lo
rpido *(que) escribe Susana!
neut. quickly that writes S. (=163a)

Adverbials of quantity in the XP position are expressed as cmo/cunto

how/how much or as lo. As in the preceding examples, lo requires that que
be overt:
(164)

a. Cmo/cunto (que) escribe Juan!

how (much) (that) writes J.
How (much) Juan writes!
b. Lo *(que) escribe Juan!
neut. that writes J. (=164a)

Quantied NPs in the XP constituent are introduced by cuntos/cuntas how

many. In peninsular and some Latin American dialects, the quantied NP
can be introduced by qu what, or the feminine determiner la:
(165)

a. Cuntos premios (que) gan Susana!

how many prizes (that) won S.
How many prizes Susana won!
b. Qu de premios (que) gan Susana!
what of prizes (that) won S.
What (a lot) of prizes Susana won!
c. La de premios *(que) gan Susana.
the of prizes
that won S.
The (amount) of prizes Susana won!

As the preceding examples show, que is optional where XP contains a form

equivalent to the interrogative. Que is required where neuter lo or feminine la
appears in XP. Note that the nal two examples in (165) contain partitive NPs
(introduced by de). The partitive form, as well as the feminine form of the
determiner in (165c), suggests that the head of the NP is a non-overt feminine
noun cantidad quantity.

1.6

Subordinate clauses

1.6.1

Argument clauses

Argument clauses, sometimes called nominal clauses, function as

subject or complement (direct, indirect or prepositional) of a governing

Overview of the grammar

55

predicate. The complementizer que that precedes declarative subordinate

clauses if the clause is nite (subjunctive or indicative). Innitival clauses have
no overt complementizer:
(166)

a. Quieren
[*(que) Sandra vaya
pronto].
want-pr.3rd.pl. that S.
go-pr.subj. soon
They want *(that) Sandra go soon.
b. Quieren
[(*que) ir
pronto].
want-pr.3rd.pl. that go-inf. soon
They want (*that) to go soon.

Campos (1993:63) observes that clauses which are complements of certain

verbs (of fear, will and desire) allow omission of que, but only if there is no
overt pre-verbal subject:
(167)

a. Temo
[estn
enojados].
fear-pr.1st.sg. be-pr.subj.3rd.pl. angry
Im afraid (they) are angry.
b. *Temo
[ellos estn
enojados].
fear-pr.1st.sg. they be-pr.subj.3rd.pl. angry

Complementizer omission is subject to dialectal variation and to varied judgments according to register. For some speakers, (167a) is impossible in colloquial Spanish.
Argument clauses may contain an indirect question or exclamative if the
governing predicate so selects:
(168)

Me pregunto
[(que) por qu sali
Mara].
CL wonder-pr.1st.sg. (that) why
leave-pa.3rd.sg. M.
I wonder (that) why Maria left.

(169)

Mira [qu alta

(que) ests].
look-I how
tall (that) be-pr.2nd.sg.
Look at how tall (that) you are!

Notice in both (168) and (169) the complementizer que is optional. In (169)
the complementizer follows the XP focus of exclamation. This mirrors the
pattern for direct exclamations discussed in the previous subsection. The
pattern in (168), however, diers from direct questions, where the complementizer does not surface.56
Argument clauses exhibit case markers (prepositions) parallel to corresponding NP arguments. Subject clauses are never preceded by prepositions,

56

Plann (1982) shows that under certain verbs of communication such as decir say,
the overt complementizer is necessary for a true interrogative interpretation. Under
such verbs, if que is suppressed, the interpretation is pseudo-interrogative.

56

The syntax of Spanish

nor are clauses which occur as direct objects of a verb, in standard dialects.57
Clauses which function as the indirect object of a verb are preceded by the
Dative preposition a. These clauses are not clitic-doubled:
(170)

(*Le)
obligaron
a Pedro [a
terminar el
proyecto].
CL(Dat.) obliged-pa.3rd.pl. PA P.
to(Dat.) nish-inf. the
project
They forced Pedro to nish the project.

Clauses which are complements of adjectives and nouns are preceded by de

of, parallel to NP counterparts:
(171)

a. Patricio est orgulloso *(de) su hermana.

P.
is
proud *(of)
his sister.
b. Patricio est orgulloso *(de) que su hermana hizo un gol en
P.
is
proud (*of)
that his sister
made a goal in
ese partido.
that game
Patricio is proud (*of) [that his sister scored a goal in that game].

1.6.2

Relative clauses

Relative clauses, traditionally called adjective clauses (because they function

as modiers of nouns), have been analyzed as sharing certain basic properties
with interrogatives. Compare (172a), (172b):
(172)

a. A quin conoci Juan?

PA whom met
J.
Whom did Juan meet?
b. la persona [a quien conoci Juan] 58
the person PA whom met
J.
the person whom Juan met

In (172a), the interrogative a quin is understood as the complement of the

verb, and lls a gap corresponding to the missing complement. Likewise, in
(172b), the relativized phrase a quien is related to a direct object gap within the
relative clause. The central dierence between the two constructions, of course,
is that the relative pronoun has an antecedent, la persona, while interrogative
57

58

In some dialects, the preposition de of is inserted before clauses which are direct
complements of verbs. This phenomenon, called dequesmo, is considered substandard.
By convention, the interrogative forms are orthographically dierentiated by an
accent on the stressed syllable when they are homophonous with relative pronouns
or the complementizer que. The convention does not signify a dierence in strong
versus weak (clitic) status.

Overview of the grammar

57

phrases do not. Consequently, the manner in which reference is derived diers

in the two constructions.
There are two signicant grammatical dierences between interrogatives
and relative clauses in Spanish. One concerns constituent order, illustrated in
(173):
(173)

a. *A quin Juan vio?

PA whom J.
saw
Whom did Juan see?
b. la persona [a quien Juan vio]
the person PA whom J.
saw

As noted previously, sentences with interrogatives corresponding to arguments of the verb require subjectverb inversion; as shown in (173b), relatives
do not.
The second dierence concerns the form (including overtness) of relativized
constituents, as opposed to corresponding interrogative phrases. The form of
relativized constituents varies depending on: (a) whether the relative clause is
restrictive or non-restrictive, (b) the grammatical function of the relative
phrase within the relative clause, and (c) the features (gender, number, deniteness and, in some cases, [ ]) of the antecedent. The following discussion will summarize properties of relatives with overt antecedents, where the
relativized constituent corresponds to an argument.59
Consider rst a restrictive relative in which the relativized phrase is the
subject of the relative clause:
(174)

a. el hombre [que trabaja aqu]

the man
that works here
b. *el hombre [quien trabaja aqu]
the man
who works here

The relative pronoun quien is impossible as subject; this contrasts with the relativized object above in (173b), and also with the relativized subject of a nonrestrictive relative:
(175)

ese hombre, [quien/que trabaja

aqu]
that man,
who/that work-pr.3rd.sg. here
that man, who/*that works here

For some speakers, quien is preferred in non-restrictive relative clauses.

Direct object relativization of a [+ ] argument has the forms in
(176):
59

For discussion of other relative clause types, including innitival relatives and free
relatives, see Campos (1993:97.), Plann (1980).

58

The syntax of Spanish

(176)

a. el hombre [que conoc

ayer]
the man
that meet-pa.1sg. yesterday
the man that I met yesterday
b. el hombre [a quien
conoc
ayer]
the man
PA whom-sg. meet-pa.1sg. yesterday
the man whom I met yesterday
c. el hombre [al
que conoc ayer]
the man
PA+the that met.1sg. yesterday
the man that I met yesterday
d. el hombre [al
cual conoc
ayer]
the man
PA+the which meet-pa.-1st.sg. yesterday
the man that I met yesterday

Notice in (176a) that que is not preceded by personal a, but other forms in
(176) are. This suggests that (176a) contains a covert form of a quien, and that
que is the complementizer. For some speakers, non-restrictive relatives pattern
the opposite way, disallowing covert a quien:
(177)

esos hombres, [a quienes conoc

ayer]
those men,
PA whom meet-pa.1st.sg. yesterday
Those men, that I met yesterday

(178)

(*)esos hombres, [que conoc

ayer] (=177)
those men
that meet-pa.1st.sg. yesterday

The relative forms in (176) are not sensitive to the deniteness of the antecedent (cf. un hombre que conoc a man that I met; un hombre a quien conoc
a man whom I met).60
Recall that the Spanish denite determiners show number/gender agreement. In sentences whose antecedent is invariant with respect to
number/gender features (e.g., algo something, nada nothing) the neuter
60

None of the relative constructions are sensitive to Case-matching of the gap with
the antecedent. In the following examples, Accusative/Nominative and Nominative
/Accusative are paired:
(i)

(ii)

El hombre [al
que conoc
ayer]
present su
trabajo hoy.
the man
PA+the that meet-pa.1st.sg. yesterday presented his
paper today
The man that I met yesterday presented his paper today.
Esta maana conoc
al
hombre que present su trabajo
ayer
this morning meet-pa.1st.sg. PA+the man
that presented his paper
yesterday
This morning I met the man that presented his paper yesterday.

Overview of the grammar

59

form lo replaces the denite article (cf. No dijo nada con lo cual yo est de
acuerdo He didnt say anything with [lo+] which I agree).
Restrictive relativization of [-HUMAN] direct objects requires que; personal a is absent, and forms with quien, el que and el cual are impossible:
(179)

a. el cuaderno [que Juan perdi]

the notebook that J.
lost
b. *el cuaderno [al/el
que Juan perdi]
the notebook PA+the/the that J.
lost
c. *el cuaderno [al/el
cual Juan perdi]
the notebook PA+the/ the which Juan lost

(180)

a. *un
a
b. *un
a

cuaderno [el
notebook the
cuaderno [el
notebook the

que Juan perdi]

that J.
lost
cual Juan perdi]
that J.
lost

Relative constituents corresponding to indirect objects and other prepositional phrases require a preposition. An unspecied covert relative pronoun is
possible only with [- ] antecedents; compare (181d), (182):
(181)

a. la persona
[de quien hablamos]
the person-f.sg. of whom speak-pa.1st.pl
the person of whom we spoke
b. la persona
[de la
que hablamos]
the person-f.sg. of the-f.sg. that speak-pa.1st.pl.
the person that we spoke of
c. la persona
[de la
cual hablamos]
the person-f.sg. of the-f.sg. which speak-pa.1st.pl.
the person that we spoke of
d. *la persona
[de que hablamos]
the person-f.sg. of that speak-pa.1st.pl.
the person that we spoke of

(182)

a. la cuestin [de que hablamos]

the issue-f.sg. of that speak-pa.1st.pl.
the issue that we spoke of
b. la cuestin [de la
que hablamos]
the issue-f.sg. of the-f.sg. that speak-pa.1st.pl.
the issue that we spoke of
c. la cuestin [de la
cual hablamos]
the issue-f.sg. of the-f.sg. which speak-pa.1st.pl.
the issue that we spoke of

As in other constructions, the indirect object generally requires a clitic double,

e.g. la persona [a la que *(le) mand la carta] the person to whom I sent the
letter.

60

The syntax of Spanish

Relativization of possessives is shown in (183). The relative pronoun agrees in

number and gender with the object possessed, rather than with the antecedent:61
(183)

a. el hombre [cuya
esposa
est en Cuba]
the man
whose-f.sg. wife-f.sg. is
in Cuba
the man whose wife is in Cuba
b. el artculo
[a cuyos
autores
conoc
ayer]
the article-m.sg PA whose-m.pl. authors-m.pl. meet-pa.1st.sg.
yesterday
the article whose authors I met yesterday
c. el curso
[para cuyos
exmenes estudiaron
mucho]
the course-m.sg. for whose-m.pl. exams
study-pa.3rd.pl. a lot
the course for whose exams they studied a lot

The selection of subjunctive versus indicative mood in relative clauses

depends on features of the antecedent. Non-existing, negated and non-specic
antecedents trigger a subjunctive relative clause; otherwise the relative clause
is indicative. This is illustrated below for a negative antecedent (184a), which
triggers subjunctive, versus an indenite specic antecedent (184b), which
triggers indicative:
(184)

a. No conozco
(a) nadie
que hable
diez
idiomas.
not know-pr.1st.sg. PA anybody that speak-pr.subj.3rd.sg. ten
languages
I dont know anyone who speaks ten languages.
b. Conozco
a alguien que habla
diez idiomas.
know-pr.1st.sg. PA someone that speak-pr.ind.3rd.sg ten languages
I know someone that speaks ten languages.

1.6.3

Temporal adjunct clauses

Finite temporal clauses introduce secondary relationships of ordering

between events. A simple nite main clause expresses a relation between the
occurrence of a state or event and a deictic moment now, often referred to
as Speech-time, following Reichenbach (1947). For example, in (185),
(185)

61

Mara cant/canta/cantar el sabado.

M.
sang/sings/will sing Saturday.

There is also a possessive relative with que su instead of cuyo:

(i)

el artculo que sus autores los conoc

ayer
the article that their authors CL meet-pret.1st.sg. yesterday
the article whose authors I met yesterday

Overview of the grammar

61

the event of Marias singing precedes, coincides with or follows Speech-time.

Temporal adjunct clauses do not alter or further specify the tense interpretation of the simple clause, but instead order the event in relation to other
events:
(186)

a. Mara cant antes de que llegramos.

M.
sang before of that arrive-pa.subj.1st.pl.
Maria sang before we arrived.
b. Mara cant mientras escriba.
M.
sang while
write-pr.ind.3rd.sg.
Maria sang while he/she wrote.
c. Mara cant despus de que llegamos.
M.
sang after
of that arrive-pa.ind.1st.pl.
Maria sang after we arrived.

The clauses in (186) exhibit sequence-of-tense, that is both clauses bear a

[+PAST] tense. Although Spanish dierentiates present and future tense
morphology, sequence-of-tense requires only agreement with respect to
[PAST]:
(187)

a. Mara cantar mientras escribo.

M.
sing-fut. while
write-pr.ind.1st.sg.
Maria will sing while I write.
b. *Mara cantar despus de que sal.
M.
sing-fut. after
of that leave-pa.ind.1st.sg.
Mara will sing after I left.

Temporal clauses may also express an ordering relationship between an

events onset or end and a secondary event:
(188)

a. Estudian
espaol desde hace
un ao.
study-pr.ind.3rd.pl. Spanish since do-pr.ind.3rd.sg. one year
They have studied Spanish for a year.
b. Conozco
a Susana desde mi niez.
know-pr.ind.1st.sg. PA S.
since my childhood
Ive known Susana since my childhood.

In (188), the temporal clause species a time which marks the onset of the
main clause activity or state. In the following example, the endpoint of the
main clause event is modied by the temporal clause:
(189)

No acepto
la propuesta hasta que me
escuches.
not accept-pr.ind.1st.sg. the proposal until that CL(Dat.) listenpr.subj.2nd.sg.
I wont accept the proposal until you listen to me.

62

The syntax of Spanish

In (189), the time of the event of the temporal clause marks the end of the
predicate no acepto la propuesta.62 A similar interpretation results if the temporal clause is introduced by mientras (que) as long as:
(190)

No acepto
la propuesta mientras (que) no me
escuches.
not accept-pr.ind.1st.sg. the proposal while
that not CL(Dat.)
listen-pr.subj.2nd.sg.
I wont accept the proposal as long as you dont listen to me.

In (190) as in (189), the time at which me escuches you listen to me occurs

marks the termination of not accepting the proposal.

1.6.4

Gerundive and participial adjuncts

Gerundive clauses contain a verb in the form of a present participle,

i.e., bearing the non-agreeing ax -ndo -ing. The participle appears in
clause-initial position:
(191)

a. habiendo terminado la reunin

have-prt. nish-pprt. the meeting
having nished the meeting
b. *la reunin habiendo terminado

The predicate of gerundive clauses may contain any elements normally possible for non-nite verb phrases, including a full range of auxiliaries, negation,
subject-oriented adverbs and clitics. As in other constructions, clitics follow
the non-nite verb:63
(192)

62

63

a. leyndolo
cuidadosamente
read-prt.+CL(Acc.) carefully
reading it carefully
b. *lo leyendo cuidadosamente

By inference the polarity of the main clause is reversed at the point in time marked
by (the completion of) the adjunct clause event. For further discussion of negation
in Italian counterparts to these contexts see Tovena (1996).
The occurrence of clitics shows that the verb is not nominalized. Clitics do not occur
with derived nominals:
(i)

a. Mara lo
ley.
M.
CL(Acc.) read
Maria read it.
b. su lectura
del
poema
her reading-f.sg. of+the poem
her reading of the poem
c. *su lo
lectura
her CL(Acc.) reading
her reading it

Overview of the grammar

63

Overt subjects of gerundives are Nominative, never Genitive, and must be

post-verbal:
(193)

a. estando t
en la sala
be-prt. you(Nom.) in the room
you being in the room
b. *t estando en la sala
c. *estando tuyo
en la sala
be-prt. your(Gen.) in the room
your being in the room

As adjuncts, gerundive clauses may be more or less closely linked to the

main predicate. In the following examples, # represents obligatory pause
intonation:
(194)

a. Fuimos
a cenar [hablando de la pelcula].
go-pa.1st.pl. to dine speak-prt. of the lm
We went to eat talking about the lm.
b. Fuimos
a cenar# habiendo salido
el sol.
go-pa.1st.pl. to dine
have-prt. come-out the sun
We went to eat, the sun having come out.

In sentence (194a), which does not require pause intonation, the subjects of
the clauses are coreferential and the event of the gerundive clause temporally
includes the event of the main clause. The gerundive clause in (194a) might be
described as a subject-oriented secondary predicate. In (194b), the subjects of
the two clauses are not coreferential, and the adjunct event does not (necessarily) temporally include the main clause event.
Participial clauses, like gerundive clauses, contain a clause-initial participle
and a Nominative subject. However, the participle in this instance is a past
participle, and these participles show number/gender agreement with the
subject:
(195)

a. terminada
la reunin
nish-pprt.f.sg. the meeting-f.sg.
the meeting nished
b. *la reunin terminada64
c. *[terminado
la reunin]
nish-pprt.m.sg. the meeting-f.sg.

As (195b) shows, the subject cannot precede the participle; failure of agreement in (195c) is also ungrammatical.
Participial clause predicates do not display full Verb Phrase structure. The
64

Bello ((1847) 1971:315, para.1178e) notes that the subject may in exceptional cases
precede the participle, as in Esto dicho that said.

64

The syntax of Spanish

participle may not be negated, it does not bear clitics, and auxiliaries are not
possible:
(196)

a. *[No terminado el proyecto], Juan sigui

trabajando.
not nish-pprt. the project,
J.
continued working
b. *manddole
el paquete (a
Juan)
send-pprt.-CL(Dat.) the package (to(Dat.) Juan)
c. *habido
terminado el proyecto
have-pprt. nish-pprt. the project
d. *sido
publicado
el artculo
be-pprt. publish-pprt. the article

Number/gender agreement makes participial clauses appear to be formed on

the passive rather than the past participle. The verbs which appear in participial clauses, however, include both those which can be passivized and others
which cannot. Perfective predicates, those which eect a change of state or
location which delimits the event, can appear in participial clauses. Example
(197) illustrates the contrast between transitive predicates with aected and
those with non-aected arguments:
(197)

a. vendidas
las ores
sell-pprt.f.pl. the ower-f.pl.
b. pintada
la casa
paint-pprt.f.sg the house-f.sg.
c. roto
el vaso
break-pprt.m.sg. the glass-m.sg.
d. ?vista
Mara
see-pprt.f.sg. M.
e. ?reconocido
el extranjero
recognize-pprt.m.sg. the foreigner-m.sg.
f. *amada
Mara
love-pprt.f.sg. M.

A similar contrast is observed among unaccusative verbs (those verbs whose

subjects behave as though they are generated as objects). Participial clauses
may be formed only on those verbs which eect a change of state or location:
(198)

a. ?venidos
los huspedes
came-pprt.m.pl. the guest-m.pl.
b. ?llegados
los estudiantes
arrive-pprt.m.pl. the student-m.pl.
c. ?salida
Mara
leave-pprt.f.sg. M.
d. *faltados
los comestibles
lack-pprt.m.pl. the foodstu-m.pl.
e. *quedadas
las tortillas
remain-pprt.f.pl. the tortilla-f.pl.

Overview of the grammar

65

Participial clauses are temporally related to the main predicate, albeit indirectly. The participial clause describes a state which precedes the onset of the
main clause event. However, this temporal connection is not strict. Consider
for example (199):
(199)

[Vendido
el coche],
salimos
a celebrar.
sell-pprt.m.sg. the car-m.sg. go(out)-pa.1st.pl. to celebrate-inf.
[The car sold], we went out to celebrate.

The event of selling the car precedes (and triggers) the event of going out to
celebrate. However, the two events seem to be independent of each other, at
least in the sense that they are viewed as distinct intervals, rather than as parts
of a single interval. There are arbitrarily many other events that could occur
between the two.

1.6.5

Innitival adjuncts

Innitival adjunct clauses are complements of prepositions, a upon;

on, de of, con with and sin without:
(200)

a. [Al
encontrar el artculo], lo le.
on+the nd-inf. the article CL(Acc.) read
On nding the article, I read it.
b. [De venir
Mara], haremos
paella.
of come-inf. M.
make-fut.1st.pl. paella
If Maria comes, well make paella.
c. [Con protestar t], no ganan
nada.
with protest-inf. you not gain-pr.3rd.pl. anything
With you protesting, they wont gain anything.
d. [Sin
saberlo
yo], los nios
salieron
a la calle.
without know-inf.CL I
the children go-pa.3rd.pl. to the street
Without my knowing it, the children went out in to the street.

Clauses introduced by a (which is contracted with determiner el to form al)

are temporal or aspectual modiers of the main clause. For some speakers,
they may also express cause. The event of the innitival clause precedes the
onset of the event of the main clause (or causes it). The innitival event must
be non-stative, if it is a temporal modier, as in (201a); if it is a causal modier, a stative is possible, as in (201b):
(201)

a. *[Al
parecer que iba
a llover], nos quedamos
en casa.
on+the seem-inf. that go-pa. to rain
CL stay.pa.1st.pl. at home
On seeming that it was going to rain, we stayed home.
b. [Al
ser francs Juan], no le pidieron
pasaporte.
on+the be French J.,
neg. CL ask.pret.3rd.pl. passport
As Juan is French, they didnt ask for his passport.

66

The syntax of Spanish

When the subject of the adjunct is not overt, it may be understood as coreferential with the main clause subject. However, coreference is not obligatory, as
shown in (202):
(202)

[Al
dejar
de llover] salimos
a caminar.
on+the stop-inf. of rain-inf. leave-pa.3rd.pl. to walk-inf.
Since it stopped raining, we went out to walk.

The predicates of a clauses display characteristics of full Verb Phrase structure, including the possibility of adverbs, clitics and passive morphology:
(203)

a. [Al
ser
publicado
el artculo], se lo mandamos
a
Julio.
on+the be-inf. publish.pprt. the article
CL CL send.pret.1st.pl. to
J.
On the articles being published, we sent it to Julio.
b. [Al
mandrselo
Jos], lo ley.
on+the send-inf.+CL(Dat.)-CL(Acc.) J.,
CL read.pret.3rd.sg.
On Joss sending it to him/her, he/she read it.

Copular verbs are possible, perhaps with a preference for a causal interpretation:
(204)

a. ?[Al
ser piloto (Roberto)] hicimos
un viaje.
on+the be pilot
make.pret.1st.pl. a trip
On (Roberts) being a pilot, we took a trip.
b. Al
ser piloto Roberto, pudo
salvar
el avin.
on+the be pilot R.
manage.pret. save-inf. the airplane
Since Roberto was a pilot, he managed to save the airplane.

De-clauses, as in (200b) above, are conditional in interpretation. These

clauses may have overt or non-overt subjects, and coreference between the
adjunct and main clause subject is not obligatory:
(205)

[De haber
pblico], actuaremos.
of have-inf. audience perform-fut.3rd.pl.
If theres an audience, well perform.

The predicate of de-clauses shows characteristic elements of the Verb Phrase:

it may be negated and accepts clitics and auxiliaries: de no haberlo entendido
if s/he hasnt understood it.
De-innitival clauses alternate with relativized degree phrases:
(206)

Mara se
enferm
de
lo mucho que comi.
M.
CL-(inc.) sicken-pa.3rd.sg. from that much that eat-pa.3rd.sg.
Maria got sick from how much she ate.

Overview of the grammar

67

The nite clause may have a non-coreferring subject, and may or may not be
intonationally separated from the main clause:65
(207)

Mara se deprimi (#)

de
lo mucho que llovi.
M.
CL depress-pa.3rd.sg. from that much that rain-pa.3rd.sg.
Maria got depressed from how much it rained.

Innitivals introduced by con are described by Campos (1993:175) as concessives (having a negative conditional despite interpretation). They typically co-occur with the adverb aun even, or with a negated main clause:
(208)

[(Aun) con estudiar] no aprendes

mucho.
even with study-inf. not learn-pr.2nd.sg. a lot
Even if you study, you dont learn a lot.

Con-innitivals admit overt nominative subjects, as shown in (209):

(209)

[Aun con protestar t]

no tendrs
xito.
even with protest-inf. you(Nom.) not have-fut.2nd.sg. success
Even if you protest, you wont succeed.

Non-overt subjects of con-innitivals are necessarily coreferential with that of

the modied clause, as shown by the degraded status of (210):
(210)

?*Aun con hacer buen tiempo, no llegu

a la ocina a
tiempo.
even with make good weather not arrive-pa.1st.sg. to the oce on
time
Even with the good weather, I didnt arrive at the oce on time.

The predicate of con-innitivals shows characteristics of full Verb Phrase

structure, accepting clitics and auxiliaries, as shown in the following example
from Campos (1993:175):
(211)

(Aun) con haberse

levantado temprano
even with have-inf.CL(3rd.sg.) arise-pprt. early
(even) having gotten himself up early

Innitivals introduced by sin without alternate with nite clauses. The

innitive is selected if the clausal subjects are coreferential; otherwise a nite
clause occurs:
65

De also introduces an adjectival small clause with similar interpretation:

(i)

a. [de
lo tan contento que estaba]
from that so happy
that be-pa.3rd.sg.
from how happy s/he was
b. [de
lo contento que estaba]
from that happy
that be-pa.3rd.sg.

68

The syntax of Spanish

(212)

a. Saqu
el libro [sin
mirarlo].
take(out)-pa.1st.sg. the book without look(at)-inf.+CL(Acc.)
I took out the book without looking at it.
b. Saqu
el libro [sin
que Juan lo
mirara].
take(out)-pa.1st.sg. the book without that J.
CL(Acc.) look(at)pa.subj.3rd.sg.
I took out the book without Juans looking at it.

The predicate of sin-innitivals shows full structure of a verb phrase; the predicate may include auxiliaries and clitics. Because sin is a negative element, an
additional negation is possible only with contrastive stress:
(213)

?Sin
NO conversar, nos entendemos.
without NOT talking,
we get along.
We get along without NOT talking.

1.7

Syntactic dialects

This section summarizes several forms of dialectal variation found in

Latin American Spanish, drawing on data summarized in Lipski (1994). Some
information on syntactic variation in peninsular dialects is found in Alvar
(1996).

1.7.1

Clitic-related variation

Latin American dialects display several variations in the grammar of

clitics. In all dialects, pronominal and anaphoric direct object full phrases cooccur with a clitic, which is referred to as a clitic double:
(214)

a. Mara lo
vio
a l.
M.
CL(Acc.) saw-pa.3rd.sg. PA him(Acc.).
Maria saw him.
b. Mara se
vio a s misma.
M.
CL(Acc.Re.) saw PA Acc.re-3rd.sg.
Maria saw herself.

In some dialects, clitic-doubling occurs with non-pronominal direct objects,

including inanimate objects, as in (215b):
(215)

a. Lo
conozco
a Juan.
CL(Acc.) know-1st.sg. PA J.
I know Juan.
b. T lo
tienes
la

(Argentina; Lipski 1994:174)

direccin.
(Bolivia; Lipski 1994:191)
you CL(Acc.m.sg.) have-2nd.sg. the address(f.sg.)
You have the address.

Overview of the grammar

69

The example cited in (215b) has a non-agreeing clitic double: the clitic has the
unmarked masculine singular form, although the full phrase direct object is
feminine. Non-agreement of clitics seems to occur commonly in areas of
direct object clitic doubling. These are found in the Ro de la Plata region of
Argentina and Uruguay, in Bolivia, Ecuador, Peru and Mexico. For discussion
of direct object clitic doubling, see Jaeggli (1982), Lujn (1987), Suer (1988),
Kayne (1994) and references cited.
Another clitic-related phenomenon that overlaps with direct object cliticdoubling is the absence of clitics in contexts where they are normally required,
such as preposed phrases, as in (216), and ordinary pronominals, giving rise
to null objects, as in (217):
(216)

Su bebito tambin [lo]

tena.
(Per; Lipski 1994:325)
her baby also
[CL(Acc.)] had-imp.3rd.sg.
She also had her baby.

(217)

(Aqu estn los medicamentos.) Cmo

Here are

the medicines.

How

has trado?
(Bolivia; Lipski 1994:191)
have you brought (them)?

In (217), the antecedent of the null direct object is a denite NP. This type of
null object is found in Andean Spanish (Bolivia, Colombia, Ecuador and
Per). In other dialects, the clitic would be required in (217). Null objects are
generally possible only if they are indenite, as in (218b):
(218)

a. Compraste caf?
You bought coee?
b. S,
compr.
Yes, I bought (some).

A third phenomenon related to clitics concerns the use of the direct object
clitic as a double with prepositional or clausal complements:
(219)

a. Se
lo
fue de viaje.
CL(re.) CL(Acc.) went of trip
(S)he left on a trip.
b. Lo
temo
que se

(Honduras Lipski 1994:272)

muera.
(Nicaragua Lipski 1994:292)
CL(Acc.) fear-1st.sg. that CL(re.) die-subj.3rd.sg.
Im afraid that he will die.

The use of clitics as doubles of PPs and clauses has been recorded in northern Latin American regions: Colombia, Honduras, Nicaragua and Mexico.
One nal clitic-related phenomenon that Lipski mentions in his survey is the
reduplication of clitics in contexts where they could appear in one position
or another:

70

The syntax of Spanish

(220)

a. No la
he podido conocerla.
(Bolivia; Lipski 1994:191)
not CL(Acc.) have been able know-inf.+CL(Acc.)
I havent been able to meet her.
b. Me
est castigndome.
(Per; Lipski 1994:191)
CL(Acc.) is
punish-prt.+CL(Acc.)
(S)he is punishing me.

1.7.2

Negation-related variation

Three types of negation-related variation have been attested in dierent areas. In the Ro de la Plata region of Argentina and Uruguay, pre-verbal
negative phrases like nadie nobody co-occur with negation:
(221)

Nadie no est.
nobody(not) is-pr.3rd.sg.
Nobody is there.

As discussed in 1.2.7 above, the co-occurrence of no not with such phrases

is standardly possible only if the negative phrase is post-verbal: No est nadie,
but Nadie est Nobody is there.
A second negation-related form of variation is the reduplication of no:
(222)

a. No hablo

ingls

no.

(Colombia (Pacic coast) Lipski 1994:215)

not speak-1st.sg. English not
I dont speak English.
b. Nosotros no vamos
no.
(Dominican Republic Lipski 1994:242)
we
not go-1st.pl. not
Were not going.

This phenomenon is said in Lipski (1994:242) to be frequent in vernacular

Dominican Spanish.
A third type, noted in Bosque (1980), is null negation with hasta until,
attested in Colombia and Mexico:
(223)

Vamos hasta que termine.

go-pret.1st.pl. until that end.pr.subj.3rd.sg.
We wont go until it ends.

1.7.3

Intensive ser

The phenomenon referred to as the intensive ser construction is

illustrated in (224):
(224)

a. Lo

hice

fue

en el verano.
(Colombia Lipski 1994:215)
CL(Acc.) do-pa.3rd.sg. be-pret.3rd.sg. in the summer
I did it in the summer.

Overview of the grammar

71

en la esta.
(Panama Lipski 1994:301)
CL(Acc.) meet-pret.1st.sg. be-pret.3rd.sg. at the party
Where I met him was at the party.
c. Me
jaba
era
en la luz.
CL(1st.sg.) pay attention-imp.1st.sg. be-imp.3rd.sg. to the light
What I was paying attention to was the [trac] light.
(Panama Lipski 1994:301)

b. Lo

conoc

fue

The glosses given for (224b) and (224c) suggest that intensive ser is a form of
cleft sentence marker. Kany (1951) documents the phenomenon as occurring
in Colombia and Ecuador as far back as the mid nineteenth century. Lipski
mentions it as a recent phenomenon in Venezuela, and as becoming quite frequent in Panama City.

1.7.4

Possessives

It was noted in Section 1.2.2 above that the Spanish genitive construction is marked either by a possessive form (mis libros, los libros mos my
books) or by a possessive phrase introduced by de (el retrato de Josena
Josenas portrait). Dialectal variation takes various forms:
(225)

a. de la Mara su casa
(Bolivia Lipski 1994:194)
of the M.
her house
Marias house
b. mi casa ma
(Honduras Lipski 1994:272)
my house my
my house
c. hijo de un su pap
(El Salvador Lipski 1994:259)
son of a his father
his fathers son

In (225a), the de-phrase co-occurs with possessive su.66 Note that the dephrase precedes the genitive pronoun. In (225b), there is doubling of the possessive marker, and in (225c), the possessive form co-occurs with an indenite
article. This last type of co-occurrence is noted in Lipski to occur in El
Salvador, Guatemala and in southern Mexico.

1.7.5

Subject-related variation

Several forms of variation related to the clausal subject are prevalent

in the Spanish of the Caribbean. The two most common phenomena are the
66

For discussion, see Camacho, Paredes and Snchez (1995).

72

The syntax of Spanish

occurrence of pre-verbal pronominal subjects in Wh-questions, as in (226),

and in innitives, as in (227):
(226)

Qu t
quieres?
(Cuba Lipski 1994:233)
what you(Nom.) want-2nd.sg.
What do you want?

(227)

a. antes de yo

salir

de mi pas

(Colombia Lipski 1994:215)

before of I(Nom.) leave-inf. of my country
before I leave my country
b. para yo
hacer eso
(Puerto Rico Lipski 1994:335)
for I(Nom.) do-inf. that
for me to do that

These phenomena are common in Cuba, the Dominican Republic, Puerto

Rico and Venezuela, and are also found in Panama. Pre-verbal subjects of
innitives are found in these same areas, as well as in Colombia and sometimes
in Ecuador.
One analysis of this dialectal dierence is in terms of the clitic status of
subject pronouns. As described earlier, overt pronouns in most Spanish
dialects are strongly stressed, and behave in all syntactic respects like full
Noun Phrases. It has been suggested (Lipski 1977; Contreras 1989) that, in
Caribbean Spanish, subject pronouns have unstressed variants which behave
like clitics, as is true of French subject pronouns.
However, Suer (1994) presents data which suggest that, at least for some
speakers, the absence of inversion is not restricted to pronominal subjects. For
these speakers, sentences like (228) are not completely ungrammatical:
(228)

Qu libro Mara quiere?

(Suer 1994)
which book M. want-pr.3rd.sg.
Which book does Maria want?

The role of weakened verbal inection in Caribbean dialects has also been discussed in relation to properties of the subject (Suer 1986; Toribio 1993,
1996).
As noted earlier, the phenomena discussed here are dialect variants of Latin
America. Some of these variants occur also in peninsular dialects. For
example, negative doubling like that in (221), and possessive doubling with
denite article + possessive form (la mi vaca the my cow) are found among
other variants in Leonese Spanish (Alvar 1996:156).

The Noun Phrase

2.1

Introduction

This chapter will examine the structure of the category traditionally

referred to as the Noun Phrase (NP). We will consider the NP from two perspectives. First, from a phrase-external perspective, we will discuss the distribution
of NP, and present a brief overview of the principles of grammar which account
for this distribution. These matters are taken up in 2.2 for argument NPs, and
in 2.3 for predicative NPs. We then turn to examination of the internal structure of NP. In 2.4, general structural properties of phrases which are established
by X-theory are introduced. This includes relations between the head and its
complements and adjuncts, and the general properties of these constituents.
Then in Sections 2.52.7, we will focus on speciers of NP. In 2.5, the structure
and function of NP determiners will be discussed; in 2.6, we examine elements
that are traditionally referred to as pre-determiners and post-determiners.
Section 2.7 nishes the discussion of speciers with an overview of speciers of
predicative NPs. Section 2.8 will address one of the central issues with respect
to constituent order within NP: the order of the head, complements and
adjuncts relative to each other. Finally, Section 2.9 will summarize the general
conclusions of the chapter.

2.2

The distribution of argument NPs

Principles of X-theory allow us to infer the presence of NP whenever

a noun is present (since all NPs are headed by nouns, and every noun (No) projects X structure (NP)). Therefore, in the following examples, the italicized
nouns head NPs:
(1)

a. Juan hizo pan.

Juan made bread.
b. Mara es doctora.
Maria is (a) doctor.
73

74

The syntax of Spanish

The NPs in (1) do not all have the same status with respect to their function
in the sentence. Consider the contrast between the post-verbal NPs, pan in (1a)
and doctora in (1b). In (1a), the two NPs in the sentence (Juan, pan) have separate, independent reference; in (1b) however, the two NPs (Mara, doctora)
do not have separate reference. The NP headed by doctora does not refer to an
individual, but instead is a property of the NP Mara.
The contrast between the post-verbal NPs in (1a) and (1b) can be described
as a distinction between arguments and predicates. In (1a), [NP pan] is an argument of hizo; in (1b), [NP doctora] is not an argument of some predicate, but is
a predicate itself. Only arguments are capable of independent reference.
Consequently, pan can be understood as having reference distinct from Juan,
but doctora cannot be understood as having reference distinct from Mara. In
this section, we will focus on the distribution of argument NPs. Below, the discussion will begin with a descriptive summary of contexts in which argument
NPs can occur; then we will review two subtheories of grammar which
account for this distribution. Predicative NPs will be discussed in 2.3.

2.2.1

The descriptive generalizations

NP cannot function as anything other than an argument or predicate.

This is shown by the impossibility of an adjunct NP in (2a):
(2)

a. *Juan hizo pan piloto.

J.
made bread pilot
b. Juan hizo pan descalzo.
J.
made bread barefoot

In (2a), the extra NP piloto is neither an argument nor a (primary) predicate, and its presence makes the sentence ungrammatical. Example (2a)
cannot be understood as meaning that Juan, who is a pilot, made bread. This
contrasts with (2b), with an Adjective Phrase, descalzo, as adjunct.1
We have assumed that the presence of a noun indicates the presence of NP,
and that an NP is interpreted either as an argument or a predicate. NP cannot
be arbitrarily inserted in a sentence such that it has no status as either an
argument or a predicate. This implies that in those cases where an NP is possible, there are specic semantic and/or grammatical relations that make the
1

Certain adjuncts appear to be NPs:

(i)

Juan hizo pan ayer.

Juan made bread yesterday.

Larson (1985) has argued, for similar examples in English, that what appear to be
NPs functioning as adverbs (predicates) are PPs headed by a covert preposition.

The Noun Phrase

75

appropriate interpretation possible. To see this more clearly, consider again

(1a) versus (1b). In (1a), pan can only be interpreted as an argument, not as
a predicate, while in (1b), doctora can only be interpreted as a predicate.
However, this is clearly not due to inherent characteristics of these nouns, as
a change in context shows:
(3)

a. Hay piloto en el avin.

is
pilot in the plane
There is a pilot in the plane.
b. La masa se hizo pan.
The dough became bread.

In (3a), piloto refers to an individual; in (3b), pan is predicative. It does not

have reference independent of la masa. The interpretation of a given NP as an
argument or a predicate is therefore not due to lexical properties of the NP
itself, but depends on the context in which the NP occurs.
In subsequent sections, it will be useful to distinguish argument NPs from
predicative NPs, since, as we will see, the dierence in function can aect the
internal composition of the phrase. With this in mind, let us look briey at how
phrase-external factors determine the status of NP as an argument or predicate.

2.2.2

Theta-theory

Traditional grammars describe the distribution of arguments in

terms of the transitivity of the predicate to which an argument is related.
Intransitive predicates select only a single argument, transitives select two
arguments, and ditransitives select three arguments. These are illustrated
below with verbal predicates:
(4)

INTRANSITIVES
a. Juan durmi.
Juan slept.
b. Los nios bailan.2
The children dance.
c. Llega el tren.
The train is arriving.

(5)

TRANSITIVES
a. Mara compr pan.
Maria bought bread.

Note that the verb bailar dance appears to be a transitive verb in examples like:
Mara bail el tango Maria danced the tango. The NP el tango does not have the
same status as the complements of transitive verbs such as those in (5). It is a
cognate object, i.e., a phrase which repeats the meaning of the verb itself.

76

The syntax of Spanish

b. Susana construy una casa.
Susana built a house.
c. Los estudiantes saben la respuesta.
The students know the answer.

(6)

DITRANSITIVES3
a. Pedro le mand un paquete a Jos.
Pedro sent a package to Jos.
b. El perro le dio un mordisco al hueso.
The dog gave the bone a bite.

Informally stated, the notion of transitivity describes the number of arguments that are required for completeness of the meaning of the predicate. For
example, the ditransitive verb mandar send in (6a) requires one individual
to carry out the action (Pedro), another to undergo the action (el paquete), and
a person or location indicating an endpoint of the sending (Jos). The number
of arguments that a given predicate requires is a lexical property of the predicate. This means that the number of arguments that will be required (as in
(4)(6)), is not predictable from anything other than the meaning of the particular lexical item itself. The lexical entry for each predicate includes specication for transitivity, often called predicate argument structure, or simply
argument structure. A lexical items argument structure species the number
of required arguments. The items argument structure must be satised when
it is inserted in a syntactic derivation, ensuring that the predicate is not used
in structures in which there are too few or too many arguments.4
In the Principles and Parameters framework, argument structure is
expressed in lexical entries in terms of one or more semantic roles assigned by
3

The term ditransitive is normally used for predicates that have both a direct and
an indirect object, as well as a subject. There are also predicates that occur with a
direct object and a locative prepositional phrase, such as Susana dej el lpiz en la
mesa Susana left the pencil on the table, which have three arguments, but are not
traditionally considered ditransitives.
In early versions of transformational grammar (Chomsky 1965), argument structure
was given in the form of subcategorization frames. Predicates comprise subcategories according to the contexts in which they can be inserted, as in (i):
(i)

a. dormir:
b. comprar:
c. mandar:

__
__ NP
__ NP PP

Subcategorization frames of the type in (i) do not give the subject argument, only the
VP-internal arguments. The subject was assumed to be provided automatically by
phrase-structure rules. Current proposals dier as to whether or not the subject
(external argument) is distinguished in some way in lexical entries.

The Noun Phrase

77

a given predicate. For example, a ditransitive verb such as mandar send

assigns an Agent role (the sender), a Theme or Patient (the thing sent) and a
Goal (the endpoint); the intransitive verb bailar selects only an Agent, etc.:
(7)

a. mandar send:
b. bailar dance:
c. comprar buy:
d. saber know:

Agent, Theme, Goal

Agent
Agent, Theme
Experiencer, Theme

The specication of these roles, called Thematic Roles, or Theta-roles, indicates in the lexical entry both the number of arguments required by a predicate, and also the particular semantic role that each argument has in relation
to the predicate. For this reason, a given NP such as Mara in (8) can have
dierent semantic roles, depending on the particular Theta-role assigned to
the NP by some predicate:
(8)

a. Mara bail.
Maria danced.
b. Mara sabe la respuesta.
Maria knows the answer.
c. Juan le mand un paquete a Mara.
Juan sent a package to Maria.

(=Agent)
(=Experiencer)
(=Goal)

In each sentence, there must be a match between the roles specied in the
lexical entry of the predicate (e.g. (7a)) and the number of positions in the syntactic structure to which a role could be assigned. In the sentences in (8), each
verb has the number of arguments that matches the number required in the
verbs lexical entry. If there is a mismatch, ungrammaticality results:
(9)

a. *Mara le sabe la respuesta a Juan.

Maria knows the answer to Juan.
b. *Juan mand.
Juan sent.

To account for the status of (8) versus (9), it is assumed that lexical insertion
of some item into a syntactic derivation involves the assignment, or discharge, of the Theta-roles associated with the lexical entry. At some level(s)
of syntactic representation, it is necessary to check the context in which a
predicate occurs against its lexical entry. The derivation will be well formed in
relevant ways if every Theta-role of a predicate is assigned to some NP (or
other argument category, such as a clause), and if every NP is assigned a
Theta-role. The sentences in (8) satisfy this requirement, because every NP
present has a role specied in the lexical entry. The sentences in (9) fail because
there are missing NPs or extra NPs.

78

The syntax of Spanish

Theta-roles are assigned under sisterhood.5 Theta-role assigners include the

head of phrase (V, in (10)), and higher bar-level projections of the head. A
Theta-role assigned by the head is its internal argument; a Theta-role
assigned by V is an external argument.
(10)

a.

b.

VP

NP

(Agent) V
comprar

NP
(Theme)

NP

(Agent)

V
dormir

In (10a), the Theta-role Theme is assigned to the sister of the verb comprar,
so this internal argument NP is interpreted as having the semantic function of
Theme of comprar. The role of the external argument (the Agent in (10a)
and (10b)) is assumed (e.g., in Chomsky 1981) to be assigned compositionally
by the verb together with any objects. This implies that it is assigned by the V
node to its sister, rather than by the verb alone.6
5

The sisterhood relation can be formalized as mutual c-command. A node x c-commands a node y if the rst branching node which dominates x also dominates y. In
(i),

(i)

VP

NP 1

NP 2

the verb c-commands NP2 but not NP1. It c-commands NP2 because the rst branching node which dominates the verb (V) also dominates NP2. NP2 c-commands the
verb also, because the rst branching node that dominates NP2 also dominates the
verb. Therefore the verb and NP2 are in a relation of mutual c-command. The verb
does not c-command NP1, because the rst branching node which dominates the
verb, V, does not also dominate NP1. NP1 is c-commanded by V.
There may be exceptions to mutual c-command as a condition on Theta-role
assignment. One case is Theta-role assignment to the complement of ditransitive
verbs, to be discussed in Chapter 3 (Section 3.6).
Evidence for the compositionality of Theta-role assignment is based on pairs like the
following:

The Noun Phrase

79

It follows from the above assumptions that an NP that is interpreted as an

argument can be generated only in those positions to which a Theta-role can be
assigned. These positions are called argument positions, or A-positions. An NP
could not be generated in a position other than an A-position and be interpreted
as an argument, since it would have no semantic role, or relation to any predicate. Thus, the structure (11b) corresponding to (11a) cannot be well formed.
(11)

a. *el libro Juan compr

the book Juan bought

b.

NP
el libro

NP

Juan

V
compr

(i)
(ii)

a.
b.
a.
b.

John rolled down the hill.

John rolled the ball down the hill.
John broke his arm.
John broke the glass.

(John=Patient)
(John=Agent)
(John=Patient)
(John=Agent)

In (ia), intransitive roll assigns Patient to its subject, but in (ib), transitive roll assigns
Agent to its subject. The status of external arguments as assigned a compositional
Theta-role is in a sense problematic, because it is inconsistent with the view that
Theta-roles are an idiosyncratic property of lexical items. Compositionality of
assignment implies that a particular semantic role cannot be determined on the basis
of knowledge of the lexical entry alone. One solution to this problem is the idea that
knowledge of the semantic role assigned to a given argument does not derive exclusively from lexical items, but is based in part on a primitive set of relations known as
the thematic hierarchy. For example, (iii) shows a partial hierarchy:
(iii)

Agent < Theme < Experiencer < Goal

If a lexical item assigns Theme to its internal object, it can then only assign to its
external argument a role that is higher on the thematic hierarchy, not a role that is
lower. The thematic hierarchy is, on this view, a universal set of restrictions on possible lexical items. It follows that the full array of Theta-roles does not have to be specied for each predicate in the lexicon, and compositionality is to be expected. The
thematic hierarchy then participates in predicting both the content of particular roles
and the distribution of roles relative to each other (the mapping of roles) in syntactic structure. For discussion of the thematic hierarchy, see Jackendo (1972).

80

The syntax of Spanish

In (11b), the NP el libro is not in a position to which a Theta-role could be

assigned, consequently it cannot be interpreted as an argument of compr.
The lexical properties of the predicate are not matched, or satised, by this
structure.
So far, we have looked only at examples in which A-positions are lled by
argument NPs. Suppose, however, that a verb does not assign any Theta-role
to an A-position. Such a position can be occupied by an NP, but that NP is
not related to a predicate, hence cannot have reference. Examples are the nonreferential or pleonastic elements in the English examples in (12):
(12)

a. It seems that the apples are ripe.

b. There were unicorns in the garden.

Pleonastics are typically invariant morphologically, and may be present only

to satisfy structural or other grammatical requirements in a derivation.
Corresponding sentences in Spanish never have an overt pleonastic, but the
presence of a covert pleonastic is suggested by invariant agreement on the
verb:
(13)

a. Parece/*n que las manzanas estn maduras.

b. Haba/*n unicornios en el jardn.7

(=12a)
(=12b)

Summarizing to this point, we have seen that argument NPs are generated
only in positions to which a Theta-role could be assigned (A-positions). If a
role is actually assigned, NP is interpreted as an argument with a particular
semantic role in relation to the predicate that assigns the Theta-role; if no role
is assigned, NP is pleonastic, and has no reference. Argument NPs cannot be
generated in non-A-positions, since they will have no semantic interpretation
in relation to any predicate.

2.2.3

Abstract case

The above brief description of argument NPs describes only the

basic or original position of NPs the position in which NP receives its
interpretation as semantically related to a predicate. Other processes can move
NPs, so that the derived position of NP may dier from its basic, or canonical, position. The moved, or derived, positions of NPs may be forced by
other NP licensing requirements. One such requirement for NPs is (abstract)
7

Some dialects of Spanish allow plural agreement in examples like (13b), parallel to
English (12b). The occurrence of agreement with the post-verbal NP suggests that
the pleonastic is a temporary placeholder for the semantic subject of the clause,
and that the features of the semantic subject are raised to subject position in the
course of the derivation.

The Noun Phrase

81

Case, such as Nominative or Objective Case. The idea behind abstract Case
is that, just as pronominals have dierent forms (yo, me, m) according to their
grammatical function, so also do other NPs have an abstract feature which
identies their grammatical function even though the NP does not overtly
express this feature. Case theory has undergone several revisions, so that recent
formulations dier signicantly from earlier ones. In the course of our discussion in subsequent chapters, the conditions under which Case is assigned will
be considered in more detail. For the moment, it will suce to outline two
Case-theoretic generalizations: (A) that Case is assigned by certain heads, in
part determined by the category of the head, and in part by lexical properties
of the head. Verbs and prepositions assign Objective Case, and the nite head
of the clause, INFL, assigns Nominative Case. Not every member of these categories has a Case feature to assign, as we will see below. (B) Objective Case,
assigned by V or P, is assigned to an NP that the head governs, which is usually
a sister constituent.8 In a sequence like: compr el libro bought the book or
con un amigo with a friend, the head (V or P) assigns Case to its sister. INFL,
however, assigns Case to the NP in its Specier (i.e., the subject of the clause).
These preliminary generalizations are summarized in (14):
(14)

Case Assignment (Simplied):

a. lexical heads (V and P) assign Case to a sister
b. INFL assigns Case to an NP in the Specier of IP

This early formulation of Case theory assumed that only phonetically overt
NPs require Case, and the requirement for Case was analyzed as a lter operating in the phonological component of the grammar. More recent versions
of Case theory have considered Case to be realized also on covert NPs, and
the Case lter has been reinterpreted as a component of the syntactic licensing of NPs.
18

Theta-roles (discussed above in 2.3.2) were also described as assigned to a sister constituent. In note 5 this relation was described technically as mutual c-command. Case
assignment is more liberal, in that the NP that is assigned Case need not ccommand the Case-assigning head. This is illustrated in (i), where the bracketed
phrase is the sister of the verb:
(i)

Consideran [AP a Mara muy inteligente].

They consider Maria very intelligent.

The bracketed phrase is an adjective phrase (or small clause) which is assigned a
Theta-role by the verb under mutual c-command. The verb also assigns Case to the
NP. Here, however, mutual c-command does not obtain: the verb c-commands NP,
but NP does not c-command the verb, because the rst branching node which dominates NP (AP) does not dominate the verb. Case-assignment in such instances is
referred to as exceptional Case marking (ECM).

82

The syntax of Spanish

It was noted above that the requirement for Case can trigger movement of
NPs from the position in which they are assigned a Theta-role. This type of
NP movement is illustrated for the boldfaced NPs in (15a), (16a):
(15)

a. [ e ] fue obligada ella a salir

was forced
she to leave
b. [ella] fue obligada a salir
she was forced
to leave

(16)

a. [ e ] parece [ ella haber salido ]

seems she to have left
b. [ella] parece [ haber salido ]
she seems
to have left

In the (a) examples above, the NP ella is not governed by a verb that can assign
Objective Case; neither the passive participle obligada in (15) nor the verb
parece in (16), for example, assigns Objective Case. This is due to morphological or lexical properties of these items. (See Chapter 3, Sections 3.5 and 3.7.)
Consequently, NP must move to the Specier of IP, where it can be assigned
Nominative Case by the Inectional head of the clause.
Summarizing, argument NPs originate in A-positions, the position in which
NP receives a semantic role as an argument of a predicate. This role must be
made visible by abstract Case, with the consequence that sometimes NP must
move in order to satisfy its Case requirement. Similarly, an interrogative NP
must move to satisfy its visibility as an operator. As we have seen, NPs cannot
be generated arbitrarily in satellite positions since, in such cases, NP will fail
to have a Theta-role, and will not have any type of argumental interpretation.
Since every constituent of a sentence must have an interpretation, derivations
with such unrelated NPs will not be grammatical.

2.3

Predicative NPs

The preceding discussion of predicate argument structure focused on

NPs as arguments. NPs can also have the distribution and interpretation of
predicates, similar to other predicate categories such as Adjective Phrases:
(17)

a. Mara es inteligente/doctora.
M.
is intelligent/(a) doctor
b. Los estudiantes parecen inteligentes/genios.
the students
seem
intelligent/geniuses
c. Los profesores consideran inteligente/genio
a Juan.
The teachers consider
intelligent / (a) genius PA J.
The teachers consider Juan intelligent/a genius.

The Noun Phrase

83

In (17a) and (17b), the post-verbal NPs/APs lack independent reference. Their
interpretation is predicative, rather than argumental. In (17c), the phrase inteligente/genio is also predicative, but the phrase a Juan is not. The predicative
NPs and APs in (17) assign a Theta-role to an NP, such as Juan, in (17c), which
is interpreted as the subject of the predicate. This implies that predicative NPs
are not generated in A-positions, since they assign, rather than receive a Thetarole. The distribution of predicative NPs (and other predicate categories) is
then complementary to that of arguments. The visibility of the predicate category is licensed through its syntactic association with an A-position. This
relation is referred to as Predication. To illustrate the licensing of the elements
of predicative NPs, (17a) derives from a structure like (18).
V

(18)

NP

es

NP

Mara

doctora

In (18), NP consists of the predicate N and its single argument, the NP Mara.
Both the predicate and the argument must satisfy grammatical licensing requirements that make them visible for interpretation. The argument NP, Mara, is
assigned a Theta-role by N, and is made visible by abstract (Nominative) Case,
once it moves to the position of clausal subject. The predicate N satises its
lexical requirement by assigning a Theta-role to the NP Mara. The predicate is
licensed by the relation of Predication. In essence, Predication completes the
interpretation of the predicate by associating it with its subject.9 Predication is
shown as coindexing of the subject and predicate.
19

The semantic basis for the Predication requirement is that predicates express sets of
individuals, and Predication identies members of that set. For example, in (i),
(i)

Mara es inteligente
M.
is intelligent

The predicate inteligente is understood as the set of individuals with the property of
being intelligent. Predication identies Mara as asserted to be a member of that set.

84

The syntax of Spanish

(19)

NP

NPi

Ni

Mara

doctora

In (19), Predication is satised because there is an NP with which the predicate A/N is coindexed. If no NP were present as a sister to A/N, Predication
would fail and the derivation would be impossible.10
Predicative NPs are not standardly analyzed as requiring Case. Abstract
Case has been considered to have a function in relation to the visibility of
Theta-roles. Predicative NPs would not be expected to require Case, since they
are not theta-marked. Examples like those in (20) support this claim:
(20)

a. Juan es (*a) mi mejor amigo.

J.
is PA my best friend
Juan is my best friend.
b. Soy
yo/*m.
be-1st.sg. I/*me
It is I/*me.

Example (20a) shows that predicative NPs do not co-occur with personal a.
This is expected if predicative NPs are not assigned Case, since personal a is
restricted to Objective contexts. In (20b), the predicative pronoun yo takes
Nominative form. This appears to indicate that the predicative NP has Case;
however, an alternative is that the predicative yo in (20b) agrees in Case with the
(covert) subject of the clause. Both sentences in (20) have a subject that is assigned
Nominative Case by the inectional head of the clause. In (20a), the subject is
Juan; in (20b), the subject is the covert pleonastic pronoun corresponding to
English it. The occurrence of Nominative as the form of post-verbal predicative
NP then suggests that the predicate shares, or agrees with, the Case of its
subject, rather than being assigned a separate Case by an independent governor.11
10

11

Williams (1982) argues that NP does not have a phrase-internal subject. That is, NP
has no A-position specier.
The assumption that predicate NPs do not require Case leaves unanswered one fact
about their distribution. That is, as was noted in Section 2.2 above, Predicative NPs
cannot occur as secondary predicates that is, in adjunct positions, as illustrated by
the ungrammaticality of sentences like Juan hizo pan piloto Juan, a pilot, made
bread. If predicative NPs required Case, these sentences would be automatically
excluded, since there is no Case assigner for the secondary predicate.

The Noun Phrase

85

Notice that the verb agrees in person/number with the predicative NP, not with
the covert pleonastic it.
Summarizing, the distribution of predicative NPs is complementary to
argument NPs. Predicative NPs do not occur in A-positions, but occur in positions that can be linked to A-positions. The licensing process by which this
linking occurs is Predication. Predicative NPs are not assigned an independent
Case. Their morphological Case forms can be derived by agreement with the
clausal subject.

2.4

The constituents of NP

In this section, we will introduce several types of constituents that cooccur with a noun to form a Noun Phrase. We will focus on the distinctive
properties of these constituents. Subsequent sections will examine in closer
detail some of the ner structure associated with them. Let us begin with a
preliminary structural description of the phrase, and of three types of constituents that can be generated with a head noun: complements, adjuncts and
speciers. Let us look rst at a phrasal skeleton which shows several types
of nodes within NP.
(21)

(WP)

(YP)

(ZP)

In (21), N (or No) is the category label of a lexical item that constitutes the
head of phrase N.12 N and N are projections of the head that is, nonlexical category nodes which share the category features of the head. Because
these nodes are projected from No, they could not be present in the absence of
a head noun. In other words, there can be no phrase without a head. Also,
because these nodes share categorial and other features of the head, the
phrasal node N (= NP) is necessarily the same category as the head. Thus, all
12

N can also have internal structure, for example in compounds. Internal structure can
also represent sub-lexical morphological relations.

86

The syntax of Spanish

phrases are endocentric. An NP can only be headed by a noun, an AP can only

be headed by an adjective, etc.
The N and N nodes may branch so as to dominate additional constituents
in (21), the optional elements WP, YP and ZP. It has been argued (Kayne
1981) that a node may have a maximum of two branches, so phrase markers
observe a condition of maximal binary branching. The optional constituents
in (21) are in dierent congurations relative to the head and to the phrase as
a whole. ZP is a sister of the head; because this is a position to which a Thetarole can be assigned, ZP is an argument or complement position. YP is in a
position to which a Theta-role could not be assigned, and YP could therefore
only be a non-argument. YP is in a position occupied by adjuncts (optional
modiers). Notice that YP is both a sister of N and is dominated by N. The
duplication of a bar-level projection such as this marks an adjunction structure. A constituent YP is adjoined to N by a process of copying the N node,
creating a position for an additional branch. (There is no intrinsic limit on the
number of adjuncts in a phrase, so N could be duplicated further, creating a
more complex NP.) WP is in the Specier of XP the position dominated by
the maximal projection of the phrase. The following subsections present a preliminary overview of these constituent types.

2.4.1

Complements

A noun may have a complement, in other words an argument that is

assigned a Theta-role by the head noun. As noted above, a complement could
occupy the position of ZP in (22), but not the position of YP, since YP is not
a sister of the head.
(22)

(YP) (non-complement: adjunct)

(ZP) (complement)

Examples of noun complements are the bracketed constituents in (23):

(23)

a. la destruccin [de la ciudad]

the destruction of the city
b. tu
creencia [en la justicia]
your belief
in the justice
your belief in justice

The Noun Phrase

87

c. el deseo [de llegar

temprano]
the desire of to arrive early
the desire to arrive early
d. la apariencia [de que ganamos el partido]
the appearance of that win.1st.pl. the game
the appearance that were winning/we won the game

Notice that in (23), the bracketed phrases are complements of the nouns
destruccin, creencia, etc. This is independent of the status of the whole NP,
which is normally an argument of another predicate. For example, the phrase
la destruccin de la ciudad in (23a) could be the complement of a verb: Vieron
la destruccin de la ciudad They saw the destruction of the city. The whole
phrase is the argument of the verb; within NP, the phrase de la ciudad is the
argument of destruccin.
Notice also in (23) that all of the complement constituents are introduced
by a preposition in these examples, de or en. The occurrence of the preposition can be related to the Case licensing of the complement. Recall from 2.2
that a theta-marked complement must be assigned Case; unlike verbs, nouns
are not Case assigners, or at least do not assign Case directly to a complement,
but do so via a prepositional Case marker. Notice that the verbal counterpart
of (23a), for example, does not have the preposition: Destruyeron (*de) la
ciudad They destroyed (*of) the city.13
Looking still at the bracketed complements in (23), we see that, in addition
to the preposition, these complements contain either NP (23a,b) or a clause
13

Chomsky (1986), proposes that Nouns (and Adjectives) do assign Case to their complements, but the nature of the Case that is assigned diers from that assigned by
Verbs, Prepositions and INFL. Nouns and Adjectives are suggested to assign
inherent Case, a Case that is linked to Theta-role assignment. This contrasts with
the structural Cases, which can be assigned to an NP, without it being thetamarked by the Case assigner. This is illustrated in (i):
(i)

Consideran [inteligente a Pedro].

consider
intelligent PA Pedro
They consider Pedro intelligent.

In (i), the verb theta-marks the bracketed constituent, an adjectival small clause. The
adjective theta-marks Pedro, the external argument of the small clause. The verb
assigns Case to this NP, as is shown by the occurrence of Personal a. We will see
below that small clauses do not occur as complements of nouns, a fact which can be
attributed to the hypothesis that a noun can only inherently Case-mark that is,
assign Case in conjunction with Theta-role assignment.
Notice that example (23b) does have a preposition in the verbal counterpart: creer
en la justicia to believe in justice. This illustrates that some verbs do not assign
Case to their complement, but instead select a complement type that contains a Case
assigner, namely the preposition.

88

The syntax of Spanish

(23c,d). Nouns do not freely select complements of other categories. This is

illustrated below by contrasts between the (a) and (b) examples:
(24)

a. Pedro se siente [AP contento].

P.
CL feels
happy
Pedro feels
happy.
b. *el sentimiento contento
the feeling happy

(25)

a. Consideran
[a Juana inteligente].
consider.3rd.pl PA J.
intelligent
They consider Juana intelligent.
b. *la consideracin [a/de Juana inteligente]
the consideration PA/of J.
intelligent

(26)

a. Vi
[llegar a Pedro].
saw.1st.sg. arrive PA P.
I saw Pedro arrive.
b. *la vista de llegar a/de Pedro
the sight of arrive PA/of P.
the sight of Pedro arrive

The verbs in (24a) and (25a) select AP and adjectival small clause complements, in (26a), a verbal small clause (cf. 1.3.1). The corresponding nominals
disallow these complements.
Not all nouns assign Theta-roles. Those which do are of the following types:
(a) nouns derived from verbs or adjectives, illustrated in (27a), and picture
nouns, illustrated in (27b):
(27)

Theta-assigning nouns:
a. Deverbal and de-adjectival nouns: destruccin (destruir)
destruction (cf. destroy); prueba (probar) proof (prove);
apelacin (apelar) appeal (appeal); delidad (el) faithfulness
(faithful); claridad (claro) clarity (clear)
b. Picture nouns: foto picture, photograph; retrato portrait; libro
book; artculo article; historia story; idea idea

The nominals in (27a) select arguments that correspond to those of the

related verb or adjective. If the verb or adjective theta-marks a PP complement, that complement is typically headed by a unique preposition, and the
complement of the nominal will also be a PP headed by the same preposition
(e.g. creer en to believe in creencia en belief in; apelar a to appeal to
apelacin a appeal to). If the related verb theta-marks an NP, then the complement of the noun selects the default preposition de (e.g., destruir la
ciudad destroy the city destruccin de la ciudad). The argument structure of

The Noun Phrase

89

picture nouns can be attributed to their relational meaning. The entities

which they denote (pictures) have the characteristic of representing objects.
These nouns can take as a complement a phrase which expresses the represented object. Other types of common and proper nouns do not have argument structure.

2.4.2

Adjuncts

As noted above, adjuncts are assumed to be generated as sisters of N,

in the position of YP in (22), repeated below.
(22)

(YP) (non-complement: adjunct)

(ZP) (complement)

The following types of constituents occupy this position:

(28)

Adjective Phrase:
a. una comida [muy buena]
a very good meal
b. un color [ exquisito]
an exquisite color
c. un poltico [el a los principios democrticos]
a politician faithful to democratic principles

(29)

Prepositional Phrase:
a. la solucin [en ese libro]
the solution in that book
b. un regalo [para Juan]
a gift for Juan
c. el estudiante [de Caracas]
the student from Caracas
d. una mquina [para reparar]
a
machine for to x
a machine to x

(30)

Clause:
a. el libro [que lemos]
the book that we read
b. una ciudad [que Mara ha visitado]
a city that Maria has visited

90

The syntax of Spanish

(31)

Small Clause:14
a. una persona [respetada por todos]
a person respected by everyone
b. los nios [jugando en la calle]
the children playing in the street

These adjunct constituents are licensed by predication. The subject of the

predication is the noun, or a higher projection of it. For example, in (29a), the
adjunct en ese libro in that book is predicated of solucin solution; in the
sequence la solucin correcta en ese libro the correct solution in that book
the PP could be predicated either of the head noun or of the noun together
with its adjunct.
Adjectival adjuncts dier from the others listed above in two respects. One
is that adjectives agree with the head noun in number and gender. Second, some
adjectives occur in positions within the Noun Phrase other than the canonical adjunct position discussed above. They may be pre-nominal, as in (32):
(32)

Adjective Phrase (pre-nominal):

a. un mero soldado
a mere soldier
b. un viejo amigo
an old friend
c. la pura fantasa
the pure fantasy

They also occur between the head noun and its complement:
(33)

a. la destruccin completa de la ciudad

the destruction complete of the city
b. tu
creencia absoluta en la justicia
your belief
absolute in justice

The position in which an adjective appears depends both on the nature of the
adjective itself the subcategory of adjective and on the adjectives interpretation in a given NP. A few adjectives, such as mero in (32a), are always prenominal. A large class of adjectives, called qualitative adjectives, such as
viejo and puro in (32b,c), can be either pre- or post-nominal. Qualitative adjectives dier in their interpretation depending on their position. In pre-nominal
position, they are appositive. For example in (32b), un viejo amigo is someone
who is old as a friend (i.e., a long-time friend), not necessarily someone who
is old. In post-nominal position, these adjectives are restrictive: un amigo viejo
14

The participial and gerundive adjuncts illustrated here are sometimes referred to as
reduced relatives. Their interpretation is analogous to full relative clauses (e.g., los
nios que estn jugando en la calle the children who are playing in the street).

The Noun Phrase

91

is someone who is a friend and is old. A third class of adjectives, relational

adjectives, which express relations of various types, such as origin (e.g. un estudiante mexicano a Mexican student) or material composition (e.g. una sustancia qumica a chemical substance) are always post-nominal.
Notice that the order of constituents shown in the examples in (33) is not
predicted to be possible as a basic order. This is so because the complement is
not the closest constituent to the head, and is therefore not a sister of it so it
could not be assigned a Theta-role. The order: headadjunctcomplement in
these examples implies that a constituent has been displaced from its Dstructure position. This topic will be taken up in Section 2.5.

2.4.3

Distinguishing complements from adjuncts

It is often not obvious at rst glance whether a post-nominal PP is a

complement of the noun or an adjunct. Consider the examples in (34), (35):
(34)

a. una
a
b. una
a
c. una
a

silla [de
Francia]
chair from France
silla [en el jardn]
chair in the garden
silla [para Jos]
chair for J.

(35)

a. un estudiante [de sintaxis]

a student
of syntax
b. un estudiante [de pelo negro]
a student
with hair black
a student with black hair
c. un estudiante [de
Mxico]
a student
from Mexico

Unfortunately, there are no straightforward syntactic diagnostics that dierentiate complements of a head noun from adjuncts. The preceding discussion
has provided some tools that can help to deduce the status of a PP in examples such as these. Recall from 2.4.1 that only certain types of nouns have argument structure: derived nominals and picture nouns. With this in mind, we
may suppose that, in (34), none of the bracketed phrases are complements of
the noun, because the head noun silla is not of the type which has arguments.
The noun estudiante in (35), however, could have a complement, if it is a deverbal nominal related to the verb estudiar. But not all of the PPs in (35) are complements of the head. One way to decide whether any of these PPs is a
complement is to convert the nominal to its verbal form, and compare the
interpretations. In (36b), the PP has the same relation to the nominal estudiante as the NP sintaxis has to the verb estudiar in (36a):

92

The syntax of Spanish

(36)

a. estudiar sintaxis
to study syntax
b. un estudiante de sintaxis
a student
of syntax

In both cases, the complement is the Theme of the head, i.e., the object of
study. In (35b) and (35c) on the other hand, the PP is not interpreted as something that undergoes study, but instead indicates an additional property of the
noun. By this criterion (a semantic one), only the PP in (35a) is a complement
of the head. A second criterion that supports this analysis is the typical
order of constituents if more than one is present:
(37)

a un estudiante de sintaxis de Mxico

b. un estudiante de sintaxis de pelo negro
c. ??un estudiante de Mxico de sintaxis
d. ??un estudiante de pelo negro de sintaxis

Examples (37c) and (37d) show marked orders. They are improved by
adding pause intonation between the PPs. These examples are revealing of the
status of the PPs, because complements are generated as sisters of the head,
and should therefore be expected to appear as the closest constituent to the
head, unless some additional operation has altered the order. Such operations
could derive sequences in (37c,d), but, in doing so, the intonation would
change. Based on the contrast between (37a) and (37c), it is possible to claim
that de Mxico is an adjunct, while de sintaxis is a complement. Likewise, de
pelo negro is an adjunct in (37b), given the contrast with (37d).
It is perhaps useful to point out a contrast between English and
Spanish with respect to diagnostics for noun complements. In English, onepronominalization replaces only N, which means that adjuncts can be sisters
of one, but complements cannot:
(38)

a. the student from Argentina and the one from Mexico

b. *the student of phonology and the one of syntax

Example (38b) is ungrammatical because one has replaced only the head, and
not the complement; (38a) is well formed because from Mexico is an adjunct,
i.e., a sister of N, which one has replaced. In Spanish, there is no overt
proform corresponding to English one, but there is an analogous construction
with an ellipted N (or a covert proform):
(39)

el estudiante de Granada y
el de Mxico
the student
from Granada and the from Mexico
the student from Granada and the (one) from Mexico

This construction does not, however, replace N exclusively, as is shown by the

grammaticality of (40):

The Noun Phrase

(40)

93

el estudiante de sintaxis y
el de fsica
the student
of syntax and the of physics

Because complements and adjuncts behave identically under ellipsis, this construction does not provide evidence supporting the structural distinction
between the two.

2.4.4

NP speciers

In early work on X-theory, it was shown that there is a third type of

constituent which can accompany the head of a phrase. These elements are
speciers, a heterogeneous class of constituents which are typically phraseinitial, and which dier from complements and adjuncts in certain ways. A
summary of Speciers of NP is given in (41):
(41)

a. Possessives: mi libro my book, tu chaqueta, your jacket

b. Demonstratives: ese vaso that glass, esta pelcula this movie
c. Denite determiners: el cine the movies, los aviones the airplanes
d. Indenite determiners: un cuchillo a knife, unas guerras some wars
e. Cardinals: dos captulos two chapters, cien pesos (a) hundred pesos
f. Quantiers: algn estudiante some student, toda persona every
person
g. Interrogatives: cules nios which children, cuntos libros how many
books

Although many of the items in (41) are similar to certain types of adjuncts,
particularly adjectives, speciers of NP must be dierentiated from typical
adjuncts (modiers), in that, in certain contexts, NP must have some type of
specier:
(42)

a. Mi/un libro est en la mesa.

my/a book is
on the table
b. *Libro est en la mesa.
book is
on the table

The obligatoriness of the specier suggests that it has some function for the
licensing of NPs, a point to which we return in 2.5. In recent years, there have
been considerable advances in analyzing the syntax of these speciers.
Sections 2.52.7 discuss these constituents and their analysis in detail.
Summarizing the main points of this section, we have seen that X-theory
makes several claims about the structure of Noun Phrases. Every NP is headed
by a noun, and may contain other types of constituents: complements, adjuncts
and speciers. Complements are sisters of the head, and are PPs or clauses.
Adjuncts typically follow complements, although adjectival complements can

94

The syntax of Spanish

also be pre-nominal or post-nominal. Speciers are a heterogeneous class,

whose structure will be discussed below.

2.5

Determiners of argument NPs

We turn now to discussion of the Speciers of argument NPs. Recall

from 2.2 that NP can function either as an argument or as a predicate, and its
distribution and licensing vary accordingly. It was noted there that the status
of NP as an argument or predicate can also aect the internal structure of the
phrase. As we will see, the determiner system is centrally aected. In this
section and in 2.6 below, speciers of argument NPs will be discussed. We then
return to predicative NPs in 2.7.

2.5.1

Distribution of speciers
Argument NPs allow all of the specier types listed in (41) above:

(43)

a. Mi/ese/el/un libro est en la mesa.

My/that/the/a book is on the table.
b. Dos libros estn en la mesa.
Two books are on the table.
c. Algunos libros estn en la mesa.
Some books are on the table.
d. Cuntos libros estn en la mesa?
How many books are on the table?

Speciers that have denite interpretations (possessives, demonstratives and

denite determiners) cannot co-occur with each other in pre-nominal position:
(44)

a. *el mi libro
the my book
b. *ese su libro
that his book

Based on the uniqueness of the pre-nominal specier position for these determiners, earlier versions of X-theory analyzed NP as having a unique
Determiner position as a daughter of X, the maximal phrasal projection
(see (45)). This structure generally predicts that any NP will have a single position for speciers, which occurs as the rst (leftmost) constituent of the
phrase. However, determiners of the rst subclass can co-occur if one is postnominal with strong stress:

The Noun Phrase

95

(45)

Det

el

libro
(46)

a. el libro tuyo
the book your
the book of yours
b. el libro ese
the book that
THAT book

The grammaticality of sequences in (46) shows that it is not the items themselves that must be unique within NP, but rather that there is a single prenominal position in which a determiner of this class can surface. This is
problematic for a purely phrase structure account of the distribution of determiners as in (45) since, in order to generate sequences like (46), the category
Det must be possible in post-nominal positions, as shown in (47):
(47)

Det

Det

el

ese

libro

Given this possibility, additional constraints are needed to exclude phrases

with only a post-nominal determiner (*libro ese).
Another issue is that Speciers of other types can co-occur pre-nominally.
Consider the sequences in (48), which contain pre-determiners and postdeterminers in boldface:
(48)

a. los varios libros

the several books
b. todos esos libros
all
those books

96

The syntax of Spanish

c. unos tres libros
some three books
d. cules de mis libros
which of my books
e. ninguno de esos libros
none
of those books

Traditional (and generative) accounts of these sequences have typically analyzed the extra speciers as modiers of a separate category, such as predeterminer, Quantier or Adjective. Under this approach, the Determiner
position is unique, but it can co-occur with modiers that specify quantity. For
example, (48a) and (48b) might have the structures shown in (49a) and (49b).
(49)

a.

Det

los

b.

varios

Det

todos esos

libros

libros

Notice that varios in (49a) is analyzed as an Adjective, based on its position

between the determiner and the noun a position where adjectives may typically appear while todos is analyzed as a Quantier. Todos could not be an
adjective, since adjectives cannot appear before a Determiner (*buenos los
libros good the books). An analysis based on the category of these speciers does account for (a) their distribution, and (b) their co-occurrence with
Det. However, one generalization remains unexpressed: the fact that these
items can also satisfy the requirement for a determiner. That is, they can also
appear without a separate determiner, unlike other optional modiers:
(50)

a. Varios libros estn en la mesa.

several books are on the table
b. *Buenos libros estn en la mesa.
good
books are on the table

Another issue is that this approach does not account for the appearance of the
Case marker de in examples like (48d), (48e). We return to the topic of preand post-determiners in Section 2.6.

The Noun Phrase

97

Summarizing to this point, we have seen that speciers of NP are not

strictly optional, like other modiers of nouns. This suggests that speciers
have a particular function in licensing the NP, so that, without a specier, the
NP is not fully interpretable as a phrase. However, we have seen that the specier function cannot be accounted for simply by positing a Determiner
position as in (45), since certain subclasses of speciers can co-occur with
each other as pre- and post-nominal constituents, and others can co-occur
pre-nominally. In 2.5.2 and in 2.6, we will consider ways of accounting for
these co-occurrences.
There is one further point to be noted with regard to the distribution of
speciers: the obligatoriness of speciers of argument NPs depends on
phrase-internal as well as phrase-external factors. Phrase-internally, the
choice of noun is relevant; unmodied proper nouns do not co-occur with
speciers in most dialects of Spanish:
(51)

*el Juan lleg.

The Juan arrived.

Phrase-external factors are also relevant. Argument NPs in certain positions

in a sentence can lack a specier:
(52)

a. Cantaron canciones toda la noche.

sang
songs
all the night
They sang songs all night.
b. *Canciones fueron cantadas toda la noche.
songs
were sung
all the night
Songs were sung all night.

In (52a), canciones is the object of the verb; in (52b), it is the subject of the
passive sentence. The possibility of an absent specier thus depends on the
surface position of NP. There are two possible analyses of NPs without overt
speciers. One is that there is simply no Det position present; the other is that
there is a Det position present lled by an empty (covert) specier. One argument supporting the latter analysis is that NPs lacking overt determiners such
as (52a) have essentially the same interpretation as though a specier were
present in this case, the indenite determiner unos some. If the NP has an
interpretation other than indenite, an appropriate specier must be present:
(53)

Cantaron *(las) canciones que aprendieron en Mlaga.

sang
*(the) songs
that learned
in Malaga
They sang the songs that they learned in Malaga.

In (53), where the object NP has a denite interpretation, the Det position
must be lled. This points to an analysis of (52a) in which the specier position is lled by a covert specier.

98

The syntax of Spanish

2.5.2

The DP hypothesis

We saw above that there are two limitations of the analysis of speciers in (45) above, where speciers are generated under a Det node as a daughter of N. One problem is that speciers are not strictly optional, but the
analysis in (45) predicts that the Det node, like other adjunct positions, should
be optional. The second problem is that, although the specier is in some cases
unique, there are also instances of co-occurring speciers. These problems
(and related ones) have been addressed in recent years under a theory of functional categories, which explores structural and grammatical relations
between the lexical categories (nouns, verbs and adjectives) and related grammatical or functional elements such as speciers. A foundational work in this
vein is Abney (1987), which proposes that D (=determiner) heads a phrase DP
(Determiner Phrase). D selects NP as its complement, as shown in (54).
(54)

DP

NP

los

libros

This structure posits a headcomplement relation between Det and NP;

however, because D is a functional category rather than a lexical category, it
does not select NP as a complement NP is not theta-marked by D. Here, the
headcomplement relation is a functional relation. This function can be
thought of in the following way: assuming the NP itself to be basically predicative, the determiner function translates the NP into an expression which
has reference, i.e., refers to denite or indenite individuals.15
15

Higginbotham (1985) proposes a mechanism by which this function is carried out,

which he terms theta-binding. This mechanism is a form of variable binding.
Recall that the licensing of predicative NPs (2.3) was stated in terms of Predication,
the syntactic relation between subject and predicate. Theta-binding is similar to
Predication in certain respects: both processes complete or saturate the predicate via a grammatical relationship with a constituent external to the predicate. The
processes dier at least syntactically, in that Predication operates between full

The Noun Phrase

99

The DP hypothesis accounts for the failure to co-occur of denite, indenite and demonstrative speciers, on the assumption that these are all generated under D, with features of either [+ ( )] or [- ]. The DP
hypothesis also provides a means of accounting for the co-occurrence of denite determiners with post-nominal, but not pre-nominal, possessives and
demonstratives:
(55)

a. los libros esos/mos

the books those/mine
b. *los esos/mis libros
the those/ my books

This contrast can be accounted for on the assumption that strong (stressed)
and weak (unstressed) forms dier structurally, and perhaps categorially.
Suppose, for example, that weak forms are simple determiners, and are always
generated in the head of DP. This head is a non-lexical category, and items
inserted there may be devoid of word-level stress. Strong forms must therefore
be generated elsewhere, perhaps as adjuncts, as shown in (56).16
(56)

DP

los

NP

NP

XP

estos

libros

It is possible that XP (the category of the post-nominal strong form) is also

DP, and that, due to its position within NP, it can inherit phrasal stress associated with the NP.

16

phrases, while theta-binding is a relation between a (functional) head and a lexical

head.
The structure shown in (56) distinguishes NP (a noun phrase) from N, in line with
the traditional three levels of structure of phrases assumed under X-theory. It is frequently assumed in more recent work that lexical categories have only two projections, the head and X level, so that X is non-distinct from phrase.

100

The syntax of Spanish

The structure (56) accounts for the surface order and co-occurrence of
demonstratives and possessives with denite determiners, but does not explain
why the phrase is well formed with two speciers. Here, however, the more
articulated structure of DP, which has a specier position of its own, is useful.
We may suppose that the denite determiner los in (56) is a purely formal
element, akin to pleonastic pronouns. In recent work, it has been proposed
that such elements must be eliminated in the course of a derivation, because
they have no semantic interpretation. If they were to remain, the sentence
would contain a semantically uninterpretable element. This replacement must
occur at an abstract level of representation referred to as Logical Form,
which is the form of a sentence which provides the input to semantic interpretation. If a pleonastic determiner occurs then, it must be eliminated in favor
of a true determiner. In (56), the post-nominal strong form is such an element.
The manner in which it becomes the determiner in Logical Form is by movement to the Specier of the higher DP leaving a trace (t) in its original position, as shown in (57).
(57)

DP

DP i

estos

[+DEF ]

NP

NP

DP

ti

libros

Here, the post-nominal demonstrative DP has moved to the Specier position

of the higher DP. Once there, its presence allows the pleonastic to be eliminated.
The notion of replacement of denite determiners may also provide a
basis for explaining the absence of determiners with proper names. Compare
(58a,b):
(58)

a. los libros
the books
b. *el Juan
the J.

The Noun Phrase

101

This contrast suggests that another means of replacing a denite determiner

is by movement of the head noun to the head of DP (see (59)).
(59)

DP

NP

Ni

Juan

N
ti

Here, the head of NP has moved, adjoining to D. There are various possible
explanations for the fact that common nouns do not undergo this movement.
These might appeal either to dierences in the inherent features of common
versus proper nouns, or to dierences in the structure associated with the two
types of nouns.17 For example, some proper nouns may be inherently denite
singular, while common nouns are not inherently specied for deniteness or
for number. It must be borne in mind, however, that the explanation for the
contrast must be language-specic, or even dialect-specic, as is the contrast
between (58a) and (58b).
Summarizing to this point, we have seen that the DP hypothesis makes two
signicant claims about the structure of determiners in relation to NP. First,
DP, like other categories, has full phrasal structure, including its own Specier,
head and complement positions. Second, the structural relationship between
the determiner and NP is a headcomplement relation. Movement out of the
NP complement is possible, both for XP and Xo (phrases and heads). On the
hypothesis that some functional elements must be eliminated, the order and
co-occurrence of determiners with denite interpretations can be accounted
for.

17

Notice that the movement of a specic subcategory of nouns (proper nouns) is

similar to the phenomenon of V-movement in English, where only the subcategory
of Aux can move to INFL. The movement analysis of proper nouns is due to
Longobardi (1994).

102

The syntax of Spanish

2.6

Pre- and post-determiners

The theory of functional categories discussed above has led to the

investigation of other speciers as heads of functional phrases. This implies
that the actual structure of NP now DP may contain further phrasal projections between DP and NP, a possibility that remains under investigation.
One candidate is a phrase associated with expressions of number and quantity that we will refer to as a Quantier Phrase (QP).18 The hypothesis that QP
is a separate phrase between DP and NP may account for the co-occurrence
of denite and indenite determiners with speciers of quantity. One such
specier is the cardinals, such as in (60), generated as in (61).
(60)

a. unos
some
b. estos
these

(61)

tres
three
tres
three

libros
books
libros
books

DP

QP

unos/estos

NP

tres

libros
18

The Quantier Phrase is proposed for Italian in Giusti (1991). (The analysis presented in the text departs from Giustis proposal, both with respect to the hierarchical position of QP and with respect to the analysis of post-determiner
quantiers.) In addition to the Quantier Phrase, it is argued by Ritter (1991) that
there is a separate phrase for number (NumP), under which the plural ax is generated. For analysis of the NumP in Spanish, see Parodi (1994).

The Noun Phrase

103

If cardinals were generated as heads of D, rather than Q, it would be impossible to account for sequences in which both are present, such as (60). On the
other hand, if cardinals were generated simply as adjuncts (for example, as
adjectives) it would be dicult to account for the fact that cardinals can act
alone as determiners. However, if cardinals are analyzed as functional items,
their distribution can be accounted for along lines similar to those of determiners discussed in 2.5.
(62)

[Tres libros] estn en la mesa.

three books are on the table

(63)

DP

QP

Qi

tres

NP

ti

libros

In (63), the Q tres has adjoined to D, replacing the determiner. Notice that DPs
such as tres libros in (62) must be analyzed as having a determiner of some
type, rather than simply a pre-nominal adjunct. This is so since the pre-verbal
subject position is one which disallows bare NPs, i.e., DPs without any overt
determiner. This is shown by the contrast between (62) and (64):
(64)

*[buenos libros] estn en la mesa.

good
books are on the table

Elements such as varios several, muchos many and pocos few, which may
either appear alone as determiners or co-occur with a denite or indenite
determiner, are other candidates for heads of QP.

104

The syntax of Spanish

The structure of QP may also account for the distinctive properties of todos,
as opposed to quantiers like varios and pocos. Recall that todos behaves like
a pre-determiner, in that it precedes rather than follows a determiner, as in
(65), and co-occurs with other quantiers, as in (66):
(65)

a. todos los/estos libros

b. *los/estos todos libros

(66)

todos los diez / muchos libros que

compraste
all
the ten / many
books that (you) bought

These generalizations are reminiscent of the co-occurrence of denite determiners discussed above in 2.5. Recall that there it was suggested that these cooccurrences can be accounted for on the assumption that functional elements
can be generated either as a head or as a full phrase. Let us extend that analysis to quantiers. Suppose QP has a specier position in which a phrasal
expression of quantity can be generated. Let us take todos to be such a specier of QP.
(67)

DP

QP 1

los

QP 2

NP

(diez)

libros que compraste

todos

In (67), QP1 can be headed by a quantity expression, such as diez ten, underlying (66), or by an abstract [+Q] head, as in the structure underlying (65a).
To derive the surface order of (65a) or (66), todos moves to the specier of DP,
as shown in (68).
Under the assumptions adopted so far, there are several types of evidence
that todos is generated as QP2 in (68), that is, as a specier of QP1, rather than
as the head of QP1. One type of evidence is the co-occurrence of todos with

The Noun Phrase

(68)

105

DP

QP i

todos

QP 1

los

QP 2
ti

NP

(diez)

libros que compraste

both D and Q, as in (68). This implies that there is a separate phrasal node
dominating todos.19 Another type of evidence is the phenomenon of quantier oat, or Q-oat. This is illustrated in (69):
(69)

Los estudiantes dicen todos que el examen fue difcil.

the students
say
all
that the test
was dicult
The students all say that the exam was dicult.

In (69), todos appears separated from the DP (los estudiantes) with which it is
interpreted. Assuming that todos is generated together with DP, its separability can be accounted for on the basis of XP (phrasal) movement: the QP dominating todos can move from the Spec of DP to other Specier positions in the
clause.20 Q-oat is impossible for heads of QP, which undergo only head movement, and cannot move to phrase-level specier positions:21
(70)

19

20
21

*los estudiantes dicen varios/tres que ...

the students
say several/three that ...
The students several/three say that...

The evidence which supports analyzing todos as a Specier of QP is also consistent

with the possibility that todos is generated directly as a Specier of DP, instead of QP.
For discussion of Q-oat, see Kayne (1975), Sportiche (1988).
Ambos both behaves like todos with respect to Q-oat:
(i)

Los estudiantes dijeron ambos que el examen fue difcil.

the students
said
both that the exam
was dicult
The students both said that the exam was dicult.

It is also like todos in lacking a partitive form: *ambos de los libros both of the
books. Unlike todos, ambos cannot precede the denite article: *ambos los libros
both the books.

106

The syntax of Spanish

A third type of evidence supporting a structural distinction between todos

and other quantiers is the distribution of the partitive construction:
(71)

a. muchos/varios/cinco de los libros

many/few/ve
of the books
b. *todos de los libros
all
of the books

In (71a), those speciers that were analyzed as heads of QP in (61) above can
be followed by a de+DP, rather than NP. As (71b) shows, todos cannot occur
in this construction. This contrast may follow from the structural dierence
between todos and other quantiers discussed before, shown in (68) above.
Notice that quantiers other than todos are transitive (not in the sense that
they assign a Theta-role, but in the sense that they select an NP complement).
Todos, on the other hand, is a Specier of QP, which does not select a complement NP.22 There are several possible ways of analyzing the structure of partitives, of which we will briey consider one. Suppose that in the partitive
construction the de-phrase is not in complement position, but that complement is instead occupied by a covert NP, understood as ones (see (72)).
(72)

QP

NP

NP

muchos

de los libros

This structure accounts for the absence of an overt NP following the quantier in the partitive construction. The impossibility of a null NP following
todos in (71b) would then be related to the fact that todos is not a head that
selects a complement, therefore it cannot select the null NP ones.23
In summary, the postulation of QP as a phrase that can intervene
between DP and NP accounts for the possibility that certain sequences of NP
determiners are possible. Additionally, the phrasal structure of QP, with an
22

23

Cada each behaves like todos in this respect. It lacks a partitive (*cada de los estudiantes each of the students versus cada estudiante each student), but it diers
from todos in other respects. It does not precede a determiner, and does not undergo
Q-oat.
As an alternative, it may be possible to analyze the partitive phrase as a specier, or
external argument of the quantier:

The Noun Phrase

107

independent specier position, accounts for a number of contrasts between

the pre-determiner todos and other quantiers.

2.7

Specifiers of predicative NPs

The preceding discussion presented the structure in (73) for speciers

of NP.
(73)

DP

(Spec)

QP

(QP)

NP

DP is assumed to be obligatory, and QP is optional. (Therefore, DP can select

either a QP or an NP as its complement.) DP is the locus of specication for
(i)

QP

DP

los libros

NP

muchos

The correct surface order would result from movement of Q to D:

(ii)

[DP muchosi [QP de los libros ti [NP e ]]]

An analysis like (ii) correctly predicts that the partitive construction is incompatible
with sequences of a denite or indenite determiner followed by muchos (*esos
muchos de los libros). Notice also that in (ii), de is a marker of partitive Case assigned
by the quantier, unlike the adjunct analysis, where de is the head of a true prepositional phrase.

108

The syntax of Spanish

(in)deniteness, and QP is the locus of expressions of quantity or amount.

Recall also from the preceding discussion that expressions of quantity are in
fact determiners, in the sense that they move to D or DP. In this section, we
consider the distribution of determiners and quantiers in predicative NPs.
Under the DP hypothesis described above in 2.5, recall that there is a functional relation between D and its complement NP which converts the predicative category NP to a referring expression. Given this view of determiners,
predicative NPs might be expected to dier from argument NPs with respect
to the necessity of a determiner, since predicative NPs do not refer, but are
predicated externally of a subject, as discussed in 2.3.1. As we will see below,
predicative NPs do not generally require determiners; however determiners do
surface in predicate NPs under specic conditions. We will also see that predicative NPs do not allow as broad a range of determiners as are possible for
arguments.
Recall that instances of predicative NPs are the boldfaced items in (74):
(74)

a. Mara es doctora.
M. is (a) doctor
b. Ese estudiante parece
genio.
that student
seems (a) genius
c. Consideran amigo a Juan.
consider (a) friend PA J.
They consider Juan a friend.
d. Eligieron a Juan presidente.
elected PA J.
president
They elected Juan president.

Notice that these singular ([-plural]) items are not preceded by a determiner.
This contrasts with corresponding singular argument NPs, which generally
require a determiner of some type:
(75)

a. Mara reconoci a *(una) doctora.

M.
recognized PA (a)
doctor
b. Ese estudiante conoce a *(un) genio.
that student
knows PA *(a) genius

The contrast between (74) and (75) suggests that predicative NPs are in fact
NPs, not DPs. Predicative NPs may, however, appear with a determiner, if one
is needed independently, for example to support a comparative:
(76)

a. *Juan es mejor
amigo.
J.
is better/best friend
Juan is a better / the best friend.
b. Juan es el mejor amigo que tengo.
J.
is the best friend that have.
Juan is the best friend that I have.

The Noun Phrase

109

c. Juan es mi mejor amigo.

J.
is my best friend.
Juan is my best friend.

In (76), the comparative mejor induces a requirement for an overt determiner.

Several approaches might be pursued in accounting for the contrast between
modied and unmodied predicative NPs. One is that they dier in their interpretation, and this dierence is reected in their category: DP versus NP.
Alternatively, it may be that both are DPs, and that the presence or absence of
an overt D is determined by phrase-internal relations.
Let us turn now to those determiners that have been analyzed as generated
under QP. This class of determiners is uniformly absent for predicative NPs,
whether modied or not:
(77)

a. Ellos
they
b. Ellos
they

son (*muchos/varios/pocos/tres) pilotos.

are
many/several/few/three
pilots
son los (*muchos) pilotos que
conoc ayer.
are the many
pilots that (I) met
yesterday

Finally, notice that it appears at rst glance that the QPs todos and ambos can
co-occur with predicative NPs:
(78)

Ellos son todos/ambos pilotos.

they are all/both
pilots

Although the position occupied by todos in (78) appears to be a specier of

the predicative NP, notice that todos is related to ellos, not pilotos. That is, (78)
is a variant of (79):
(79)

Todos ellos son pilotos.

all
they are pilots
All of them are pilots.

The discontinuity of [todos ellos] is then a result of Q-oat. This is conrmed

by the fact that where the subject NP has its own quantier, todos becomes
impossible:
(80)

*Muchos de ellos son todos pilotos.

many
of them are all
pilots

To summarize, we have seen above that predicative NPs have overt determiners only if modied, and do not take any QPs.

2.8

Constituent order within NP

The last topic that we address with respect to NP concerns the order
of adjectival adjuncts in NP relative to the head and complements. Recall from

110

The syntax of Spanish

2.4 that the preliminary phrasal skeleton introduced there, repeated below
as (81), generates the basic order summarized in (82).
(81)

(WP)

N
(82)

(YP)

(ZP)

Head Complement Adjunct

Most adjuncts, including PPs, APs, clauses and small clauses occupy the
canonical adjunct position at S-structure. APs, however, have greater
freedom of position with respect to the head and complements. Certain APs
may be pre-nominal as well as post-nominal, as illustrated in (83):24
(83)

a. una comida muy buena

a
meal
very good
a very good meal
b. una muy buena comida
a
very good meal

Furthermore, APs can also appear between the head noun and a complement:
(84)

la destruccin completa de la ciudad

the destruction complete of the city

In this section, we will consider how these orders are derived.

Let us take as a starting point the alternative orders in (83), and consider
several ways of generating these two orders. One possible analysis is that each
order is base-generated in the same order that appears on the surface. That is,
each order is generated from a separate D-structure, which implies that
adjuncts can have more freedom of position than was given in the basic
phrasal skeleton in (81). Suppose instead that adjuncts could be generated
either before or after the N to which the adjunct is adjoined, as shown in (85):
24

Pre-nominal APs cannot contain complements:

(i)

a. un competente pianista
a competent pianist
b. *un competente en jazz pianista
a competent in jazz pianist

This generalization also holds in English, as noted in Emonds (1976). This is shown
by the ungrammaticality of the gloss for (ib).

The Noun Phrase

(85)

b.

DP

a.

una

DP

AP

muy buena

111

una

AP

muy buena

comida

comida

On this analysis, the contrast in order between complements (which must

follow the head) and adjuncts (which can precede or follow the head) could
derive from theta-theory: theta-marking by the head may be from left to
right in head-initial languages. However, this analysis has several drawbacks. First, it does not account for the fact that adjuncts other than APs
must be post-nominal. Consequently, the structure shown in (85b) would
generate ungrammatical sequences, for example with pre-nominal PPs or
other types of adjuncts. Second, recall from Section 2.4 that qualitative
adjectives (those which can be both pre-nominal and post-nominal) have
dierent interpretations depending on their order. Post-nominally, they are
restrictive, while pre-nominally their interpretation is appositive. The analysis in (85) does not provide any way of accounting for this dierence in
interpretation.
A variation of this analysis overcomes the last drawback noted above. Prenominal APs (with an appositive interpretation) can be analyzed as Speciers
of NP (daughters of N) rather than as adjuncts. Example (85b) would then
have the structure shown in (86).
(86)

DP

una

AP

muy buena

comida

112

The syntax of Spanish

The idea underlying this analysis is that the level at which a non-argument is
attached plays a role in determining how the constituent is interpreted in the
phrase. Jackendo (1977) argued that restrictive modiers are lower in the
structure than are non-restrictive modiers. A restrictive modier may be
assumed to receive its restrictive interpretation via the adjunction structure
from the additional N that is present. A non-restrictive modier is simply a
specier, without an additional N.
Let us now return to the third position in which APs can occur between
the head and its complement, illustrated in (84), repeated below:
(84)

la destruccin completa de la ciudad

the destruction complete of the city

Consider rst the possibility that this order is base-generated (in other words,
that the surface order is the same as the D-structure order). If we continue to
assume that only binary branching is permitted, the structure would be as in
(87).
(87)

DP

NP

la

PP

AP

completa

de la ciudad

destruccin

Here, the AP completa is a post-nominal adjunct, and the complement PP is

incorrectly also adjoined. This structure cannot be correct, because the PP
is not a sister of the head, and therefore cannot be theta-marked.
A plausible alternative to base-generating the structure shown above is to
suppose that the complement has undergone rightward movement from a Dstructure position adjacent to the head (see (87)).

The Noun Phrase

(88)

113

DP

NP

la

PP

AP

PP

destruccin

de la ciudad

completa

On this analysis, the PP is a sister of the head at D-structure, and can be thetamarked by it.
Although the preceding analysis achieves an adequate description of the
surface order of constituents in NP, it has been questioned on several grounds,
both theoretical and empirical. One theoretical point that is a topic of ongoing
inquiry is whether there is any rightward movement at all (Kayne 1994). Leaving
aside this issue, which takes us beyond the matter at hand, let us consider two
related empirical issues. First, the possibility for nounadjectivecomplement
order is common, not just in Spanish, but in Romance generally, and diers
from other languages and language families, as can be illustrated easily in
English, which disallows: *the destruction complete of the city. This cross-linguistic contrast suggests that there is a syntactic parameter at work. Since
parameters are not construction-specic, but involve general lexical properties
of the language, it would be desirable to account for this particular order in as
general a manner as possible. Furthermore, notice that the Romance/English
parameter is not restricted to Noun Phrases, but is reected also in the order of
constituents in clauses:
(89)

a. Destruyeron
completamente la ciudad.
destroyed.3rd.pl. completely
the city
They completely destroyed the city.
b. Mara escribi ayer
una carta.
M.
wrote yesterday a letter
Maria wrote a letter yesterday.

114

The syntax of Spanish

(90)

a. *They destroyed completely the city.

b. *Maria wrote yesterday a letter.

The sentences in (89) show that, in Spanish, an adverb can intervene between
a verb and its complement, while corresponding examples in (90) show that in
English, this order is ungrammatical. Presumably, the account of NP constituent order should generalize to clausal constituent order. In fact, research on
clausal word order has led the way to a unied analysis of these phenomena.
Following Emonds (1978), the contrast in (89) versus (90) is due to the movement of main verbs to the inectional head of the clause INFL, shown in
(91).
(91)

IP

NP

Mara INFL

VP

escribi AdvP

ayer

NP
una carta

On this derivation, which will be discussed in Chapter 4, an adverb is generated in the specier of VP, and the verb moves to its left. The absence of verb
movement in English is conrmed by the impossibility of fronting a verb in
questions (e.g., *Wrote Mary a letter?).
Likewise, it has been proposed that the position to which nouns move must
be the head of a functional category that is associated with NP (Malln 1989;
Cinque 1992). That category must be lower than the DP structure because
the noun is still to the right of any determiners. The functional category in
question has been argued to be associated with the nouns inectional features
of number and gender. We will refer to these features as an Agreement Phrase
(AgrP), since these features are shared by the noun and its agreeing modiers.
On this analysis, the structure of (84) is as shown in (92).

The Noun Phrase

115

DP

(92)

la

AgrP

Agr

NP

AP

destruccin completa

PP
de la ciudad

The noun is not in its original position the position in which it theta-marks
its complement. It has moved to a higher (functional) head that is associated
with the nouns inection. It has been argued that the relative richness or
strength of agreement in Romance languages explains why this order is possible in Spanish (as in other Romance languages) but not in English. In
English, the weakness of the Agreement Phrase does not attract the noun
to this position (except at the level of logical form). Consequently, the order
nounadjectivecomplement does not occur in English.
As a nal point, let us return briey to the analysis of the pre-nominal
adjective position, illustrated in (93):
(93)

a. la completa destruccin de la ciudad

the complete destruction of the city
b. una muy buena comida
a
very good meal

Recall that the structure of (93b), shown in (86) above, was analyzed as having
an adjective in the NP specier position. Assuming now that there is an AgrP
above NP, there are now two possible ways of deriving pre-nominal adjective
order. One possibility is that the noun is in its base position, rather than in the
head of AgrP, and the adjective is in the NP specier. On this analysis, movement of the noun to Agr is optional. A second possibility, however, is that the
noun has moved to the head of AgrP, and that the adjective has also moved,
to the specier of AgrP.
Summarizing the main points of this section, we have seen that there is
a subclass of adjective, qualitative adjectives, that can be generated in two

116

The syntax of Spanish

positions: (a) in standard post-nominal adjunct position, where they are interpreted as restrictive modiers; and (b) in specier of NP position, where they
receive an appositive interpretation. The order headadjectivecomplement
cannot be generated directly from a D-structure in which the adjective follows
the head since, on this derivation, shown in (87) above, the head could not
theta-mark its complement. Two ways of deriving this order by movement
were considered. One analysis is rightward movement of the complement; a
second analysis, which unies NP constituent order with clausal constituent
order, is leftward movement of the noun. As in clauses, the position to which
the noun moves has been proposed to be a functional category associated with
inection (here, nominal inection for number and gender). Dierences
between English and Romance with respect to this order have been explained
in terms of the relative richness or strength of nominal inection in the two
language families.

2.9

Summary

In this chapter, the noun phrase has been examined from two perspectives. From the phrase-external perspective, we considered the distribution
and interpretation of NP, as well as the principles that license NPs. In 2.2 and
2.3, two types of NPs were discussed: argument NPs and predicative NPs.
Argument NPs are licensed by Theta-role assignment and by Case. Since the
heads that assign Theta-roles are not necessarily Case assigners, NP may move
in the course of a derivation to satisfy a requirement for Case. Predicative NPs
are not licensed by Theta-roles, but are predicates that are interpreted via
Predication. Section 2.4 introduced preliminary characteristics of the internal
structure of the phrase. These include (a) the relations between the head and
the phrase (headedness, endocentricity), and (b) the types of constituents that
co-occur with a noun to form a Noun Phrase: complements, adjuncts and
speciers. Sections 2.52.7 focused on the determiner system. In Section 2.5,
the distribution of determiners and their analysis as DPs was introduced.
Section 2.6 then discussed pre-determiners and post-determiners: constituents that can co-occur with determiners, or can function as determiners
themselves. These elements were analyzed as functional elements generated in
a Quantier Phrase. Section 2.7 summarized the determiner system of predicative NPs, which has overt determiners when modied, but cannot have independent quantiers. Section 2.8 returned to lower NP structure, and
considered the derivation of NPs that contain pre-nominal APs as well as APs
that intervene between the head and its complement. These orders have been

117

The Noun Phrase

analyzed in terms of movement of the head noun to the left to the head of
a functional category that we have called AgrP. This category bears inectional features of number and gender, and is analogous to the inectional head
of clauses. A noun phrase such as (94) then may include the functional projections shown in (95):
(94)

todas estas varias interesantes categoras estudiadas

all
these several interesting categories studied

(95)

DP

QP

todas

estas

QP

QP

AgrP

varias

AP

Agr

interesantes

Agr

NP

AP

categoras
N

AP

estudiadas

The Verb Phrase

3.1

Introduction

In Chapter 2, we studied the Noun Phrase from two points of view.

First, from a phrase-external perspective, we discussed the contexts in which
NP is found, and introduced subtheories that account for NP distribution.
We then examined the internal structure of the phrase, describing the basic
structural and functional relations among constituents that co-occur with a
noun to form NP (or DP). In this chapter and in Chapter 4, we will consider
these same issues with respect to the Verb Phrase (VP). We will begin the discussion in this chapter with an overview of the distribution of VP, and
provide a preliminary description of principles that account for this distribution (Section 3.2). We will see that there are two grammatical relations that
restrict the distribution of VP: the relation between VP and the clausal
subject, and the relation between VP and Tense. Subsequent sections of this
chapter will be concerned with relations between the head of the phrase and
its arguments. Section 3.3 discusses the external argument, or subject;
Sections 3.43.7 describe complements or internal arguments. Although we
will not provide a detailed account of the properties of verbal arguments, we
will see that the subtheories introduced in Chapter 2 provide a means of
structurally distinguishing verbal arguments from non-arguments. In
Chapter 4, we will take up several additional issues related to the structure of
the Verb Phrase, including how clitics and auxiliary verbs are related to the
verb and its arguments.

3.2

The distribution of VP

This section is concerned with the contexts in which Verb Phrases are
found. We begin in 3.2.1 with a summary of environments in which VP is generated, and then discuss in 3.2.2 two factors that determine this distribution.
118

The Verb Phrase

3.2.1

119

VP as primary predicate of the clause

Consider rst the contrast between sentences (1a) and (1b):

(1)

a. Susana descans.
Susana rested.
b. *Susana cansada
Susana tired

These sentences dier in the category of their main predicate. In (1a), the predicate is the VP headed by the verb descans, while in (1b), assuming that no
verb has been ellipted, the main predicate is an Adjective Phrase headed by
cansada. This contrast in grammaticality illustrates the rst generalization
with respect to the distribution of VP, namely that a VP and only VP occurs
as the primary predicate in clauses.1 Notice that this contrast would be dicult to account for on semantic grounds, since adjectives and verbs are both
predicative categories that can describe states, and they can be similar in their
argument structure. The dierence between VP and AP (likewise, NP and PP)
must be a syntactic dierence. A simplied (surface) structure for (1a) is shown
in (2a), and the distributional generalization is given as (2b). In (2), the clause
is IP (Inectional Phrase), whose head, INFL, contains features for Tense
(and Agreement), and VP is the sister of the INFL node. Let us refer to this
position as the primary predicate of the clause.
(2)

IP

a.

DP

Susana

INFL

VP

TENSE

[+FINITE ]

descans

b. (Only) VP occurs as a sister of INFL (as primary predicate of a

clause).

This generalization is common cross-linguistically. Some languages, however, such as

Japanese, appear to allow categories other than VP as main clausal predicates.

120

The syntax of Spanish

As we will see below, VP does not occur in adjunct positions, where it corresponds to a secondary predicate of a clause, as in (3), where there is a VP
in an adjunct position (VP2) that is not contained in a separate clause.
Adjuncts can of course contain VPs, but, in such cases, there is evidence that
the VP is a sister of a separate IP node, so that the VP within the adjunct conforms to (2b). This is illustrated by the contrast between (4a) and (4b):
(3)

IP

VP1

IP

NP

INFL
(4)

VP 1

a. Susana dej la sala [cansada].

Susana left the room [tired ].
b. *Susana dej la sala [descansaba/descansar]
Susana left the room [(she) rested/to rest]

In (4a), there is a clausal adjunct or secondary predicate, the AP cansada. In

(4b), the secondary predicate is a VP, which is ungrammatical in both nite
and non-nite form. Notice that (4b) becomes grammatical if the adjunct is
introduced by a preposition like para in order:
(5)

a. Susana dej la sala para descansar.

Susana left the room in order to rest.
b. Susana dej la sala para que Jos descansara.
Susana left the room in order that Jos rest.

Para selects as its complement a clause, as is shown by the presence of the overt
complementizer que in (5b). In this context, either an innitive or nite (subjunctive) form of the verb is possible, because VP is again related to an IP in
the subordinate clause.
We have seen that VP is obligatory as the primary predicate of the clause,
or IP, as shown in (1) above. VP is impossible as a secondary predicate of a
clause, as shown by the contrast in (4) above. A VP may appear within an
adjunct only if, within that adjunct, it is the primary predicate of a separate
IP node. There are certain adjuncts, such as those illustrated in (6), that supercially resemble VPs rather than full clauses:

The Verb Phrase

(6)

121

a. Llegado Jos, empez la esta.

Jos arrived, the party began.
b. Habiendo terminado la leccin, Susana sali.
Having nished the lesson, Susana left.

These participial adjuncts might appear to be a constituent closer in structure to VP than to IP. However, comparable Italian constructions are argued
in Belletti (1990) to have clausal structure. One argument supporting this
conclusion concerns the order of the participle relative to an overt subject,
such as Jos in (6a). An overt subject of a participial clause must follow the
participle:
(7)

*Jos llegado, empez la esta.

Jos arrived, began the party

Belletti (1990) argues that the order participlesubject can be explained on the
assumption that the participle has undergone head movement to an empty
complementizer position, as shown in (8) (with a simplied structure):
(8)

[CP llegado [IP Jos [VP ]]]

The presence of the complementizer position as a landing site for verb movement then provides indirect evidence that the structure is a clause, just as the
overt complementizer indicates clausal structure in (5b).
There is one quite clear exception to the generalization that VP appears only
as a primary predicate. In clauses with auxiliary verbs, sequences of VPs are
possible, as in (9b):
(9)

a. Los nios comieron.

The children ate.
b. Los nios han comido.
The children have eaten.

Assuming both the auxiliary and the participle to be verbs, (9b) must have two
VPs, one dominating each verb. There have been numerous proposals in the
literature as to the structure of sequences with auxiliaries, and we will defer
fuller discussion of their structure and function until Chapter 4. Here, the
point of interest is that, if such sequences contain two verbs (and therefore
two VPs), but only one INFL node, it follows that one of these VPs is in some
position other than that of primary predicate. Two possible analyses of the
structural relation between an auxiliary and the following participle are
shown in (10).
In (10a), the VP dominating the auxiliary is structurally the primary predicate (sister of INFL), and the participle is also the sister of a head but the

122
(10)

The syntax of Spanish

a.

b.

VP 1

INFL

INFL

VP

VP 2

VP

han

comido

comido

han

head is the auxiliary verb. In (10b), the auxiliary VP is analyzed as a specier

of the participle, the latter being in primary predicate position. Both of these
structures require some loosening of the description of VP distribution.
Either the VP can occur as the sister of (a) INFL or (b) an auxiliary verb
(structure (10a)); or VPs headed by certain verbs auxiliaries may be speciers or adjuncts of a primary predicate. Finally, note that any analysis of auxiliaries must be able to accommodate multiple auxiliaries:
(11)

Esta manzana puede haber estado siendo comida.

This apple may have been being eaten.

Another possible exception to the distribution of VP as a primary predicate

is constructions with verbs that are informally described as semi-auxiliaries.
These verbs are main verbs according to standard diagnostics (for example,
unlike auxiliaries, they select their own arguments). However, when followed
by an innitival complement, the sequence of verb+innitive behaves syntactically like sequences of auxiliary followed by main verb. These predicates
include aspectual and volitional verbs like (12), and causative and perception
verbs such as (13):
(12)

a. Mara lo
quiere cantar.
M.
CL(Acc.) wants to sing
Maria wants to sing it.
b. Pedro lo
volvi
a copiar.
P.
CL(Acc.) returned to copy
Pedro copied it again.

(13)

a. Juan se lo hizo escribir (a Pedro).

J. CL(Dat.) CL(Acc.) made write (to P.)
Juan made him (Pedro) write it.

The Verb Phrase

123

b. Mara lo
vio pasar.
M.
CL(Acc.) saw happen
Maria saw it happen.

Predicates of the type in (12), referred to as Restructuring predicates, continue to be topics of investigation. Although there has been controversy as to
the underlying structure of these sequences, there is consensus that in the
supercial structure (and the logical form) of these sentences, the innitive
behaves like a constituent of the matrix clause. (See Chapter 6, Section 6.5.)
Summarizing, we have seen above that the contexts in which VP occurs are
restricted: the principal context is as a sister of INFL, i.e., as the primary predicate of a clause. VP does not occur as an independent adjunct. The clearest
exception to this generalization is auxiliary verbs.

3.2.2

Licensing VP

In this section, we will restrict our attention to VP as primary predicate, that is, in the position of sister of INFL, as shown in (14).
(14)

IP

DP

INFL

VP

Here, VP is licensed both by its relation to the subject DP and by its relation
to INFL. Let us consider these in turn. The relation between VP and the
subject is via Predication. Recall from Chapter 2 that a predicative NP is
licensed by Predication a relation between the NP and an argument that is
interpreted as its subject. Likewise, VP is predicated of a subject. In (14), VP
is in a conguration that satises Predication. Although the VP and the
subject DP are not mutually c-commanding, (Chapter 2, Section 2.2.2), they
satisfy a slightly looser congurational requirement: mutual m-command.2
2

M-command is dened as in (i):

(i) m-commands i does not dominate and every maximal projection that
dominates dominates .
In (14), the subject and VP m-command each other, since neither dominates the other
and the only maximal projection that dominates one dominates the other the IP.

124

The syntax of Spanish

The principle that enforces Predication for VP is referred to as the Extended

Projection Principle (EPP), which states that every clause must have a subject.
Because VP is the primary predicate in clauses, the EPP enforces Predication
for VP.
Although the subject of a clause is not always overt, dierent types of
clauses provide evidence that a subject constituent is present.3 This can be
illustrated for nite clauses in Spanish, which unlike in English allow
covert subjects subjects that lack phonetic content:
(15)

a. Mary walked.
b. *Walked.

(16)

a. Mara camin.
Maria walked.
b. Camin.
walk.past.3rd.sg.
You/she/he walked.

The contrast between the grammaticality of (16b) and the ungrammaticality

of (15b) in English is described in terms of the presence versus absence of
null subjects. Although one might be tempted to analyze (16b) as lacking a
subject DP, there is evidence that (16b) has a DP in the specier of IP which
simply lacks phonetic content, as shown in (17).
(17)

IP

DP

pro
[3rd.]
[sg.]

INFL
[PAST]

VP

camin

It is well beyond the scope of the present discussion to show that clauses must have
subjects. We will touch on this topic in Chapter 4. In-depth discussion is to be found
in Rothstein (1983).

The Verb Phrase

125

Here, the subject position is occupied by a DP that has syntactic features, but
no overt phonetic features. One type of evidence supporting this analysis is
that the VP in (16b) is interpreted as an event performed by an Agent, which
means that a Theta-role has been assigned to some constituent. There are also
features carried by DPs beyond the person and number features that are
overtly marked on the verb. These are semantic features such as [+ ],
which allow the DP to satisfy the semantic selectional restrictions of the predicate. The features of the subject constituent also allow it to participate in
various grammatical processes. For example, the subject can serve as the antecedent of a reexive (e.g., Hablaba consigo mismo He talked to himself ). If
an appropriate antecedent for a reexive is not present in the structure, the sentence becomes ungrammatical, as in: *El problema fue resuelto por s mismo
The problem was solved by himself.
Returning to the main point, the obligatoriness of the clausal subject under
the EPP has the eect of ensuring that VP will be licensed by Predication.
However, this does not by itself automatically account for the ungrammaticality of VP as an adjunct, since predicative adjuncts also undergo
Predication. The restricted distribution of VP must therefore be accounted for
on some other basis. There is, as noted above, another constituent in IP with
which VP is related: the head, INFL, bears Tense features that are related to
the inectional morphology of the verb. Although the precise nature of the
relationship between INFL and V is a topic of ongoing research (and is a
point to which we return in Chapters 4 and 5), we can describe the relation
here in informal terms. The Tense features of INFL ([ ], [ ]) participate in determining the interpretation of VP, because these features
provide a temporal specication or temporal location for the event
expressed by the VP. It is possible to think of these features of INFL as a type
of determiner for VP. On this view, we might draw an analogy between the
INFL/VP relation in (14) and the Det/NP relation that was discussed in
Chapter 2. In fact, IP is standardly analyzed as a functional category,
although, as noted above, the precise nature of its functional relation to VP
is under active investigation.
We have considered here the conditions that account for the occurrence of
VP in primary predication contexts. Recall, though, that VP can also occur as
a sister of certain verbs, including auxiliaries and semi-auxiliaries. To unify
these contexts with those discussed above, it may be possible to analyze these
auxiliary verbs as having grammatical features that are similar in relevant ways
to Tense features. This would allow them to provide a temporal specication
for VP just as INFL does. In other words, these verbs might be analyzed as

126

The syntax of Spanish

functional categories that share certain crucial properties with INFL. These
issues will be taken up in Chapter 4. Below, we will turn to consideration of
one facet of phrase-internal syntax, namely the relations between the verb and
its arguments.

3.3

The external argument of VP

We saw above in 3.2 that, due to the EPP, a subject constituent must
be present in the Specier of IP and, once that constituent is present,
Predication co-indexing occurs. In this section, we will take a closer look at the
derivation of the subject. We will see below that the interaction of thetatheory and Case theory determines the position of the subject at D-structure
(the initial syntactic representation, or deep structure) and S-structure (the
representation after overt movement, or surface structure). One issue that
we will be concerned with is the conditions under which a subject is assigned
a Theta-role by the predicate. Recall from the discussion of theta-theory in
Chapter 2 that a DP is an argument (i.e., can have reference) only if it is
assigned a Theta-role. As is illustrated by the contrast between (18a) and
(18b), not all predicates assign a Theta-role to a subject:
(18)

a. Los nios leyeron los libros.

The children read the books.
b. Parece que los nios leyeron los libros.
(It) seems that the children read the books.

In (18a), the VP headed by leyeron assigns a Theta-role (Agent) to the DP los

nios. In (18b), the VP headed by parece does not assign any Theta-role to the
subject, and the covert pronoun can only be interpreted as pleonastic (non-referential). This pleonastic pronoun is grammatically necessary to satisfy the
EPP (and hence, Predication). The VPs headed by leyeron read and parece
seem dier with respect to their argument structure: only the former has the
lexical property of assigning a thematic role to an external or subject argument. The rst topic that we will address below in 3.3.1 is the relationship
between this lexical property and the D-structure syntactic representation. We
will then see how theta-assignment interacts with Case to derive the S-structure
representation. Then, in 3.3.2 we will consider a variant of (18b) in which a
pleonastic has not been inserted, as shown in (19):
(19)

Los nios parecen haber ledo los libros.

The children seem to have read the books.

The Verb Phrase

127

In (19), the main clause has an S-structure referential subject, los nios
despite the fact that the predicate parece does not assign any external Thetarole.

3.3.1

Theta-role assignment and Case assignment to the external

argument

To explore Theta-role assignment to the external argument, consider

again predicates such as leer which assign a Theta-role, setting aside for the
moment predicates like parecer, which do not. As noted above, the assignment
of Theta-roles by individual verbs is largely idiosyncratic. The fact that argument structure is not predictable means that it must be specied in the verbs
lexical entry. The lexical entry for leer, for example, would include the role
Agent (los nios in (18a)), and the role Theme (los libros in (18a)).
Furthermore, the lexical entry must specify a structural context in which each
role is assigned. The verb leyeron assigns the role Theme to its complement
(or internal argument), and the role Agent to its subject (or external
argument). These roles are not interchangeable. We cannot say: *los libros
leyeron (a) los nios, meaning that the children read the books. The lexical
entry for leer must not only specify the roles assigned by the verb, but must
also dierentiate its external from its internal role(s). This can be specied in
the lexical entry as shown in (20):
(20)

leer: V

Theme Agent(External)

In (20), the Agent role has been identied as an external argument.

Arguments not so specied are internal arguments, or complements of the
verb.
Theta-theory characterizes the conguration in which Theta-roles such as
those in (20) are syntactically realized on, or assigned to, a constituent. Internal
arguments are generated usually as sisters of the verb; their Theta-role is
assigned by a head to a sister of the head (see (21)). We will see in later sections
that Theta-role assignment is not restricted to sister constituents, but is possible as long as the head c-commands the complement. The head c-commands
the complement if the rst branching node dominating the verb (here, V) also
dominates the complement. Constituents that are sisters both c-command
each other. In (21), DP also c-commands the head, because the rst branching
node that dominates DP also dominates V. Mutual c-command is not essential
to theta-marking, but the head that assigns a Theta-role must c-command the
constituent to which the Theta-role is assigned.

128

The syntax of Spanish

(21)

DP (Theme)

leyeron

los libros

But what about the external arguments Theta-role? In earlier versions of

transformational grammar, it was assumed that the subject of a clause was
generated in the specier of IP, in which case it is structurally distant from the
verb more distant even than adjuncts of VP. More recently, it has been argued
that the external argument of the verb is generated within the VP. In particular, the external argument position is analyzed as generated in the Specier of
VP, as shown in (22). Here, there is no DP in the specier of the IP in D-structure; the Agent role is assigned to a DP that is in the specier position of VP.
This position is distinct from complement position, in that the Agent DP is not
a sister of the head. It was noted above that the verb assigns a Theta-role to a
sister constituent. In (22), the external argument is a sister of V, and Chomsky
(1986) has suggested that external Theta-roles are actually assigned by V
that is, compositionally by the verb together with its sister constituent.4 If this
is correct, then all Theta-roles are assigned to a sister: in (22), the Theme role
is assigned by the verb to its sister; the Agent role is assigned by V to its sister
in the Specier of VP. This analysis of theta-assignment to the external argument is referred to as the VP-internal subject hypothesis.
The structural conditions under which Theta-roles are assigned then determine the position of the external argument at D-structure. Let us now consider the derivation to S-structure. Recall from Chapter 2 that DP must be also
assigned Case. The verb does not by itself assign Case to its external argument.
The Case of a subject (Nominative Case) is instead assigned by the head of
the clause, INFL, to a DP in its specier position. Consequently, the external
argument in the Specier of VP must move to the Specier of IP (see (23)).
This movement then produces a surface structure in which the subject is in its
14

As noted in Chapter 2, external roles are assigned compositionally, as illustrated in

(i):
(i)

a. John broke the window.

b. John broke his arm.

(External role: AGENT)

(External role: THEME)

These sentences illustrate that the role of the subject of the verb break can vary
depending on features of the complement.

The Verb Phrase

(22)

129

IP

INFL

VP

DP

(Agent)
los nios V

DP (Theme)

leyeron

los libros

canonical position, i.e., the Specier of IP. Note that, unlike English, the
clausal subject in Spanish does not always precede the verb in declaratives, a
dierence that might be attributed to a parametric dierence in Case assignment. (See also Chapter 5 on clausal constituent order.)
(23)

IP

DP

los nios

INFL

VP

DP

DP (Theme)

leyeron

los libros

Summarizing to this point, we have dierentiated two types of Theta-roles:

internal roles, assigned to complement positions, and external roles, assigned
by V to an argument in the Specier of VP. Movement of that argument to
the position of clausal subject, or Specier of IP, is triggered by Case,
because INFL is the assignor of Nominative Case.

130

The syntax of Spanish

3.3.2

Subject-to-subject raising

The derivation of the clausal subject discussed above characterizes

the external argument as related to two dierent constituents. It is thetarelated to V (i.e., is assigned its Theta-role by V); and it is Case-related to
INFL (i.e., is assigned Case by INFL). At S-structure then, the subject does
not occupy just a single position, but forms a chain consisting of two DP positions: the Specier of VP (the position to which a Theta-role is assigned), and
the Specier of IP (the position to which Case is assigned). Analyzing the derivation of the clausal subject in this way provides a natural account for alternations like (24a) versus (24b):
(24)

a. Parece que los nios leyeron los libros. (=18b)

b. Los nios parecen haber ledo los libros.
The children seem to have read the books.

In (24a), the subject of the main clause is the covert pleonastic pronoun corresponding to English it. As noted above, this pronoun has no reference
because parecer does not assign any Theta-role to it. Nevertheless, in order to
satisfy the EPP (and Predication), the covert pronoun is inserted in the
Specier of IP, where it is assigned Case, and where it also triggers
subjectverb agreement. The main clause in (24a) has a grammatical subject
one that has grammatical features, and Case, but it has no semantic content,
which follows from the fact that it is not assigned a Theta-role. Now consider
(24b). The subject of the main clause does have semantic content, which means
that it must bear a Theta-role. However, this role cannot have been assigned
by parecer, which we have seen in (24a) does not assign an external role. How
can the role in (24b) have been assigned? Notice that a verb in the subordinate
clause assigns an Agent role. Notice as well that in (24b), unlike (24a), the subordinate clause is not nite. What these facts suggest is that the Agent DP is
generated in the Specier of the VP of the subordinate clause, and it has
moved to the Specier of IP in the main clause, where it is assigned Case. The
D-structure of the subordinate clause is shown in (25a), the S-structure in
(25b):
(25)

a. [parece [IP INFL [VP los nios [haber [ledo los libros]]]]]
b. los nios INFL [parecen [IP INFL [VP [haber [ledo los libros]]]]]

The construction illustrated in (25) is referred to informally as Subject-tosubject Raising or simply Raising, because the external argument of a subordinate clause has undergone movement to a higher clause (i.e., has
raised). This movement of DP is triggered by Case: the DP moves from its
theta-position in the lower clause to the Specier of the matrix IP where it is

The Verb Phrase

131

assigned Nominative Case. This movement also satises the EPP for the main
clause, since, once this movement has taken place, the matrix IP has a subject.
Finally, notice that Raising cannot occur in structures like (24a), where the
subordinate clause is nite, although we will not explore here the reasons why
it is impossible. Notice, however, that if movement were to occur, the moved
NP would have Case twice: the Nominative Case from its original position in
the subordinate clause, and the Nominative Case from INFL of the main
clause.

3.4

Complements of V: prepositional complements versus adjuncts

We turn now to the structural position that is closest to the verb,

namely the sister of the head, or complement position, shown as XP in (26):
(26)

XP (complement; sister of V)

This is a position that is theta-marked by the verb. It can be occupied by complements of dierent categories, such as those illustrated in (27):
(27)

a. Pedro dijo que los nios leyeron los libros.

(CP) (=clause)
Pedro said that the children read the books.
b. Susana pate la pelota.
(DP)
Susana kicked the ball.
c. Juan habl con el vecino.
(PP)
Juan talked with the neighbor.

All of the complements in (27) are generated in the position XP in (26), which
is the position that is theta-marked by the verb. One of the rst questions that
arise concerning the analysis of complements is simply: how do we know
whether a constituent that follows the verb is a complement at all? On what
basis are complements dierentiated from adjuncts? To illustrate the problem,
consider the PPs following the verb hablar in (28):
(28)

a. Juan habl de poltica.

Juan spoke about politics.
b. Juan habl de nuevo.
Juan spoke of new
Juan spoke again.

As we will see below, there are diagnostics that can be applied to a given constituent to determine its relation to the head. These will show that in (28a), the

132

The syntax of Spanish

PP de poltica is a complement of habl; in (28b), the PP de nuevo is an adjunct,

which is a sister of V. Their respective structures are shown in (29a,b).
(29)

VP

a.

b.

VP

PP

PP

habl

de poltica

de nuevo

habl

Diagnostics which distinguish PP complements from PP adjuncts include:

(a) hacerlo do so replacement, and (b) substitutability of the preposition.
Hacerlo is a proform which replaces a V constituent, i.e., the head and its
complement. Let us illustrate this rst with the verb patear and its DP complement:
(30)

Pedrito pate la pelota y Jos lo hizo tambin.

Pedrito kicked the ball and Jos did so too.

In the second conjunct of (30), lo hizo is a proform that has replaced a second
occurrence of the constituent pate la pelota. The antecedent of this proform
is the V (pate la pelota) in the rst conjunct. Like English do so, hacerlo may
also replace a higher V, one which dominates V and an adjunct. To see this,
consider rst (31a), whose VP structure contains two V nodes, as shown in
(31b).
(31)

a. Pedrito pate la pelota con entusiasmo.

Pedrito kicked the ball with enthusiasm.

V1

b.

V 2

PP

DP

pate

la pelota

con entusiasmo

The Verb Phrase

133

Hacerlo can replace either of the V constituents in (31b) as shown by the two
examples in (32), where the antecedent of hacerlo is boldfaced:
(32)

a. Pedrito pate la pelota con entusiasmo pero Jos lo hizo distradamente.

Pedrito kicked the ball with enthusiasm but Jos did so distractedly.
b. Pedrito pate la pelota con entusiasmo y Jos lo hizo tambin.
Pedrito kicked the ball with enthusiasm and Jos did so too.

In (32a), based on the understood antecedent of hacerlo, we can say that it

replaces V2 in the structure (31b) of the rst conjunct. Since the adjunct is
not included, we can say that it has been left behind rather than being
replaced. In (32b), lo hizo can be interpreted as replacing V1, so that the
adjunct is included in the material replaced by hacerlo.5
Summarizing to this point, we have seen that hacerlo can replace any V in
a structure, so that an adjunct phrase may either be replaced or left behind.
We will now see that hacerlo must replace a V, never just a verb: a complement cannot be left behind. This is shown by the ungrammaticality of (33):
(33)

*Pedrito pate la pelota y Jos lo hizo el jugete.

Pedrito kicked the ball and Jos did so the toy.

The intended reading of (33) is that hacerlo replaces only pate in the rst conjunct, leaving behind the complement. The ungrammaticality of (33) shows
that hacerlo replaces only V, not V.
Let us now apply the hacerlo test to examples with prepositional phrases, as
in (34):
(34)

Juan habl de poltica de nuevo.

Juan spoke of politics again.

If the PP de nuevo is an adjunct in (34), it should be able to be left behind under

hacerlo replacement. Likewise, if the PP de poltica is an adjunct, it should be
able to be left behind. If either of these constituents is a complement, however,
leaving it behind should produce ungrammaticality. The relevant examples are
shown in (35)-(36):
(35)

Juan habl de poltica de nuevo y Jos lo hizo por primera vez.

Juan spoke of politics again and Jos did so for the rst time.

(36)

*Juan habl de poltica y Jos lo hizo de economa.

Juan spoke of politics and Jos did so of economics.

The grammaticality of (35) shows that, since the PP de nuevo can be left
behind, this PP is an adjunct. The ungrammaticality of (36) shows that the
5

Sentence (32b) is actually ambiguous. Lo hizo can be interpreted as replacing either

V in the rst conjunct.

134

The syntax of Spanish

PP de poltica must be a complement of the verb, since it cannot be left

behind by hacerlo replacement. The hacerlo test therefore provides one type
of evidence for distinguishing prepositional complements from adjuncts. The
only limitation on the use of this diagnostic is that, because hacerlo itself is an
active predicate, it cannot be used if the predicate of the rst conjunct is
stative:
(37)

*Juan se parece a Jos, y Pedrito lo hace tambin.

Juan resembles Jos, and Pedrito does so too.

Consequently, the status of complements versus adjuncts of statives cannot

be diagnosed using the hacerlo test.
A second test that distinguishes prepositional complements from adjunct
PPs is the ability of the preposition to be replaced by other prepositions. An
adjunct headed by a preposition can typically be replaced by a dierent
adjunct, headed by a dierent preposition:
(38)

Recib un mensaje . . .
I received a message . . .
a. de Jos.
from Jos.
b. para Susana.
for Susana.
c. por telfono.
by phone.
d. en la biblioteca.
in the library.
e. durante la conferencia.
during the lecture.

By contrast, the preposition that heads a PP complement is typically xed,

and replacing the preposition produces ungrammaticality:
(39)

a. So con/*de Mara.
I dreamed about (lit.: with)/*of Maria.
b. Cont con/*en tu ayuda.
I counted on (lit.: with)/on your help.
c. Insisti en/*con tu ayuda.
S/he insisted on/*with your help.

The prepositions in (39) cannot be replaced freely by other prepositions, which

is one indicator that these prepositional phrases are complements. (The
hacerlo test conrms these examples as complements.)
Certain predicates do allow substitution of their complement preposition,
but only within a certain range. This is illustrated by the grammatical versus
ungrammatical alternatives in (40):

The Verb Phrase

(40)

135

a. Puse el bolgrafo en / sobre / detrs de / debajo de / encima del escritorio.

I put the pen in / on / behind / under / on top of the desk.
b. *Puse el bolgrafo para/con el escritorio.
I put the pen for/with the desk.

The substitutability of prepositions in (40a) appears to be an exception to the

generalization given above. However, notice that all of the grammatical prepositions in (40a) express location, while the ungrammatical preposition con in
(40b) does not. The range of substitutions that are grammatical can be
accounted for on the assumption that the Theta-role assigned by poner is a
Locative role, and that all of the prepositions in (40a) have a Locative feature.
Predicates that theta-mark a complement with a Locative role generally allow
this range of substitution.
We have seen above that complement PPs can be distinguished from adjunct
PPs in terms of their behavior under hacerlo replacement, and with respect to
the range of substitutions of the preposition. These dierences can be
accounted for in terms of the structural dierence between complements and
adjuncts: only the former are sisters of the head. This dierence in turn follows
from theta-theory. A complement PP is a sister of the head, because it must be
in this position in order to be theta-marked by the verb. An adjunct PP cannot
be a sister of the head, because if it were, it would be subject to interpretation
as a complement so the position of sister to the head is uniquely a complement position. An adjunct PP therefore must be higher in the structure, where
it is a sister of V, and dominated by V (shown, for example, in (31b) above).

3.5

Complements: direct object DPs

We will now consider properties of DP complements. We saw above

that two diagnostics distinguish PP complements from PP adjuncts. It is to be
expected that DP complements should also pattern dierently from adjuncts,
and we will see below that this is in fact the case. This is expected because DP
complements, like PP complements, are theta-marked by the verb, and therefore are sisters of the head. Adjuncts, however, are higher in the structure, generated as sisters of V. However, we will also see that, in some respects, DP
complements pattern dierently from PP complements.
One of the diagnostics introduced above, hacerlo replacement, can be
applied to DP complements. In fact, it was shown above in Section 3.4 that DP
complements dier from adjuncts under hacerlo replacement (cf. (32) versus
(33)). Several additional properties of direct object DPs distinguish them
from adjuncts. These properties also dierentiate DP complements from PP

136

The syntax of Spanish

complements. Before introducing these properties, let us briey consider why

DP and PP complements might be expected to behave dierently in some
respects. Theta-role assignment could not be relevant, since both DPs and PPs
are theta-marked by the verb. One way in which DP and PP complements
dier is that DP complements are assigned Case directly by the verb. This
Case is referred to as Objective (or Accusative) Case. A direct object DP then
is a complement that is related to the verb both by Theta-role assignment by
V and by Accusative Case assignment. A prepositional complement is linked
to the verb by Theta-role assignment only. Consequently, DP complements
should have certain properties attributable to the Case relation that are not
expected for PP complements or adjuncts. These properties, which will be discussed below, include: (a) clitic doubling in certain cases; (b) passivization;
and (c) co-occurrence with Personal a.
The rst property that is particular to DP complements is their co-occurrence
with clitics under certain conditions. DP complements dier from PP complements in this respect: PP complements do not co-occur with clitics, nor do
adjuncts introduced by prepositions.6 One context in which DP complements
co-occur with a clitic is if the DP is a reexive or reciprocal (i.e., an anaphor),
or a pronominal. The clitic agrees with (or doubles) the grammatical features
of the DP as shown in (41):
(41)

a. Susana *(se)
pate a s misma.
S.
CL-re. kicked PA herself
Susana kicked herself.
b. Susana *(lo)
pate a
l.
S.
CL-DO kicked PA him
Susana kicked him.

In (41a) the object of the verb patear is the reexive phrase a s misma; in (41b),
the object is the pronoun a l. Both sentences are ungrammatical without the
clitic. Adjuncts, on the other hand, do not co-occur with a clitic, even if they
contain an anaphor or pronominal. Two types of adjuncts which illustrate this
are emphatic reexives and benefactives introduced by para for. First, notice
that these two types of phrases are indeed adjuncts, as is shown by their behavior in the hacerlo construction:
(42)

EMPHATIC REFLEXIVE:
Susana resolvi el problema por s misma, y Pedro lo hizo con ayuda.
Susana solved the problem by herself, and Pedro did so with help.

The form of Spanish clitics was summarized briey in Chapter 1. We will not consider the derivation of clitics used in illustrations below, since this topic will be considered further in Chapter 4.

The Verb Phrase

(43)

137

BENEFACTIVE:7
Compr un coche para Juan, y Marta lo hizo para Pedro.
I bought a car for Juan, and Marta did so for Pedro.

The boldfaced constituents can be left behind under hacerlo replacement, indicating that they are not complements of the verb. Observe now that these constituents do not co-occur with a clitic:
(44)

a. Susana resolvi el problema por s misma.

Susana solved the problem by herself.
b. *Susana se resolvi el problema por s misma.
Susana CL(re.) solved the problem by herself.

(45)

a. Compr un coche para l.

I bought a car for him
b. *Le compr un coche para l.
CL I bought a car for him.

Nor do prepositional complements occur with a clitic double:

(46)

a. Juan (*se) habl consigo mismo.

Juan CL(re.) spoke with himself.
b. Mara (*le/lo) habl con l.
Maria CL(Acc./Dat.) spoke with him.

A second context in which DP complements co-occur with a clitic is the left

dislocation construction, in which a constituent interpreted as discourse topic
7

Benefactives can also be introduced by the preposition a, as in (i):

(i)

Le
compr
un coche a Juan.
CL(Dat.) bought-1st.sg. a car
for J.
I bought Juan a car.

In this construction, the benefactive behaves like a complement according to several

tests, including the hacerlo test:
(ii)

*Susana le compr un coche a Jos, y Pedro se lo hizo a ella.

Susana bought Jos a car, and Pedro did so for her.

Notice that the corresponding English benefactive can also behave like a complement, in that it undergoes Dative shift (I bought a car for her / I bought her a car). In
both languages, these quasi-complements can still be dierentiated from complements that are theta-marked by the verb under other diagnostics, such as passivization. In neither language can the benefactive become the subject of a passive, as
shown by (iii) and its gloss:
(iii)

*Jos fue comprado un coche.

Jos was bought a car.

For discussion of the dierent types of indirect objects, see Strozer (1976), Demonte
(1994a).

138

The syntax of Spanish

is in pre-clausal position (see Chapter 5 for discussion). As shown in (47), Left

dislocation of a DP complement is ungrammatical without a clitic double:
(47)

a. La respuesta, Susana *(la)

sabe.
the answer, S.
(CL(Acc.)) knows
The answer, Susana knows it.
b. El problema, Mara *(lo)
resolvi.
the problem, M.
(CL(Acc.)) solved
The problem, Maria solved it.
c. La pelota, Pedrito *(la)
pate.
the ball,
P.
(CL(Acc.)) kicked
The ball, Pedrito kicked it.
d. A Pedrito, su mam *(lo)
bes.
PA P.
his mother (CL(Acc.)) kissed
Pedrito, his mother kissed him.

Compare the above examples with left-dislocated adjuncts in (48ac) and the
prepositional complement in (48d):
(48)

a Para Juan, (*le)

compr
un coche.
for J.
(CL(Dat.)) bought-1st.sg. a car
For Juan, I bought a car.
b. El martes, (*le/lo)
compr
un coche.
the Tuesday (CL(Dat./Acc.)) bought-1st.sg. a car
Tuesday, I bought a car.
c. Con el vecino,
Mara (*le/lo)
cant.
with the neighbor, M.
(CL(Dat./Acc.)) sang
With the neighbor, Maria sang.
d. Con el vecino,
Mara (*le/lo)
habl.
with the neighbor, M.
(CL(Dat./Acc.)) spoke
With the neighbor, Maria spoke.

In these examples, Left Dislocation is possible, but becomes ungrammatical

with the clitic.
Summarizing to this point, clitic doubling of a DP complement is required
in certain contexts, but clitic-doubling of PP complements and adjuncts does
not occur. This dierence can be accounted for on the hypothesis that a clitic
is a reex of a Case relation between a verb and its complement. If the verb
does not assign Case to a constituent, then a clitic should never double that
constituent.
A second property of DP complements is that they can surface as the
subject of passive clauses:
(49)

a. Mara resolvi el problema.

Maria solved the problem.

The Verb Phrase

139

b. El problema fue resuelto (por Mara).

The problem was solved (by Maria).
(50)

a. Jos invit a Mara.

Jos invited Maria.
b. Mara fue invitada (por Jos).
Maria was invited by Jos.

Neither adjuncts nor prepositional complements are grammatical as the

subject of a passive:
(51)

a. Compr un coche para Juan.

I bought a car for Juan.
b. *Juan fue comprado un coche.
Juan was bought a car.

(52)

a. Hablamos con los vecinos.

(We) spoke to the neighbors.
b. *Los vecinos fueron hablados (con).
The neighbors were spoken (with).

It is standardly assumed that the movement of a DP object to subject position

in passive clauses is attributable to Case. The passive participle of a verb assigns
a Theta-role to its complement, but cannot assign Case to it, because Case is
absorbed, i.e., taken up by the participial ax. In a structure such as (53),
(53)

DP

compr+ado
buy
ax

el coche
the car

the Case feature of the verb is assigned to the ax (-ado), leaving the DP complement without Case. Because this DP must be assigned Case, it must move
to a position in which it can be assigned Case by another head. As discussed
in 3.3 above, one head which can assign Case is INFL, which assigns
Nominative (subject Case). The DP object undergoes movement to the
Specier of IP, where it is assigned Nominative (see (54)). Consider now the
fact that prepositional complements do not undergo passivization. A verb that
selects a prepositional complement does not have an Accusative Case feature
to assign to its sister, since it selects a PP, not a DP complement. Therefore in
a structure like (53), the passive participle would not have a Case feature to

140

The syntax of Spanish

assign to the participial ax. Therefore, Case is never absorbed by this ax,
and the participle would not be a well-formed constituent.8
A third property of DP complements is their co-occurrence with Personal
a, typically if the DP is interpreted as [+ ] and [+]:9
(54)

IP

DP

INFL

VP
fue resuelto

(55)

a. No veo *(a) Mara.

I do not see PA Maria.
b. No veo (*a) el problema.
I do not see PA the problem.

(56)

Buscamos (a) una secretaria.

We are looking for PA a secretary. (a specic person)
We are looking for any secretary.

In (55a), the animate, specic complement requires Personal a; in (55b), the

inanimate object cannot co-occur with Personal a. Example (56) is grammatical with or without Personal a, and its interpretation varies accordingly: with
Personal a, the complement is interpreted as [+]. As shown below,
Personal a co-occurs only with direct complements of verbs. It does not occur
in other contexts, such as within adjuncts:10
18

19

10

The availability of such passives in English, on the other hand, appears to be related
to the property of preposition-stranding in English, which has been suggested
(Stowell 1981) to indicate that English prepositions undergo reanalysis with their
governing verb. Reanalysis appears to be a device which rescues the absorption of
Case, transmitting it to the ax.
As pointed out in Chapter 1, Personal a can also occur with inanimate objects, if an
inanimate object is personied e.g., llamar a la muerte to call death or if both
subject and object are inanimate e.g., El invierno sigue al otoo Winter follows
autumn.
The occurrence of Personal a with DP complements raises several questions, including the nature of the morpheme, why it occurs only with complement DPs, and why
there are restrictions on its occurrence, such as animacy and specicity. See Lois
(1982) and Torrego (1998) for detailed discussion of Personal a.

The Verb Phrase

(57)

141

a. Marta resolvi el problema por s misma.

Marta solved the problem herself.
b. *Marta resolvi el problema por a s misma.
Marta solved the problem by PA herself.

Summarizing, we have seen that DP complements behave dierently from

adjuncts with respect to all of the diagnostics examined. Among these, certain
of these properties are common to DP and PP complements; others are particular to DPs; the latter have been suggested to be related to Case. The properties discussed above are summarized in (58)(59):
(58)

Complement (PP and DP) properties:

a. cannot be left behind under hacerlo replacement;
b. specic markers (non-replaceable prepositions, Personal a).

(59)

Complement DP properties:
a. Clitic doubling in certain contexts;
b. Co-occurrence with Personal a;
c. Become grammatical subject of passives.

3.6

Indirect objects

We turn now to indirect objects. We will see below that these complements also have properties that distinguish them from adjuncts, and properties that distinguish them from prepositional complements and direct
objects. In 3.6.1, similarities between direct and indirect objects will be presented; in 3.6.2, dierences between direct and indirect objects are summarized, and in 3.6.3, dierences between indirect objects and prepositional
complements.
Properties of indirect objects will be illustrated below using two types of
predicates: ditransitive verbs (verbs of transfer) such as those in (60), and
simple transitive verbs of contact, such as those in (61). (The indirect object
is shown in boldface):
(60)

a. Mara le
mand un paquete a Pedro.
M.
CL(Dat.) sent
a package to P.
Maria sent a package to Pedro.
b. Susana le
ensea matemticas a Jos.
S.
CL(Dat.) teaches math
to J.
Susana teaches math to Jos.

(61)

a. Juan le
golpe la nariz a Eduardo.
J.
CL(Dat.) hit
the nose to E.
Juan hit Eduardos nose.

142

The syntax of Spanish

b. Beatriz le
frot
los dedos a su hija.
B.
CL(Dat.) rubbed the ngers to her daughter
Beatriz rubbed her daughters ngers.

The examples in (60) and (61) all have three DPs: (1), a subject, the
Nominative Case-marked argument, which in these examples precedes the
verb; (2), a direct object that is assigned Objective (Accusative) Case by the
verb, as discussed above in 3.4; and (3) the boldfaced indirect object, which is
typically analyzed as receiving Dative Case. Dative Case is marked morphologically by (a) the morpheme a preceding the DP and, frequently, (b) a Dative
clitic (le in all of the above examples), which agrees with the indirect object in
person and may agree in number.11 The two verb classes illustrated above dier
from each other with respect to the Theta-role assignment, a point to which
we will return in 3.6.2.

3.6.1

Similarities between direct and indirect objects

Indirect objects behave like direct objects with respect to their behavior under hacerlo replacement and with respect to clitic doubling in the contexts discussed above. The examples in (62) show that an indirect object cannot
be left behind under hacerlo replacement:
(62)

11

a. *Mara le
ense historia a Pedro, y
Susana (se) lo
hizo a Jos.
M.
CL(Dat.) taught history to P.,
and S.
(CL(Dat.))
did so to J.
Maria taught history to Pedro, and Susana did so to Jos.
b. *Mara le
at las manos a Pedro, y
Susana (se) lo
M.
CL(Dat.) tied the hands to Pedro and Susana (CL(Dat.))
hizo a Jos.
did so to Jos.
Maria tied Pedros hands, and Susana did so to Jos.

Dative clitics in the 3rd person do not show number agreement if a 3rd person direct
object is also present:
(i)

(ii)

Susana les
mand un paquete.
S.
CL(Dat.3rd.pl.) sent
a package
Susana sent them a package.
Susana se
lo
mand (a ellos).
S.
CL(Dat.3rd.) CL(Acc.3rd.sg.) sent
(to them)
Susana sent it to them.

In (i), Dative les is plural in form (compare sg. le), but the form se which occurs with
sequences of IO(3rd. person) DO(3rd. person) does not have a plural *ses form.

The Verb Phrase

143

Recall from 4.3 that a complement cannot be left behind under hacerlo
replacement, because hacerlo replaces only V, not V, and complements are
daughters of the lowest V in the structure. The ungrammaticality of sentences
in (62) shows that these indirect objects pattern like other complements and
unlike adjuncts with respect to hacerlo replacement.
Indirect objects pattern with direct objects with respect to clitic doubling in
contexts discussed previously:
(63)

a. Susana *(se)
mand un paquete a s misma.
S.
(CL(re.)) sent
a package to herself
Susana sent a package to herself.
b. Jos *(le)
mand un paquete a ella.
J.
(CL(Dat.)) sent
a package to her
Jos sent her a package.

(64)

a. Mara *(se)
golpe
M.
(CL(re.)) hit
Maria hit her own nose.
b. Mara *(le)
golpe
M.
(CL(Dat.)) hit
Maria hit his nose.

(65)

la nariz a s misma.
the nose to herself
la nariz a l.
the nose PA him

a. A Pedro, Susana *(le)

mand un
To P.,
S.
*(CL(Dat.)) sent
a
Susana sent a package to Pedro.
b. A Pedro, Susana *(le)
golpe la
to P.,
S.
*(CL(Dat.)) hit
the
Pedro, Susana hit his nose.

paquete.
package.
nariz.
nose

Examples (63) and (64) show that reexive and pronominal indirect objects
require an appropriate form of the Dative clitic; omission of the clitic results
in ungrammaticality. Likewise, a left-dislocated indirect object requires a
clitic, as shown by the ungrammaticality of the sentences in (65) without the
Dative clitic.12
12

These facts are perhaps not surprising. Indirect objects typically co-occur with a
Dative clitic even when the indirect object is an ordinary DP, such as a proper name,
and it is often assumed (and taught) that the Dative clitic double is obligatory.
However, indirect objects subdivide with respect to the obligatoriness of the clitic, as
discussed in Strozer (1976). For one type of indirect object, the clitic is in fact optional:
(i)

Susana va
a mandar(le) un paquete a Jos.
S.
is going to send(CL) a package to J.

Demonte (1994, 1995) studies these alternations in detail. She shows that the
absence of the clitic has systematic grammatical and semantic properties, corresponding roughly to the alternation in English between directindirect object order
and the Dative-shift construction (e.g. Susan sent Jos a package).

144

The syntax of Spanish

Summarizing, we have seen above that indirect objects behave like direct
objects with respect to hacerlo replacement and clitic doubling. Taken by
themselves, these facts lead to the hypothesis that direct and indirect objects
should have the same analysis: theta-marked by the verb, and assigned Case
by the verb. On this analysis, the only dierence between direct and indirect
objects would be described in terms of the Case that the verb assigns: direct
objects are assigned Accusative, and indirect objects assigned Dative.
However, we will see below that this analysis is not fully adequate, because
there are respects in which indirect objects dier from direct objects, which this
analysis does not predict.

3.6.2

Dierences between direct and indirect objects

Indirect objects dier from direct objects in several ways. Let us begin
with two diagnostics discussed above in relation to the direct object: Personal
a and passivization.
The absence of Personal a with indirect objects is perhaps not surprising.
Its use with indirect objects would generate sequences like (66):
(66)

*Juan le
mand dinero a a su hermano.
J.
CL(Dat.) sent
money to PA his brother
Juan sent his brother money.

Second, indirect objects do not undergo passivization:

(67)

a. *Pedro (le) fue mandado un libro (por Juan).

Pedro was sent a book (by Juan).
b. *Pedro (le)
fue golpeado la nariz.
P.
CL(Dat.) was hit
the nose
Pedro was hit the nose.

The failure of indirect objects to undergo passivization is typical of Romance,

in contrast with English.13
A third property of indirect objects that distinguishes them from direct
objects is that various types of constituents can be marked with Dative
morphology and display the complement properties described above, even
though they are not theta-marked by the verb. One such case is benefactive
phrases, which can be expressed either by a PP introduced by para, as in (68a),
or as an indirect object, as in (68b):14
13

14

As noted in Chapter 1, there are individual exceptions, such as the verb preguntar
ask, which in some styles (particularly journalese) allows the indirect object to be
passivized: El presidente fue preguntado... The president was asked...
Several other instances of Dativization are discussed in Masullo (1992).

The Verb Phrase

(68)

145

a. Susana compr un coche para sus amigos.

Susana bought a car for her friends.
b. Susana les
compr un coche a sus amigos.
S.
CL(Dat.) bought a car
to her friends
Susana bought a car for her friends.

The PP in (68a) behaves like an adjunct, but the Dative form in (68b) behaves
like a complement, as shown by their behavior under hacerlo replacement:
(69)

a. Susana compr un coche para sus amigos, y Pedro lo hizo para

sus hermanos.
S.
bought a car
for her friends and P.
did so for
his brothers
Susana bought a car for her friends and Pedro did so for his brothers.
b. *Susana les
compr un coche a sus amigos, y Pedro
se lo hizo a sus hermanos.
S.
CL(Dat.) bought a car
to her friends and P.(CL(Dat.))
did so
for his brothers
Susana bought a car for her friends and Pedro did so for his brothers.

In (69a), the adjunct phrase introduced by para can be left behind under
hacerlo replacement. The corresponding phrase in (69b), introduced by a and
doubled by a clitic, behaves like a complement: it cannot remain behind under
hacerlo replacement.
The adjunct behavior of benefactives indicates that they are not assigned a
Theta-role by the verb. Nevertheless, a benefactive DP can be Dativized,
which means that in the course of the derivation, it can be licensed in a complement position. This accounts for the ability of these phrases to occur with
a Dative clitic double, as in (68b), and for their complement behavior with
respect to hacerlo replacement, as in (69b). Adjuncts cannot, however, surface
as direct objects. Alternating with Mara cant para sus amigos Mara sang
for her friends, there is no *Mara los cant meaning Mara sang for them.15
Another type of Dativized indirect object that is not theta-marked by the
verb is the indirect object of verbs of contact, which have been illustrated
throughout the preceding discussion. These indirect objects alternate with
direct objects, as shown by the phrase sus amigos his friends in (70):

15

In constructions with semi-auxiliaries like causative verbs, Accusative Case (like

Dative Case) can be assigned to a DP that is theta-marked by a following innitive.
For example:
(i)

Juan hizo escribir a Pedro.

J.
made write
PA P.
Juan made Pedro write.

146

The syntax of Spanish

(70)

a. Jos les
golpe las rodillas a sus amigos.
J.
CL(Dat.) hit
the knees to his friends
Jos hit his friends knees.
b. Jos golpe a sus amigos en las rodillas.
J.
hit
PA his friends on the knees
Jos hit his friends on the knees.

In (70a), the phrase sus amigos is an indirect object, in (70b), a direct object.
The Dativization of this phrase is perhaps described most easily in relation
to the analysis of (70b). In this example, the verb hit is a simple transitive verb,
which assigns the role Theme to its object, sus amigos, and assigns
Accusative Case to that phrase. The phrase en las rodillas is an adjunct of location in (70b), as shown by its behavior under hacerlo replacement:
(71)

Juan golpe a Eduardo en la nariz, y Pedro lo hizo en la oreja.

Juan hit Eduardo on the nose, and Pedro did so on the ear.

In (71), the PP can be left behind by hacerlo replacement, indicating that it is

a sister of V, not of V in other words, it is higher in the structure than complements.
Now consider (70a), with the same verb, but which has both a direct object
and an indirect object. This verb is not thematically ditransitive. It still assigns
a single Theta-role Theme to its direct object. Here, however, the direct object
is a DP that expresses a part of an individual, while the indirect object
expresses the whole individual. This particular kind of part/whole relation
is known as a relation of inalienable possession. The indirect object is the
inalienable possessor of the direct object body part. Because the two DPs are
not separate entities, they could not have two separate thematic roles assigned
by the verb. (If they did, they would be expected to refer to separate individuals, rather than a part/whole relation.) They must therefore comprise a single
argument of the verb, which expresses the part/whole relation via a complement-internal theta-marking relation. As shown in (72), a role of Possessor
is assigned by the body-part noun.
(72)

DP

la

DP

nariz

a Eduardo

(assignment of Possessor role by N)

The Verb Phrase

147

The DP in (72) is in turn assigned a Theta-role by the verb (see (73)).

V

(73)

DP

golpe

la nariz
a Eduardo

(assignment of Theme role by V)

We have seen so far that the inalienable possessor is not theta-marked by the
verb, but by the body-part noun, within the complement DP. This possessor
is Dativized, in the sense that, in the course of the derivation, it comes to be
grammatically related to the verb. It bears Dative morphology, which nouns
cannot assign, and behaves like a complement of the verb in other relevant
respects, as examples throughout this section have shown. The Dativized
object is syntactically identical to the theta-marked object of ditransitive
verbs.
Summarizing, we have seen above that indirect objects have properties that
are not shared by direct objects. Indirect objects do not co-occur with Personal
a, do not undergo passivization, and the Case morphology of the indirect
object may be assigned to dierent constituents. The Dative constituent may
be: (a) a theta-marked complement of the verb, as in the case of verbs of transfer; (b) an adjunct, particularly a benefactive; or (c) an argument of the direct
object DP (an inalienable possessor).

3.6.3

Theta-marking and Case assignment to indirect objects

In this section, we will address some issues related to the analysis of

indirect objects, including: (a) theta-marking of the indirect object of ditransitives, and (b) the description of Dative a. In previous discussion, it has been
shown that complements are generated as sisters of the head of phrase. This
assumption has a role in accounting for the behavior of complements under
hacerlo replacement, and in accounting for the conditions under which Thetarole assignment occurs. Theta-marking of complements was described above
as a relation between a head and its sister constituent. If a verb theta-marks
two complements, direct and indirect objects, as is the case for ditransitive
verbs, the simplest structure consistent with the foregoing assumptions is that
the head and the two complements are all sisters, as shown in (74). The
problem with this structure is that it violates a widely accepted restriction
against ternary branching. The V in (74) is ternary (with three branches

148

The syntax of Spanish

descending from V).16 If ternary branching is excluded, there must be an

additional node dominating one complement or the other. One alternative is
(75). This structure is consistent with binary branching, but it presents other
problems. It is exceptional with respect to theta-marking, since the indirect
object is not a sister of the head, and therefore cannot be assigned a Thetarole directly by the head. Furthermore, it predicts that indirect objects should
behave like adjuncts, rather than complements. For example, this structure
leads to the expectation that the indirect object should be able to be left behind
under hacerlo replacement, because there is a V lower in the structure that
hacerlo could replace. Structure (75) is thus inadequate with respect to
accounting for the properties of indirect objects.
(74)

DP
V
(direct object) (indirect object)
(75)

DP

DP (indirect)

DP (direct)

The paradox of indirect object structure has been given another solution
in work by Richard Larson (1988). Under Larsons proposal, a ditransitive
verb actually has two parts, or phrasal heads, each of which is in an appropriate structural relation to one of the complements. Adapting the analysis
somewhat, we may take the D-structure to be as shown in (76). Here, the verb
dar is a sister of the indirect object, which is assigned a Theta-role by the verb.
There is a separate head of phrase in the structure, which is empty at Dstructure. It is lled in the course of the derivation via movement of the verb,
as in (77). In this higher position, the verb can assign a Theta-role to the direct
object. Although the verb is not a sister of the direct object, it c-commands it
(cf. 3.2), since the rst branching node that dominates the verb also dominates

16

Larson (1988) and Demonte (1995) present arguments (beyond the scope of the
present discussion) proposing that the direct and indirect objects are not symmetrical (mutually c-commanding), as they would be if the structure in (74) were correct.

The Verb Phrase

149

the direct object. (Notice that the indirect object is shown in (77) as a DP. An
alternative, to be discussed below, is that it is a PP headed by Dative a.)
(76)

DP

DP (direct)

dar
give
(77)

V
dar

DP
(direct)

DP
(indirect)

Because the conditions for theta-marking of both complements are satised

in (77), these complements should behave dierently from adjuncts, since
adjuncts are not c-commanded by the verb. Both complements are predicted
to be replaced by hacerlo, and not left behind: hacerlo cannot replace any of
the lower V nodes in (78), because none of these nodes dominates the (entire)
verb. Since the verb necessarily undergoes movement to the higher V position
(to theta-mark its direct object), hacerlo can replace only this higher constituent. Therefore, hacerlo cannot leave behind either the direct object or the indirect.
Let us now turn to the issue of Case assignment to the indirect object, and
in particular to the status of Dative a. The question that arises with respect to
this morpheme is whether it is a true preposition or whether, like Personal a,
it is a reex of Case assignment by the verb. Neither of these analyses accounts

150

The syntax of Spanish

straightforwardly for the properties of indirect objects discussed above.17

Consider rst the hypothesis that Dative a is a preposition, as shown in the
partial structure (78b), corresponding to the VP for (78a).
(78)

a. Juan le
regal un libro a Jos.
Juan CL(Dat.) gave a book to Jos
b.
VP

DP
Juan

regal

DP

un libro V

PP

DP

Jos

Here, the indirect object is a PP, headed by Dative a, which assigns Case to its
DP complement. This analysis is compatible with several of the properties of
indirect objects discussed above. If correct, it accounts for ways in which indirect objects pattern with PP complements rather than prepositions with
respect to the absence of Personal a and impossibility of passivization. It also
correctly predicts that Dative a does not behave like Personal a. A second
dierence is shown by the contrast between (79) and (80), where the adjective
phrase is predicated of the object:
(79)

17

Vi a [mi amiga enojada].

saw PA my friend angry
I saw my friend angry.

The peculiarities of Dative a have been the topic of numerous studies, and there is
little consensus as to whether Dative a is a preposition or not. For detailed discussion of Datives see Strozer (1976) and Masullo (1992). An alternative analysis of
theta-marking of the indirect object is presented in Zubizarreta (1987). She proposes
that the Theta-role of indirect object is assigned by Dative a and by the verb. On this
analysis, the verb and preposition are a complex predicate in the lexicon.

The Verb Phrase

(80)

151

*Mand una carta a [mi amiga enojada].

sent
a
letter to my friend angry
I sent a letter to my friend angry.

The direct object in (79) accepts a secondary predicate as a modier, the indirect object in (80) does not. Demonte (1986) shows that indirect objects
pattern with PPs with respect to secondary predicates, a fact which she argues
follows from their prepositional phrase structure, as distinct from that of
direct objects preceded by Personal a.
There are then several types of evidence that support an analysis of Dative
a as a preposition. However, this analysis requires some auxiliary hypothesis
to account for the dierences between indirect objects and other PP complements, including the occurrence of clitic doubling and Dativization phenomena. These phenomena suggest that the PP headed by Dative a comes
into construction with the verb in terms of a grammatical relation that is
independent both of Theta-role assignment and Case assignment. Note that
the alternative, according to which Dative a is not a Preposition, would require
auxiliary hypotheses to account for dierences between Dative a and Personal
a, as well as other dierences between direct and indirect objects.

3.6.4

Summary

In this section, similarities and dierences between indirect objects

and other complements were presented. Let us summarize the main generalizations outlined in the discussion. First, we saw in 3.6.1 that indirect objects
pattern generally with complements, not with adjuncts, with respect to the
basic structural diagnostic of hacerlo replacement. Indirect objects are not,
however, identical in their behavior to either prepositional or direct objects.
Unlike PP complements, they require clitic doubling under the same conditions as do direct objects. However, in other respects indirect objects dier
from direct objects: they disallow passivization, and Personal a diers from
Dative a in crucial respects, as discussed in 3.6.3. Indirect objects present interesting issues with respect to both theta-marking and Case. Constituents that
are not theta-marked (such as benefactives) can be Dativized, and even the
conditions under which ordinary indirect objects of ditransitive verbs are
theta-marked require a more abstract analysis of the Verb Phrase than had
been previously assumed. Finally, with respect to Case-marking, it was shown
that Dative a is best analyzed as a preposition which assigns Case to the Dative
DP. This analysis leaves as an open question why clitic doubling occurs with
these objects, and why Dativization is as robust as it is, compared with other
complements.

152

The syntax of Spanish

3.7

Complements of unaccusative verbs

Section 3.3 above discussed the external argument of VP: the argument that normally corresponds to the subject of the clause. According to the
analysis given there, an external argument is generated in the Specier of VP,
as a sister of V (which assigns a compositional Theta-role to the DP). The Dstructure position of the external argument is shown by the DP position in
(81). From this position, the external argument moves to the Specier of IP,
where it is assigned Nominative Case by INFL. Note that the external argument in (81) is structurally distinguished from complements, which are sisters
of the verb (or at least c-commanded by it, as discussed in 3.6.3). This structural dierence between external and internal arguments is supported by diagnostics that are sensitive to the structural relation between the argument and
the head. In (82) for example, we see that the external argument can, naturally,
remain behind under hacerlo replacement:
(81)

VP

DP

V
V

(82)

Mara ley el diario, y Pedro lo hizo tambin.

Maria read the paper, and Pedro did so too.

Unlike complements, the external argument disallows bare NPs:18

(83)

*Nios leyeron estos libros.

Children read these books.

There are, however, certain verb classes whose grammatical subjects do not
behave in expected ways relative to diagnostics which distinguish external
from internal arguments. One of these classes is presentational verbs,
described as such because they introduce the (existence or presence of) DP
into the discourse. This includes verbs like llegar arrive, venir come, aparecer appear, salir come/go out, and negatives of them, such as faltar
lack, desaparecer disappear.19 As shown in (84) and (85), these verbs allow
bare NP subjects, but only following the verb:
18
19

An exception is coordinate constructions, as noted in Chapter 1.

The verb classes whose behavior is discussed here are not particular to Spanish.
Cross-linguistically, roughly the same predicates display properties that dierentiate
their subjects from ordinary subjects. For discussion, see Levin (1993), Levin and
Rappaport-Hovav (1994); for Romance, see Tortora (1997) and references cited there.

The Verb Phrase

(84)

a. Llegaron estudiantes.
arrived students.
Students arrived.
b. *Estudiantes llegaron.
Students arrived.

(85)

a. Faltan tomates.
lack
tomatoes
Tomatoes are lacking.
b. *Tomates faltan.
Tomatoes are lacking.

153

Intransitive verbs that are not members of this class do not allow bare NP subjects:20
(86)

a. *Cen gente (a las ocho).

dined people (at eight oclock)
People dined (at eight oclock).
b. *Gente cen (a las ocho). (=86a)

(87)

a. *Tosen nios.
cough children
Children are coughing.
b. *Nios tosen. (=87a)

In these respects, the subject of these predicates pattern with complements.

Yet, in other respects they are supercially subjects, not objects. Like other
subjects, their pronoun form is Nominative, and they agree with the verb in
person and number. Unlike objects, they do not co-occur with Personal a, and
they do not require clitic doubling. In Romance languages other than Spanish,
additional diagnostics dierentiate these arguments from ordinary external
arguments. For example, in French and Italian, these predicates take a dierent auxiliary, one corresponding to be rather than have in the compound
past tense:21
20

It may be possible to override the prohibition on bare NPs as subjects of this verb
class by placing the sentence within a discourse frame such as the one shown in (i):
(i)

21

A. Qu pas?
What happened?
B. Primero, abrimos el restaurante, despus, cen gente, y nalmente
cerramos como a las doce.
First, we opened the restaurant, then people dined, then nally we
closed around midnight.

Corresponding arguments in Italian and French pattern like complements with

respect to the distribution of certain clitics. For detailed discussion, see Burzio
(1986).

154

The syntax of Spanish

(88)

a. Molti studenti hanno telefonato.

Many students have telephoned.
b. Molti studenti sono arrivati.
Many students have (lit.: were) arrived.

(89)

a. Jean a mang.
Jean has eaten.
b. Jean est arriv.
Jean has (lit: was) arrived.

English also has diagnostics that distinguish presentational subjects for

example, they can occur in the there construction: There arrived many students;
*There ate many students.
Building on work by Perlmutter (1978), these predicates have been analyzed
as assigning a Theta-role only to an internal argument, as shown in (90b), the
D-structure corresponding to (90a).
(90)

a. Llegaron los estudiantes.

b.

VP

DP

llegaron

los estudiantes

Unlike other verbs that select an internal argument however, these verbs do
not assign Accusative Case to their complement hence, their description:
unaccusative verbs. The argument must have Case, however, and in order to
satisfy the Case requirement, the DP moves to the specier of IP, where it is
assigned Nominative (see (91)).
When this movement takes place, the DP has the surface properties normally associated with subjects: Nominative Case, subjectverb agreement,
and other properties associated with the Specier of IP position. At the same
time, the subject is still linked to the position in which its Theta-role is
assigned: the complement of V. This is the position that is responsible for the
object-like properties of the DP.
Recall from above that bare NPs are grammatical only if the subject follows
the predicate: Llegaron estudiantes; *Estudiantes llegaron Students arrived.
This contrast suggests that the bare NP is still in the complement position. The
question arises, in these derivations, as to how the NP gets Case, if it does not

The Verb Phrase

(91)

155

VP

DP
los INFL
estudiantes

VP

DP

llegaron

move to the Specier of IP. One solution to this problem proposed in Burzio
(1986) is that Nominative Case is transmitted from subject position to the
NP in object position. Case transmission is made possible due to the presence
of a null pronoun in subject position:
(92)

[IP proi INFL [VP [V llegaron estudiantesi ]]]

There arrived students.

An alternative analysis, according to which unaccusative verbs assign a Case

distinct from Accusative, is proposed in Belletti (1988). On this analysis, the
verb llegaron in (92) assigns Partitive case to its complement.
Unfortunately, Spanish does not provide clear diagnostics that support analyzing the post-verbal subject as remaining in object position, as in (92).
Hacerlo replacement, at rst glance, seems suggestive:
(93)

a. Los profesores llegaron y los estudiantes lo hicieron tambin.

The teachers arrived and the students did so too.
b. *Llegaron profesores y lo hicieron estudiantes tambin.

The contrast between (93a) and (93b) seems to show that once the DP has
moved to the Specier of IP, as in (93a) with a pre-verbal subject, the subject
can be left behind under hacerlo replacement. In (93b), the post-verbal subject,
lled by a bare NP, cannot be left behind, apparently indicating that the DP
is still in complement position. However, the results of this test are clouded by
the fact that (93b) may be ungrammatical for another reason: the proform
hacerlo imposes certain semantic requirements on its subjects essentially, an
Agent reading. The unaccusative verb llegar does not assign an Agent role to
its internal argument, but rather a Theme role. Therefore, the ungrammaticality of (93b) may be due to the mismatch between the type of argument selected
by llegar (Theme) and the type of argument selected by hacerlo (Agent). Why
then is (93a) grammatical? One possibility is that verbs like llegar have two

156

The syntax of Spanish

lexical entries: an unaccusative entry, and a second, Agentive entry, which

selects an Agent as an ordinary external argument. Since the role of Agent can
only be an external role, assigned compositionally by VP, then (93a) is possible. Both conjuncts have an Agentive verb. In (93b) however, hacerlo would
be analyzed as Agentive, but the internal argument of llegaron in the rst conjunct could not be an Agent, since it is an internal argument, as shown by its
bare determiner.
Summarizing, we have seen in this section that certain intransitive verbs
have surface subjects that behave according to some diagnostics (which vary
cross-linguistically) like complements i.e., like sisters of the head. The exceptional behavior of these predicates has been accounted for by analyzing the
verb as unaccusative: a verb which selects an internal argument, but which
does not assign Case to that argument. The argument undergoes movement to
the Specier of IP, where its Case requirement is satised by assignment of
Nominative by INFL. The D-structure position of the subject is the position
that is responsible for complement-like behavior of the subject.

3.8

Summary

In this chapter, we have considered certain aspects of the syntax of the Verb
Phrase. We began with a review of the distribution of VP, which occurs typically as a sister of INFL, as the primary predicate of a clause. Unlike AP, VP
does not occur as a secondary predicate or adjunct, unless the adjunct is itself
a clause. The licensing of VP that accounts for its distribution was related to
Predication, which is obligatory for clauses, and the TenseV relation, whose
nature is less well studied. Subsequent sections of the chapter examined relations between the verb and its arguments. Dierences between arguments of
the verb and adjuncts (or modiers) were discussed, as were dierences among
various arguments. Section 3.3 discussed the derivation of external arguments,
which are theta-marked in the Specier of VP, and move to the Specier of IP
to be assigned Case. Predicates that do not theta-mark an external argument,
such as parecer seem, can nevertheless have a referential subject as a result
of movement (Raising) of a subject from a subordinate clause. The topic of
subject position and licensing is of course more complex than was suggested
here. We return In Chapters 4 and 5 to this topic. Sections 3.43.6 discussed
properties of PP complements, direct objects and indirect objects, each of
which behaves dierently from adjuncts according to some diagnostics, and
which also behave dierently from each other, particularly with respect to their
Case-relation to the verb. Finally, in Section 3.7, a separate class of predicates

The Verb Phrase

157

known as unaccusatives was introduced. The surface subject of these VPs

does not behave like an ordinary external argument, but shares certain properties with complements. It was shown how these properties could be
accounted for on the hypothesis that the surface subject is generated in complement position, but is not assigned Accusative by the unaccusative verb.
The DP therefore moves to the Specier of IP, and is assigned Nominative
Case by INFL, or receives Case in object position either by Case transmission, or by assignment of Partitive Case by the verb.

VP-related functional categories

4.1

Introduction

In Chapter 3, our discussion focused on two aspects of VP syntax: the

distribution of VP, and phrase-internal structural relations between the verb
and its arguments. In this chapter, we will examine several grammatical processes which participate in licensing VP constituents at S-structure. To introduce the issues that will be taken up in this chapter, let us begin by reviewing
the D-structure discussed in Chapter 3 for a transitive predicate such as (1).
(1)

a. Juan ley el diario.

J.
read the newspaper
b.
IP

INFL

VP

DP

Juan

DP

ley

el diario

Given the similarity between the D-structure order of constituents in (1b) and
the surface form of the sentence (1a), it might appear that little needs to be
said about the derivation once the D-structure is formed. In fact, one might
question whether IP is necessary at all: conceivably, a clause could be analyzed
as consisting of nothing more than the VP itself. However, there are reasons
158

VP-related functional categories

159

why the IP must be present. The standard approach to the VPIP relation that
has been adopted in recent generative literature rests on the distinction
between lexical and functional categories discussed in Chapter 2. IP is
assumed to be a functional category which has a role in licensing constituents
of VP. IP is, in this sense, a functional extension of VP (Grimshaw 1991),
much as DP is a functional extension of NP. One of the functional relations
between INFL and VP has already been introduced in Chapter 3 (Section 3.3):
INFL is the head which assigns (Nominative) Case to the clausal subject.
Movement of DP to the Specier of IP is necessary to satisfy this abstract
Case requirement, so the derivation of (1b) includes DP movement:
(2)

IP

Juan i

INFL

VP

DP

ti

DP

ley

el diario

In this chapter, we will discuss three other movements which aect the licensing of VP constituents. In Sections 4.2 and 4.3, the relation between Tense features of INFL and V will be examined. In 4.2, we will see that the distribution
of VP-adverbs provides evidence that V has raised to INFL at S-structure.
Then in 4.3, several further aspects of V-to-INFLNFL movement are
addressed. We will discuss the temporal licensing of the VP by Tense features,
how this licensing operates, and what implications it has for the analysis of
clauses with auxiliary verbs. We will suggest that there is a second functional
category in the VP extended projection which participates in licensing of
verbs: Aspect. Sections 4.4 and 4.5 will then introduce two additional constituents: clitics (4.4) and negation (4.5), which have also been analyzed in recent
work as functional heads. Our discussion in this chapter leaves aside the licensing of one constituent: the subject. For the purposes of this chapter, we will
assume that the subject occupies the Specier of IP, as discussed in Chapter 3

160

The syntax of Spanish

(Section 3.3). We return in Chapter 5 to a fuller discussion of the distribution

of the subject.

4.2

VP-adverbs and the verb/tense relation

In this section, we will focus on the distribution of VP-adverbs. As the

discussion will show, the distribution of VP-adverbs has been argued to
provide evidence which bears on the relation between V and INFL. To narrow
our discussion, it will be useful to distinguish between sentence adverbs and VPadverbs. Sentence adverbs modify (or have scope over) the entire proposition
expressed by a clause. They are standardly analyzed as dependents of IP (the S
node in earlier terminology). We will refer to them as IP-adverbs below. There
are several semantic sub-types of IP-adverbs, a matter which we leave aside
here. IP-adverbs include PPs such as por supuesto of course, sin embargo
nevertheless, non-derived adverbs such as quizs perhaps, and a number of
-mente -ly adverbs such as naturalmente naturally, realmente really, obviamente obviously. These adverbs typically correspond to adjectives which
select clausal subjects. For example, Obviamente le el diario Obviously I read
the newspaper corresponds to Es obvio que le el diario It is obvious that I
read the newspaper.1 IP-adverbs can occur in the positions shown in (3):
(3)

a. Probablemente Mara ley ese libro.

Probably Maria read that book.
b. Mara probablemente ley ese libro.
Maria probably read that book.
c. (?)Maria haba probablemente ledo ese libro.2
Maria had probably read that book.

Some adverbs can modify more than one type of constituent. For example, naturalmente naturally is ambiguous between sentential modication, with the reading it
is natural that x, and VP modication, with a manner interpretation (e.g., Estos
tomates maduraron naturalmente These tomatoes ripened naturally).
The order: auxiliaryadverbparticiple may vary in acceptability according to the
particular auxiliary used. Separation of haber from a following participle is generally worse than separation of estar or passive ser from a following participle:
(i)
(ii)
(iii)

Mara estaba probablemente leyendo ese libro.

M.
was
probably
reading that book
El libro haba sido probablemente prohibido por la censura.
the book had been probably
banned
by the censors
*El libro haba probablemente sido prohibido.
the book had probably
been banned

Given the acceptability of (i) and (ii), one might conclude that IP-adverbs can intervene between constituents of the VP, and perhaps the status of (iii) reects an

VP-related functional categories

161

d. (?)Mara haba ledo probablemente ese libro.

Maria had read probably that book.
e. (?)Mara haba ledo ese libro probablemente.
Maria had read that book probably.

As shown in (3ab), IP-adverbs occur grammatically before or after the

subject, and only marginally following the auxiliary (3c), the participle (3d)
and the entire VP (3e). For (3d,e), it is not clear whether IP-adverbs are fully
grammatical with normal intonation contours, although pause intonation is
not obligatory for adverbs in these positions. Example (3e) is grammatical
with a pause between the object and the adverb, in which case it is presumably
in a clause-peripheral (dislocated) position. We will focus here on the contrast between (3a,b) and (3c,d). Let us suppose that adverbs can adjoin to X
projections (as in (4)). Here, the adverb is adjoined either to IP or to I. If these
are the only positions in which IP-adverbs can be generated, (3a) and (3b) are
accounted for, as are the ungrammaticality of (3d) and (3e). However, the
order in (3c) remains as a question: there, the adverb intervenes between the
auxiliary and the following participle. We will return to this derivation in (4.3),
where we address the structure associated with auxiliary verbs.
(4)

IP

IP

(Adv)
probablemente

DP
Mara
(Adv)

probablemente
I

VP

VP-adverbs are adverbs that modify the event (or state) expressed by VP, or
some constituent of it. These include the adverb types illustrated in (5):3

idiosyncrasy of the perfective construction, as suggested in Zagona (1988). However,

if auxiliaries ser and estar are analyzed as raising verbs, their complement IPs would
be expected to select IP-adverbs.
Just as some IP-adverbs can also modify other constituents (see note 1), some VPadverbs listed in (5) can modify constituents other than VP. For example, the Extent
adverbs can modify quantiers and adjectives: Leyeron [apenas tres pginas] They
read barely three pages; Es [casi ciego] He is almost blind.

162

The syntax of Spanish

(5)

VP-Adverbs:4
a. Time: ayer yesterday, hoy today, ahora now, maana tomorrow,
anteayer the day before yesterday, frecuentemente often, antes before,
an still/yet, todava still, ya already,, etc.
b. Place: aqu here, all there, lejos far, cerca near, abajo below,
afuera outside, etc.
c. Extent/degree: casi almost, apenas barely, meramente merely, slo
only, etc.
d. Manner: bien well, mal badly, rpido quickly, quedo quietly,
fcilmente easily, etc.
e. Quantity: mucho a lot, poco little, demasiado too much, menos
less, etc.

The classes shown in (5) reect various types of modication of VP, and, as
we will see below, these types behave roughly as classes with respect to their
distribution. However, it should be noted that members of each class may
dier from one another with respect to other features. For example, aqu
here and afuera outside dier with respect to deixis; afuera outside and
lejos far away dier with respect to specicity. These features can aect the
distribution of particular items (cf. note 4). We will focus our discussion on
the general patterns of distribution of the classes in (5). In 4.2.1, the distribution of these adverbs relative to the subject, verb and complements is summarized. Sections 4.2.2 and 4.2.3 discuss the derivation of the occurring orders,
particularly in relation to the position of the verb at S-structure.

4.2.1

Distribution of VP-adverbs relative to S-V-O

VP-adverbs can occur in any of the positions marked by x in (6):5

Each of these types can be sub-classied. For example, Place adverbs may be [specic] (e.g., aqu here is [+ ], lejos far is [- ]). Such features aect
the behavior of particular items within a class. For example, only [+ ] Place
adverbs can be clause-initial:
(i)

a. Juan conoci a su amigo lejos/all.

Juan met his friend far away / there.
b. All Juan conoci a su amigo.
There Juan met his friend.
c. ?*Lejos Juan conoci a su amigo.
Far away Juan met his friend.

In addition to the orders given in the text, the constituent order pattern (i) is possible,
as illustrated in (ii):
(i)
(ii)

Adv V- Subject Object

a. Siempre dice Juan la misma cosa.
always says J.
the same thing

VP-related functional categories

(6)

163

x Subj x V x Obj -x

However, not all sub-classes can occupy all positions. Let us rst consider the
post-verbal positions, before and after a complement. As shown in (7), all
adverbs except Extent adverbs can occupy one or both post-verbal positions:
(7)

Post-verbal adverbs
a. Los trabajadores recibieron ayer/ya
el
sueldo.
(Time)
The workers
received yesterday/already their salary.
a Los trabajadores recibieron el sueldo ayer/ya.
b. ?Juan conoci all a su mejor amigo.6
(Place)
J.
met
there his best friend.
b Juan conoci a su mejor amigo all.
c. *Los estudiantes terminaron apenas el examen.
(Extent)
*The students
nished
barely the exam.
c.*Los estudiantes terminaron el examen apenas.7
d. Mara ley cuidadosamente el diario.
(Manner)
M.
read carefully
the newspaper.
d. Mara ley el diario cuidadosamente.
e. Susana ama mucho a su hija.
(Quantity)
S.
loves much
her daughter.
e.Susana ama a su hija mucho.

The range of adverbs which can appear in pre-verbal positions is more

restricted, as shown in (8) and (9):
(8)

Subject Adverb V Object

a. Los trabajadores ya/ayer recibieron el sueldo.
The workers already/yesterday received their salary.
b. Juan all conoci a su mejor amigo.
Juan there met his best friend.
c. Los estudiantes apenas terminaron el examen.
The students barely nished the exam.

(Time)
(Place)
(Extent)

b. Poco hablan ellos de ese tpico.

little speak they of that topic
c. All cuenta la gente historias interesantes.
there tell
the people stories interesting
There the people tell interesting stories.

This order is taken up in Chapter 5, where variation in the position of the subject is
described.
The occurrence of all between the verb and its complement may be possible only if
the complement is dislocated that is, with an intonation shift indicating that it has
been moved from complement position.
The example is ungrammatical on an Extent reading of the adverb. Apenas can have
a manner interpretation with great diculty. On this reading, apenas can follow
the complement.

164

The syntax of Spanish

d. *?Mara cuidadosamente ley el diario.
Maria carefully read the paper.
e. *Susana mucho/demasiado ama a su hija.
Susana a lot/too much loves her daughter.

(9)

Adverb Subject V Object

a. Ya/ayer los trabajadores recibieron el sueldo.
Already/yesterday the workers received their salary.
b. All Juan conoci a su mejor amigo.
There Juan met his best friend.
c. *Apenas los estudiantes terminaron el examen.
Barely the students nished the exam.
d. *?Cuidadosamente Mara ley el diario.
Maria carefully read the paper.
e. *Mucho/demasiado Susana ama a su hija.
A lot/too much Susana loves her daughter.

(Manner)
(Quantity)

(Time)
(Place)
(Extent)
(Manner)
(Quantity)

As shown in (8) and (9), adverbs of Manner and Quantity cannot immediately
precede the nite verb, and neither these nor adverbs of Extent can appear in
the pre-subject position.

4.2.2

Post-verbal adverbs

In this section, we will focus primarily on VP-adverbs which appear

between the verb and its complement. VP-nal adverbs represent the expected
order, the position in which adjuncts of all types appear.8 The problems raised
by: [V adverb object] order are the same as those discussed in Chapter 2
(Section 2.8) with respect to the order: [N adjective complement]. The
central problem is that Theta-theory predicts that complements should be
closer to the head than non-complements, since the head must c-command a
constituent in order to theta-mark it. That prediction is borne out by the
surface order in English: the equivalent of (7d) for example, *Maria read carefully the newspaper, is ungrammatical. However, in Spanish like Romance
generally VP-adverbs can intervene between the verb and its complement.
There are two general ways of accounting for the derivation of [V adv
object] order: (a) base-generation of the surface order, or (b) by movement of
one or more constituents. Let us rst consider how the derivation would
proceed under each of these approaches. The base-generation approach
assumes that the constituents in (9) are in the same conguration at D-structure
and at S-structure. On this approach, the constituent: [ley frecuentemente el
8

Some verbs take adverbs as complements, such as sentirse bien/mal to feel

well/bad(ly), vestirse bien to dress well, etc. These adverbs are sisters of the verb,
rather than adjuncts.

VP-related functional categories

165

diario] would be generated with the adverb as a sister of the verb, as in (10a) or
(10b).
(10)

a.

Adv

b.

DP

ley frecuentemente el diario

V
ley

DP

Adv el diario
frecuentemente

In (10a), V is a ternary-branching structure, so that both the adverb and the

direct object are sisters of the verb. In (10b), the adverb is the sister of the verb,
and the direct object is higher in the structure. This diagram observes the constraint on binary branching the restriction that any node can have at most
two branches descending from it.
A second approach to the derivation of [V adv object] constituent order
is via movement of the subject and the verb to the left, as shown in the derivation in (11). On this approach, the adverb is generated in a position adjoined
to VP (or alternatively, moved there from a VP-nal position). Two other VP
constituents have moved: the external argument has moved to the Spec of IP
(as discussed in Chapter 3), and the verb has moved to the head of IP.
Both of the approaches outlined above derive the correct order of constituents, but they dier in several important ways. Consider rst the basegeneration analysis illustrated in (10a). One drawback to this analysis is that
it violates the constraint on binary branching (Chapter 2, Section 2.8). A
second drawback is that it obscures the structural distinction between arguments and adjuncts, as noted above. The structure (10b) on the other hand
observes the constraint on binary branching, and does make a structural distinction between the complement and the adjunct. However, the distinction is
the opposite of that predicted by theta-theory: in (10b), the adjunct is ccommanded by the verb, while the complement is not. The movement analysis avoids these problems, since in (11), the complement is a sister of the verb,
and the adjunct is generated in a VP-adjoined position. The two approaches
also make dierent claims as to the position of the verb at S-structure. The
base-generation analysis assumes that, at S-structure, the verb is still in its Dstructure position as the head of VP; the movement analysis shown in (11)
analyzes the verb as occupying a position outside VP, in the head of IP.
The movement analysis (Emonds 1978) claims that, in Romance, both main
verbs and auxiliary verbs move to INFL. In English, on the other hand, only

166
(11)

The syntax of Spanish

a. D-structure

IP

INFL

VP

VP

Adv

frecuentemente

DP

Mara

b. S-structure

DP

ley

el diario

IP

Maria i

INFL

leyj

VP

Adv

frecuentemente

VP

DP

ti

DP

tj

el diario

auxiliaries and main verb be move to INFL. One type of evidence supporting
this analysis is found in the contrasts between Spanish and English with
respect to question inversion. In Spanish, both main verbs and auxiliaries can
move to clause-initial (pre-subject) position in questions:

VP-related functional categories

(12)

167

a. Ley Juan el diario?

read J.
the newspaper
Did Juan read the newspaper?
b. Est Juan leyendo el diario?
is
J.
reading the newspaper
Is Juan reading the newspaper?
c. Haba Juan ledo el diario?9
had
J.
read the newspaper
Had Juan read the newspaper?

In English, only auxiliaries can be inverted:

(13)

a. *Read Juan the newspaper?

b. Is Juan reading the newspaper?
c. Had Juan read the newspaper?

The contrast between (12a) and (13a) can be accounted for on the assumption
that question inversion involves movement of INFL to pre-clause position.
Unless a verb can move to INFL, it cannot undergo inversion in questions. In
both English and Romance, auxiliaries move to INFL, hence the grammaticality of inverted auxiliaries in the (b) and (c) examples.
This analysis also accounts for another contrast between Spanish and
English. In Spanish, VP-adverbs can follow both auxiliaries and main verbs
(cf. note 2):
(14)

a. (?)Juan haba frecuentemente ledo el diario.

J.
had often
read the newspaper
b. Juan ley frecuentemente el diario.
J.
read often
the newspaper

In English, adverbs can follow an auxiliary verb only:

(15)

a. Juan had often read the newspaper.

b. *Juan read often the newspaper.
c. Juan DID often read the newspaper.

Although we have not yet discussed the internal structure of VPs with auxiliaries, let us assume minimally that they originate somewhere within VP. On

The present perfect form of the compound tense does not allow fronting of the auxiliary by itself:
(i)

a. ?*Ha Juan ledo el diario?

Has Juan read the newspaper?
b. Ha ledo Juan el diario?

Suer (1987) suggests that this contrast is due to the clitic status of the form ha.

168

The syntax of Spanish

this assumption, the distribution of adverbs in (14) and (15) is accounted for
directly by the movement analysis. In (14), one verb, either an auxiliary or
main verb, moves to INFL, where it can be followed by a VP-adjoined adverb.
In (15a), an auxiliary verb has also moved to INFL, and can be followed by a
VP-adjoined adverb. In (15b), the main verb cannot move to INFL, hence it
cannot be followed by a VP-adjoined adverb. In (15c), the pleonastic verb do
occupies INFL (as is supported by its ability to be inverted in questions), and
can be followed by a VP-adjoined adverb.
Summarizing, we have seen that the movement approach analyzes the verb
as exterior to the Verb Phrase at S-structure. The hypothesis that any verb can
move to INFL in Spanish, while V-to-INFLNFL movement in English is
restricted to auxiliaries, accounts for dierences between the two languages
with respect to verb-fronting in questions, following adverbs.
With respect to the position in which adverbs are generated, it can be
assumed that all of the adverb classes that can occupy post-verbal position (all
except Extent adverbs) can be generated in one of two ways: (a) in VP-nal
position (adjoined to V or VP), or (b) in a VP-initial position (adjoined to V
or VP). On this analysis, there are two ways of accounting for the impossibility of Extent adverbs in these positions. One possibility is that they are generated in these same positions, but must themselves undergo movement;
alternatively, they may be generated higher in the structure.

4.2.3

Pre-verbal adverbs

Recall from 4.2.1 that adverbs of Time, Place and Extent can be preverbal the latter, obligatorily. Given the analysis in 4.2.1, according to which
the verb occupies INFL at S-structure, it follows that these adverbs must also
be VP-external at S-structure. And on the assumption that the subject occupies the Specier of IP, the pre-verbal adverbs must occupy an IP-adjoined or
I- adjoined position.
Recall that these are the positions in which IP-adverbs appear (cf. (4)). One
approach to the derivation of these adverb orders would be to assume that a
VP-adverb can move from a VP-internal position, adjoining to I or IP. If this
analysis is correct, there must be a semantic restriction on this movement,
since adverbs of Manner and Quantity cannot undergo it. Alternatively, it
may be that the adverbs which appear in pre-verbal positions can be interpreted as a type of IP-adverb, ones which are event-oriented. This hypothesis
nds some support in that an adverb like all cannot be pre-verbal unless it
has scope over the entire event:

VP-related functional categories

(16)

IP

Adv

(all)

IP

DP

Juan

Adv

(all)

(17)

169

INFL

VP

DP

conoci

a su mejor
amigo

a. Juan all conoci a su amigo.

Juan there met his friend.
b. (*)Juan all dispar la echa.
Juan there shot the arrow.

In (17a), the adverb has scope over the entire event: it describes the location
of the entire event, including the location of the subject. In (17b), all can also
have scope over the entire event, but there is also a reading on which it does
not. On this reading, the adverb species only the location of the arrow, and
only at the end of the event. The adverb does not have scope over the subject.10
The analysis of pre-verbal adverbs as event-oriented IP-adverbs does,
however, present diculties: on this analysis Extent adverbs would be analyzed as exclusively IP-adverbs, since they do not appear in any post-verbal
10

Notice also that a pre-verbal Place adverb has scope over negation, while a postverbal adverb is ambiguous:
(i)

a. Juan all no conoci a su amigo. Juan there didnt meet his friend.
b. Juan no conoci a su amigo all. Juan didnt meet his friend there.

In (ia), the Place adverb is outside the scope of negation: there was an event in a
location specied by all, and the event was a non-meeting. In (ib), negation can
have scope over the locative, with the reading: It is not the case that Juan met his
friend there.

170

The syntax of Spanish

positions. However, the fact that they cannot be clause-initial (pre-subject)

remains unexplained. Also, the fact that Time adverbs can appear in all positions in VP as well as in IP, with no apparent dierence in scope or restrictiveness, is not explained.
Summarizing the main points of this section, it was shown in 4.2.1 that VPadverbs can occur in any position before or after the subject, verb and object
of a transitive clause. However, not all types of adverbs appear in all of these
positions. In 4.2.2, the post-verbal positions were analyzed. It was shown that
the occurrence of adverbs between the verb and its object is unexpected under
the assumptions that have been presented previously as to the relative positions of complements and adjuncts. However, this order is only problematic
on the assumption that the verb is within VP at S-structure. The hypothesis
that the verb has moved to INFL explains the order of adverbs, and explains
relevant contrasts between Spanish and English. In 4.2.3, the analysis of preverbal adverbs was discussed. On the assumption that the verb is in INFL,
these adverbs are I or IP adjuncts at least at S-structure, although not necessarily at D-structure. Although this analysis accounts for the pre-verbal
positions for VP-adverbs, it does not provide any immediate explanation for
dierences in distribution of the various sub-classes of VP-adverbs.

4.3

Auxiliary verbs, tense and aspect

As noted above, the hypothesis that V moves to INFL in Spanish

accounts for the order of VP-adverbs relative to the verb and its complements.
Also, relevant contrasts between English and Romance follow, on the hypothesis that English main verbs (except be) do not move to INFL. Two aspects of
this analysis have not been addressed to this point, however. First, the question arises as to what triggers V-to-INFL movement. That is, if the verb did
not move, what type of principle would be violated? Second, what is the
parameter that dierentiates English and Romance? Finally, why do auxiliary
verbs seem to be immune to the parameter? That is, auxiliary verbs move to
INFL in both English and Romance; it is only main verbs which cannot move
to INFL in English. Although these topics remain under active investigation,
we will summarize below recent proposals that have been advanced to account
for these phenomena. Section 4.3.1 outlines the approach to V-to-INFLNFL
movement proposed in Chomsky (1993), which analyzes movement as a
means of satisfying the checking of features associated with functional categories. This analysis accounts for parametric variation, such as the V-toINFLNFL parameter, in terms of the strength of the functional feature in

VP-related functional categories

171

question. In 4.3.2, we turn to the structure of auxiliary verbs and their movement characteristics. As we will see, the feature-checking approach to verbal
licensing leads to the hypothesis that there is a functional category in addition
to INFL which participates in the licensing of VP: Aspect.

4.3.1

Triggering V-to-INFL movement: feature checking

Let us now consider the question of what triggers movement of a verb

to INFL. To begin, let us examine more closely what the INFL head consists
of. It is assumed to include features for Tense and for Agreement, as shown in
(18). The TENSE features of INFL are its specication for niteness and, if
nite, for a specic tense (e.g., Past). The AGR features of INFL correspond
to the person and number features of the subject, features which appear as
part of the verbal inection in nite clauses. Both of these feature sets are
functionally related to VP constituents. The AGR features of INFL are
related to the Case of the subject DP, as will be discussed further in 4.4.3 and
in Chapter 5. The Tense features of INFL provide a temporal specication for
the event denoted by the VP, as discussed in Chapter 3 (Section 3.2). INFL is
thus a category which carries out functions in relation both to DP and to VP
as a whole. Following Pollock (1989), much recent work has analyzed
Agreement and Tense as separate heads, each dominated by a separate phrasal
node. Here, the two heads will be analyzed as a unit.
(18)

INFL

INFL

AGR
[PERSON]
[NUMBER]

VP

TENSE
[ FINITE]
[ PAST]

The mechanics of linking up INFL with DP and V have evolved along

with broader developments in syntactic theory. Early analyses assumed INFL
(or its predecessor, the Aux node) to contain the verbal axes for Tense and
Agreement. For example, the D-structure of (19a) might be as shown in (19b):
(19)

a. Ellos cantaron.
they sing-pa.3rd.pl.
They sang.
b. [ellos [INFL -aron] [VP [V cant- ]]]
they
-pa.3rd.pl. sing

172

The syntax of Spanish

The ax would then be merged with the verb-stem via ax-hopping (movement of the ax onto the verb stem) or via movement of the verb-stem to
INFL. In recent work, following Chomsky (1993), it has been assumed that
all words are fully inected as they enter the syntax. A verb is then already
inected for Tense and Agreement features. On this approach, INFL contains
abstract features, rather than the ax itself:
(20)

[ellos [INFL [+ ] [3rd.pl.NOM.] [VP cantaron]]

they
sing-pa.3rd.pl.

The role of INFL is to check the features of V and DP, rather than to assign
features to them. Feature-checking is the mechanism by which functional
heads satisfy the syntactic requirements associated with their functions. For
example, [+ ] INFL checks the [+ ] feature of the verb. If the features match, this licenses INFL relative to its function of temporally specifying VP. If features do not match, or if features remain unchecked, the
derivation cannot produce a syntactically well-formed sentence. INFL must
therefore attract to it categories which have the right features to match its own
functional features. The categories which INFL must attract include DP,
related to its Case function, and V, which it species temporally. Thus, a more
detailed analysis of INFL would include these categorial features: a set of Dfeatures, which attracts a DP, and a set of V-features, which attracts a verb:
(21)

AGR:
[]
[ ]
[]
D

TENSE:
[ ]
[ ]
V

In the course of a derivation, the V-features of TENSE can be checked once

a verb moves to INFL, leaving a trace (t) in its original position, as shown
in (22).
(22)

INFL

VP

Vi

INFL

[ FINITE]

[ FINITE]

ti

VP-related functional categories

173

Example (22) shows one of the congurations in which feature checking can
occur: between a functional head and an Xo adjoined to it. Feature checking
can also occur between a functional head and an XP in its specier position.
An example of this latter relation is the DPINFL relation which checks Case
and other D-features of the subject. In (23), the D-features of INFL are
checked, if the Specier of IP contains a DP whose features match the Dfeatures of INFL.
(23)

IP

DP i

VP

INFL

Vj

INFL

tj

tj

The approach described above predicts that the derivation of every clause
(in every language) involves movement of a DP to the Specier of IP, so that
the D-features of INFL are checked, and movement of a verb to TENSE,
satisfying feature-checking of the temporal specication of the verb. Thus,
all languages have the structure shown in (23) at some stage of their derivation. However, we saw in 4.2.1 that there is a parametric dierence between
English and Spanish with respect to V-to-INFL movement. Here, we will
consider briey one formulation of the parameter which distinguishes the
two languages. Chomsky (1993) proposed that parametric variation in Sstructure representations is due exclusively to language-particular dierences in the properties of functional categories, and characterizes these
dierences in terms of feature strength. In the case of V-to-INFL movement, the parameter resides in the strength of the V-feature of INFL: in
English, the V-feature of INFL is weak, so that verbs do not move to
INFL until late in the derivation, in the covert syntax between S-structure
and logical form. In Spanish and other Romance languages, the V-feature of
Tense is strong, and a verb must be attracted to Tense by S-structure. This
formulation of the parameter accounts for the contrast between English and
Romance with respect to the position of main verbs in nite clauses. However,
it does not provide a straightforward account of English auxiliaries: if the

174

The syntax of Spanish

V-feature of INFL is weak in English, no verb is expected to occupy INFL

at S-structure. It was shown above, however, that English auxiliaries do
occupy INFL, based on the distribution of VP-adverbs with English auxiliaries, and their movement to clause-initial position in questions. In eect,
auxiliaries seem to be immune to the English/Romance V-to-INFL parameter, a point to which we return once we have considered the derivation of
auxiliaries in more detail.

4.3.2

Auxiliary verbs

There are three issues that arise with respect to the derivation of
clauses containing sequences such as (24):
(24)

Susana [ha estado estudiando].

S.
has been studying

One question is, how are the auxiliary verbs structurally related to other VP
constituents, particularly to the following verb? A second question is, how are
the inectional features of each verb in the sequence licensed? A characteristic of auxiliary verbs like those in (24) the aspectual auxiliaries is that the
form of the following participle is rigid: it is necessary therefore to account for
the grammaticality of (24), versus the ungrammaticality of *Susana ha estudiando Susana has studying. Finally, what features of auxiliaries dierentiate them from main verbs, such that their dierences with respect to
V-to-INFL movement are accounted for?
With respect to the rst issue, the structural relation between the auxiliary
and the following participle, there have been two general approaches in the
generative literature. Early work within X-theory analyzed auxiliary verbs
(other than the copula) as Speciers or Adjuncts of VP. For a sentence such
as (24), the auxiliaries would be generated as in (25) or (26). These approaches
to auxiliary verb structure take note of the fact that the main verb, which
heads the entire VP, is the constituent which is selectionally dominant. The
main verb, not the auxiliary, is the theta-marking head, and other adjuncts
must be selectionally compatible with features of the main verb. These analyses also have in common a structural characterization of auxiliaries: unlike
the main verb (estudiando studying), the auxiliaries are not dominated by a
separate VP. Auxiliaries are structurally minor categories, lacking full
phrasal structure. This way of describing the dependence of auxiliaries on a
main predicate is inadequate, though, because auxiliaries do have their own
phrasal structure. They can introduce their own adjuncts, and make available
more specier positions than are present in clauses without auxiliaries. This is

VP-related functional categories

175

illustrated by the distribution of elements such as no, and the oated quantier todos. Consider rst the contrast in (27):
(25)

VP

Spec

ha

estado

estudiando
(26)

VP
V

ha

estado

estudiando
(27)

a. Juan no estaba NO estudiando.

J.
not was
NOT studying
Juan wasnt NOT studying.
b. *Juan no NO estudiaba.
J.
not NOT studied
Juan didnt not study.

Typically, only a single instance of no not is possible in a given clause, as

shown by the status of (27b). A second negative is possible, with strong stress,
but only if an auxiliary is present, as in (27a). This contrast suggests that no
may be generated as a Specier of VP. On this hypothesis, (27b) would be
ungrammatical because there is only one VP, so only one Specier position in
which no could be generated. In (27a), there are two VPs, hence two Specier
positions in which no can be generated.
Likewise, the oated quantier todos (cf. Chapter 2, Section 2.6) can cooccur with a second quantier, but only in the presence of an auxiliary:

176

The syntax of Spanish

(28)

a. Mis amigos estaban siempre todos estudiando lingstica.

my friends were
always all
studying
linguistics
My friends were always all studying linguistics.
b. *Mis amigos estudiaban siempre todos lingstica.
my friends studied
always all
linguistics (=28a)

If an auxiliary is present, as in (28a), both constituents are grammatical, an

indication that a second VP is present. In the derivation of (28a), a quantier
occupies the specier of each of the two VPs. The quantiers appear to be
adjacent, as a result of movement of the auxiliary to INFL. In (28b), there is
no auxiliary verb, hence no second VP, so only one VP specier position is
available. These contrasts indicate that auxiliary VPs are full phrases, despite
their absence of argument structure.
The second approach to auxiliaries analyzes them as being in a head
complement relation with the following constituent (see (29)).
(29)

IP

INFL

VP

ha

VP

estado

VP

DP

Susana

estudiando

VP-related functional categories

177

This analysis overcomes the limitations of the analyses described above. The
core VP, headed by the main verb, together with its arguments, is the lowest
VP constituent in the structure. This VP is the complement of the VP headed
by estar, which in turn is the complement of the VP headed by ha. Each VP
has an independent Specier, so multiple negation and quantiers are possible, as in (27a) and (28a). The derivation involves movement of the subject
DP through the Specier positions of each VP to the Specier of IP. The
highest verb, ha, will move to INFL.
One possible drawback to the structure in (29) is that it does not explain
the morphology of the participle following the auxiliary. The traditional
transformational account of participial morphology is based on axhopping. However, given the assumptions outlined in 4.3.1 above, an axhopping analysis is no longer viable, if words enter a derivation fully inected.
Assuming a feature-checking approach to the licensing of verbal morphology
as discussed in 4.3.1, the complement must be a more complex constituent.
On this approach, the participial morphology would be associated with the
checking of features for another functional head. We return to this point
below.
Summarizing to this point, we have seen above that auxiliary verbs head
their own VP, and the participle is the head of a phrase which is structurally
the complement of the auxiliary. However, the hypothesis that the complement category is a VP is problematic, in that it fails to account for the licensing of the participles morphology. Let us then assume that the structural
relations shown in (29) are correct, but that the auxiliarys complement may
be some category other than VP.
This brings us to the second issue raised at the beginning of this section:
how is the morphology of the participle licensed? To see how this might be
accomplished, let us recall how verbal morphology is licensed in simple
clauses: the verb is attracted to INFL by the V-feature of Tense, and feature
agreement with respect to features such as [+ ], [] allows the features of the verb and INFL to be checked, licensing both of these heads. What
appears to be needed for participles is an additional INFL-like node, which
checks features of the participle. Suppose this node is INFL, as shown in (30):
(30)

[ . . . INFL [auxiliary [INFL [ participle . . . ]]]]

Given a D-structure like (30), the participle would move to the lower INFL,
checking participle features, and the auxiliary would move to the higher
INFL, likewise checking its features. If the structure in (30) is correct, it provides an immediate explanation for the licensing of the participle, and does so
in a way that also explains why IP-adverbs can follow auxiliaries: because the

178

The syntax of Spanish

complement of the auxiliary is IP. However, there are both empirical and conceptual diculties with the notion that auxiliaries select an IP complement. If
the clause were an IP, its INFL node should be specied for [ ], and
the participle should then conform to the syntax of [ ] verbs. This prediction fails, however, with respect to such properties as independent negation
and clitic placement. Compare the raising verb + innitive in (31) with the
auxiliary + participle sequences in (32):11
(31)

a. Juan parece [no saber

la respuesta].
J.
seems not to know the answer
b. Juan parece [saberlo].
J.
seems to know-CL(DO)
Juan seems to know it.
c. *Juan parece [lo saber].
J. seems CL(DO) to know (=31b)

(32)

a. *Juan ha [no estudiado ese captulo].

J.
has not studied
that chapter
b. *Juan ha [estudidolo].
J.
has studied+CL(DO)
Juan has studied it.
c. *Juan ha [lo
estudiado]
J.
has CL(DO) studied
Juan has studied it.

In the clausal (IP) complement of parecer seem in (31), the IP has standard
innitival properties: it can be negated, and clitics associated with the innitive must be enclitics (follow the verb), as shown by the contrast between (31b)
and (31c). The participle construction in (32a) is ungrammatical with independent negation and with clitics, whether enclitic, as in (32b), or proclitic, as in
(32c). The impossibility of independent negation and clitics associated with
the participle in (32) suggests that the participle is not specied either as
[+ ] or [- ], and is therefore not likely to be an IP.
The problems noted above for the INFL hypothesis correlate with a conceptual problem with respect to the temporal interpretation of participles: it
implies that participles have a value for niteness, which is problematic with
respect to its temporal construal. We will give a brief overview of the nature
11

The data in (32) with auxiliary haber have also apply to progressive estar, with one
exception. Unlike perfect participles, progressive participles do admit enclitics:
(i)

Juan est [estudindolo].

J.
is
studying+CL(Acc.)
Juan is studying it.

This fact suggests that progressive estar might be analyzed as a raising verb of the
restructuring class.

VP-related functional categories

179

of the problem, arguing that the morphology of the participle is not Tenserelated, but Aspect-related.
The [ ] feature of INFL species the location of an event on the
time-line. A [+ ] event is interpreted as past, present or future in relation to another time, typically the utterance time of the sentence. A
[- ] event is not specied for a temporal location. However, this partition does not apply well to aspectual + participle constructions. In particular,
the participle is not interpreted as either nite or non-nite, but is interpreted
as an aspect of a more complex set of states. To illustrate, compare the (a)
and (b) examples in (33) and (34):
(33)

a. Juan
J.
b. Juan
J.

vive en Madrid.
lives in Madrid
ha vivido en Madrid.
has lived in Madrid

(34)

a. Juan escribe una novela.

J.
writes a
novel
Juan is writing a novel.
b. Juan ha escrito una novela.
J.
has written a
novel

In the (a) examples with a simple present tense, the event is interpreted as
present or cotemporaneous with the utterance-time of the sentence. In the
(b) examples, although the clause is still present tense, the participle is not
systematically ordered relative to the utterance-time. Example (33b) is actually ambiguous as to whether Juan lives in Madrid at utterance-time or not;
(34b) is unambiguous: the event of writing a novel precedes utterance time.
Comparison of (33a) and (33b) suggests that what the auxiliary + participle
morphology does is split the event into a set of states: the state of Juans living
in Madrid and a state of Juan at the present time; these can be thought of as
aspects of the event, which in (33b) may or may not be cotemporaneous. In
(34b), the two states are: Juans writing a novel, and Juan at the present time.
In this case, the two states are not cotemporaneous: the writing precedes the
state of Juan at the present time. Based on this generalization, it is reasonable
to suppose that the function associated with participial morphology is a function which translates an event into a part, or subevent, of a complex event.
This generalization extends to progressives:
(35)

a. Juan
J.
b. Juan
J.

vive
lives
est
is

en Madrid.
in Madrid
viviendo en Madrid
living
in Madrid

In (35a), the clause describes a single state of Juan: his living in Madrid; in
(35b), the clause describes two states of Juan: Juan living (at place x), and

180

The syntax of Spanish

Juans living in Madrid. Progressive morphology can be described here as

locating the second state as a subpart of the rst.
If the preceding description of participles is accurate, then the aspectual
auxiliaries can be analyzed as subcategorizing an Aspect Phrase (AspP) which
contains functional features corresponding to perfective and progressive
aspect (e.g. [ ], [ ]).12 The participle itself, which heads VP at Dstructure, can be analyzed as undergoing feature checking by an Aspect head,
in a manner analogous to the INFLV relation, as shown in (36). To derive the
S-structure of (36), the auxiliary haba moves to INFL, and the participle estu(36)

a. Susana haba estudiado.

S.
had studied

b.

IP

INFL

VP

haba

AspP

Asp

Asp

VP

[+PERF]

DP

Susana

estudiado
12

Lema (1991) analyzes the auxiliary itself as an Aspectual head. This analysis obviates the need for a feature [+ ], which is adopted in Belletti (1990).

VP-related functional categories

181

diado moves to the head of the AspP, where its morphological features are
checked. The subject moves through Specier positions up to the Specier of
IP, where D-features are checked. On this analysis, IP-adverbs like probablemente would be analyzed as adjuncts of IP or of AspP: the heads which license
V-features. This analysis reduces the distribution of VP to two positions, both
shown in (36): as sister of INFL, and as a sister of Asp. In Chapter 3 (Section
3.2), it was noted that the canonical position in which VP occurs is as a sister
of INFL, and the primary exception to this generalization is that VP can be a
sister of an auxiliary verb. On the analysis shown in (30), VP is uniformly generated as a complement of a functional category which checks the morphology of the head.
Finally, let us return to the issue of the V-to-INFL parameter. Recall from
4.3.1 that the analysis of the parameter in terms of the strength of the Vfeature of INFL accounts for the dierence in surface constituency of main
verbs in Spanish versus English. Unexpectedly, however, English auxiliaries
move to INFL in overt syntax, just as do Spanish auxiliaries. The immunity
of English auxiliaries to the INFL-parameter is dicult to reconcile with
the assumption that the strength of the V-feature of INFL is responsible for
the contrast between English and Romance main verbs. If, however, there is
an additional functional category, AspP, which intervenes between INFL and
main verbs, then it is possible that the parameter lies in the strength of the Vfeature of this head. To see how this might be executed, suppose the sequence
of heads is as in (37):
(37)

INFL (V) Aspect V

Suppose now that the V-feature of INFL is strong in both languages, so V is

attracted to INFL at S-structure. If an auxiliary is present, it moves, in both
Spanish and English. What if no auxiliary is present? In this case, if Aspect is
absent, main verbs should move to INFL in both languages. However,
suppose that Aspect is always present, sometimes with minus values for its
features just as INFL is present, even with minus values for its [ ]
feature.13 Then the sequence of heads is as in (38):
(38)

13

INFL Aspect V

The possibility that AspP is always present raises a question as to the distribution
of multiple negation and quantiers discussed previously (cf. (27), (28)). These were
excluded from structures which contain only one verb, based on the claim that there
is no second specier position in which to generate a second element. If AspP is
always present, the question that must be resolved is whether or not it allows independent speciers.

182

The syntax of Spanish

In this structure, a main verb must rst move through Aspect before it can
move to INFL. If this structure is correct, it is possible to account for the contrast between auxiliaries and main verbs in terms of the strength of the Vfeature of Aspect. Suppose that, in Spanish, Aspect has a strong V-feature. A
main verb will move to Aspect, checking its V-feature, then move to INFL,
checking the V-feature of Tense. In English, the V-feature of Aspect is weak.
Main verbs cannot move to Aspect in the overt syntax, and therefore cannot
move to INFL. However, in English, INFL has a strong V-feature: this feature
must be checked before S-structure, but it cannot be checked by movement of
a main verb. It can only be checked by a modal or pleonastic do elements
which are standardly assumed to be generated in INFL.14
On the analysis outlined above, auxiliary verbs are immune to the parameter which dierentiates English and Romance because they, unlike main
verbs, can be inserted in the structure higher than Aspect. This leads to the
question of what function the Aspect head carries out in relation to VP such
that it must be present in the functional structure above VP regardless of
whether the verb has participle morphology or not. A possible answer to this
question is available on an analysis of Aspect as having the same type of internal structure as is assumed for INFL. Recall that INFL has both V-related
features and D-related features, the latter associated with the checking of
subject features, including Case, person and number. It may be that Aspect is
structured in the same way, and that its D-features are associated with the
object position, rather than the subject position. Tense and Aspect are then
the heads which license the grammatical subject and object respectively. We
will return to this point below as we examine the derivation of object clitics.
For the present discussion, the relevant point is that auxiliaries can be viewed
as verbs which lack both an internal argument, and (perhaps therefore) intrinsic aspectual features. Unless they are inected for aspect, they do not require
an Aspect head to dominate their VP. Besides perfective haber have, items
that might be analyzed as auxiliary verbs include progressive estar, passive ser,
and the copulas, ser and estar. Perfective haber and progressive estar are standardly analyzed as aspectual auxiliaries, for reasons discussed above. The
remaining items are non-aspectual, and there is no clear semantic generalization that covers them, or that unies them with aspectual auxiliaries. In fact,
these verbs dier from main verbs only in a negative sense: they do not
select arguments, and therefore never constitute the primary (semantic) pred14

A remaining problem is the main verb be. This verb moves to INFL, but cannot be
assumed to be generated above Aspect, because it can co-occur with aspectuals.

VP-related functional categories

183

icate of a clause.15 However, because these verbs all behave alike with respect
to such phenomena as V-to-INFL movement in English, it has often been
assumed that they, together with the modals (and pleonastic do in English),
comprise a subcategory of verbs distinguished by the feature [+ ].
If this feature exists as a subcategory of verb, then it is possible that English
INFL could be analyzed as attracting the feature [+ ], rather than
the feature [+]. As noted in Lema (1991), however, this feature does not seem
to play any signicant cross-linguistic role in the theory of functional categories.
Summarizing, Section 4.3.1 outlined the feature-checking analysis of Vto-INFL movement. On this analysis, lexical items are inserted in syntactic derivations fully inected. The inectional features of a lexical category
in this case the verb enter into a checking relation with the features of
a functional head in this case, INFL. Thus, V-to-INFL movement is triggered by the requirement that inectional features be checked. Cross-linguistic variation with respect to the presence or absence of overt movement
is accounted for in terms of the strength of the functional features which
must be checked. Only strong features are checked prior to S-structure. The
contrast between Spanish and English with respect to V-to-INFL movement was accounted for in terms of the strength of the V-feature of INFL.
On this account, the movement of auxiliary verbs to INFL remains unexplained. In 4.3.2, the derivation of clauses with auxiliaries was discussed.
It was argued that the structural relation between an auxiliary and the following participial phrase is a headcomplement relation, and that the following participial phrase cannot be simply a VP. It must also contain a
functional category which checks participial features. It was suggested that
this functional category is not INFL, but Aspect Phrase. This hypothesis
accounts for the form and interpretation of the participial phrase, and the
distribution of IP-adverbs. Finally, if AspP is present in all clauses, not
only those with auxiliaries, the relative strength of the V-feature of Aspect
may be responsible for the contrasting behavior of English auxiliaries and
main verbs.

15

There is one exception to this generalization, following Snchez and Camacho

(1995). They argue that one sub-type of copular ser, equative ser (e.g. Juan es el
hombre de azul Juan is the man in blue), subcategorizes an argument DP.

184

The syntax of Spanish

4.4

Clitics

In this section, we will examine the derivation of clauses with clitics.

One of the central issues for the derivation of clitics concerns their relation to
the arguments of the verb. In (39) for example,
(39)

a. Mara compr ese coche.

Maria bought that car.
b. Mara lo
compr.
M.
CL-DO bought
Mara bought it.

the transitive verb comprar buy occurs with a DP complement, as expected,

in (a), but in (b) the complement position is not lexically lled. Instead, the
verb is preceded by a clitic. One question raised by this alternation is: how is
the transitivity of the verb satised in (39b)? It seems clear, in informal terms,
that the clitic corresponds in some way to the missing direct object, an intuition which is supported by the fact that (39b) is ungrammatical without the
clitic. In precisely what formal way the clitic is associated with the object of the
verb has been a topic of long-standing debate among generative grammarians.
In 4.4.1 and 4.4.2 we will summarize two traditional approaches to the derivation of clitics. One is the movement approach, which assumes that the clitic
originates in the canonical object position, and undergoes movement to its
surface position associated with the verb. The second is the base-generation
analysis, which assumes that clitics originate in a clitic position associated with
the verb. For each of these approaches, we will consider their advantages and
drawbacks, and, in the course of the discussion, we will consider properties of
clitics which have motivated these approaches. In 4.4.3, we will discuss recent
proposals of a third type, based on the hypothesis that clitics are functional
categories.

4.4.1

The movement approach to clitics

Recall that the central question raised by the alternation in (39) is how
to account for the contrast between the distribution of full phrase complements, as in (39a), and cliticized complements, as in (39b). The distribution of
full-phrase complements is as expected. Assuming that Spanish is a headinitial language, a verb theta-marks a DP complement to its right: The derivation of (40) involves movement of V-to-INFL, as discussed in 4.2 and 4.3,
and the complement DP surfaces to the right of the verb. One aspect of this
derivation has not been discussed yet: the complement DP must also satisfy
the requirement for Case. In Chapters 2 and 3, object Case was described as

VP-related functional categories

185

assigned by the verb. On this analysis, the object DP is fully licensed in its Dstructure position, and its surface order is accounted for. However, under the
feature-checking theory outlined above, it is expected that Case features are
checked by a functional category. We will set this issue aside, and return to it
in 4.4.3. For the moment, let us assume that, whatever the mechanics of object
Case feature checking, it does not impinge on the relative ordering of the verb
and its phrasal complement. What about the derivation of (39b)? Its surface
form has no overt DP in complement position, yet the verb is interpreted as
transitive, and the clitic has person and number features which correspond to
the interpreted complement: 3rd person singular. One approach to the derivation of (39b) has been to assume that its D-structure is just like (40), except
that the complement DP is a clitic pronoun (see (41)). The S-structure of (41)
involves movement of the clitic to the left of the verb. Early accounts of clitic
movement (Emonds 1975; Quicoli 1976) generally attribute the need for movement to the fact that the clitic is an unstressed morpheme which must attach
to another constituent.16 There have been several proposals regarding the
landing site of clitic movement (i.e., the position to which the clitic moves).
The traditional (and most prevalent) analysis assumes that the clitic attaches
to the verb when it undergoes movement (see (42)).
(40)

DP

compr

ese coche

bought

that car

(41)

16

DP

compr

lo

The clitic status of lo is shown by the fact that it cannot be phonologically separated
from a verb. For example, in answer to the question Qu compraste? What did you
buy?, the answer Lo It is impossible. (Compare: Qu compraste?, What did you
buy? Eso. That.)

186

The syntax of Spanish

(42)

DP

ti

CL1

lo

compr

Adjunction of the clitic to the verb is motivated on grounds that the verb is
the clitic host, i.e., the constituent on which the clitic is phonologically
dependent. On this analysis, (42) represents an intermediate stage in the derivation. Once the clitic adjoins to V, the complex verb would then move to
INFL. An alternative version of the movement hypothesis (Kayne 1989) analyzes the landing site of clitic movement as INFL. On this analysis, the verb
rst moves to INFL, then the clitic adjoins to the V+INFL complex (see (43)).
The advantages of the movement analysis are straightforward: rst, it analyzes complements as having a uniform D-structure, as shown by the similarity between (40) and (41). Second, the movement analysis attributes clitic
movement to a property which is well motivated, the status of the clitic as a
clitic. Movement of the clitic in the syntax allows the clitic morpheme to be
structurally adjacent to its host. The contrast between the S-structure position
of full phrases and of clitics is therefore explained. A third advantage of the
movement analysis is that it provides a natural account of the fact that clitics
can appear in construction with a verb other than the theta-marking head:
IP

(43)

DP i

Mara

INFL

INFL t i

CLk
lo

VP

Vj
compr

INFL V

DP

tj

tk

VP-related functional categories

(44)

a. Mara lo
haba comprado.
M.
CL(DO) had bought
Maria had bought it.
b. Mara lo
est comprando.
M.
CL(DO) is
buying
Maria is buying it.

(45)

Mara lo
quiere comprar.
M.
CL(DO) wants to buy
Maria wants to buy it.

187

In (44), the clitic appears in construction with an auxiliary. In (45), it appears

in construction with a main verb (quiere), not the verb with which the clitic is
thematically associated (comprar). That lo is interpreted as a complement of
comprar is conrmed by the contrast between (45) and (46):
(46)

a. *Mara
Maria
b. *Mara
Maria

lo
CL(DO)
lo
CL(DO)

haba comprado
had bought
quera comprar
wanted to buy

un
a
un
a

libro.
book.
libro.
book.

In (46), the non-nite verb selects its own complement, un libro, and the clitic
cannot be added. This shows that in (44) and (45), which are parallel in other
respects, the clitic is interpreted as the complement of the non-nite verb. It
follows that when a clitic appears in construction with a verb other than the
theta-marking head, there must be an empty category related to the clitic
lower in the structure. This empty category is provided automatically under
the movement analysis, because movement always leaves a trace or empty
category in the position vacated by the moved constituent.
One problem for the movement analysis is the phenomenon of clitic doubling, illustrated in (47), for direct objects, and (48), for indirect objects:
(47)

(48)

a. Mara se
M.
CL(DO.re.)
Maria saw herself.
b. Mara lo
vio
M. CL(DO) saw
Maria saw him.

vio a s misma.
saw PA herself
a l.
PA him

a. Pedro se
envi una carta a s mismo.
P.
CL(IO.re.) sent a
letter to himself
Pedro sent a letter to himself.
b. Pedro le
envi una carta a ella/Mara.
P.
CL(IO) sent a
letter to her/M.
Pedrol sent a letter to her/Maria.

Direct and indirect object clitics co-occur with full-phrase reexives and pronominals; indirect object clitics (and direct objects, in some dialects) co-occur

188

The syntax of Spanish

also with ordinary referential arguments. Clitic doubling is problematic for the
movement analysis, because if the clitic originates in the theta-marked position, there is no obvious source for the full phrase. One solution to this
problem (Hurtado 1989b) is the hypothesis that the full phrase is generated as
an adjunct, which is linked indirectly to the clitic. Another solution, to be discussed further below in 4.4.3, is to generate the [clitic + phrase] as a complex
constituent.17
An additional problem for the movement analysis is that certain clitics do
not alternate with full phrases:
(49)

a. Los nios
se comieron los dulces.
the children CL ate
the sweets
The children ate up the sweets.
b. El barco se hundi.
the boat CL sank
The boat sank.
c. Juan se
parece a Pedro.
J.
CL(re.) resemble PA P.
Juan resembles Pedro.

(Aspectual se)

(Middle se)

(Inherent Reexive se)

The aspectual clitic se in (49a) emphasizes the completedness of the event.

Middle se in (49b), on the gloss given, is standardly analyzed as a derivational morpheme which marks the deletion of the Agent Theta-role from the
verbs argument structure (Saltarelli 1994). The clitic in (49c) is an Inherent
Reexive. It does not correspond to an argument of the verb, and cannot
co-occur with a full phrase (Lujn 1976). Because these clitics do not correspond to arguments, there is no obvious full phrase source for them. The fact
that they appear at S-structure in the same positions as argumental clitics
implies that not all clitics which appear in clitic positions have arrived there
by movement.
Summarizing, one approach to the derivation of clitics is by movement
from DP positions. This approach allows a uniform D-structure position for
complements, since both full phrases and clitics are assumed to originate in
complement positions. It can also provide a natural account for the Sstructure distribution of clitics, on the assumption that clitics move in order
to be able to attach to a host, a requirement which follows from their clitic
status. The movement analysis also accounts naturally for the appearance of
clitics in construction with verbs other than the theta-marking head, since all
instances of clitic placement are derived by movement. This analysis, however,
does not easily accommodate the phenomenon of clitic doubling, and does
not extend to clitics other than those that are argument-related.
17

For a critique of these solutions to clitic doubling, see Suer (1988) and Franco
(1993).

VP-related functional categories

4.4.2

189

The base-generation approach

The second approach to clitics is the base-generation approach. As its

name implies, this approach claims that clitics do not move from full-phrase
argument positions, but are base-generated as a constituent of a complex verb
(Rivas 1977; Strozer 1976; Jaeggli 1982; Borer 1984). On this analysis, fullphrase complements are generated exactly as described in 4.4.1: a DP to the
right of the verb is the theta-marked argument, but if the complement is a
clitic, the argument requirements of the verb are satised in a dierent manner.
The precise properties attributed to the clitic, and to the clitic + verb complex,
vary in dierent studies. One approach is illustrated by the structure (50b).
(50)

a. Mara lo
compr.
M.
CL(DO) bought
Maria bought it.
b.
V

DP i

CL i

lo

compr

(pronoun
features)

On this approach, the DP that is a sister of the complex verb is the argument
position; it is theta-marked by the verb, exactly as in non-clitic, full-phrase
complement constructions. The clitic is formally linked to the DP by an agreement process, shown in (50b) as co-indexing.18 The clitic then agrees with DP
in person and number.
The crucial dierence between clitic and non-clitic complements is the
manner in which the verb assigns its Case. Normally, the verb assigns Case to
DP; but if a clitic is present, the clitic is assigned object Case, because it overtly
represents the grammatical features of the argument.
The primary advantage of this approach over the movement analysis is that
it provides a natural account for the phenomenon of clitic doubling. Because
the clitic is not generated in argument position, the DP position can either be
occupied by a covert pronoun or by a full phrase. When a full phrase is present,
the DP can receive Case by two mechanisms. On one view, the agreement relation causes the Case feature assigned to the clitic to be shared by the clitic
18

Notice that the co-indexing in (50b) is a superscript, which is distinct from the subscript notation that links a moved constituent with its trace.

190

The syntax of Spanish

and the DP. Another hypothesis is that clitic doubling is possible only if an
additional Case assigner is present, such as a preposition, like Dative a, and
Personal a. A second advantage to this approach is that it captures a generalization concerning the distribution of covert pronouns. In null subject languages, covert subject pronouns are grammatical, a fact which is standardly
attributed to the richness of subjectverb agreement morphology, which
allows the person and number features of the covert pronoun to be identied.
The analysis of clitics outlined above extends this generalization to objects:
the agreement between the clitic and DP allows the person and number features of the covert object pronoun to be recovered. Likewise, both subject and
object DPs can be overt, which follows from the sharing of Case/Agreement
features. Furthermore, this approach is compatible with the existence of clitics
which do not correspond to arguments. [ + ] compounds can be
formed in the lexicon, either as a result of derivational lexical processes such
as middle formation, or by free insertion of clitics like aspectuals or inherent
reexives. Because these clitics do not correspond to arguments, they need not
be licensed by Case/Agreement in the same way as argument DPs.
The primary drawback to the base-generation approach is that it predicts
that clitics cannot undergo movement at all. This is so because the [ +
] constituent is analyzed as a compound verb, and, as such, is expected to
behave like a single lexical item. Parts of the verb, like other compounds, are
therefore predicted not to move apart from each other once they are inserted
in a derivation. This is problematic for the base-generation hypothesis because
at S-structure, clitics do not necessarily appear in construction with the verb
with which they are selectionally related, as illustrated in (44)(45) above.
Summarizing, the base-generation approach analyzes the clitic as inserted
in the syntax as a component of a compound verb, not in the argument position theta-marked by the verb. This analysis provides a natural account of
clitic doubling, and of covert pronominals. However, the base-generation
analysis illustrated here predicts incorrectly that clitics should not move away
from the verb to whose argument they are grammatically related, since parts
of compound lexical items cannot otherwise undergo independent movement.

4.4.3

Clitics as functional heads

We have seen above that both of the traditional approaches to the

generation of clitics capture certain of their core properties, but each has signicant empirical drawbacks as well. In this section, we will briey examine
how the evolving role of functional categories in the grammar may shed light
on the apparent paradox presented by the movement versus base-generation
problem presented by clitics.

VP-related functional categories

191

One line of research in this area develops the hypothesis that clitics are NPrelated functional items (Uriagereka 1992; Belletti 1995). This hypothesis is conceptually related to the movement analysis, according to which clitics originate
in argument positions. Recall that one drawback to the movement analysis is
that it cannot easily accommodate clitic doubling, because it generates the clitic
as an object pronoun, as shown in (41), repeated below. If the DP in (41) is a pronominal, the phrase has no internal structure, since, normally, pronominals do
not co-occur with determiners, quantiers, or other modiers. The clitic should
then constitute the entire argument. Suppose, however, the clitic is not a pronominal, but is instead a head of DP, as in (51). On this analysis, the problem of
clitic doubling is easily resolved, because the determiner, unlike pronominals,
does co-occur with additional material internal to its complement NP.19
(41)

DP

compr

lo

(51)

DP

Det

lo

A second line of research that is conceptually related to the base-generation

approach is the hypothesis that clitics are base-generated, not as a part of the
verb, but as a head of a functional category that is VP-related. Thus far, we have
discussed two such categories: INFL and Aspect head. It was suggested in 4.3
that the Aspect head is present in all clauses, and contains features for [], []. To see how clitics may be associated with these
19

There are several possible approaches to the internal structure of the NP containing
the full phrase double of the clitic. One possibility is to suppose that the clitic determiner subcategorizes a PP; another possibility is that the full phrase is a DP, with
structure analogous to determiner doubles such as (i) (cf. Chapter 2, Section 2.5) :
(i)

el hombre ese
the man
THAT
that man

192

The syntax of Spanish

projections, recall from 4.3.1 that INFL is analyzed as containing two sets of features: Tense features and Agreement features, as shown in (21), repeated below.
(21)

AGR
[]
[ ]
[]
D

TENSE
[ ]
[ ]
V

These features enter into checking relations with two VP constituents: with the
verb, and with the subject DP. This AGR portion of INFL has been called
Agreement-subject or Agr-s. It has been argued that subject clitics are in
fact generated in this AGR portion of INFL. It has been assumed as well that
there is an object-related AGR head which checks Case and other features of
the object, and that object clitics are generated in this head (Franco 1993). Let
us suppose that Agreement-object (Agr-o) is associated with the function of the
Aspect Phrase:
(52)

AGR-object
[]
[]
[]
[]
D

Aspect
[ ]
[ ]

This complex is functionally related to the verb and its object in a manner
analogous to the INFLVP relation. That is, Aspect has a V-feature, which
triggers movement of V-to-Aspect for V-feature checking, and the direct
object is also attracted to the Specier of the AGR-o phrase for D-feature
checking. This is illustrated in (53) (for simplicity, we omit the subject and the
IP portion of the structure):
(53)

a. compr ese coche

bought that car

AspP

b.

DP j

Asp

ese coche
Asp

Vi
compr

Asp V

ti

DP

tj

VP-related functional categories

193

In (53b), the verb has moved to the Aspect head, whose V-feature is checked.
The verb will undergo further movement higher in the structure, eventually
reaching INFL, as discussed previously. The DP has moved from its thetaposition to the Specier of Aspect Phrase, checking the D-features of Agr-o,
in a manner structurally analogous to the movement of a subject to the
Specier of Agr-s associated with Tense.
In a clause with an object clitic, the clitic is assumed to instantiate the Agro head. Notice that the features of Agr-o include a gender feature. This feature
is independently motivated, since passive participles show gender agreement.
The Agr-o complex now contains the feature array displayed by object clitics,
and may be the D-structure position in which object clitics are generated.
Relevant portions of the structure for (54a) are shown in (54b). In (54b), the
clitic is generated as the Agr-o head of Aspect. Its features are checked by
movement of DP to the Specier of Aspect Phrase. The DP in (54) is a covert
pronoun, but, under dialect-specic conditions, could also be an overt phrase,
in clitic-double constructions.
(54)

a. lo compr
AspP
b.

DP j

Asp

pro

Asp

compr

Asp V

lo

ti

DP

tj

This analysis, like the base-generation analysis, generally approaches clitics

as V-related, rather than D-related, in the sense that they are generated in a
VP-related functional head. Like the traditional base-generation analysis
described above, clitics are related to the argument via an agreement relation.
Here, that relation is a Specierhead relation, and the clitic is a realization of
the Case-checking function in the VP extended projection. An advantage of
this analysis is that it explains why Spanish pronominal clitics occur in clauses
but not in other categories, such as within Noun Phrases. Since DP does not
have Tense and Aspect, any Agr Phrases which occur within other categories
do not have the same checking functions as these. Unlike the traditional basegeneration analysis discussed in 4.4.1, the analysis in (54) does not generate

194

The syntax of Spanish

the clitic as part of a compound verb. This approach potentially overcomes

the principal drawback to the base-generation analysis, since the Aspect head
can, in principle, move separately from the verb.
Two nal observations are in order with respect to the functional analysis
of clitics. First, notice that the analysis of clitics illustrated in (54) is also compatible with the Determiner analysis of clitics illustrated in (51). That is, the
clitic could be analyzed as the constituent which moves to an abstract Agr-o
constituent in Aspect, satisfying the D-feature checking of that head. On this
analysis, movement of the clitic would be accounted for on the basis of Dfeature checking. Second, neither of the analyses outlined here provides an
obvious explanation for movement of clitics to INFL, where they precede a
nite verb. However, the analysis in (54) would suggest that an account in
terms of the relation between Tense and Aspect might be explored.
Summarizing the main points of this section, two traditional approaches to
the analysis of clitics were introduced: the movement analysis, which analyzes
clitics as originating in the position of full phrases; and the base-generation
analysis, which analyzes clitics as generated as a constituent of a compound
verb. Each of these analyses captures some properties of clitics, but each has
limitations. In 4.4.3, it was shown how an analysis of clitics as functional
heads, either D-related or V-related, potentially overcomes the most severe
limitations of earlier analyses.

4.5

Negation

We turn now to clausal negation. In this section, we will discuss recent

approaches to negation and Negative Concord, based on the hypothesis that
negation is a functional category in the VP/IP projection. We will begin in
4.5.1 with a discussion of simple negation and the Neg Phrase hypothesis.
In 4.5.2, we consider the phenomenon of Negative Concord, and how the
analysis of Negation as a functional category has been applied to its analysis.

4.5.1

Clausal negation with no not

The morpheme no not appears in designated S-structure positions.

As shown in (55), it precedes the rst verb in the clause, whether auxiliary or
main verb, and whether nite or non-nite:
(55)

a. Juan no ley el diario.

J.
not read the newspaper
Juan did not read the newspaper.

VP-related functional categories

195

b. Juan no haba ledo el diario.

J.
not had read the newspaper
Juan had not read the newspaper.
c. *Juan haba no ledo el diario.
J.
had not read the newspaper
d. Juan quera [no leer
el diario].
J.
wanted [not to read the newspaper]
e. Juan quera [no estar leyendo el diario].
J.
wanted not to be reading the newspaper
f. *Juan quera [estar no leyendo el diario].20
J.
wanted to be not reading the newspaper

In (55a) and (55b), no precedes the nite verb; it cannot follow an auxiliary, as
shown in (55c). No also precedes the rst innitive, whether it is an auxiliary
or main verb. A second generalization concerning the distribution of no is that
it must be adjacent to the verb. Only clitics can intervene between no and the
verb; other constituents cannot:
(56)

a. *No Juan ley el diario.

not J.
read the newspaper
Juan didnt read the newspaper.
b. *Juan no ayer
ley el diario.
J.
not yesterday read the newspaper
Juan didnt read the newspaper yesterday.
c. Juan no lo
ley.
J.
not CL(Acc). read
Juan didnt read it.

The uniform distribution of no in clauses has led to the assumption that it

is generated in a position higher than INFL, since nite verbs occupy INFL
at S-structure. Laka (1990) argues that Spanish no is generated as the head of
a functional category Sigma Phrase, which alternates with the emphatic
armative s yes. Following much recent research, I will refer to this category as Neg Phrase. Let us see what the D-structure of a sentence like (55a)
would be, assuming that Neg is higher than INFL. As a rst approximation,
consider (57), where we omit the Aspect Phrase, for ease of presentation. To
derive the S-structure, the nite verb moves to INFL, deriving the order: No
ley Juan el diario. This order is grammatical, with the subject following the

20

Example (55f) is possible if understood as containing constituent negation rather

than clausal negation. This reading is brought out by adding a continuation such as
the one in (i):
(i)

Juan quera estar no leyendo el diario, sino descansando.

Juan wanted to be not reading the newspaper, but resting.

196

The syntax of Spanish

verb. To derive a pre-verbal subject, the external argument moves to the

Specier of IP. Notice, however, that in (57) the Specier of IP is lower than
the Neg Phrase. Movement of the subject would incorrectly derive: *No Juan
ley el diario. This problem is resolved by the Split INFL hypothesis
(Pollock 1989), which, as noted briey in Section 4.3.1, analyzes as separate
heads the Agreement and Tense features of INFL. Following Belletti (1990),
the Agr-s phrase is assumed standardly to precede Tense. Once these heads are
separate, Neg can be between them:
(57)

NegP

Neg

Neg
no

IP

INFL

VP

DP

Juan

(58)

DP

ley

el diario

Agr-s Neg Tense

Assuming that the nite verb moves to Tense, the correct order of Negation
relative to the verb is derived. A pre-verbal subject will also be the rst constituent, since the subject DP will move to the Specier of the Agr-s phrase,
thereby preceding negation. Adverbs must also be prohibited from adjoining
to the right of Negation in other words, adverbs must adjoin to the Agr-s
phrase, not to Tense Phrase.
Two comments on this approach to negation are in order. First, the hypothesis that the negative morpheme is base-generated in the Neg Phrase leads to
the assumption that the position of the Neg Phrase in universal grammar can
be parametrized. In English, negative not follows INFL:

VP-related functional categories

(59)

197

a. John has not read the newspaper.

b. *John read not the newspaper.
c. John did not read the newspaper.

Assuming that English auxiliaries move to INFL at S-structure, but main

verbs do not, the order in (59) suggests that the Neg Phrase is below INFL,
and perhaps below auxiliaries:
(60)

English: Agr-s Tense Neg

A second issue concerns the position of Agr-s and of subjects generally,

which will be taken up in Chapters 5 and 6. The status of Agr-s (and likewise,
Agr-o) as a separate phrasal head (as shown in (58)) has been questioned in
some recent studies. Furthermore, assuming that it is a separate head, its position relative to Tense has been debated. If the Split INFL hypothesis should
be shown incorrect in future research, then an alternative mechanism is needed
to capture the position of pre-verbal subjects.

4.5.2

Negative Concord

The term Negative Concord concerns the interpretation of negation in the presence of a class of items referred to as n-words, such as nadie
nobody/anybody, nunca, nothing/anything, nada nothing/anything,
etc., and NPs introduced by ningn no, as in ningn libro no book. As indicated by the glosses, these items are only sometimes negative in meaning. This
is illustrated for nadie in (61):
(61)

a. Nadie vino.
Nobody came.
b. No vino nadie.
not came anybody
Nobody came.

In (61a), pre-verbal nadie is interpreted as a negative, glossed as nobody. In

(61b), nadie co-occurs with no, but these are construed as a single negative.
Example (61b) does not mean nobody didnt come (i.e., everybody came);
it is identical to (61a) in meaning. Informally speaking, the two constituents
function together as a discontinuous negative under certain conditions. The
conditions for Negative Concord are, rst, that either no or an n-word must be
pre-verbal, as shown in (62):
(62)

a. Nadie dijo nada.

nobody said anything
b. Juan no dijo nada.
J.
not said anything
Juan didnt say anything.

198

The syntax of Spanish

c. *Juan dijo nada.
J.
said anything

In (62a), an n-word is pre-verbal, and, in (60b), no is pre-verbal. Example (60c)

is ungrammatical without a pre-verbal negative. Second, only one negative
constituent can be pre-verbal, as shown in (63):
(63)

a. No vino nadie.
not came anybody
Nobody came.
b. *Nadie no
vino.
Nobody didnt come.
c. *Nadie nada
dijo.
nobody anything said

Negative Concord has been analyzed in terms of a Specierhead licensing

relation between the n-word and the head of Neg Phrase (Haegeman and
Zanuttini 1991; Uribe-Etxebarra 1994). This can be cast in terms of the
feature-checking approach to functional categories described in 4.3.1. Assume,
for example, that the Neg Phrase contains an abstract feature for [+ ]:
(64)

NegP

Neg

[+NEG]

TP

This feature must attract a constituent specied for [+ ], and feature

checking is the mechanism for grammatically licensing clausal negation. The
assumption that this feature is strong accounts for the fact that one negative
constituent must be pre-verbal. It must attract either no or an n-word, the
items which have Neg features to check. To derive (62a), nadie moves to the
Specier of NegP, and checks the strong feature of the abstract functional
head, as in (65).
(65)

NegP

nadie
[+NEG]

Neg

[+NEG]

TP

To derive (62b), no is inserted (not moved from a lower position), and it then
checks the feature of the abstract head. Alternatively, no might be analyzed

VP-related functional categories

199

as a phrase, which, like n-words, is in the Specier of NegP. These alternatives

are shown in (66). In (66a) no is a head. It is required in order to check a
feature of the abstract [+ ] head of NegP. In (66b), no is analyzed as a
phrase. On either analysis, both the pre-verbal distribution of no, and the
uniqueness of the pre-verbal negative constituent are accounted for. The
impossibility of multiple negatives in pre-verbal position is accounted for on
the assumption that constituents only move to check features, and cannot
move otherwise. Once a negative constituent has checked the strong feature
of Neg, no other constituent can move to NegP before S-structure. The distribution of Negative constituents is thus analogous to interrogatives, an
insight captured in Haegeman & Zanuttinis proposal. Another fact which
must be accounted for is that, while n-words can be post-verbal, no cannot.
The functional analysis of no accounts for this by base-generating no as the
head of Neg Phrase. This is also accounted for on a feature-checking
approach, on the assumption that no is base-generated in Neg Phrase in one
of the positions shown in (66).
(66)

b.

NegP

a.

[+NEG]

TP

no
[+NEG]

Neg

no
[+NEG]

Neg

Neg

NegP

TP

[+NEG]

Although the functional analysis of Neg Phrase may account for the distribution of negative constituents at S-structure, there are several details of
negation that require further analysis. One issue concerns the parameter that
distinguishes Negative-Concord and non-Negative-Concord languages. If it
is assumed that the licensing of n-words operates uniformly across languages, it remains to be explained why Spanish n-words like nadie and nada
can be genuine negatives when pre-verbal, but not separate negatives when
post-verbal.21 Also, Bosque (1980) notes that n-words can be licensed in
certain contexts which lack an overt negative. This is shown by the contrast
in (67):
21

The Negative-Concord parameter has been attributed to lexical properties of nwords. However, what these properties are precisely has been debated extensively.
For discussion, see Suer (1995) and Bosque (1980).

200

The syntax of Spanish

(67)

a. Mara canta mejor que ninguno de nosotros.

M.
sings better than any
of us
b. *Mara canta tan bien como ninguno de nosotros.
M.
sings as well as
any
of us

In (67a), the negative constituent is contained within a comparative clause

with an ellipted predicate; it cannot be assumed to be licensed internal to that
clause, as is shown by (67b), which is ungrammatical for many speakers.
Summarizing, we have seen in 4.5.1 that a functional analysis of negation
accounts for the distribution of negative no, assuming that it is generated in a
Neg Phrase, which is higher in clause structure than the S-structure position
of the verb. This requires a Split INFL analysis, with Neg Phrase intervening
between Agr-s and Tense. In 4.5.2, we saw that the distribution of n-words is
accounted for, on the assumption that the Neg head has a strong feature which
attracts a negative constituent to it before S-structure.

4.6

Summary

This chapter has examined several aspects of the syntax of the VP

in relation to the functional categories which license VP constituents. The
distribution of VP-adverbs was examined in Section 4.2. The order of the
verb in relation to adverbs shows that verbs move out of VP to INFL in the
derivation to S-structure. Section 4.3 introduced the feature-checking
approach to V-movement, and examined the structure associated with auxiliary verbs, and movement of auxiliaries. That section proposed that the functional heads which attract V include INFL (specically the Tense features of
INFL) and Aspect. The Aspect head was suggested to be necessary to license
participle morphology in clauses with aspectual auxiliaries, and was suggested to be present in all clauses, just as INFL is. This hypothesis provides
one way of accounting for parametric variation with respect to movement of
main verbs. Section 4.4 discussed the derivation of clitics. The two traditional
analyses of clitics, movement and base-generation, each account for certain
properties of clitics, but do not provide natural explanations for others.
Recent approaches to clitics have analyzed them as functional categories,
either in NP-related or VP-related functional projections. On both of these
approaches, the Aspect/Agr-o head may play a central role, either as the position in which clitics are generated, or the position to which they move. These
analyses can relate the movement of clitics to the D-feature checking of
objects, although movement of clitics to INFL is not naturally accounted for.
Nevertheless, each of these approaches provides a more adequate description

VP-related functional categories

201

of the syntax of clitics than was possible under early formulations. Section
4.5 discussed negation. Analysis of negation as a functional category in IP
was shown to account for the distribution of negative no and for basic properties of n-words.

Subjects, topics and declarative constituent

order
5.1

Introduction

In this chapter, we will take up the issue of the order of the clausal
subject relative to the verb and objects in declaratives. In Chapter 1 it was
noted that Spanish declaratives are generally grammatical with either preverbal or post-verbal subjects:
(1)

a. Mara compr un coche.

M.
bought a car
b. Compr Mara un coche.
c. Compr un coche Mara.

(S-V-O)
(V-S-O)1
(V-O-S)

Stated informally, the central question to be addressed here is: what properties of Spanish are responsible for this exibility in constituent order, as
compared with a language like English, where subject order is xed? This
question has been investigated actively from the earliest stages of the principles and parameters framework. Below, we will discuss some of the central
issues that bear on this problem, and consider informally the main lines of
investigation that have been explored within the principles and parameters
framework.
In previous chapters, the analysis of the clausal subject has been touched
on only briey. It was noted in Chapters 3 and 4 that the subject is generated
VP-internally, and is assumed to move to the Specier of IP to satisfy its Case
requirement. Nominative Case is assigned, or checked (cf. Chapter 4,
Section 4.3.1), by the head of IP, INFL:
1

It was noted in Chapter 1 that at least for some speakers, V-S-O order for sentences
like (1b) is not as fully acceptable as S-V-O or V-O-S. However, V-S-O becomes
acceptable if the subject is non-agentive, as in sentences like: Sufri el paciente dolores
terribles The patient suered terrible pains. Since V-S-O order is available for at
least some predicates, it will be assumed here to be grammatical.

202

Subjects, topics and declarative constituent order

(2)

[IP DPi
INFL
NOM < Case

203

[VP ti [V V . . . ]]]

The movement shown in (2) derives the S-V-O order that is observed in declaratives in English. This particular type of movement has been referred to as Amovement that is, movement to an argument position, as discussed in
Chapter 3. The main question that will be addressed below is whether the same
form of movement occurs in Spanish. A related issue is, how are the alternative orders in (1b) and (1c) generated? Finally, we will consider the relationship between the exibility of subject order and the grammaticality of
covert subjects.
We begin the discussion in Section 5.2 with an introduction to basic issues
related to the licensing of the subject in a language like Spanish, in which the
subject displays freer constituent order than in xed subject order languages. As the discussion will show, evolving assumptions about movement
make it dicult to account for the licensing of Spanish subjects in a way parallel to xed-order languages. We conclude that Spanish does not have the
same type of A-movement that derives S-V-O order in English. We will then
explore some of the alternative ideas that have been proposed regarding what
determines the position of the subject. Beginning in 5.3, we consider the traditional notion that constituent order is sensitive to the discourse informational content of constituents. That is, notions such as old versus new
information, Theme-Rheme, or Topic-Comment analysis of sentences
have been claimed to aect the order of constituents in declaratives. Section
5.3 will introduce two of these notions, Focus and Topic, which appear to
be signicant for characterizing constituent order. We will then go on to review
two approaches to the derivation of pre-verbal constituents as aected by
Topic-hood. Section 5.4 explores the hypothesis that features related to discourse information content are grammaticalized as functional features of the
clause in Spanish, and, consequently, they may trigger movement of certain
constituents to pre-verbal positions. Section 5.5 considers Topic constituents
called dislocated Topics, of which there are two types with distinct syntactic properties. Both of these types of dislocated constituents have been argued
to be adjuncts, rather than speciers, and not derived by movement. Finally in
Section 5.6, we turn to the issue of the relationship between exible subject
order and null subjects. We begin with a description of the Null Subject
Parameter (Rizzi 1982; Jaeggli 1982) as formulated within the Government
and Binding framework. We then consider some alternative formulations of
the parameter.

204

The syntax of Spanish

5.2.

The problem of free subject order

As noted above, the main question that we consider here is how the
clausal subject is licensed in Spanish, and, in particular, why it is possible for
subjects to appear either pre-verbally or post-verbally in declaratives. In this
section, we will discuss why the optionality of pre-verbal subjects is not
easily accounted for. The main issue is that, as the principles and parameters
framework has evolved, the operation of move alpha has become more constrained. Initially, movement was considered to be entirely optional: particular movement might or might not occur. As long as no principle came to be
violated in the structure derived by the movement, both results were expected
in principle to be ne. More recently, many authors (following Chomsky
(1993)) have assumed that movement is to be avoided if it is not necessary. This
is so because it seems that grammars are constrained by principles economy.
Because movement adds complexity to a derivation, it has been assumed in
much recent research that movement cannot occur unless it must, to avoid violating some principle. As a result, the theory leads to the view that optional
movement does not actually occur.
In this section, we will see how this evolving view of movement aects the
analysis of subjects. We begin the discussion in 5.2.1 with a review of assumptions that have been introduced already in Chapter 4 concerning the role of
functional categories in triggering movement. Then in 5.2.2, we apply these
assumptions to the analysis of the clausal subject, under the additional
assumption that movement is impossible unless it is triggered by a feature of
a functional category. As we will see, the framework as outlined to this point
does not provide a satisfactory analysis for the exibility of order of the
subject.

5.2.1

Review of the role of functional categories

In Chapter 4 (Section 4.3.1), several assumptions were introduced

regarding how functional categories contribute to the licensing of constituents
in a derivation. We will review these assumptions here, focusing on the following three characteristics of functional categories such as INFL:
(3)

Characteristics of Functional Categories

a. Check features such as Case, rather than assign features;
b. Attract a category with the appropriate feature to check;
c. Parametric variation: features are either strong or weak.

Let us again illustrate these in relation to the features of INFL. Recall that
INFL contains features for Tense and Agreement, as shown in (4). Both of

Subjects, topics and declarative constituent order

205

these feature sets are functionally related to VP constituents: AGR features of

INFL are related to the subject DP. The Tense features of INFL are functionally related to the verb. In Chapter 4 we focused on the verbal features of
Tense, and their relation to movement of V-to-INFL. Recall that the notion
of INFL as a functional category provided an alternative to the process of
Ax-hopping. That is, instead of having a verb stem hook up with its ax
in the syntax, as in (5),
(4)

INFL

INFL

AGR
[PERSON]
[NUMBER]
[CASE]
D
(5)

VP

TENSE
[ FINITE]
[ PAST]
V

[ellos [INFL aron ] [VP [V cant- ]]]

they
-pa.3rd.pl. sing

we assume instead that a verb is already inected before it enters the derivation. On this approach, INFL does not assign the Past ax to the verb.
Instead, INFL contains abstract features rather than the ax itself:
(6)

[ellos [INFL [+FINITE] [3rd.pl.NOM. ] [VP cantaron]]]

they
sang

Assuming an analysis like (6) rather than (5), the role of INFL is to check
that the appropriate features are present on the verb. This checking can
occur only in a local environment, and the result is that the verb must move
close enough to the features of INFL for checking to take place. In this sense,
INFL indirectly attracts the verb. Once V-to-INFL takes place, feature
checking takes place. For our purposes, we can think of checking as a kind of
feature agreement, although the technical aspects of checking are more
involved than is described here.
So far, we have reviewed the notion that functional categories like INFL
contain abstract features which must match, or be checked against, the features of an appropriate element in the derivation. We also saw that, in order
for feature checking to take place, it can be necessary for a constituent that is
too far away from INFL to move, so that it is close enough for this to happen.
In this sense, the functional category attracts a category to it.

206

The syntax of Spanish

The last point to consider is how parameters occur. Recall from the discussion of the previous chapter that language-particular variation in feature
checking is due to the strength of the features of functional categories. In
the case of V-to-INFL movement, the parameter resides in the strength of the
verbal features of INFL: in Spanish, this feature is strong, which means that
it must be checked early in the derivation in eect, before the point that we
have referred to as S-structure. In English, this feature is weak, which means
that it cannot be checked early in the derivation. In fact, it cannot occur until
after the S-structure representation is determined, so that it is a covert movement. The parametric dierence in word order can be illustrated as follows.
From a common deep structure, shown schematically in (7a), we have the
two supercial orders, (7b) and (7c):
(7)

a. [IP INFL [VP SUBJECT [V V OBJECT]]]

b. [IP [INFL Vi + INFL] [VP SUBJECT [V ti OBJECT]]] (overt movement)
c. [IP INFL [VP SUBJECT [V V OBJECT]]]
(no overt movement)

In (7b), the verb has moved to INFL as an overt process, due to the strong
verbal feature of INFL. In (7c), the verb has not moved to INFL, since the
verbal feature of INFL is weak. The verb in (7c) would move covertly, in the
Logical Form component of the syntax. Notice that, if no other processes
aected the derivation in (7), the two languages in question would have dierent basic word order. The language illustrated in (7b), which has a strong
verbal feature in INFL would have: V-S-O order as its basic order. The language illustrated in (7c), which has a weak verbal feature in INFL, would have
S-V-O order as its basic declarative order. However, these derivations represent only part of the picture. The INFL features that are related to the subject
have not been specied for strength. Below, we will consider this component
of INFL in relation to the question of subject order.

5.2.2

Movement of the subject to the specier of IP

Let us now consider how the mechanisms described above aect the
subject constituent. We saw above that there is a relationship of feature checking between INFL and two related lexical categories: the verb, and the subject.
We showed how the strength of the verbal feature determines whether or not
the verb moves to INFL (at least, as an overt process). We turn now to the corresponding features of INFL that are related to the subject. The subjectrelated features of INFL include the agreement features (person, number) and
the Case feature. This set of features will be considered as a group, and, for

Subjects, topics and declarative constituent order

207

convenience, we will refer to them as noun-features or N-features, just as

the features of Tense are verb-features.2
Just as the verb features of INFL are specied as either strong or
weak,so are the N-features of INFL. There are two possible outcomes for
the subject, according to the strength of INFLs N-features. These are shown
by the schematic derivations in (8). The verb is shown as moving overtly to
INFL, as was discussed for Spanish in Chapter 4 (Section 4.3):
(8)

a. [IP INFL [VP SUBJECT [V V OBJECT]]]

b. [IP [INFL Vi + INFL] [VP SUBJECT [V ti OBJECT]]]
c. [IP SUBJECTj [INFL Vi + INFL] [VP tj [V ti OBJECT]]]

Beginning with an initial structure (8a), the verb moves to INFL, and the
strong verb features of INFL can be checked. The N-features might be either
strong or weak. Example (8b) depicts a language in which the N-features are
weak, and the subject does not move, at least as an overt movement. Example
(8c) depicts a language in which N-features are strong, and the subject moves
to the Specier of IP.
Consider now why neither of the options in (8b) or (8c) is adequate for
Spanish. If the N-features of INFL are analyzed as strong, it would be
expected that movement of a subject (as a movement in the overt syntax)
would be obligatory, since this is the characteristic of strong features: they
must undergo feature checking before the point at which surface word order
is xed. On the other hand, if the N-features of INFL were weak, then
movement of the subject should be impossible, since weak features do not
require checking before surface word order is xed. Both of these options lead
to an expected xed order, which is problematic for Spanish. That is, the
mechanisms described cannot account for the grammaticality of the various
orders that are possible for the subject:
(9)

a. Mara compr un coche.

M.
bought a car
b. Compr Mara un coche.
c. Compr un coche Mara.

(S-V-O)
(V-S-O)
(V-O-S)

If the N-features of INFL were strong, then both (9b) and (9c) should leave
the strong N-feature of INFL unchecked prior to S-structure. If the N-features
of INFL were weak, then (9a) should be impossible. Notice that there are two
ways of deriving a surface order for (9a), both of which are inconsistent with
2

A more technically precise way of referring to these features is as D-features, since

the category that bears them is a DP, not an NP.

208

The syntax of Spanish

the feature strength of INFL features for Spanish. One way would be to do no
movement, leaving the verb as the head of VP, and the subject in the Specier
of VP. This would be impossible, since as we discussed in Chapter 4, the verb
features of INFL are strong in Spanish. On the other hand, (9a) could be
derived by analyzing both the verb features and the N-features of INFL as
strong. On this analysis, the surface structure would have the verb in INFL,
and the subject NP in the Specier of IP. This revives the problem of N-features. If they are strong, the grammaticality of (9b) and (9c) are unexpected.

5.2.3

Summary

As we have seen here, the approach to constituent order that is based

on feature strength, taken together with the view that movement is impossible
unless it is absolutely necessary, seems inadequate to handle situations such as
the one illustrated here, in cases like (9). Here, we have a constituent whose
order is free in declaratives, while in other languages it is xed.
This apparent paradox has led to two lines of investigation within the principles and parameters framework, which will be considered below. One possibility is that there are additional functional categories in the clause, and
features other than the N-features of INFL can be responsible for triggering
movement of constituents. In particular, it has been suggested that a functional category that is active in Spanish is one that is related to the discourse
function of constituents. We discuss this concept in 5.3, then in 5.4 consider
how it may illuminate properties of the subject. A second possibility that has
been explored is that, since movement is so tightly constrained, it is possible
that alternations like those in (9) point to a derivation that does not have
movement. This possibility is considered in 5.5.

5.3

Discourse roles: Focus and Topic

Several traditions in the grammatical literature have discussed the

information load borne by dierent constituents of a sentence. Terms such
as old and new information, theme and rheme, topic and comment have
been used to describe the contribution of various sentence constituents to the
ow of information within the discourse. In this section two central concepts
related to information structure will be introduced: Focus, and Topic. We
begin with the notion of Focus, in 5.3.1, then, in 5.3.2, we will consider one
subcase of non-Focal material, that of Topic.

Subjects, topics and declarative constituent order

5.3.1

209

Focus

The discourse context within which a sentence is framed determines

the discourse role for sentence constituents. For example, in the exchange in
(10), the (a) sentence provides a discourse context for the answers in (b)(d):
(10)

a. A quin vio Jos?

Who did Jos see?
b. Jos vio a Pedro.
Jos saw Pedro.
c. Vio a Pedro.
(He) saw Pedro.
d. A Pedro.
Pedro.

The question in (10a) asks for information about the individual whom Jos
saw. The answers in (b)(d) provide this information. In the answers, the only
constituent which provides new information is the direct object, Pedro. The
subject and verb are given or previously known by virtue of the context
in (10a). Comrie (1989) refers to the new information as the Focus of the
sentence, and all other constituents as Non-Focus. Zubizarreta (1998) takes
the relevant division to derive from the discourse notion of presupposition.
What the speaker and hearer assume to be true at the time that the sentence is
uttered is presupposed; elements of a sentence which are not presupposed
are the Focus of a sentence. Zubizarreta represents this distinction in terms of
the feature [ ]. As is illustrated by the question and answer pairs in (11)-(13),
any constituent, or the entire sentence, may be the Focus. The Focus constituent in the (b) sentences is bracketed:
(11)

(12)

(13)

a. Adnde fue Jos?

Where did Jos go?
b. Jos fue [FOCUS a casa].
Jos went [home].
a. Qu hizo Jos?
What did Jos do?
b. Jos [FOCUS fue a casa].
Jos [went home].
a. Qu pas?
What happened?
b. [FOCUS Jos fue a casa.]
[Jos went home.]

In (11) and (12), the question in (a) determines what is presupposed (hence,
Non-Focal) in the answer. In (13), the question provides no presupposition (or

210

The syntax of Spanish

only that something happened); likewise, in discourse-initial contexts, an

entire sentence may be a Focus.
Summarizing, sentences are assumed to contain some [+ ] material,
and may also contain [- ] material, although this is not always the case,
as in examples like (13b). The question arises as to how the grammar identies
focal material in a sentence, and how the partition between Focal and NonFocal material is marked. Addressing the second point rst, in Spanish, the
partition between Focal and Non-Focal material is not explicitly marked, as
is shown by the fact that the (b) sentences in (13) may be identical, although
they dier in their Focus bracketing. Although there is no explicit division
between [+ ] and [- ] constituents, there is nevertheless a systematic association between [+ ] and intonation, and between these and
word order, as argued in Zubizarreta (1998). To illustrate the relationship
between Focus and intonation, note rst that the normal position for the intonation peak (Nuclear Stress) in non-emphatic declaratives is the rightmost3
stressed syllable of the predicate:
(14)

a. Jos fue a CAsa.

Jos went HOME.
b. Su hermano comi una manZAna.
His brother ate an apple.
c. Mara baiL.
Maria danced.

This primary stress cannot fall on a [- ] constituent.4 This is shown by

the impossibility of (14a), repeated below, as an answer to (15):
(15)

Quin fue a casa?

Who went home?

(14a)

*Jos fue a CAsa.

Jos went HOME.

The question in (15) establishes fue a casa as presupposed, or [- ]. This

material will then also be [- ] in the answer, (14a). The only [+ ]
constituent of (14a) is its subject:
(16)

[[+ ] Jos

[[- ] fue a casa]]

Zubizarreta (1998) argues that the crucial context is hierarchical: the most deeply
embedded constituent is stressed.
This generalization is termed the Focus Prosody Correspondence Principle. See
Zubizarreta (1998:38) and references cited.

Subjects, topics and declarative constituent order

211

Primary stress then falls on a [- ] constituent, casa. For (14a) to become

a possible answer to (15), either its intonation or its word order can be altered,
as shown in (17): 5
(17)

a. JoS fue a casa. (emphatic stress on Jos)

b. Fue a casa Jos. (Nuclear Stress on Jos)

In (17a), word order remains the same, but intonation changes: an emphatic
stress, rather than Nuclear Stress, is assigned to the [+ ] constituent. In
(17b), intonation remains the same, but the order of subject and predicate is
reversed, so that the [+ ] subject is in position to be assigned Nuclear
Stress.6
Summarizing, we have seen above that a sentence may consist of both
[+ ] (new, or non-presupposed) and [- ] (old or presupposed)
constituents. The scope of [+ ] may vary for a given sentence according to the context in which the sentence is uttered, as illustrated in (11)(13).
Although word order is not exclusively determined by Focus, the two are
systematically related. Non-emphatic (Nuclear) stress falls on the rightmost
(hierarchically lowest) word stress in a clause, and this stress must coincide
with [+ ]. The alignment of stress with [+ ] may produce variation in either stress or word order.

5.3.2

Topic

The notion of TOPIC, like Focus, is determined relative to a discourse context. Comrie (1989) describes Topic as the constituent that
expresses what the sentence is about. For example, given the context (18),
(18)

Qu pas con Jos?

What happened to Jos?,

the reply will be a sentence whose Topic is Jos:

(19)

a. [TOPIC Jos] se fue.

Jos went away.
b. Eligieron presidente a [TOPIC Jos].
They elected Jos president.

Example (17a) may be more natural than (17b), at least for some speakers. In this
respect, it appears to pattern with V-O-S (with true transitive predicates). See note
10.
Zubizarreta (1998) proposes that the reordering of subject and predicate like that in
shown in (17b) is triggered by the [] status of the predicate. The movement is termed
p-movement.

212

The syntax of Spanish

The notion of aboutness that is generally taken to underlie the concept of

Topic is based on the presuppositions of speaker and hearer. In the dialogue
beginning in (18), it is presupposed that something happened to Jos; the
phrase [DP Jos] thus is presuppositional, and in the context of the question
(18), is the natural discourse Topic of answers in (19). In contexts other than
information questions, where a Topic is not determined, one possibility is that
a sentence will not have a Topic determined by previous discourse. The paradigm case is discourse-initial sentences (out of the blue statements) and sentences answering questions like Whats new?; such sentences have no Topic.
In other circumstances, where there is a previous discourse with presuppositional background, a Topic may be newly introduced. These newly set
Topics may be preceded by such phrases as: en cuanto a as for, hablando de
speaking of, etc. These phrases are sometimes referred to as Left
Dislocations or hanging Topics, and have some properties that are not
shared by all Topic constituents.
Topic constituents, whether newly set or not, can be syntactically dislocated. For example, (20) is an alternative to (19b):
(20)

(Hablando de) Jos, lo eligieron presidente.

(Speaking of) Jos, (they) elected him president.

In (20), the constituent (Hablando de) Jos is followed by a juncture and,

optionally, by a pause.7 Dislocated constituents are unambiguously interpreted as Topics. This is shown by the fact that left-dislocated constituents
must be specic in interpretation, which follows from the status of Topics as
a subcase of non-focal, or presupposed, material.8 Compare the (a) and (b)
examples below:
(21)

Dislocated Topics can also appear to the right of the clause:

(i)

a. Algunos estudiantes leyeron ese captulo, pero no s cules.

Some students read that chapter, but I dont know which (ones).
b. ? *Algunos estudiantes, leyeron ese captulo, pero no s cules.
Some students, (they) read that chapter, but I dont know which ones.

Lo
eligieron presidente, a Jos.
CL(DO) elected president, PA J.
They elected him president, Jos.

Right-dislocated Topics dier from left-dislocated ones in some respects. For

example, the Topic in (i) cannot be preceded by hablando de speaking of.
Zubizarreta (1989) follows Reinhart (1982) in dening aboutness relative to the
notion context set (Stalnaker 1978), the set of propositions that the speaker takes to
be true at a given point in the discourse.

Subjects, topics and declarative constituent order

(22)

213

a. Ningn estudiante ley ese captulo.

No student read that chapter.
b. ?*Ningn estudiante, ley ese captulo.
No student, (s/he) read that chapter.

The (a) examples show that non-specic pre-verbal subjects are grammatical.
In the (b) examples, however, where the subject is set o by comma intonation
and is obligatorily interpreted as Topic, the non-specic Topic is ungrammatical.
Summarizing, a Topic characterizes what a sentence is about. Topics are
selected from among the elements which are presupposed at a given point in
a discourse. In the context of information questions, the Topic of the answer
is narrowly identied; in other contexts, the speaker may choose one (or more)
such elements. Topics may be identied by appearing in a clause-peripheral
position, intonationally separated from their comment. We turn now to the
derivation of constituent order in declaratives, considering in particular how
the Topic status of constituents may be analyzed as aecting order.

5.4

[Topic] Movement to the specifier of IP

In this section we consider the derivation of pre-verbal subjects as an

instance of movement of a Topic constituent to the Specier of IP. This
approach claims, in essence, that a pre-verbal subject moves to the Specier of
IP not specically in order to participate in checking of N-features of INFL,
but in order to check a feature related to its discourse role. The empirical generalization that this analysis captures is that, as a result of movement, declaratives may have the order: XPPredicate, where XP is a Topic whether a
subject, or a non-subject. Examples of non-subject topics are illustrated in
(23) (from Zubizarreta 1998):
(23)

a. Todos los das compra Juan el diario.

every
day buys
J.
the paper
Juan buys the newspaper every day.
b. El primer da de escuela deber acompaar cada madre a su
hijo.
the rst
day of school should accompany each mother PA her
child
Each mother must accompany her child the rst day of school.
c. Ayer
present Mara su renuncia.
yesterday handed-in Maria her resignation
Yesterday Maria handed in her resignation.

214

The syntax of Spanish

d. En este bar escribi Max su primera novela.
in this bar wrote Max his rst
novel
In this bar, Max wrote his rst novel.

Below, section 5.4.1.discusses the derivation of sentences like (23); then 5.4.2
will compare this type of movement with other cases of movement. On the
assumption that [ ] is a functional feature, 5.4.3 considers whether it
occupies INFL or a higher head.

5.4.1

[Topic] Movement

Koopman and Sportiche (1991) proposed that one source of parametric variation across languages is the manner in which Nominative Case is
assigned. In English-type languages, INFL assigns Case only to the Specier
of IP; in Spanish-Italian-type languages, INFL assigns Case to a DP to its
right, in its VP-internal position, as shown in (24). Here, INFL assigns Case
to DP in the Specier of VP. On this analysis, the Specier of IP is not an Aposition. Following on Koopman and Sportiches work, other authors have
noted that the Specier of IP is then available as a landing site for A-bar
movements. Goodall (1991) proposed that IP can be the landing site for Whconstituents, a matter to which we return in Chapter 6 (Section 6.2).
(24)

Nominative Case assignment in Spanish/Italian

IP

INFL

VP

DP

(Case)

It has also been proposed that IP can function as a Topic Phrase that is,
that the Specier of IP can be the landing site for Topic constituents.9 The
idea of IP as a Topic Phrase is that movement of a constituent to the
Specier of IP may be triggered by a [ ] feature in INFL, rather than
9

For dierent versions of this approach, see Malln (1992), Zubizarreta (1998).

Subjects, topics and declarative constituent order

215

by the N-features discussed in 5.2.2. This implies that either the subject or
another constituent must move to the Specier of IP, if this feature is
strong. Ignoring N-features for the moment, other features of INFL would
be as shown in (25):
(25)

[IP XP [INFL [+TO P I C ] V-features] [VP DP V . . . ]]

[+ ]

If the subject constituent is the Topic, then the subject will of course be preverbal, as shown in (26):
(26)

Topic is the subject:

[IP Marai [INFL [+TO P I C ] comprj] [VP ti tj el coche]]
[+]

Where a constituent other than the subject is the Topic, the subject remains in
its base position:
(27)

Topic is an adverb/PP:
[IP Ayer [INFL [+TO P I C ] comprj] [VP Mara tj el coche]]
[+ ]

In (27), movement of the adverb to the Specier of IP satises checking of the

Topic feature of INFL; the subject need not move, and therefore is not
expected to move. These two derivations then account for S-V-O order in
declaratives and also for the XP-V-S-O pattern that was illustrated in (23)
above.
It is possible in principle to extend the approach outlined above to generate
clauses in which the subject is a VP-nal constituent. In such clauses, if the
predicate is presupposed, it is a potential candidate for Topic constituent, as
in (28b), derived as in (29):
(28)

a. Quin compr el coche?

who bought the car?
b. Compr el coche [Focus Mara].

(29)

Topic is predicate (Agr-o/Aspect Phrase):

[IP [compr el coche] i [INFL [+TO P I C ] V-features ] [VP Mara ti]]
[+ ]

The order in (29) is derived by movement of the [+] predicate either

the VP itself or its functional projection, Aspect Phrase (Chapter 4, Section
4.3). A possible objection to this approach is that it does not allow V-to-INFL
movement. This is problematic since, as discussed above in 5.2, the verbal features of INFL are strong. If the verb did not move, presumably this would
violate the requirement that the strong verbal feature be checked at the point
in the derivation that corresponds to S-structure. This does not seem to be

216

The syntax of Spanish

unsolvable, however, since the verbal features of INFL could perhaps be

checked, since the verb is the head of VP, and VP is the Specier of IP. The
Specierhead relation is a typical conguration in which feature checking
(like feature assignment under government) is possible. In Chapter 6, we will
mention another means of generating V-O-S orders, via movement referred to
as Scrambling.10
Finally, let us return to the issue of checking of the N-features of INFL.
Suppose a Topic constituent other than the subject occupies the Specier of
IP, as in (27) or (29). This position cannot then also be lled by the subject.
The question then is how are the N-features of INFL checked? One possibility is that the N-features of the subject undergo head movement (to INFL),
rather than phrasal movement (to the Specier of IP). That is, on the assumption that N-features are weak, they might be supposed to adjoin to INFL to
the head of IP rather than moving as a phrase to the Specier of IP. A second
possibility, which is compatible with the assumption that N-features are
strong, is that some property of the subject agreement features of the verb
allow it to check both V-features and N-features of INFL. A third alternative
is that the [ ] feature which triggers preposing is not a feature of IP, but
of a higher functional category, as shown in (30):
(30)

[XP Fo [IP Subject [INFL VP]]]

[]

On this hypothesis, the N-features of the subject can be checked in the standard way in the Specier of IP.
Summarizing, on the hypothesis that [] is grammaticalized as a functional feature, it is expected that a broader range of constituents can move to
the Specier of IP than is possible in languages which lack this feature. This
expectation is borne out, as shown by the grammaticality of XP-V-S-O order.
The variety of constituents that can be Topics accounts in principle for the
variety of positions in which the subject surfaces.
10

Zubizarreta (1998) argues that leftward movement of [VO] as a constituent is possible in Italian, but not in Spanish. She proposes that VOS order is derivable via
leftward adjunction (or scrambling ) of the object, once V has moved to INFL:
(i)

a. [IP INFL [VP Subj [V Obj]]]

b. [IP V+INFL [VP Subj [tv Obj]]]
c. [IP V+INFL [VP Obji [VP Subj tv ti ]]]

The steps in the derivation shown in (i) are: V-to-INFL, in (ib); and adjunction of
the Object in (ic). Ordoez (1997) proposes this movement (see Chapter 6, Section
6.4), but argues also for leftward movement of [VO] as a constituent under certain
conditions.

Subjects, topics and declarative constituent order

5.4.2

217

The [Topic] feature

We saw above that the hypothesis that [] is a functional feature

accounts for general patterns of constituent order in declaratives which
contain Topics. Here, we mention questions that arise under an analysis of
[ ] as a functional feature of the clause. One issue concerns whether
[ ] behaves in a way that is expected for functional features. On the
assumptions that have been outlined here, it would be expected that every IP
which bears a [ ] feature should show obligatory preposing of exactly
one Topic constituent. The movement of one such phrase should be obligatory, since, if there were none, the [ ] feature would remain unchecked.
The movement should occur only once, since after checking of the [ ]
feature of INFL took place, any further movement of Topic constituents
should be unnecessary, and therefore impossible.
Topics diverge from this expectation in that clauses may have more than one
Topic, as illustrated in the embedded clauses in (31):
(31)

a. Me
dijeron que Caterina, en verano, no se
pone esos
zapatos.
CL(IO) said
that C.,
in summer,
she wont put on those
shoes.
They told me that Caterina, in the summer, she wont put those shoes
on.
b. Jos prometi que en agosto, para
descansar, iremos a la
playa.
J.
promised that in August, (in order) to rest,
well go to the
beach.

Sequences of Topics are not expected to be possible, since once the feature is
checked by movement of one constituent, no further movement should be
possible. There is another construction with which [] constituents can
be compared, which are restricted to one movement per clause. This is movement of Contrastive Focus (CF or Focus) constituents. In the following
examples, the CF constituents are shown in capitals:
(32)

a. En VERANO no se pone esos zapatos.

(It is) in summer that (she) wont put those shoes on.
b. Esos ZAPATOS no se pone en verano.
those shoes (she) wont put on in the summer
It is those shoes that she wont put on in summer.
c. *Esos ZAPATOS en VERANO no se pone.
It is those shoes in summer that she wont put on.

Examples (32a) and (32b) are derived by preposing, triggered by a [ ]

feature; in (32c), we see that preposing of more than one such constituent

218

The syntax of Spanish

produces ungrammaticality. This is likewise true of interrogatives, as will be

discussed in Chapter 6, and in general of constructions in which there is movement of a constituent to a non-argument position (Chapter 2, Sections
2.22.3). Insofar as Topic-preposing diers from this pattern, it diverges from
expectations.
The Contrastive Focus construction illustrated in (32) displays order and
intonation which dier from ordinary declaratives. Since declaratives with
Topic constituents do not have the same characteristics, it is likely that Topic
preposing is not derived by the same type of process. This conclusion is reinforced by an additional characteristic of Topic preposing that diverges from
other constructions involving movement to a non-argument (or A-bar)
position. Topics allow (or require) a clitic double, while other such movements
disallow it. Again, the contrast is illustrated relative to [ ] constituents:
(33)

a. Esos ZAPATOS no se (*los) pone en verano.

It is those shoes (she) wont put (*them) on in the summer.
b. Esos zapatos, no se *(los) pone en verano.
Those shoes, (she) wont put *(them) on in the summer.

The movement of Focus constituents illustrates properties that are typical of

movement to A-bar positions. Although movement to these positions is not
discussed until Chapter 6, it is useful to note here that Topics do not share
these properties, and are therefore likely to be derived by other mechanisms.
Notice also that, if Topic movement were parallel to other cases of movement
triggered by a functional feature, we would expect the constituent order:
XPVP, where the constituent preceding the verb has a denable discourse
role. Assuming that the verb moves to INFL, the verb should then mark a partition between Topic/non-Topic, etc.11 This leaves two generalizations unexplained. One is that sentences can have the verb as the initial constituent,
which suggests that a Topic is not obligatory. Also, when a phrase moves to
pre-verbal position, we cannot predict its discourse function. A pre-verbal
subject may be a Topic, but it need not be. Thus, the hypothesis of []
feature checking does not account satisfactorily for the order and interpretation of subjects.
Summarizing, we have seen that the analysis of [ ] as a functional category of the clause captures certain properties of declarative word order.
However, this type of approach seems to predict that Topic constituents are
11

This pattern may in fact be more general than it appears. Insofar as there may be
other processes which aect constituent order, these may obscure the generalization
in question. One such process that is motivated in Zubizarreta (1998) is p-movement: a local reordering which allows Nuclear Stress to align with Focus material.
See Section 5.3.1.

Subjects, topics and declarative constituent order

219

unique, and that they should behave in a way similar to Focus constituents. In
Section 5.5 below we will discuss an analysis of pre-verbal constituents that
does not rely on movement or feature checking. We return then to the question of whether that analysis obviates the need for movement of the type discussed above.

5.4.3

Topics and neg constituents

We turn now to one nal issue with respect to the movement analysis
of Topic constituents: their order relative to negative constituents. As the data
in (34) show, a pre-verbal subject must precede negative no not and n-words
like nunca never:
(34)

a. *No Juan cant eso.

not J.
sang that.
Juan didnt sing that.
b. Juan no cant eso.
c. *Nunca Juan cant eso.
never J.
sang that
Juan never sang that.
d. Juan nunca cant eso.

Recall from Chapter 4 that a Neg Phrase has been proposed for Spanish which
is structurally higher than IP. If the Neg Phrase analysis is correct, then it
appears that Topic constituents must move to the Specier of a Phrase higher
than IP, as in (35); otherwise, they should follow NegP, as in (36).
(35)
(36)

[FP Topic Fo
[NegP
Nego

5.4.4

Summary

[NegP Nego [IP Verb . . . ]]]

[IP Topic Verb . . . ]]

To summarize, one approach to the derivation of pre-verbal phrases,

including subjects, is to assume that a feature other than N-features of the
subject attracts a constituent to the Specier of IP. It has been suggested in
various studies that information structure is relevant for constituent order. In
this section we have considered the hypothesis that a feature of this type, a
[ ] feature, is a functional feature of clauses, and, as such, can trigger
movement of a subject or other constituent to the Specier of IP. This hypothesis provides a natural account for the occurrence of constituents other than
the subject in pre-verbal position, as well as for the exibility of subject
order relative to the verb and objects. This approach is not without certain
problems, however. In particular, a theory-internal question arises as to

220

The syntax of Spanish

whether [ ] is appropriately analyzed as a functional feature, given that

Topic constituents are not unique. The question also arises as to the source of
dierences between the movement that attracts Topics and other A-bar movements. Finally, it was noted that, if there are functional projections above IP
in clause structure, then Topics must move to those upper projections, given
the relative order of Topics and other pre-verbal constituents such as
Negation. In 5.5.3 below, we discuss an alternative proposal for deriving preverbal subjects (and other constituents) which resolves some of these issues.

5.5

Dislocated topics

It was noted in Section 5.3 above that Topic constituents may be dislocated, in which case they appear in clause-peripheral positions, as in (20),
repeated below.
(20)

(Hablando de) Jos, lo eligieron presidente.

(Speaking of) Jos, (they) elected him president.

Such constituents are intonationally marked by a juncture and, optionally, by

a pause. Cinque (1990) has argued that there are two sub-classes of dislocated
Topics, with distinct syntactic properties. In 5.5.1 below, the properties of
these two sub-classes will be discussed, and, in 5.5.2, we discuss their derivation, which has been argued to be by base-generation in dislocated position,
rather than by movement from the clause-internal position in which the Topic
is interpreted. In 5.5.3, we consider the hypothesis that this base-generated
adjunct analysis of Topics may subsume non-dislocated Topics. That is, on
this view, pre-verbal subjects and other pre-verbal Topic constituents have
been proposed to be syntactically dislocated constituents in general, whether
or not they are intonationally marked as dislocated.

5.5.1

Two types of dislocated topics

Dislocated Topics in Spanish, like Italian (Cinque 1990), are of two

types: Left Dislocations (LD), and Clitic Left Dislocations (CLLD).
Their properties are summarized in (37) and (38) respectively:
(37)

Left Dislocations (LD):

a. the dislocated constituent must be an NP (DP)
b. the coreferential element may be an overt phrase, pronoun or epithet
c. appear in root clauses only
d. dislocated constituent may be preceded by topicalizing expressions

Subjects, topics and declarative constituent order

221

e. dislocated constituent need not display grammatical and selectional

connectivity with the coreferential element
f. dislocation is non-recursive
g. insensitive to syntactic islands
(38)

Clitic Left Dislocations (CLLD):

a. the dislocated constituent is not restricted to NP (DP)
b. the coreferential element cannot be an overt category
c. may appear in both root and embedded clauses
d. the dislocated constituent cannot be preceded by topicalizing
expressions
e. dislocated constituent must display grammatical and selectional
connectivity with the coreferential element
f. dislocation is recursive
g. sensitive to strong islands only

These properties are illustrated below; unless otherwise noted, examples are
from Olarrea (1996).
The LD and CLLD constructions dier with respect to the category of the
dislocated constituent. In the LD construction in (39) and (40), only a DP is
grammatical, while in CLLD in (41), other phrases are possible:
(39)

a. Juan, no me acuerdo de l.
J.,
not CL remember of him
Juan, I dont remember him.
b. *De Juan, no me acuerdo de l.
*of J.,
not CL remember of him
Of Juan, I dont remember him.

(LD)

(40)

a. Juan, lo
vimos a l
en la esta.
J.
CL(DO) saw
PA him at the party
Juan, we saw him at the party.
b. *A Juan, lo vimos a l en la esta. (=40a)

(LD)

(41)

a. De Juan, no me acuerdo.
of J.,
not CL remember
Of Juan, I dont remember.
b. A Juan, lo
vimos en la esta. (=41a)
PA Juan, CL(DO) saw
at the party
Juan, we saw him at the party.

(CLLD: cf. (39b))

(CLLD: cf. (40b))

In (39) and (40), a DP is dislocated. These become ungrammatical if the dislocated DP is preceded by preposition or Personal a. In (41), however, the dislocated constituent can be PP or a DP preceded by Personal a. These dierences
correlate with a second property: in (39) and (40), the dislocated phrase is associated with an overt coreferential phrase, while in (41), the coreferential phrase
is not overt. The only potentially overt constituent is a clitic and this is possible

222

The syntax of Spanish

only if the dislocated constituent allows for a clitic. In (41a), the dislocated constituent corresponds to a PP complement of the verb, for which there is no corresponding clitic in Spanish; in (41b), the dislocated constituent corresponds to
the direct object of the verb, and the corresponding clitic is present and
required. As noted in (37b), the coreferential phrase may be an overt phrase of
various types: non-pronominal, pronominal or an epithet:
(42)

a. El baloncesto, ese deporte le

encanta

tu

hijo.
(overt phrase)
the basketball that sport
CL(IO) enchants PA your child
Basketball, that sport is loved by your son.
b. Miles Davis, l s que me
fascina (tonic pronoun)
M.
D.
he yes that CL(IO) fascinates
Miles Davis, he is indeed fascinating to me.
c. (En cuanto a) Pedro, parece que el desgraciado se lleva
con
as for
P.,
seems that the bastard
CL gets along
with
todo el mundo, inclusivo con el enemigo. (Zubizarreta 1998)
everyone
including with the enemy
As for Pedro, it seems that the bastard gets along with everyone,
including with his enemy.

The CLLD coreferential phrase cannot be overt; it must be an empty category,

although this category may require a clitic:
(43)

a. En Juan, no es posible conar (*en l).

In J.,
not is possible trust
in him
Juan, it is impossible to trust.
b. A Mara, no la
v nunca (*a ese chica) tan enfadada.
PA M.
not CL(DO) saw never PA that girl
so irritated
Maria, I have never seen (*that girl) so irritated.

The LD occurs in root clauses only, while the CLLD may be in root and
embedded clauses:
(44)

a. Sin embargo,
Bernardo, estoy segura que nadie confa
en ese idiota.
on the other hand B.
(I) am sure
that nobody had
condence
in that idiot
On the other hand, I am sure that, as for Bernard, nobody had
condence in that idiot.
b. *Sin embargo,
estoy segura que Bernardo, nadie confa
en ese idiota. (=44a)
on the other hand (I) am sure
that B.,
nobody had
condence in that idiot.

Subjects, topics and declarative constituent order

223

Only the LD construction can be preceded by topicalizing expressions:

(45)

a. En cuanto a Antxon, l no va a terminar su tesis.

As for Antxon, he will not nish his thesis.
b. *Te he dicho que en cuanto a Antxon, lo
vi ayer.
(I) have told you that, as
for Antxon, CL(DO) saw yesterday
I told you that, as for Antxon, I saw him yesterday.

In (45a), the LD in the root clause may be preceded by a topicalizing expression; in (45b), the embedded CLLD cannot appear with such an expression.
The examples in (46)(48) illustrate the three nal characteristics of LD: in
(46), the connectivity between the dislocated phrase and its coreferential
phrase; in (47), the non-recursion of LD; in (48), the insensitivity of LD to
syntactic islands:
(46)

a. Nosotros, nadie nos

ha visto.
we(Nom.) nobody CL(DO) has seen
We, nobody has seen us.
b. El ordenador, yo odio esas mquinas infernales.
The computer, I hate those infernal machines.

(47)

a. En cuanto al dictador y al pueblo, ste lo repudia a aqul .

(Contreras 1978)
As for the dictator and the people, the latter hates the former
b. *Juan, el libro, l no lo ha comprado.
Juan, the book, he hasnt bought it.

(48)

a. Hablando de Freaks, un amigo que ha visto esa pelcula me

ha dicho que es magnca.
speaking of Freaks, a friend that has seen that movie CL(IO)
has said that is great
Speaking of Freaks, a friend that saw that movie said that it was great.
b. (En cuanto a) el Sr. Gonzales, que Mara lo haya invitado sorprendi a
todo el mundo. (Zubizarreta 1998:188)
As for Mr. Gonzales, that Mara invited him surprised everyone.

The connectivity referred to with respect to (46) concerns the absence of

necessary agreement between the dislocated phrase and its coreferential position within the clause with respect to features such as category, Case, and
number/gender agreement. In (46a), the dislocated phrase diers from its
coreferential phrase with respect to Case; in (46b), the dislocated phrase is
masculine singular, while its coreferential phrase is feminine plural. In (47),
the non-recursiveness of LD is illustrated. In (47a), there are two LD constituents, but these are constituents of a single conjoined Topic; in (47b), where
the two are not conjoined, the resulting sentence is ungrammatical. In (48), the
examples illustrate that LD is not sensitive to islands: in both examples, the

224

The syntax of Spanish

coreferential phrase is contained within a strong island in (48a) a relative

clause, in (48b) a sentential subject.
CLLD diers in each of these respects, as shown in (49)(51):
(49)

a. *Nosotros, no nos
han dicho nada.
we
not CL(DO) have said anything
We, they didnt say anything to us.
b. *El ordenador, las
odio.
the computer CL(DO3rd.f.pl.) hate
The computer[masc.], I hate them[fem.].

(50)

a. Ese libro a Pedro no se

lo
dio nadie.
that book, to P.,
not CL(IO) CL(DO) gave nobody
That book, to Pedro, nobody gave it to him.
b. A Mara esa pelcula no le
interesa.
to M., that movie, not CL(IO) interests
To Maria, that movie, it doesnt interest her.

(51)

a. ??A Mara alguien que le

dio un regalo no me
salud en la esta.
to M.
someone that CL(IO) gave a gift
not CL(DO)
greeted at the party
Maria, someone who gave her a gift didnt greet me at the party.
b. A esos espas no s
cmo se puede saber quin los
traicion.
PA those spies not know how one can
know who CL(DO)
betrayed
Those spies, I dont know how one can know who betrayed them.

In (49), the CLLD construction, unlike LD in (46) above, disallows disagreement between the dislocated phrase and the position to which it is related. In
(50), we see that CLLD does allow multiple constituents, unlike the LD construction. And nally, examples like (51a) illustrate that CLLD is sensitive to
strong islands constituents out of which extraction is generally impossible;
while (51b) shows that CLLD is not sensitive to a weak Wh-island.
Summarizing, we have seen above that Spanish displays the same two subcases of dislocated Topic constituents that Cinque identied in Italian: the LD
(Left Dislocation) construction, and the CLLD (Clitic Left Dislocation) construction. The divergent properties of the two are summarized in (37)(38)
above.

5.5.2

Derivation of dislocated topics

Let us now consider how the dislocated constituents in the LD

and CLLD constructions are generated. The dislocated Topic in the LD

Subjects, topics and declarative constituent order

225

construction has been analyzed as base-generated in a Topic phrase (Rivero

1980; Hernanz and Brucart 1987):
(52)

[TopP [(En cuanto a) Juan] TOPo [lo

vimos (a l) en la esta]]
as for
J.
CL(DO) saw
PA him at the party
Juan, we saw him at the party.

On the assumption that there is no direct grammatical link between the Topic
constituent and the following clause, several of the properties of LD follow:
the fact that the coreferential element may be overt, the absence of grammatical and selectional matching between the Topic and the coferential phrase,
and the insensitivity of the relation to syntactic islands.12 The non-recursivity
of the LD phrase would also follow if the Topic phrase enters into a checking
relation with a [ ] feature of the head. In other words, the LD Topic is
in a Specier position.
The CLLD Topic has been analyzed as a clausal adjunct. Let us illustrate
rst with a Topic corresponding to a direct object, as in (53a), with the derived
structure in (53b):
(53)

a. Juan lo
vimos en la esta.
J.
CL(DO) saw
at the party
Juan, we saw him at the party.
b. [IP Juanj [IP - [INFL lo vimos [VP proj tv proi en la esta]]]]

The Topic, Juan, is an IP adjunct. Within the clause, the coreferential phrase
is an ordinary pronominal: that is, a covert pronoun, pro, which is licensed
identied by the clitic.
Cinque (1990) has argued that the interpretive relationship between the
adjunct and the covert pronoun is established via an A-bar chain a set of
positions which together provide a complete expression: Juani, proi. The chain
is an A-bar chain if its highest position is in an A-bar position. Normally,
chains are constructed when movement takes place. The coindexing between
the moved constituent and its trace forms a chain. Here, Cinque is proposing
that a chain can be formed even in the absence of movement. The chain
accounts for the movement-like properties of CLLD. This includes the empty
category as the foot of the chain, the sensitivity of the construction to strong
islands, and the connectivity. At the same time, these chains do not display
the typical characteristics of A-bar chains that are derived by movement. One
dierence is the absence of a clitic-double for A-bar movement, illustrated by
Wh-movement in (54):
12

The occurrence of LD Topics in root clauses only may be related to their function
as the discourse-topic as discussed in 5.3: that is, this Topic is the Topic of a sentence, not of a clause.

226

The syntax of Spanish

(54)

a. Qu zapatos compr Susana?

What shoes did Susana buy?
b. Qu zapatos *los
compr Susana?13
what shoes
CL(DO) bought S.
What shoes did Susana buy them?

The impossibility of the clitic in (54), compared with its obligatoriness in

(53), suggests that the former does not involve movement of the direct
object.14
A second dierence between CLLD and typical cases of A-bar movement
is that the latter is restricted to one preposed constituent per clause. CLLD, as
noted above, is recursive:
(55)

a. *A quin qu libro le regalaste?

To whom what book did you give?
b. A quin le regalaste qu libro?
To whom did you give what book?

(56)

A Pedro, ese libro, se

lo
regal.
to P.,
that book, CL(IO) CL(DO) gave
To Pedro, that book, I gave it to him.

The preposing of only one constituent per clause is expected if the movement
is necessary for checking of a functional feature. The recursion of CLLD
Topics then suggests that these are not in Specier positions, and not derived
by movement, but are adjuncts.
A third property of dislocated Topics which dierentiates them from typical
cases of A-bar movement is that dislocated Topics do not trigger
subjectverb inversion. They are compatible with pre-verbal subjects:
(57)

Estos zapatos, Susana los compr.

Those shoes, Susana bought them.

(58)

a. *Qu zapatos Susana compr?

Which shoes Susana bought?

13

14

The contrast shown in the text does not extend to cases where the clitic is needed for
independent reasons, such as if the interrogative constituent is a Dative which is
obligatorily doubled, as in A quin le avisaste? Whom did you notify?
The contrast between (53) and (54) may be explained on the assumption that the
empty category left by A-bar movement is a variable, which, like referential
expressions, requires Case. Assuming that clitics absorb Case, as discussed in
Chapter 4 (Section 4.4), the empty category in (54b) would fail to be identiable as
variable, since it lacks Case. Then (53) is derivable only by means other than movement, since no variable could be licensed. The object position can be analyzed either
as a trace of the moved clitic or as a covert pronoun.

Subjects, topics and declarative constituent order

227

b. *ESOS zapatos Susana compr.

THOSE shoes Susana bought.

The impossibility of pre-verbal subjects in sentences like (58) appears to be a

phenomenon similar to subjectverb inversion (movement of V+INFL to
COMP). However, it has been argued that this account cannot be maintained
in Spanish, a point to which we return in Chapter 6. However the contrast
between (57) and (58) is accounted for, the informal generalization seems
clear: CLLD dislocated Topics are compatible with a pre-verbal subject, while
the A-bar movement illustrated in (58) is not.
Summarizing, neither LD nor CLLD exhibit properties of standard cases
of A-bar movement. LD may be analyzed as generation of a Topic constituent in the specier of a Topic Phrase. CLLD Topics have been argued to be
clausal adjuncts, linked to a clause-internal covert pronominal via an A-bar
chain. The Topic-pronoun chain is subject to conditions similar to those
derived by movement, accounting for similarities between CLLD and movement. However, the empty category is a pronominal, and the Topic has not in
fact been moved from a clause-internal position thus accounting for dierences between CLLD Topics and cases of A-bar movement.

5.5.3

Pre-verbal subjects as CLLD adjuncts

It has been argued that the CLLD analysis discussed above may be
extended to account for pre-verbal subjects.15 The claim underlying this proposal is that pre-verbal subjects display the properties of the CLLD construction, not of constituents which have undergone movement. On this analysis,
the pre-verbal subject in (59a) would be derived as in (59b):
(59)

a. Mara compr el coche.

Maria bought the car.
b. [IP Marai [IP proj [INFL comprj ] [VP tj tj el coche]]]

Here an overt DP, Mara, is adjoined to IP. IP does not contain a trace of
the Topic, but rather a covert pronoun, pro, generated in the standard VPinternal subject position. This pronoun is the grammatical subject. It moves
to the Specier of IP where it checks N-features of INFL. The Topic must be
licensed as having some relation to the clause. It is associated with the subject
via an A-bar chain, and is then interpreted as the antecedent of pro.
The same type of analysis can be extended to a pre-verbal adverbial:
15

For further discussion see Contreras (1991), Olarrea (1996) and references cited.

228

The syntax of Spanish

(60)

Adverb/PP-Topic:
a. Ayer
compr Mara ese coche.
yesterday bought M.
that car
[IP Ayer [IP proj [INFL comprj ] [VP Mara tj ese coche] tj ]]]
b. Ayer
Mara compr ese coche.
yesterday M.
bought that car
[IP Ayer [IP Mara proj [IP [INFL comprj] [VP tj ese coche] tj]]]]

In (60a), the adverb is the only Topic; in (60b), both the adverb and the subject
are Topics. This approach claims that the Specier of IP is a dedicated DP
position; however, weak N-features leave subjects in post-verbal position. The
only circumstance under which a subject DP that is interpreted as Topic will
be pre-verbal is via CLLD. Olarrea (1996) argues that the limited violations of
connectivity displayed by pre-verbal subjects support a CLLD analysis. In
particular, the possibility of Topicverb disagreement with respect to
[ ] in (61) suggests that the Topic is not a Specier of IP:
(61)

Los estudiantes tenemos

un alto concepto de nosotros mismos.
the students
have-1st.pl. a high opinion of usselves
Students, (we) have a high opinion of ourselves.

In (61), person disagreement is permitted, but only where the plural may
include the 1st or 2nd person indicated by verbal inection. That the verbal
inection agrees with a pro subject is supported by the form of the anaphor,
which agrees with pro, not with the Topic.
Finally, let us compare the CLLD analysis with the movement analysis discussed in Section 5.4 above. The CLLD analysis overcomes the problems that
are inherent in the movement analysis. Because pre-verbal constituents are
generated as adjuncts, their pre-verbal position does not have to be explained
in relation to a functional feature. It is therefore to be expected that such constituents are optional, and need not be unique. Second, the non-movement
properties of Topic constituents are accounted for, since the Topic is associated with a null pronominal, rather than a trace. However, there is one generalization that is not naturally accounted for under the CLLD analysis (as
an exclusive account of how pre-verbal constituents are derived). That is, the
unmarked order of constituents in neutral contexts contexts in which no
information is shared or presupposed is S-V-O. In other words, in clauses
with no Topic, a pre-verbal subject appears before the verb, not after it. If the
CLLD analysis were extended to cover these declaratives, it would lose its
account of the fact that complements are pre-verbal only if they have a Topic
interpretation. These sentences would be accounted for under the movement
analysis discussed in 5.4. Assuming that the [ ] feature is present

Subjects, topics and declarative constituent order

229

whether there is a Topic ([+ ]) or not ([-]), the derivation of

neutral sentences might also call for movement of a [- ] constituent to
satisfy the checking of the [ ] feature. Because this movement would
also accomplish checking of N-features of INFL, it is arguably more economical to move the subject constituent than to move a non-subject. The
preference for an initial subject in neutral contexts might thereby be
accounted for.

5.5.4

Summary

In this section, we have examined the properties and derivation of dislocated Topics in Spanish. In 5.5.1, we saw that there are two sub-classes of
dislocated Topics: Left Dislocations (LD) and Clitic Left Dislocations
(CLLD). In 5.5.2, the derivation of these two sub-classes was discussed. The
LD construction was analyzed as generated in the Specier of a Topic Phrase
in root clauses. The CLLD construction was analyzed as being generated by
adjunction to the clause. As the discussion showed, CLLD shares certain
properties with cases of movement, although it is not derived by movement of
the Topic. This is evident, given the cluster of dierences between CLLD and
standard cases of A-bar movement. Finally, 5.5.3 discussed an extension of
the CLLD analysis, according to which pre-verbal subjects in Spanish can be
generated as clausal adjuncts, interpreted via an A-bar chain in the same
manner as other CLLD constituents. This analysis is consistent with the
reserving of the Specier of IP as a DP position, which comes to be occupied by a subject only at the end of a derivation, once covert movement has
taken place, on the assumption that N-features are weak. Comparing this
analysis with the movement analysis of pre-verbal subjects, we considered
whether the CLLD approach obviates the need for a movement analysis of the
type discussed in Section 5.4. One context in which movement may still be
needed is sentences which have no Topic, but which have S-V-O as unmarked
order.

5.6

Subject order and the NS parameter

The nal topic that we will address in connection with declarative

constituent order is the relationship between the freedom of position of the
clausal subject and the grammaticality of null subjects. Early generative analyses hypothesized that these phenomena are related: free subject inversion,

230

The syntax of Spanish

or VP-nal subjects, arise from the same property of grammar as do null subjects, and this generalization was formalized as the Null Subject Parameter
(henceforth, NS parameter). This section will present a brief sketch of
early formulations of the NS parameter framed within Government and
Binding theory, then consider those generalizations under some more recent
assumptions.

5.6.1

The NS parameter in government and binding

In early (rule-based) versions of transformational grammar (e.g.,

Chomsky 1965), it was assumed that phrase structure rules like (62a) generate
subjects in clause-initial position at D-structure:
(62)

a. S NP (Aux ) VP
b. VP V NP

By rule (62a), a subject NP precedes the predicate, which may be expanded as

in (62b). These rules generate [ S [ V O ]] order (and constituency) directly at
D-structure. Within the principles and parameters framework, the D-structure
position of the subject is determined by theta-theory. In the Government and
Binding theory of Chomsky (1981), it is assumed that an external argument
can be assigned a Theta-role in the Specier of IP (which corresponds to the
S node in (62)). Applying that theory to Spanish, the D-structure of (63a)
would be as shown in (63b).
(63)

a. Mara compr un coche.

Maria bought a car.

b.

IP

NP 1
Mara

INFL

VP

NP 2

compr

un coche

The subject NP is assigned a Theta-role by VP. Let us refer to this as the VPexternal subject hypothesis. The subject NP in (63b) is assigned Case by

Subjects, topics and declarative constituent order

231

INFL. Case is assigned by certain heads of phrase to an NP that is governed by the head. We need not dene government here, but will rely on
an informal description. A head of phrase such as INFL governs nodes that
are within its maximal projection (IP), such as its specier and its complement although INFL governs into those phrases in a very restricted way.
For example, INFL governs its complement, VP, in (63b), but does not
govern the direct object, since the direct object has a more proximate governor: the verb.
The VP-internal subject hypothesis, discussed in Chapter 3 (Section 3.3),
posits the Specier of VP as the position in which an external argument of the
verb is theta-marked. On that hypothesis, the D-structure of (63a) is as shown
in (64).
(64)

IP

INFL

VP

DP 1
Mara

DP 2

compr

un coche

The D-structure in (64) must give rise to NP-movement, because DP1 cannot
be assigned Case in the Specier of VP. The transitive verb compr does not
assign Nominative Case; it assigns Objective (or Accusative) Case only to its
complement. Consequently, DP1 must move to a position where it is governed
by a head which can assign Case to it. DP1 moves to the Specier of IP, where
it is governed by INFL, and assigned Nominative Case. The S-structure for
(64) is then (65). The derivations in (63) and (64)(65) dier in their representation of the D-structure (theta) position of the subject. Both analyses derive
an S-structure subject in the Specier of IP. On both analyses, the Specier of
IP is the position in which Case is assigned by INFL, which governs the
subject.

232

The syntax of Spanish

(65)

IP

DP 1
Mara

INFL

VP

DP i
ti

DP 2

compr

un coche

Turning now to NS languages, a correspondence has been noted between

the grammaticality of null subject pronouns and the relative richness of
agreement in languages like Spanish and Italian, compared with languages
like French and English; the latter have relatively impoverished subject person
and number agreement morphology in their verbal paradigms. The implementation of this observation varies in dierent studies but, as we will see, two generalizations are common to several approaches:
(66)

Properties of rich agreement

a. Rich agreement allows the content of null pronouns to be identied;
b. Rich agreement aects government of the subject which aects Case.

Jaeggli (1982) and Rizzi (1982) proposed formulations of the NS parameter based on the assumption, standard at that time, that subjects are generated
in the Specier of IP. Rizzi argued that the richness of agreement in NS languages underlies a cluster of syntactic properties, including null subjects and
free inversion of the subject. Rizzi attributed these properties to the clitic-like,
pronominal character of the AGR features of INFL. Due to the clitic
pronoun-like properties of INFL, he proposed that it could absorb Case.
Stated informally the absorption of Case by AGR allows a null pronoun to
appear in the Specier of IP, as in (67). The phonetically empty pronoun in
the Specier of IP is a legitimate empty category because its content is identied by the rich agreement features of INFL. Also, the agreement relation
between the NP and the clitic-like AGR, shown in (67) as co-indexing, links
the covert pronoun to an overt lexical item.

Subjects, topics and declarative constituent order

IP

(67)

NP i
pro

INFL

VP

TENSE

AGR i
(68)

233

a. Compr un coche Mara.

bought a car
Maria

IP

b.

NP

ti

INFL

VP

NP i

VP

Mara

NP

compr

un coche

Rizzi further argued that the structure in (67) is responsible for the occurrence of subject postposing in NS languages. To derive (68a), the subject is
moved rightward, adjoining to VP, as shown in (68b). The derivation in (68b)
is incomplete, however, and would produce an ill-formed derivation if no
further processes apply. The NP trace in the Specier of IP is not legitimate,
because it is does not have an antecedent that is higher in the structure.16 The
16

The relation between an NP-trace and its antecedent is analogous to the relation
between an overt reexive or reciprocal phrase (an Anaphor) and its antecedent.
Both traces and anaphors must be c-commanded by their antecedent. A node a ccommands node b if and only if the rst branching node which dominates a also
dominates b. In (69), the NP-trace in the Specier of IP is not c-commanded by its
antecedent the VP-adjoined NP. The rst branching node which dominates the
moved NP is VP, and this node does not also dominate the trace.

234

The syntax of Spanish

second step in the process is the replacement of the trace by a covert pleonastic pronoun, as shown in (69). The structure in (69) is well formed, since pronouns, unlike traces, do not require an antecedent. The postposed NP is well
formed with respect to Case theory, because it is linked indirectly to the cliticlike AGR, which bears Case.
IP

(69)

NP i
[pro]

AGR i

INFL

VP

TNS

NP i

VP

Mara

NP

compr

un coche

Notice that only languages which have clitic-like AGR can have null pronouns, and only these languages can therefore rescue a structure like (68)
by insertion of a null pronoun. In English, a derivation like (68b) will always
turn out to be ungrammatical because, since AGR is too weak to bear
Case, a null pronoun could not be inserted to rescue the derivation. There are
other constructions which are also accounted for on the basis of the grammaticality of (69) (cf. Chapter 1, Section 1.2.8). Once this structure is generated, further processes can apply to the post-verbal subject, such as
Wh-movement. Rizzi argues that, in this position, the subject is governed by
the verb, rather than INFL. Because of this, the postposed subject behaves
like a complement with respect to its extraction patterns, not like a pre-verbal
subject.17
17

This can be illustrated by the contrast between English and Spanish with respect to
overt complementizers in interrogatives. In English, extraction of an object, but not
a subject, is compatible with an overt complementizer:
(i)

a. Who did you say that John saw?

b. *Who(m) did you say that saw John?

In Spanish and Italian, both subject and object can be extracted over an overt complementizer:

Subjects, topics and declarative constituent order

235

Jaeggli (1982) argued that AGR in NS languages moves in the syntax via
ax-hopping, and is attached to the verb at S-structure. The post-verbal
subject such as (68a) above is governed by the verb+AGR:
(70)

[IP PRO [INFL ] [VP [V compr+AGR un coche] Mara]]

Summarizing to this point, the two analyses described above derive NS

properties from the morphological and syntactic properties of the AGR
node of INFL. In Rizzis analysis, AGR is clitic-like, which allows it to bear
Case, which in turn makes null pronouns possible, and in turn participates in
licensing derivations with subject postposing. Postposed subjects are governed by the verb, and therefore behave dierently from pre-verbal subjects
in interrogatives and other related constructions. Jaegglis analysis claims
that post-verbal subjects are governed by V+AGR, once ax-hopping has
applied.
There is one theory-internal problem with the analyses described above,
which is resolved with the advent of the VP-internal subject hypothesis. The
problem concerns the use of lowering rules: subject postposing and axhopping. Both of these rules move a constituent lower into the structure, and
to the right the opposite of other NP movements in these languages, such
as object-to-subject movement in passives. The framework within which these
analyses were formulated assumed that both rightward and leftward movements are freely available. Movement itself was assumed to be unconstrained;
the derivation was excluded only if the resulting structure violated a general
principle of grammar. More recent theories have proposed restrictions on
movement which have the eect of restricting or eliminating postposing
movements. As discussed in Section 5.5 above, recent research has suggested
that movement of phrases is possible only to the Specier of functional categories, and only when required to satisfy feature checking. General principles of economy have the eect of prohibiting movement that is not
necessary. Rightward movement would then only be expected in languages
(ii)

a. A quin dijiste que Juan vio?

PA whom said that J.
saw
Who did you say that Juan saw?
b. Quin dijiste que vio a Juan?
who said that saw PA J.
Who did you say that saw Juan?

The contrast between (ib) and (iib) is argued by Rizzi to be due to the grammaticality of subject postposing. Once the subject is postposed, it is governed by the verb,
and its extraction properties pattern with complements. For details of the relation
between the overt COMP and extraction, see Rizzi (1982).

236

The syntax of Spanish

that have functional categories with rightward Specier positions. Kayne

(1994) argues that there are no rightward movements at all in UG. Given
these more restrictive views of movement, and evolving views of the conditions under which movement is possible, the subject lowering analysis is
less tenable.
The issue of lowering is resolved under the VP-internal subject hypothesis. Since the subject is generated within VP, it is not lowered when postposed.
Consider rst the derivation of a clause with a null subject such as (71a), which
derives from the D-structure (71b). As in the previous analyses, the rich agreement features are the source of the grammaticality of (71b) with a null subject.
As mentioned above, Koopman and Sportiche (1991) propose that one source
of parametric variation across languages is the manner in which Nominative
Case is assigned. In English-type languages, INFL assigns Case only to the
Specier of IP; this is associated with its non-lexical (non-rich) properties. In Spanish-Italian-type languages, INFL assigns Case to a DP to its right,
such as the covert pronoun in (71b), due to its (rich) lexical properties. The
derivation of (71b) involves Nominative Case assignment to the null pronoun
in its D-structure position.
(71)

a. Compr un coche.
(He/she) bought a car.

IP

b.

INFL

VP

DP 1
pro

DP 2

compr

un coche

Subject postposing involves adjunction to the right, as in (72). The same

lexical property of INFL which determines that it assigns Case to its right
in (71) and (72) also determines its ability to license a phonologically null
pronoun (71), and the trace of the postposed NP in (72). The complement-like
behavior of postposed subjects is also captured.

Subjects, topics and declarative constituent order

(72)

237

a. Compr un coche Mara.

IP

b.

INFL

VP

NP i

VP

NP

ti

Mara

NP

compr

un coche

Summarizing, we have reviewed several approaches to the NS parameter

under assumptions that were current within early forms of the principles and
parameters framework, commonly referred to as the Government and
Binding framework. Those analyses have two general features in common.
First, post-verbal subjects are derived by an optional movement rule. This rule
would move a subject from its D-structure position (either the Specier of IP,
under earlier assumptions, or the Specier of VP, under later assumptions),
adjoining it to the right of VP. Second, NS properties are attributed to the
lexical properties of the AGR features of INFL. AGR licenses null pronouns (either due to the absorption of Case by AGR, or due to a parameter
in how INFL assigns Case).

5.6.2

Richness of AGR and strength

We saw in 5.6.1 that early analyses of the NS parameter attributed the

properties of NS languages to the lexical properties of AGR. As discussed in
5.5, under more recent assumptions about conditions under which movement
occurs, it appears that the contrast between Spanish and English may be captured in terms of the strength of the N-features of AGR in INFL. In English,
these features are strong, so pre-verbal subjects are obligatory; in Spanish, these
features are weak, and pre-verbal subjects are not obligatory. Instead, subjects
occur in a pre-INFL position as a consequence of their Topichood.

238

The syntax of Spanish

The question arises as to whether the strong/weak distinction discussed in

5.5, if correct, accounts for the possibility of null subjects.18 We will not
attempt to answer this question here, but will point out that there is a potential description of AGR in Spanish that is compatible with strong N-features
of AGR. Belletti (1990), in discussing contrasts between French and Italian,
suggests, following Roberts (1990), that dierences between the two languages
with respect to V-to-INFL movement may be accounted for in terms of the
structure of the verb in particular, the structure of its subject agreement
morphology. She suggests that agreement can either be a sub-lexical (X-1) morpheme in languages like French, or a zero-level morpheme in Italian. Under
the more recent assumptions discussed above and in Chapter 4, according to
which lexical items enter the syntax fully formed, the contrast would be as
shown in (73).
(73)

a.

stem

b.

Agr 1

Agr

Let us suppose that (73a) represents the French/English structure of

Agreement on the verb, and (73b) the Spanish/Italian structure. In sentences
like (74),
(74)

a. Maria [INFL has [VP sung]].

b. Mara [INFL ha [VP cantado]].
Maria has sung.

INFL is occupied by an inected auxiliary in both languages, but the structure of that head diers, as shown in (73). Now suppose (contrary to what was
assumed in 5.5) the N-features of INFL are strong in both English and
Spanish (and Italian).
Since features can be checked either by a phrase in the Specier of IP, or by
head adjunction, the fronted verb can check features if it is structured as in
(73b), since is a head; but not if the verb is structured as in (73a). In this
case, the sublexical is syntactically inert, since it is sublexical. In general,
sublexical material is invisible for syntactic processes. Therefore, in English,
18

Since parameters are suggested to derive from features of functional categories, the
presence or absence of a null pronoun in a language is not straightforwardly relatable to constituent order. It has been assumed standardly in generative research that
empty categories (like pro) do not dier in fundamental ways from overt counterparts, a point which makes the systematicity of null subject pronouns perhaps dicult to relate to constituent order. For arguments that covert pronouns in fact have
dierent interpretive properties from overt pronouns, see Montalbetti (1986).

Subjects, topics and declarative constituent order

239

the strong N-features of INFL can only be checked by DP movement to the

Specier of IP. In Spanish, the agreement features of the verb check Nfeatures. IP therefore need not have a phrase in its Specier position. A preverbal subject moves to the Specier of Topic Phrase, according to the value
of Topic features.

5.7

Summary

In this chapter, we have considered the derivation of the clausal

subject in Spanish, with emphasis on accounting for the surface order of the
subject. In 5.2, we saw that optional movement is not considered to be possible in more recent versions of the principles and parameters framework.
Since movement is motivated to check features of functional heads, and
economy considerations preclude unnecessary movement, it is expected that
movement of the subject constituent to the Specier of IP for checking of Nfeatures of INFL should either be obligatory or impossible, but not optional.
The question arises as to whether there are other functional features that may
come into play in determining the position of the subject. In Section 5.3, one
candidate for such a feature was introduced, based on the notions of Topic
and Focus. In 5.4, we considered the hypothesis that a Topic feature can attract
a subject or other constituent to the Specier of IP. This hypothesis provides
a natural account for the occurrence of constituents other than the subject in
pre-verbal position, as well as for the exibility of subject order relative to
the verb and objects. However, this hypothesis incorrectly predicts that every
clause will have exactly one Topic a prediction that is not borne out. In
Section 5.5, we considered an alternative analysis of pre-verbal subjects,
according to which they may be generated as clausal adjuncts. On this analysis, pre-verbal subjects would be subsumed under Clitic Left Dislocation
structures, which share certain properties with A-bar chains, although movement is not involved. This analysis does not require the postulation of a functional feature related to Topic status of adjuncts, simplifying the system of
functional features. The movement analysis may still be needed, however, to
account for the occurrence of pre-verbal subjects in neutral contexts where
there is no Topic in the sentence.
In 5.6, we considered the properties of verbal agreement, and the relationship between subject position and the NS parameter. Early analyses of the NS
parameter attributed a cluster of properties, including null subject pronouns
and the inversion of subjects, to lexical properties of subject agreement. We
then considered this generalization in terms of more recent assumptions as to

240

The syntax of Spanish

the nature of syntactic parameters as deriving from variations in feature

strength. Although the derivation of subjects in Spanish is compatible with an
analysis of AGR (N-features) as weak, we suggested that there is an alternative available, which can treat these features as strong, and checked by a verb
in INFL.

A-movement and
Xo movement through COMP
6.1

Introduction

In this chapter, we will examine several constructions whose derivations are standardly assumed to involve movement to CP, the highest projection of clausal structure. As the discussion will show, the claim that
operator-like phrases such as interrogatives move overtly to CP in Spanish
has been debated in recent literature. In 6.2, we discuss Wh-movement, beginning with a summary of core properties, and then turning to issues of structure particularly landing sites as discussed in recent work. Section 6.3
discusses Contrastive Focus Phrases, which have been argued to be derived
by A-movement also. We again review properties of the construction and
then the derivation, with emphasis on the landing site. Section 6.4 briey
summarizes several other phenomena that have been analyzed as involving
A-movements, although these constructions lack an overt operator-like
element, or in some cases any overt movement. This section begins with
Scrambling as discussed in Ordez (1997), then introduces three constructions that have been argued to be derived via movement of a null operator:
parasitic gaps, complex adjectivals, and null indenite objects. Section 6.5 discusses Xo movement to (and through) the head of CP.

6.2

Wh-movement

Traditionally, the rule of Wh-movement is assumed to subsume the

movement of interrogative phrases in direct and indirect questions, as well as
the movement of relative pronouns in relative clauses. Here, the properties of
Wh-movement will be illustrated for interrogatives only. For discussion of relative clauses, see Plann (1980), Suer (1984), Rivero (1990) and Brucart (1994).
In this section, we will describe several characteristics of Wh-questions. In
6.2.1, two basic properties of Wh-questions are introduced: the obligatory
241

242

The syntax of Spanish

movement of a Wh-phrase, and the obligatory inversion of a verb. We will

see how, under certain assumptions, these properties follow from the WhCriterion of Rizzi (1996). Two assumptions that underlie this analysis are rst,
that inversion involves head movement to COMP, and, second, that Whmovement involves movement of a Wh-phrase to the Specier of CP. In
Sections 6.2.2 and 6.2.3, we will see why these assumptions have been questioned in recent research. Addressing inversion rst, 6.2.2 shows, following
Suer (1994), that the verb is lower in the clause, and that an additional form
of licensing must be assumed in order to account for inversion eects. In 6.2.3,
we consider evidence that calls into question the assumption that Wh-phrases
occupy the Specier of CP. The position of Wh-phrases has been the subject
of numerous studies, and our discussion will outline the solutions to the
landing site problem. As several authors have argued, another construction
(Focus Movement) may provide evidence that bears on the choice among these
alternatives. We then turn to this matter in 6.3.

6.2.1

Wh-movement and the Wh-criterion

Two central properties of Wh-movement are, rst, that a single Whconstituent appears in clause-initial position, and, second, that the position of
the verb is restricted in certain ways. The clause-initial position of Wh-phrases
is illustrated in (2):
(1)

Juan ley ese libro.

J.
read that book

(2)

a. Qu libro ley Juan?

(Direct Question)
which book read J.
Which book did Juan read?
b. Mara no sabe [qu
libro ley Juan]. (Indirect Question)
M.
not know which book read J.
Maria doesnt know which book Juan read.

In the direct question (2a) and the indirect question (2b), the Wh-phrase qu
libro which book appears in a clause-initial position, not in canonical object
position following the verb, as in (1). Movement is necessary for the sentence
to have an ordinary interrogative reading. This is illustrated by the contrast
between (2a), where the Wh-phrase has moved, and echo-questions like
(3b):
(3)

a. Speaker A:
Mara ley el diario.
Maria read the paper.

A-movement and Xo movement through COMP

243

b. Speaker B:
(*)Mara ley qu?
M.
read what
Maria read what?

The sequence in (3b) is only possible in a context like (3a), where it echoes
the previous sentence. It could not be used in a neutral context to ask a question about what Mara read.
Restrictions on the position of the verb are similar (though not identical) to
the eects of SubjectAuxiliary Inversion in English. Examples like (4) and (5)
(from Torrego 1984) show that a subject constituent cannot generally appear
between the Wh-phrase and the verb:
(4)

a. Qu queran esos dos? (Torrego 1984:103)

What did those two want?
b. *Qu esos dos queran?

(5)

a. Con quin vendr

Juan hoy?
with whom will come J.
today
With whom will Juan come today?
b. *Con quin Juan vendr hoy?

The inversion of the verb relative to the subject has sometimes been analyzed as movement to a position outside IP, typically to Co. We will see below
that extending this analysis to Spanish is problematic in several respects.
Before we look at further data, however, let us consider how the obligatory
movement of the Wh-phrase and V-fronting have been accounted for.
Rizzi (1996) argues that the order of Wh-phrases and verbs in questions
follows from the Wh-Criterion:
(6)

The Wh-Criterion
a. A Wh-operator must be in a Spechead conguration with Xo[+Wh].
b. An Xo[+Wh] must be in a Spechead conguration with a Wh-operator.
(Rizzi 1996:64)

It is standardly assumed that a [+ ] feature (or Q feature) appears on a

clause, designating it as having a question interpretation. The Wh-Criterion
ensures that when this feature is present in a clause, a Wh-operator (a Whphrase) will occur in the Specier of the [+ ] head, licensing both the operator and the [+] head. Rizzi discusses the possibility of variation in where
the [+ ] feature occurs in the clause, a point to which we return below. For
the moment, let us assume that the relevant head is COMP. Rizzi notes that
the structure required to satisfy the Wh-Criterion is (7). In languages with
overt Wh-movement, the Wh-criterion applies at S-structure, accounting for
the appearance of a Wh-phrase in a clause-initial position. (In more recent

244

The syntax of Spanish

terms, the [+ ] feature of Co would be analyzed as strong. See Chapter 5,

Section 5.2.) The conditions of the Wh-Criterion are met if a [+ ] phrase is
in Spec of CP, and a [+ ] head is in Co.
(7)

CP

Wh-op

C [+WH]

IP

The position of the verb in interrogatives may also follow from the WhCriterion. To satisfy the Wh-Criterion, Co must have a [+ ] feature. Rizzi
proposes that in main clauses (non-lexically selected clauses), the [+ ]
feature originates in INFL, and moves to COMP via INFL-to-C movement.
The English asymmetry between embedded clauses and main clauses with
respect to inversion is suggested to follow from dierences in where the [+ ]
feature is generated. In a complement clause, [+ ] is a feature of the embedded Co. Movement of INFL-to-C is thus unnecessary.
Summarizing to this point, the obligatory movement of a Wh-phrase to CP
is necessary to satisfy the WH-criterion. The phenomenon of inversion in root
(or main) clauses is also accounted for, on the assumption that [+ ] is generated on INFL in main clauses, and on Co in complement clauses. Rizzi notes,
however, that Italian and Spanish do not exhibit this same asymmetry between
main and embedded clauses:1
(8)

a. No saba qu queran esos dos.

not knew what wanted those two
I didnt know what those two wanted.
b *No saba qu esos dos queran.
I didnt know what those two wanted.

He suggests that inversion in embedded clauses like (8) may follow from crosslinguistic dierences in where [+ ] is generated. In a language such as
English, presumably [+ ] is generated on Co, so the Wh-Criterion can be
satised in embedded clauses without movement of INFL-to-C. In Italian
however (and Spanish), the impossibility of pre-verbal subjects in questions
suggests that the [+ ] feature may be generated on INFL, rather than Co.
Then movement of INFL-to-Co would be necessary to satisfy the s-selection
requirements of the matrix verb.
1

Rizzi attributes the same pattern of inversion in embedded clauses to Romanian and
Catalan as well as Spanish and Italian.

A-movement and Xo movement through COMP

245

Summarizing, the Wh-Criterion accounts for the obligatoriness of Whmovement: a Wh-phrase must move to the Specier of a category whose head
is specied as [+]. The Wh-Criterion may also explain verb inversion.
Where the [+ ] feature is generated in INFL, INFL must move to Co to
satisfy the Wh-Criterion as well.

6.2.2

Argumental agreement licensing

The preceding discussion presupposes that the landing site for Whmovement and inversion is CP. This assumption has been questioned,
however, both for Wh-phrases and for verb inversion. Here we consider the
derived position of the verb. Suer (1994) argues that the verb is not in Co, but
is lower in the clause, in INFL. Evidence supporting this analysis is found in
the order of pre-verbal adverbs relative to the verb:
(9)

a. A

quin jams ofenderas

con tus

acciones?
(Suer 1994:345)
PA whom never oend-cond. you with your actions
Who(m) would you never oend with your actions?
b. Qu idioma todava estudia Pepita en su tiempo libre?
which language still
studies P.
in her time
free
Which language does Pepita still study in her free time?

In these examples, an adverb can intervene between the Wh-phrase in the

Specier of CP and the verb. It is generally assumed that adverbs adjoin to
XPs (phrases), not to an X such as the C in (9). (Corresponding sentences in
English are ungrammatical: *Which language still does Pepita study?) The
adverbs must therefore be adjoined to IP, and the verb must not be in Co.
A second argument against movement of V to Co derives from the fact that
certain Wh-phrases are compatible with a pre-verbal subject. This is illustrated in examples like (10) from Torrego (1984:106):
(10)

a. En qu medida la constitucin ha contribuido a eso?

In what way has the constitution contributed to that?
b. Por qu Juan quiere salir antes que los dems?
Why does Juan want to leave before the others?

If the pre-verbal subject occupies the Specier of IP (or is an IP adjunct, as

discussed in Chapter 5 (Section 5.5)) it must be that the verb has not moved
higher than INFL. If the verb can remain in INFL in interrogatives like (10)
without violating the Wh-criterion, then presumably this is also possible in
other interrogatives.
Suer (1994) has proposed a licensing relation between the verb and its
arguments to account for cases like (10). She shows that the generalization

246

The syntax of Spanish

underlying the contrast between cases like (10), which allow a pre-verbal
subject, and cases like (4), (5) and (8), which do not, concerns the status of the
interrogative phrase as an argument of the verb. Non-argument interrogative
phrases are compatible with pre-verbal subjects.2 The following generalization
then must be accounted for:
(11)

a. *Argument Wh-phrase subject Verb . . .

b. Non-argument Wh-phrase subject Verb . . .

Suer proposes that this dichotomy follows from an additional form of licensing that links the verb and its arguments:
(12)

Argumental Agreement Licensing

a. Argumental Wh-phrases must be licensed through symmetric Argagreement between (=SpecC) and (=C).
b. Arg-agrees with (=V) only if and are Arg-marked and no other
Arg-marked element is closer to .

Because the Verb is in INFL in Spanish, (12a) can be satised only if Co agrees
in features with INFL.
(13)

[ CP

Wh [

C [IP

[V+INFL] ]]]

In (13), the Wh-phrase must agree with Co to satisfy (12a). However, since Co
lacks features, agreement is satised indirectly via (12b). The Wh-phrase
agrees with V. If another argument of the verb occupies the Specier of IP,
then (12b) is not satised:
(14)

[ CP

Wh [

C [IP

DP

[V+INFL] ]]]

In (14), a DP subject occupies the Specier of IP. This phrase is an argument

that is closer to V than the Wh-phrase. Therefore, the agreement relation
between the Wh-phrase and V is blocked, and the Wh-phrase is not licensed
as an argument. If the Wh-phrase were a non-argument, (12) would not be relevant, and a pre-verbal subject would be grammatical, as in (10).
Summarizing, Suers analysis accounts for the contrast between argument
and non-argument Wh-phrases with respect to the possibility of pre-verbal
subjects. Argumental Agreement Licensing claims that pre-verbal subjects are
2

Some authors, such as Goodall (1991), have challenged this descriptive generalization, by showing that adjuncts like cundo when, and dnde where, are incompatible with pre-verbal subjects.

A-movement and Xo movement through COMP

247

ungrammatical in certain questions because they block a relation between the

Specier of CP and INFL. Thus, although the verb is lower in the structure,
inversion eects similar to INFL-to-C movement are observed. An INFLto-C analysis cannot be correct, however, given the distribution of pre-verbal
adverbs in Wh-questions, and the limited availability of pre-verbal subjects.

6.2.3

Embedded questions and the landing site for Wh-movement

In embedded questions, a Wh-phrase may appear to the right of an

overt complementizer (Rivero 1978, 1980; Plann 1982) :
(15)

a. Te preguntan que para qu quieres

el
you ask(3rd.pl.) that for what want(2nd.sg.) the
They ask you what do you want the loan for.
b. Murmur
que con quin poda
murmured(3rd.sg.) that with whom could(3rd.sg.)
He asked, by murmuring, who could he go with.

prstamo.
loan
ir.
go

This possibility is restricted to indirect questions under verbs of saying. Plann

(1982) notes that the presence of the overt complementizer under these verbs
correlates with an interpretation of the sentence as a reported question. If no
complementizer is overt, the sentence is generally interpreted as a reported
assertion.
In view of cases like (15), the position of clause-initial Wh-phrases has been
suggested to be lower than the Specier of CP. There have been several proposals as to what the derived position for Wh-phrases is. One approach has
been to examine whether the structure of CP is more elaborate or articulated
than is typically assumed. Rivero (1978) argued, for example on independent
grounds, that CP must be recursive. The Wh-phrase could then be assumed
to occupy a lower Specier of CP:3
(16)

. . . [CP [Co que [CP para quin C . . . ]]]

that
for whom

A second alternative, proposed in Goodall (1991), is that the Specier of IP

may be a landing site for Wh-movement:
(17)

. . . [CP [Co que [IP para quin V+INFL . . . ]]]

that for whom

The proposed structure (17) is consistent with the evidence discussed above
concerning the derived position of the verb, as Goodall also argues.
3

The notion that CP (then S-bar) may be recursive is introduced in Chomsky (1977).
For elaboration of this hypothesis see Rizzi and Roberts (1989) and Suer (1991).

248

The syntax of Spanish

Furthermore, assuming that subjects are generated VP-internally, the typical

inversion eects are expected if V moves to INFL over the Specier of VP:
(18)

[IP Wh-phrase [INFL V+INFL [VP Subject V . . . ]]]

The overt subject would remain in the Specier of VP at S-structure, as discussed in Chapter 5 (Section 5.2).
A third alternative as to the derived position of Wh-phrases is that they
occupy the Specier of a category between CP and IP:
(19)

. . . [CP [Co que] [XP para quin Xo [IP (Subject) [V+INFL] . . . ]]]
that
for whom

The possibility that such a category is present, and is a potential landing site
for Wh-phrases has been mentioned in several studies, primarily in relation to
the analysis of Focus Phrases, to which we turn in 6.3. Goodall (1991, 1999)
argues that, even if Wh-phrases move to the Specier of IP, as in (18), there is
also evidence for movement to a higher position. Motivation for this phrase
will be discussed below in 6.3.

6.2.4

Summary

We saw in 6.2.1 that Wh-questions have two core properties:

obligatory movement of a Wh-phrase and, in certain cases, the non-occurrence of a pre-verbal subject. These properties have been analyzed as following from the Wh-Criterion of Rizzi (1996). In 6.2.2, we saw that the
possibility for pre-verbal subjects is related to the status of the Wh-phrase:
only Wh-phrases that correspond to arguments of the verb are incompatible
with a pre-verbal subject. Suer (1994) proposes that this generalization
follows from a licensing relation (Argumental Agreement Licensing)
between the verb and its Wh-argument. This relation is blocked by the occurrence of an intervening pre-verbal subject. In 6.2.3, the co-occurrence of a
complementizer and a following Wh-phrase was discussed. The issue raised
by this phenomenon is the landing site of the Wh-phrase. Possibilities that
have been discussed in the literature include movement: (a) to the specier of
a second CP, (b) to the Specier of IP, and (c) to the specier of a category
between CP and IP.

6.3

(Contrastive) Focus

In this section we will examine clauses with an initial emphatic constituent, as illustrated in (21), based on the declarative (20):

A-movement and Xo movement through COMP

(20)

Mara compr los tomates en el mercado.

Maria bought the tomatoes at the market.

(21)

a. MARA compr los tomates en el mercado.

It was Maria who bought the tomatoes at the market.
b. En el MERCADO compr Mara los tomates.
It was at the market that Maria bought the tomatoes.

249

In (21), the initial phrase has an emphatic stress, shown by capitals. In Section
6.3.1, the properties of emphatic constituents called Focus constituents
will be summarized. Section 6.3.2 will show that Focus constituents share
certain basic properties with Wh-phrases, a fact which has led to an analysis
of Focus phrases as derived by A-movement. Section 6.3.3 considers the
landing site for Focus phrases and Wh-phrases.

6.3.1

Properties of Focus phrases

The Focus constituents in sentences like (21) above have several characteristics that distinguish them from non-emphatic pre-verbal constituents.
These are summarized in (22) (Howard 1993):
(22)

Focus Phrases:
a. contain the intonational peak of the clause;
b. license contrastive extensions;
c. produce paraphrases of cleft sentences.

Let us look rst at intonation. In non-emphatic sentences, the intonational

peak of the sentence is normally within the predicate (the rightmost phrasal
stress), as illustrated in (23):
(23)

Mara compr los tomates en el merCAdo.

Maria bought the tomatoes at the MARket.

In emphatic sentences such as (21), the constituent to the left of the verb contains the intonational peak. This shift in intonation is associated with a cleft
interpretation, and with the possibility for contrastive extensions:
(24)

MARA compr esos tomates en el mercado, no Jos.

It was Maria who bought the tomatoes at the market, not Jos.

Zubizarreta (1998) describes the emphatic stress rule of sentences like (24) as
giving rise to an interpretation in which (part of) the presupposition is reasserted or denied. For example in (24), the listeners presupposition (that it was
Jos who bought the tomatoes) is denied by the speaker.4 In the literature on
4

This interpretation corresponds to the Contrastive Focus discourse role in Comrie

(1989).

250

The syntax of Spanish

Spanish, this construction has been referred to under various names, such as
Topicalization, Rhematization, Informational Focus, Focus Fronting
and Focus Movement. We will refer to it as Focus Movement.

6.3.2

Movement properties

We turn now to properties of emphatic sentences which provide evidence as to the structure and derivation of the Focus constituent. As shown
in Hernanz and Brucart (1987), Focus phrases pattern with Wh-phrases in
crucial respects. This implies that Focus phrases, like Wh-phrases, are derived
by A-movement (movement to a non-argument position).
Like Wh-movement, Focus movement triggers subjectverb inversion, as
shown by the contrast in (25):
(25)

a. ESE captulo ley por completo Josena.

that chapter read for complete J.
It was that chapter that Josena read completely.
b. *ESE captulo Josena ley por completo.=(25a)
that chapter J.
read for complete

Recall from Section 6.2 that inversion eects have been analyzed as due to the
requirements of licensing criteria: the Wh-Criterion and Argumental
Agreement Licensing.5 If Focus phrases were base-generated in initial position as clausal adjuncts, they would not be expected to trigger inversion.
Focus constituents also show other properties characteristic of Amovement. Like Wh-phrases, a Focus constituent cannot be linked to a position within an island such as a relative clause:
(26)

a. *ESE poema conozco a la mujer que escribi.

It is that poem that I know the woman who wrote.
b. *Qu poema conoces a la mujer que escribi?
Which poem do you know the woman who wrote?

In (26a), the phrase ESE poema that poem is the object of the verb escribi.
The ungrammaticality of (26a) can be ascribed to the impossibility of moving
any constituent out of the relative clause, which acts like an island for movement, as is shown in (26b) for a Wh-constituent.
Another parallel is the absence of clitic doubling for Focus phrases and for
Wh-phrases:
5

Within the Minimalist Program, these licensing requirements are subsumed under
the feature-checking requirements of functional categories, as was discussed in
Chapter 5 (Section 5.2).

A-movement and Xo movement through COMP

(27)

251

a. ESOS tomates (*los) compr Mara.

It was those tomatoes that Maria bought (*them).
b. Cules tomates (*los) compr Mara.
Which tomatoes did Maria buy (*them)?

Neither movement admits two fronted constituents in a clause:

(28)

a. *Cundo qu compr Juan?

When what did Juan buy?
b. *AYER los TOMATES compr Juan.
It was yesterday the tomatoes that Juan bought.

Furthermore, a Wh-phrase and a Focus phrase cannot be fronted in the same

clause, regardless of order:
(29)

a. *Cundo las MANZANAS compraron?

When was it the apples that they bought?
b. *Las MANZANAS cundo compraron?
It was the apples when that they bought?

Both Wh-phrases and Focus phrases can be preceded by a dislocated Topic

constituent in a main clause:
(30)

a. En octubre, POCAS manzanas compraron.

In October, few apples they bought.
b. En octubre, qu compraron?
In October, what did they buy?

Finally, both Wh-movement and Focus movement can appear in initial

position of complement clauses only if the fronted constituent is compatible
with the semantic selection of the matrix predicate. Wh-movement is only possible in clauses that can s-select indirect questions, as shown by the contrast in
(31):
(31)

a. Mara se pregunta [qu compraron en agosto].

Maria wonders [what they bought in August].
b. *Mara anunci [qu compraron en agosto].
Maria announced [what they bought in August].

Sentence (31a) is grammatical as an indirect question; (31b) is not.6 Example

(31a) admits a question interpretation because the verb preguntarse wonder
semantically selects as a complement a clause that contains an interrogative
([+ ]) feature. The verb anunciar announce, on the other hand, does not.
In complement clauses then, interrogatives are only possible if they are consistent with lexical properties of the selecting predicate. Focus movement is
6

Example (31b) is grammatical as a declarative with a free relative complement, i.e.,

Maria announced what they bought in August.

252

The syntax of Spanish

also restricted, although it is not restricted by the same feature, as noted in

Howard (1993). Focus movement is restricted according to whether the
embedded clause is assertive or presupposed (examples from Contreras
(1978)):
(32)

a. Dice que MAANA lo operan.

He says that its tomorrow that theyre operating on him.
b. *Siento que MAANA lo operen.
I regret that its tomorrow that theyre operating on him.

The verb decir say of the main clause in (32a) is an assertive verb in the
sense that it asserts the propositional content of its complement clause. The
factive verb sentir regret in (32b) is non-assertive. It presupposes the content
of its complement, rather than asserting it. In the latter context, Focus movement is ungrammatical. Although it is not clear precisely what feature is
present in assertive contexts that is absent in (32b), the contrast in (32) implies
that semantic selection pertains for Focus phrases, as it does for interrogatives.
Summarizing, we have seen that Focus movement has properties quite close
to those of Wh-movement. The only dierence between the two concerns sselection in embedded clauses. Wh-movement is s-selected by a [+ ] feature,
and Focus movement is not, although it does show evidence of s-selection by
some feature, whatever that may turn out to be.

6.3.3

Landing site

Based on similarities between Wh-movement and Focus constructions, it has been assumed that Focus constituents are derived by movement
(rather than by base-generation as pre-clausal adjuncts, for example). Some
authors (Hernanz and Brucart 1987; Campos and Zampini 1990) have analyzed both Wh-movement and Focus movement as A-movement to the
Specier of CP, which triggers verb fronting from INFL to the
Complementizer:
(33)

[CP Las MANZANASj [C comprk [IP Marai [I tk [VP ti tk tj ]]]]]

On this analysis, all constituents lower in the structure than CP, including the
subject and any IP adjuncts, follow the verb, which has moved to COMP. The
only constituents that could precede the verb are the Focus constituent in the
Specier of CP and any CP adjuncts. Dislocated Topics, for example, precede
the Focus constituent:
(34)

[CP En agosto, [CP Las MANZANASj [C comprk [IP Marai [I tk [VP ti tk tj

]]]]]]
In August, it was the apples that Maria bought.

A-movement and Xo movement through COMP

253

As with Wh-movement, recent studies have argued that the S-structure

landing site for Focus movement is lower in clause structure than CP. One
argument in support of this analysis is that in embedded clauses, the fronted
constituent a Focus constituent or a Wh-phrase appears to the right of the
complementizer:
(35)

a. Dice [CP que [ MAANA lo operan]].

He says that its tomorrow that theyre operating on him.
b. Me preguntaron [(que)[ a quin invitas]].
They asked me (that) whom youre inviting.

The structure assigned to sequences like those in (35) varies according to a

number of other assumptions: in particular, the articulation of the VP-related
functional projections in IP. In Chapter 4, the projections discussed included
Aspect Phrase, with associated Agr-o; Tense, with associated Agr-s (=IP); and
Neg Phrase:7
(36)

NegP IP (Aux) AspP VP

The landing site of Focus movement must be at least as high as NegP, since
the fronted constituent precedes no:
(37)

Dice que MAANA no lo operan.

He says that its not tomorrow that theyre operating on him.

Notice as well that a fronted Focus constituent must precede an n-word in the
Specier of NegP:
(38)

a. Dice que nunca lo operaron.

He says that never did they operate on him.
b. Dice que en ese HOSPITAL (no) lo operaron.
He says that its (not) in that hospital that they operated on him.
c. *Dice que nunca en ese HOSPITAL lo operaron.
He says that never is it in that hospital that they operated on him.
d. (?)Dice que en ese HOSPITAL nunca lo operaron.
He says that its in that hospital that they never operated on him.

The ungrammaticality of (38c) shows that the Focus constituent is not adjoined
to Neg between the head and a constituent in the Specier. The grammaticality of (38d) shows that the Focus constituent is higher than NegP, either
adjoined to it, or in the specier of a higher functional category between NegP
and CP. Between these alternatives, the evidence seems to point to the latter: if
the Focus constituent were adjoined, it would be expected that several Focus
constituents should be possible, since adjunction is generally not restricted to
7

Recall from Chapter 4 (Section 4.5) that, since pre-verbal subjects precede Negation,
Agr-s may be higher than Neg. This issue is left open here.

254

The syntax of Spanish

a single constituent. Furthermore, if the Focus constituent were adjoined, it

would not be expected that movement should be restricted to clauses with particular features s-selected by the matrix predicate. Finally, the adjunction analysis does not capture the similarities between Wh-movement and Focus
movement, since Wh-movement is uncontroversially movement to a Specier
position. These factors lead to the conclusion that there is a phrase between
NegP and Comp, whose functional features are related to Focus constituents:
(39)

COMP Foc Neg INFL (Aux) Aspect VP

Focus movement is therefore movement to the specier of FocP.8 This

hypothesis is consistent with the properties observed: only one constituent can
be fronted in a clause (like Wh-phrases and n-words Chapter 4, Section 4.5)
and fronting is possible in complement clauses only if s-selected. As for the
impossibility of both Wh-movement and Focus movement in the same clause,
the fact that WH-constituents appear to the right of Comp in certain embedded clauses suggests that Wh-phrases can appear at S-structure in the Specier
of FP if FP is specied for [+] features.

6.3.4

Summary

It was shown above that Focus movement shares properties with Whmovement, which suggests that it is derived by movement rather than basegeneration. Like Wh-movement, Focus movement has been analyzed in recent
studies as movement to a position lower than the Specier of CP, since Focus
phrases can appear to the right of a complementizer. There is some evidence
which suggests that the Focus phrase is above the IP. Since Focus constituents
can (at least marginally) precede preposed n-phrases, they must be either
adjoined to NegP or in the Specier of a higher functional category, FP
between Comp and NegP. The adjunction analysis does not capture straightforwardly the uniqueness of the Focus constituent, its similarities to Whmovement, or the fact that it is s-selected by some feature related to the
assertive value of the clause. The FP hypothesis accounts for these properties straightforwardly.

6.4

Other A-movements

This section will briey introduce four additional constructions that

have been analyzed as involving A-movements.
8

Uriagereka (1995) argues for the existence of a Focus Phrase based on the occurrence
of a Focus particle in Western Romance.

A-movement and Xo movement through COMP

6.4.1

255

Scrambling

Recall from Chapter 5 (Section 5.6) that VP-nal subjects (free

subject inversion) in Italian and Spanish are analyzed in Rizzi (1982) as
deriving from rightward movement of the subject to a VP-adjoined position:
(40)

a. Compr el diccionario Juan.

bought the dictionary J.
Juan bought the dictionary.
b. [IP ti INFL [VP [VP compr el diccionario] Juani ]]

In subsequent years, several theoretical developments have occurred which

call this analysis into question. Among these is Kaynes (1994) hypothesis that
excludes rightward movements altogether. If correct, this hypothesis implies
that V-O-S order must be derived from S-V-O order by leftward movement:
(41)

[V

Obj i [S [tv t i ]]]

Ordez (1997) argues that this reordering in Spanish is a form of movement

referred to as Scrambling. On this analysis, (40a) would be derived as shown
in (42):
(42)

[compri [el diccionarioj ti [VP Juan ti tj]]]

In (42), the verb has moved from its base position through Agr-o, and then to
Tense (V-to-INFL movement). The object, el diccionario, has also moved, to
a position outside VP, so that its surface position is left of the subject. (This
position might be either an adjoined position or the Specier of a functional
category below Tense.) The subject remains in its base position in the Specier
of VP (see Chapter 5, Section 5.2).
Ordez presents several types of evidence in favor of the analysis illustrated above. One piece of evidence concerns the hierarchical relationship
between the moved object and the subject. The Scrambling analysis, unlike the
rightward movement analysis, claims that the complement c-commands the
subject after movement. Another type of evidence concerns the relationship
between Scrambling and Wh-movement. He shows that there are restrictions
on the order of complements in multiple questions. As shown in (43b) and
(44b), a post-verbal WH-subject must precede a WH-complement in VP
(examples from Ordez 1997:52):
(43)

a. Qu le compr quin a quin?

what CL-bought who for whom
What did who buy for whom?

256

The syntax of Spanish

b. *Qu le compr a quin quin?
what CL-bought for whom who
What did for whom buy who?

(44)

a. Qu dijo quin de quin?

what said who of whom
What did who say of whom?
b. *Qu dijo de quin quin?
what said of whom who
What did of whom say who?

The judgments shown above reect multiple question readings, rather than
echo-questions. In the ungrammatical (b) examples, a Wh-phrase may be analyzed as having moved leftward via Scrambling. Ordez notes that Whphrases have been shown to resist Scrambling in certain languages (e.g.,
German). The word order patterns in (43) and (44) may therefore have a
natural explanation under a Scrambling analysis. This suggestion is supported
by the fact that the (b) examples cannot be excluded in terms of a more general
ban on movement of WH-complements across other Wh-phrases. If the
movement is Wh-movement, rather than Scrambling, no violation ensues.
Ordez cites cases from Jaeggli (1982):
(45)

a. Quin dijo qu?

Who said what?
b. Qu dijo quin?
What did who say?

In (45b), the complement Wh-phrase, qu what, has moved over the subject.
In this instance, the object has moved to a clause-initial position. The grammaticality of (45b) compared with (43b) and (44b), suggests that the latter are
excluded by some condition that is specic to Scrambling.9
Summarizing, it has been proposed that the order V-O-S is derived via (leftward) A-movement of the object. This analysis claims that the object is structurally higher than the subject, which remains in its initial VP-internal
position. Ordez (1997) argues that this analysis may account for restrictions
on the order of non-clause-initial WH-phrases in multiple questions.
9

The English equivalent of (45b) is ungrammatical:

(i)

*What did who say?

This sentence involves movement of the object across the WH-subject, and then,
after S-structure, the subject WH-phrase would move across the object. It is the
second, covert movement that produces a violation (compare: What did she say?).
This type of violation is referred to descriptively as a Superiority violation. As
Ordez notes, the grammaticality of (45) in the text shows that (43b) and (44b)
cannot be attributed directly to Superiority.

A-movement and Xo movement through COMP

6.4.2

257

Parasitic gaps

It has been observed that A-movement can under certain circumstances allow a second gap to appear. Consider the contrast between (46a) and
(46b, c):
(46)

a. Qu libro archivaste [sin

leer ]?
which book led
without read-inf.
Which book did (you) le without reading?
b. *Archivaste ese libro [sin
leer ].
led
that book without read-inf.
(You) led that book without reading.
c. *Ese libro parece haber
sido archivado [sin
leer ].
that book seems have-inf. been led
without read-inf.
That book seems to have been led without reading.

In each of these sentences, the adjunct clause is missing the object of the verb
leer read. The gap in the adjunct clause has been referred to as a parasitic gap, because its presence is dependent on the occurrence of an A-chain
in the main clause. That is, the parasitic gap in (46a) is in some way legitimated
by the presence of the (Wh-phrase-trace) chain in the main clause. This is supported by the interpretation of (46a), where the gap takes its reference from
the Wh-phrase, as well as by the ungrammaticality of (46b), where the main
clause does not contain an A-chain. In (46c), the object of the participle archivado led has undergone movement, but in this case the movement is Amovement that is, movement to an argument position: the matrix subject
position. The ungrammaticality of (46c) shows that A-chains do not license
parasitic gaps.
Chomsky (1982) argues that parasitic gaps are traces left by the movement
of a phonetically null operator:
(47)

Qu libroj archivaste tj [sin [Opj [PRO

leer ej ]]]?

which book

read-inf.

filed

without

Which book did (you) file without reading?

Evidence that supports an analysis of null operator movement for the parasitic gap is based on the relationship between the trace (the parasitic gap) and
the CP that, by hypothesis, contains the null operator. As in other cases of
A-movement, the parasitic gap cannot be contained within a syntactic
island:

258

The syntax of Spanish

(48)

*Qu libroi archivaste ti sin [Opj preguntarte cundo leer ej]?

which book led
without wonder
when read-inf.
Which book did you le without wondering when to read?

In (48), the trace of the null operator is contained within a Wh-island: a CP

whose specier is occupied by a WH-constituent. Since this position is already
occupied, the null operator cannot have moved through it. The null operator
must have moved in one step across the occupied CP:
(49)

*Qu libroj archivaste tj sin [Opj PRO preguntarte [cundo leer ej]]?

The preceding observations imply that a null operator should be able to

appear in a higher CP such as in (49) only if the operator has moved successively through lower CPs (successive cyclically). It has been noted, however,
that Spanish parasitic gaps disallow even this type of successive movement, as
is illustrated by the contrast between (50a) and (50b) (examples from Garca
Mayo and Kempchinsky (1994)):
(50)

a. Which articles did you put on reserve without convincing the students
to read?
b. *Qu artculos pusiste en reserva sin
convencer
a los
estudiantes de leer?
(=50a)
which articles put
on reserve without convince-inf. PA the
students
of read-inf.

In descriptive terms, null operator movement appears to be restricted to the

CP of the clause containing the parasitic gap. This relation is more local than
is generally the case for A-movements. Wh-phrases, for example, can move
successive-cyclically:
(51)

Qu artculo dijiste que crey Mara que insisti Pedro en que leyera
Susana?
Which article did you say that Maria believed that Pedro insisted that
Susana read?

In (51) the Wh-phrase, qu artculo which article, has moved from its base
position as the complement of leyera read, in the most deeply embedded
clause, through the specier of CPs of intermediate clauses, to its derived position in the matrix clause. Null operator movement is thus more restricted than
Wh-movement, although the two share properties of chains produced by
movement.10
10

There are a number of additional constraints on parasitic gap constructions, and a

number of aspects of the analysis of null operators that have been omitted here. For
further discussion, see Bordelois (1986) and Garca Mayo and Kempchinsky (1994),
and references cited.

A-movement and Xo movement through COMP

259

Summarizing, parasitic gaps have been argued to be derived by movement

of a null operator, creating an A-chain that is interpreted in relation to
another A-chain, such as that produced by Wh-movement. The null operator
chain exhibits the properties typical of A-movement, but is more restricted
still: the operator and its trace are restricted to the same clause.

6.4.3

Complex adjectivals

A second construction that has been analyzed as involving null operator movement is illustrated by the pairs in (52)(54) (examples from Aissen
and Perlmutter (1976:14)):
(52)

a. Ser difcil componer estas radios.

It will be dicult to x these radios.
b. Estas radios sern difciles de componer.
These radios will be dicult to x.

(53)

a. Es fcil entender los resultados.

It is easy to understand the results.
b. Los resultados son fciles de entender.
The results are easy to understand.

(54)

a. Fue imposible comer el postre.

It was impossible to eat the dessert.
b. El postre fue imposible de comer.
The dessert was impossible to eat.

In the (b) sentences above, the logical object of the innitive appears in position of the matrix subject. Aissen and Perlmutter (1976) show that the preposed NP is a clausal subject, not a topic or other adjunct. Their conclusion
is based on its subject-like behavior: the fact that it can be a null subject;
that it triggers subjectverb agreement in the matrix clause; and that it can
undergo further NP movement in raising contexts:
(55)

Estas radiosi parecen [ti ser difciles de componer].

These radios seem to be dicult to x.

These facts indicate that the initial NP in sentences like (52b)(54b) is the
subject of the main clause. This NP is also interpreted as the complement of
the verb of the embedded clause, which implies that an empty category in that
position is assigned a Theta-role by the verb:
(56)

Estas radios sern difciles [de componer e].

A central issue raised by this construction is that the preposed NP could not
have moved from the embedded object position directly to the matrix subject
position without violating principles that constrain such movement. In particular, the trace left by this movement would violate Binding Principle A, which

260

The syntax of Spanish

requires that certain traces (such as the trace of a moved NP) be bound by an
antecedent within a local environment not in the next higher clause. The
empty category in (56) does not have an antecedent that is structurally close
enough to satisfy Principle A. Furthermore, there is no way in which the movement could have occurred in a sequence of steps so that the Binding requirement could be satised. A movement such as that shown in (57) is not
consistent with the interpretation:
(57)

Estas radiosi sern difciles [de ti componer ti].

In (57), the NP estas radios has moved from object position of the embedded
clause to subject position, then raised to subject position of the matrix clause.
Each step in this derivation would be an admissible movement. However, this
derivation cannot be correct, since the subject of the embedded clause is not
interpreted as estas radios, but as an arbitrary person or persons, i.e., a null
PRO:
(58)

Estas radiosi sern difciles [de PRO componer ti].

These radios will be dicult (for one) to x.

The question which remains, then, is how the preposed NP is associated with
the position in which its Theta-role is assigned.
Chomsky (1981) proposes that the solution to this problem lies in the nature
of the movement. Rather than NP movement to the matrix subject position,
it appears that the embedded clause object is a null pronominal that undergoes movement to the specier of the embedded CP:
(59)

Estas radiosi sern difciles [CP PROi de [PRO componer ti]].

In (59), the null pronoun (which has an index dierent from that of the innitival subject), moves to the specier of CP, and is coindexed with the NP estas
radios. This proposal is problematic with respect to the NP estas radios, which
must be inserted in a position that is not assigned a Theta-role. Since it is not
related by movement to the null pronoun or its trace in the embedded clause,
the derivation should fail the Theta-Criterion already at D-Structure. To avoid
this violation, Chomsky suggests that the NP is not present at D-structure, but
is inserted in the course of the derivation to S-structure.
The hypothesis that complex adjectivals involve A-movement, and not
simple Raising (or A-movement) is supported by similarities between the
chain formed by movement of PRO to CP, and chains formed by other Amovements. One similarity is that both of these chains license parasitic gaps.
Recall from Section 6.4.2 that parasitic gaps are possible only if the null operator can be interpreted in relation to an overt operator. This is shown by the
contrast between (46a) and (46c), repeated below:

A-movement and Xo movement through COMP

(46)

261

a. Qu libro archivaste [sin

leer ]?
which book led
without read-inf.
Which book did (you) le without reading?
c. *Ese libro parece haber
sido archivado [sin
leer ].
that book seems have-inf. been led
without read-inf.
That book seems to have been led without reading.

In (46c), A-movement of the complement ese libro to the matrix subject position does not license a parasitic gap, unlike Wh-movement in (46a). Returning
to complex adjectivals, the movement shown in (59) is supported by its ability
to license a parasitic gap. Compare the complex adjectival in (60a) with Amovement in (60b):
(60)

a. Este problema es fcil de resolver

sin
examinar
detenidamente.
this problem is easy of resolve-inf. without examine-inf.
carefully
This problem is easy to solve without examining carefully.
b. *Este problema parece haber
sido resuelto sin
examinar
detenidamente.
this problem seems have-inf. been solved without examine-inf.
carefully
This problem seems to have been solved without examining carefully.

In (60a), the complement of examinar is a well-formed parasitic gap. If it is

correct that parasitic gaps are generally licensed by an A-antecedent, then
(60a) must have a null operator in the specier of the intermediate clause. By
contrast, (60b) shows that the occurrence of movement is not sucient to
license a parasitic gap, if the chain produced by movement occupies an Aposition, rather than an A-position.
A second argument supporting the analysis of complex adjectivals as
involving A-movement is that, like parasitic gap constructions, Spanish
complex adjectivals allow only clause-bounded A-movement:
(61)

*Ese libro es difcil de convencer

a los estudiantes de leer.
that book is dicult of convince-inf. PA the students
of read-inf.
That book is dicult to convince the students to read.

Whatever the explanation for this stricter form of locality, the fact that both
constructions involving A-movement of a null operator are restricted in the
same way supports a unied treatment.
Summarizing, complex adjectivals show evidence of A- movement of a null
constituent, one which takes an antecedent in a higher clause. The antecedent
is, in these cases, the subject of the matrix clause. Support for a movement
analysis derives from (a) the fact that the construction licenses parasitic gaps,

262

The syntax of Spanish

and (b) that the null operator movement shows the same highly local character as is displayed by null operator movement in parasitic gap constructions.

6.4.4

Indenite null objects

Null objects are not generally grammatical in Spanish, as the following examples (from Campos (1986)) illustrate:
(62)

a. Compr un/el libro.

I bought a/the book.
b. Lo compr.
CL(DO) bought
I bought it.
c. *Compr.
I bought.

In a context such as (63), a null object is grammatical, if it is interpreted as

indenite:
(63)

a. Compraste caf?
You bought coee?
b. S, compr.
Yes, I bought (some).

As Campos (1986) notes, (63b) is not possible in Romance languages that have
overt partitive clitics, including French, Catalan and Italian. Portuguese and
Spanish, however, do not have partitive clitics, and allow null objects.
Following Raposo (1984), Campos argues that sentences like (63) involve
movement of a null operator.
Evidence for the null operator analysis derives from the fact that the construction observes the constraints that typically apply to movement. Examples
such as (64) illustrate that where the null object is contained within an island,
it is ungrammatical:
(64)

a. Juan traer cerveza a la esta?

Will Juan bring beer to the party?
b. Su novia me dijo que traera.
His girlfriend told me that he would bring (some).
c. *Existe el rumor de que traer.
There is the rumor that he will bring (some).

Example (64a) sets the discourse context for the null object. In (64b), the null
object is contained in a complement clause. If the derivation involves movement through CP, (64b) is grammatical because the specier of the embedded
CP is available as an intermediate landing site for the operator:

A-movement and Xo movement through COMP

(65)

263

[CP Opi [IP pro dijo [CP ti que [IP pro traera ti]]]

From this position, the operator undergoes further movement to the Specier
of the matrix CP. Note that this movement (unlike the null operator of parasitic gap and complex adjectival constructions) is not clause-bounded.
Movement to the higher CP is possible. This step in the derivation is supported
by the contrast between (64b) and (64c). In the latter example, the complex NP
makes further movement to the matrix CP impossible:
(66)

*[CP

Op i [IP pro existe [DP el rumor de [CP t i que traer t i ]]]].

The observance of constraints on movement leads Campos to conclude that

null objects are derived by A-movement of a null operator. The question
remains as to why this null operator construction is not clause-bound, as are
parasitic gaps and complex adjectivals.

6.4.5

Summary

In this section, we have reviewed certain constructions that have been

analyzed as involving A-movement. These include Scrambling, which moves
an object to the left of a subject; and those constructions which have been
argued to be derived by movement of a null operator: parasitic gaps, complex
adjectivals, and null objects. Although we have only sampled the evidence
underlying these analyses, we nevertheless have seen similarities in their properties. In particular, all these constructions have been argued to involve movement, based in part on the observation that general constraints on movement
are apparent.

6.5

Head movement to (and through) COMP

In this section, we return to the topic of head movement. It has been

noted previously that, while V-to-INFL movement is characteristic of declaratives (Chapter 4, Sections 4.2 and 4.3) the evidence of non-declaratives indicates that I-to-C movement does not take place in overt syntax (Section 6.2.2).
There are, however, several clause types that have been argued to involve overt
movement of heads to (or through) COMP. These include several types of
non-nite clauses where V-to-C movement is in evidence (6.5.1) and clauses in
which clitics climb to a higher clause, moving through COMP (6.5.2).

264

The syntax of Spanish

6.5.1

V-to-INFL-to-C movement

It has been argued (Rivero 1994; Rooryck 1992; Belletti 1995) that
certain imperative clauses are derived via movement of V-to-INFL and then
I-to-C. This argument has been made for what Rivero (1994) terms true imperatives. Rivero distinguishes between true imperative constructions and surrogate imperative constructions. True imperatives are constructions with verb
forms identiable by a morphology not shared by the same person in any other
tense in the system. True imperatives are usually only 2nd person. Surrogate
imperatives are morphologically identical to the same person of an existing
tense, usually a present or an innitive, and are not restricted as to person. For
example, compare the true imperative canta sing-I-2nd.sg. Sing! and No
cantes not sing-pr-subj.-2nd.sg. Do not sing. The latter is a surrogate imperative, as it uses the morphology of an existing tense, the present subjunctive.
Several properties of true imperatives are accounted for naturally if the derivation involves overt I-to-C movement. Cross-linguistically, it is quite
common for clitics to be positioned after the verb. Compare the true imperative in (67) with the surrogate imperative in (68):11
(67)

a. Hazlo!
do-I.2nd.sg.+CL(DO)
Do it!
b. *Lohaz!

(68)

a. *No hgaslo!
not do-pr.subj.2nd.sg.
Dont do it!
b. No lo hagas!

Clitic order might be accounted for in several ways, depending on assumptions

as to how clitics are generated and moved (see Chapter 4, Section 4.4). On the
assumption that clitics occupy Agro (either by being generated there or by
movement), the order in (67) is derived by moving the verb rst to Tense and
then to C:
(69)

COMP

Agr

TENSE Agr-o (CL) [VP V]

In (69), the verb moves rst to TENSE, bypassing Agr-o, which is occupied
by object clitics. Tense then moves to COMP. Rivero (1994) assumes that
11

Rooryck (1992) notes that encliticization occurs with imperatives even in languages
which otherwise lack encliticization, such as French.

A-movement and Xo movement through COMP

265

imperatives are non-nite, so that Agr-s is inert. She assumes also that what
triggers movement of I-to-C is a feature in COMP that is associated with an
imperative operator.
A second property of true imperatives is that they are incompatible with
negation:
(70)

a. Hazlo!
do-I.2nd.sg.+CL(DO)
Do it!
b. *No hazlo!
not do-I.2nd.sg.+CL(DO)
Dont do it!
c. No lo
hagas!
not CL(DO) do-pr.subj.2nd.sg.
Dont do it!

Rivero argues that the negative head no blocks movement of the imperative
form. This blocking eect can be attributed to the operator-like status of
negative no. Neg acts like a closer antecedent for the trace of the moved verb:
(71)

[C [TNS+V] i + COP][Neg OP t i ]

The Neg head therefore blocks the antecedenttrace relationship between the
verb in COMP and its trace.
Summarizing to this point, true imperative constructions have two properties which appear to be consistent with head movement to COMP. One is the
broad cross-linguistic phenomenon of imperative encliticization, which is suggestive of V-movement. Second, the incompatibility of true imperatives with
negation is accounted for on the assumption that I-to-C movement is necessary, since Neg blocks further movement.
Other types of non-nite clauses have been suggested to be derived via I-toC movement. As noted in Chapter 1 (Sections 1.6.41.6.5), various classes of
non-nite adjunct clauses disallow pre-verbal subjects:
(72)

a. habiendo terminado la reunin

have-prt. nish-pprt. the meeting
having nished the meeting
b. *la reunin habiendo terminado

(73)

a. de venir Mara
of come-inf. M.
if Maria comes
b. *de Mara venir

266

The syntax of Spanish

(74)

a. terminada
la reunin
nish-pprt.f.sg. the meeting
the meeting nished
b. *la reunin terminada

Under the assumption that movement is purely optional, this constituent order
generalization would be suggestive of I-to-C movement, since nothing would
block the occurrence of the subject in the Specier of IP. Under more recent
assumptions that were discussed in Chapters 4 and 5, however, the impossibility of a pre-verbal subject is not necessarily indicative of I-to-C movement,
since the subject cannot always appear pre-verbally, as discussed above in 6.2.2.
The diagnostics mentioned above can shed light on the position of the verb
in these clauses: if I-to-C applies, they are expected (a) to show encliticization,
and (b) to disallow negation. Gerundive and innitival clauses, like non-nite
verbs in general, have enclitics rather than pro-clitics:
(75)

a. habindolo
terminado
have-prt.+CL(DO) nish-pprt.
having nished it
b. *lo habiendo terminado

(76)

a. de leerlo
Juan
of read-inf.+CL(DO) J.
if Juan reads it
b. *de lo leer Juan

These clauses do, however, admit negation:

(77)

(78)

a. No habiendo terminado la reunin, me qued en la ocina.

Not having nished the meeting, I stayed at the oce.
b. No estando t en la sala, habl con Susana.
You not being in the room, I talked with Susana.
a. Al no encontrar el artculo, me desesper.
On not nding the article, I panicked.
b. De no venir Mara, no hay esta.
If Maria doesnt come, there wont be a party.

It appears that in these clauses, the INFL containing V does not move overtly
to COMP, since this movement would be blocked by an intervening negative
head, as in (71) above.
Absolute (participial) clauses, however, disallow both object clitics, as in
(79), and negation, as in (80):12
12

As noted in Belletti (1990:94 .), the Italian counterparts of these participial clauses
allow object clitics, but disallow negation.

A-movement and Xo movement through COMP

(79)

a. publicado
el artculo
publish-pprt. the article
published the article
b. *publicdolo
publish-pprt.+CL(DO)
it published

(80)

a. *No terminada
la reunin, me qued
en la ocina.
not nish-pprt.f. the meeting.f. CL remained at the oce
With the meeting not nished, I remained at the oce.
b. *No vendido el coche, tena poco dinero.
not sell-pprt. the car
had little money
With the car not sold, I had little money.

267

The ungrammaticality of object clitics whatever its source might be does

not provide evidence for the verbs position. Recall that encliticization is suggestive that the verb has moved to I and then I has moved to C. Evidence from
order of the verb and clitics is thus unavailable in this case. Negation, however,
does provide evidence. As shown in (80), participial clauses do not accept
negation. In this respect, these clauses dier from participles used as adjectives, which do allow negation (e.g. un problema no resuelto a problem not
resolved). Participial clauses thus pattern with true imperatives, and may be
derived via I-to-C movement.
Summarizing, it has been proposed that clitic order and the impossibility of
negation provide evidence for the position of V in a clause. Based on these
diagnostics, there are two types of clauses that appear to be derived via V-toINFL and subsequent I-to-C movement: true imperatives and participial
clauses. A broader class of clauses (including gerundives and innitival
adjuncts) disallow pre-verbal subjects. Assuming that movement is impossible
unless it is triggered by the presence of a strong feature, the subject may remain
within VP unless some strong feature requires that it move. If the subject is
within VP in these non-nite clauses, these cases provide no evidence as to
whether I-to-C movement has applied.

6.5.2

Clitic Climbing

It was noted in Chapter 4 (Section 4.4) that clitics can sometimes

occur in construction with the verb of a higher clause, as in (81):
(81)

Mara lo
quiere comprar.
M.
CL(DO) wants buy-inf.
Maria wants to buy it.

268

The syntax of Spanish

In (81), lo it is understood as the complement of the innitive comprar

buy. However, the clitic precedes the verb quiere wants in the matrix
clause. This phenomenon is referred to informally as Clitic Climbing,
because the clitics climb out of the clause in which they are interpreted.
Because Romance clitics generally do not behave like full phrases, they are not
expected to undergo the types of movement that phrases undergo. In fact,
Clitic Climbing is fairly restricted. It occurs if the clitic originates in a clause
that is the complement of a particular class of trigger verbs. Other, nontrigger verbs do not allow clitic climbing.13 The verb parecer seem, for
example, is a raising verb, but is not a Clitic-Climbing trigger:
(82)

a. Mara parece saberlo.

M.
seems know-inf.+CL(DO)
Maria seems to know it.
b. *Mara lo parece saber.

The unusual properties of Clitic Climbing as a movement process have led to

various proposals.14 Kayne (1989) argues that Clitic Climbing is derived via
movement of clitics through COMP. That is, under certain conditions, clitics
can undergo head movement independently of a verb. The crucial steps in the
derivation involve movement of the clitic to INFL, and I-to-C movement. The
derivation of (81) would begin as shown in (83):
(83)

a. Mara quiere [CP Co [IP PRO INFL [VP comprar lo]]].

b. Mara quiere [CP Co [IP PRO lo +INFL [VP comprar t]]].
c. Mara quiere [CP lo +INFL+Co [IP PRO t [VP comprar t]]]

In (83a), the clitic is shown in the position corresponding to the complement

of comprar. In (83b), lo moves to INFL, while the innitive remains within
VP. In (83c), I-to-C movement carries the clitic to COMP. Once the clitic occupies this clause-initial position, it can cliticize to the matrix verb.
Kayne (1989) presents evidence supporting the steps shown in (83b) and
(83c). With regard to movement of the clitic to INFL even in the absence of
V-to-INFL movement Kayne notes that, although Romance clitics generally
attach to V, they do not always do so, as is illustrated by examples like (84):
13
14

For discussion of the class of trigger verbs, see Aissen and Perlmutter (1976).
Several studies have pursued the notion that Clitic Climbing is possible only if the
complement clause has undergone structural reanalysis, so that the matrix and complement clauses are merged into a single clause. Other studies have argued that the
trigger verbs are light verbs or auxiliary verbs, which select a VP complement
rather than a true clause. For discussion of these approaches, see Aissen and
Perlmutter (1976), Rizzi (1982), Kayne (1989), Moore (1991) and references cited.

A-movement and Xo movement through COMP

(84)

(*) Jean a promis

de
les bien faire.
J.
has promised for/to CL well do
Jean has promised to do them well.

269

(Kayne 1989:240)

In (84), the clitic les them is separated from the following innitive by an
adverb. He notes that while (84) is ungrammatical in Modern French, it
existed in earlier stages of French and is attested in other dialects. In such
cases, the clitic can be analyzed as having climbed to the INFL node of the
innitive. A further point with regard to movement of clitics to INFL concerns the range of languages in which Clitic Climbing is found. Kayne proposes that there is a correlation between the admissibility of Clitic Climbing
and of null subjects. Stated informally, the idea is that, in NS languages, INFL
is strong enough as a governor of VP for clitics to escape VP, moving to INFL,
independently.15 This accounts for why Clitic Climbing is common to the
many Romance languages which allow null subjects, including earlier stages
of French, but is impossible in Modern French.
Turning to the second step in the derivation, shown in (83c) above, Kayne
argues that this step is also necessary. He notes that Clitic Climbing is impossible if the complement clause is nite. This is shown in (85), where the complement is a subjunctive clause rather than an innitive:

15

Kaynes account of the correlation between null subjects and Clitic Climbing calls
for revision, under assumptions of the Minimalist Program, where government is
eliminated as a theoretical construct. Trevio (1990), discussing Clitic Climbing in
causative constructions, argues that the relationship is indirect: null subject languages are also free-inversion languages (i.e., allow VP-nal subjects), and Clitic
Climbing may be blocked unless the clitic can move through the Specier of IP. On
Trevios analysis, Clitic Climbing displays inversion eects, as suggested by the
contrast in (i):
(i)

a. La
hizo construir t
por Leonardo
CL(DO) made construct-inf. by Leonardo
(S)he had it built by Leonardo.
b. *La
hizo a Leonardo construir t
CL(DO) made to L.
construct-inf.
(S)he made Leonardo build it.
c. (?)La hizo construir t a Leonardo. (=(ib))

In (ia), the embedded clause has no argumental subject, but rather an Agent represented in a by-phrase. This contrasts with (ib), where the embedded clause has an
argument subject occupying a canonical pre-verbal (Specier of IP) position. In
(1c), the embedded clause has an argument subject, but it has undergone free inversion. This implies that the Specier of IP is empty, and is available as a landing site
for the moved clitic.

270

The syntax of Spanish

(85)

a. Mara quiere [CP que [Juan lo

compre]].
M.
wants
that J.
CL(DO) buy.pr.subj.
Maria wants that Juan buy it.
b. *Mara lo quiere [CP que [Juan t compre]].

If clitics could move directly from a lower INFL to a higher one, attaching to
the matrix verb, it would be unexpected that this movement should be
restricted to innitives. If clitics must move through COMP, the ungrammaticality of Clitic Climbing in (85b) can be attributed to the fact that COMP is
not empty, but is occupied by the complementizer que. Further evidence supporting movement of clitics through COMP concerns the interaction of Clitic
Climbing and Wh-movement. Citing Italian data discussed in Rizzi (1982),
Kayne notes that Clitic Climbing is marginally possible where a Wh-phrase
occupies the complement CP:
(86)

Non ti saprei che dire. (Kayne 1989:243)

(I) NEG CL(DAT.2nd.sg) would-know what to-say
I wont know what to say to you.

In (86), clitic ti you has moved through a [+ ] CP, whose Specier is occupied by the Wh-phrase che what. By contrast, if the Wh-expression is a head,
as in (87b), Clitic Climbing is impossible:
(87)

a. Non so se farli. (Kayne 1989:245246)

(I) NEG know if to-do-CL(DO-3rd.pl)
I dont know whether to do them.
b. *No li so se fare. (=87a)

In (87), the COMP position is lled by the [+ ] complementizer se if. The

occurrence of this element makes Clitic Climbing impossible, as in (87b). A
similar contrast may be apparent in Spanish:
(88)

a. ?* No lo
s
cmo arreglar.
(I) Neg CL(DO.3rd.sg.) know how x-inf.
I dont know how to x it.
b. *No lo
s
si
arreglar.
(I) Neg CL(DO.3rd.sg.) know whether x-inf.
I dont know whether to x it.

Example (88a) seems to be a degree better than (88b), where COMP is lled
by a [+] complementizer.
Summarizing, we have seen here that the phenomenon of Clitic Climbing
may involve head movement through the COMP of innitival clauses. Kayne
(1989) argues that clitics can move as heads to INFL without an accompanying verb, and that this movement is possible only in NS languages. He further
argues that clitics move through COMP in the process of raising to a higher

A-movement and Xo movement through COMP

271

verb. If this movement can only be to an empty COMP, several restrictions on

Clitic Climbing are accounted for, including the restriction to innitival complements, and the limited interaction of Clitic Climbing with Wh-movement.

6.6

Summary

This chapter has explored basic characteristics of constructions

whose derivation involves the upper portions of clause structure, which has
been assumed standardly to be the Complementizer Phrase. As the discussion
of 6.2 showed, the standard cases of movement to CP have been analyzed as
necessary to satisfy the Wh-Criterion of Rizzi (1996). However, in Spanish,
the descriptive generalization is more complex, since it appears that only argument Wh-phrases are incompatible with a pre-verbal subject. Another phenomenon discussed above concerns the co-occurrence of a complementizer
and a following Wh-phrase. The issue raised by this phenomenon is the
landing site of the Wh-phrase. Possibilities that have been discussed in the literature include movement: (a) to the specier of a second CP, (b) to the
Specier of IP, and (c) to the specier of a category between CP and IP.
The hypothesis that there is a functional category between CP and IP was
shown in 6.3 to be useful in accounting for the distribution of Focus phrases.
Since Focus constituents can follow a complementizer and can (at least marginally) precede preposed n-phrases, they may be analyzed as moving to the
Specier of a functional category FP between Comp and NegP. This analysis captures the uniqueness of the Focus constituent, its relationship to the
assertive value of the clause, and similarities between Focus movement and
Wh-movement.
Sections 6.4 and 6.5 summarized additional constructions that have been
analyzed as involving A-movement. Section 6.4 discussed phrasal movements, including Scrambling, which moves an object to the left of a subject;
and several constructions which have been argued to be derived by movement
of a null operator: parasitic gaps, complex adjectivals, and null objects.
Section 6.5 discussed head movement to COMP. As the discussion showed,
these movements occur only in non-nite clauses. In 6.5.1, verb-initial clauses
were examined. These included (true) imperatives and participial clauses, both
of which are incompatible with negation. This fact is explained on the hypothesis that I-to-C movement is blocked by an intervening Nego. Clauses which
require I-to-C movement are expected to disallow an (active) Neg Phrase
between INFL and COMP. Finally, Section 6.5.2 discussed Clitic Climbing, a
phenomenon that has been analyzed also as involving movement to COMP.

272

The syntax of Spanish

In this case, the movement in question is of clitics, while the verb remains VPinternal. As argued in Kayne (1989), this accounts for the cross-linguistic correlation between Clitic Climbing and null subjects, and for restrictions on
Clitic Climbing in those languages in which it is found.

References
Abney, Steven P. 1987. The English Noun Phrase in its Sentential Aspect. Ph.D. dissertation. Cambridge, Mass.: MIT.
Aissen, Judith, and David Perlmutter. 1976. Clause Reduction in Spanish in
Proceedings of the Second Annual Meeting of the Berkeley Linguistics Society,
ed. Henry Thompson and Kenneth Whistler, Vicki Edge, Jeri J. Jaeger, Ronya
Javkin, Miriam Petruck, Christopher Smeall, Robert D. Van Valin Jr. 130.
Alatorre, Antonio. 1989. Los 1,001 aos de la lengua espaola. Mexico City: Tezontle.
(2nd edn.)
Alvar, Manuel (ed.). 1996. Manual de dialectologa hispnica: El espaol de Espaa.
Barcelona: Editorial Ariel, S. A.
Arnaiz, Alfredo. 1996. N-words and Wh-in-situ: Nature and Interactions.
Unpublished Ph.D. dissertation. Los Angeles: University of Southern
California.
Belletti, Adriana. 1988. The Case of Unaccusatives Linguistic Inquiry 19, 134.
1990. Generalized Verb Movement: Aspects of Verb Syntax. Turin: Rosenberg &
Sellier.
1995. Italian/Romance Clitics: Structure and Derivation. Paper presented at the
25th Linguistic Symposium on Romance Languages (Seattle, March).
Bello, Andrs. (1847, 1971). Gramtica de la lengua castellana. Mexico: Editora
Nacional.
Bethell, Leslie (ed.). 1984. The Cambridge History of Latin America. Vol. II.
Cambridge: Cambridge University Press.
Bonet, Eulalia. 1991. Morphology after Syntax: Pronominal Clitics in Romance.
Unpublished doctoral dissertation. Cambridge, Mass: MIT.
1995. Feature Structure of Romance Clitics Natural Language and Linguistic
Theory 13, 607647.
Bordelois, Ivonne. 1986. Parasitic Gaps: Extensions of Restructuring in Bordelois
(1986), 124.
Bordelois, Ivonne, Heles Contreras and Karen Zagona (eds.). 1986. Generative Studies
in Spanish Syntax. Dordrecht: Foris Publications.
Borer, Hagit. 1984. Parametric Syntax. Dordrecht: Foris Publications.
Bosque, Ignacio. 1980. Sobre la negacin. Madrid: Ediciones Ctedra, S. A.
(ed.). 1996. El sustantivo sin determinacin: La ausencia de determinante en la lengua
espaola. Madrid: Visor Libros.
Bright, William (ed.). 1992. International Encyclopedia of Linguistics. New York:
Oxford University Press.
273

274

References

Brucart, Jos Mara. 1994. Sobre una incompatibilidad entre posesivos y relativas
especicativas in Demonte (1994b), 5186.
Burzio, Luigi. 1986. Italian Syntax: A GovernmentBinding Approach. Dordrecht:
Reidel.
Camacho, Jos, Liliana Paredes and Liliana Snchez. 1995. The Genitive Clitic and
the Genitive Construction in Andean Spanish Probus 7.2, 133146.
Campos, Hctor. 1986. Indenite Object Drop Linguistic Inquiry 17, 354359.
1993. De la oracin simple a la oracin compuesta: Curso superior de sintaxis
espaola. Washington, D. C.: Georgetown University Press.
Campos, Hector, and Mary Zampini. 1990. Focalization Strategies in Spanish
Probus 2, 4764.
Chomsky, Noam. 1965. Aspects of the Theory of Syntax. Cambridge, Mass.: The MIT
Press.
1977. On Wh-Movement in Culicover et. al., (1977), 71132.
1981. Lectures on Government and Binding. Dordrecht: Foris Publications.
1982. Some Concepts and Consequences of the Theory of Government and Binding.
Cambridge, Mass.: The MIT Press.
1986. Knowledge of Language. New York: Praeger.
1993. A Minimalist Program for Linguistic Theoryin Hale and Keyser (1993), 152.
1995. The Minimalist Program. Cambridge, Mass.: The MIT Press.
Cinque, Guglielmo. 1990. Types of A Dependencies. Cambridge, Mass.: The MIT
Press.
1992. Functional Projections and N-movement within the DP GLOW Newsletter,
28, 1213. (Abstract of paper presented at the 1992 GLOW Conference.)
Clements, J. Clancy. 1994. Notes on Topicalization and Object Drop in Spanish in
Mazzola (1994), 219237.
Comrie, Bernard. 1989. Language Universals and Linguistic Typology. Chicago: The
University of Chicago Press.
Contreras, Heles. 1978. El orden de palabras en espaol. Madrid: Ctedra.
1989. Closed Domains Probus 1, 163180.
1991. On the Position of Subjects in Perspectives on Phrase Structure: Heads and
Licensing, (Syntax and Semantics; 25) ed. Susan Rothstein. New York:
Academic Press, 6379.
1996. Sobre la distribucin de los sintagmas nominales no predicativos sin determinante in Bosque (1996), 141168.
Culicover, Peter W., Thomas Wasow and Adrian Akmajian. 1977. Formal Syntax. New
York: Academic Press.
Dcsy, Gyula (ed.). 1986. Statistical Report on the Languages of the World as of 1985.
Bloomington, Ind.: Eurolingua.
Demonte, Violeta. 1986. C-Command, Prepositions, and Predication Linguistic
Inquiry 17, 147157.
1994a. La ditransitividad en espaol: lxico y sintaxis in Demonte (1994b),
431470.
(ed.). 1994b. Gramtica del espaol. Mexico: El Colegio de Mxico, Centro de
Estudios Lingsticos y Literarios (Publicaciones de la Nueva revista de
lologa hispnica; 6).
1995. Dative Alternation in Spanish Probus 7, 530.

References

275

Emonds, Joseph. 1975. A Transformational Analysis of French Clitics Without

Positive Output Constraints Linguistic Analysis 1, 132.
1976. A Transformational Approach to English Syntax: Root, Structure-Preserving,
and Local Transformations. New York: Academic Press.
1978. The Verbal Complex VV in French Linguistic Inquiry 9, 151175.
Fontana, Josep M. 1993. Phrase Structure and the Syntax of Clitics in the History of
Spanish. Unpublished Ph.D. dissertation. The University of Pennsylvania.
Franco, Jon. 1993. On Object Agreement in Spanish. Unpublished Ph.D. dissertation. Los Angeles: The University of Southern California.
Galms de Fuentes, lvaro. 1996. Mozarabe in Alvar (1996), 97110.
Garca Mayo, Pilar, and Paula Kempchinsky. 1994. Finiteness in Romance vs. English
Parasitic Gap Constructions in Mazzola (1994), 303316.
Giusti, Giuliana. 1991. The Categorial Status of Quantied Nominals Linguistische
Berichte 136, 438454.
Goodall, Grant. 1991. Spec of IP and Spec of CP in Spanish Wh-questions. Paper
read at the 21st Linguistic Symposium on Romance Languages (University of
California, Santa Barbara, February).
1999. On Preverbal Subjects in Spanish. Paper presented at the 29th Linguistic
Symposium on Romance Languages (Michigan State University).
Green, John. 1992. Spanish in Bright (1992), IV, 5864.
Grimes, Ruth (ed.). 1988. Ethnologue: Languages of the World. Dallas, Tex.: Summer
Institute of Linguistics.
Grimshaw, Jane. 1991. Extended Projection. Unpublished MS. Brandeis University,
Waltham, Mass.
Haegeman, Liliane, and Raaella Zanuttini. 1991. Negative Heads and the NEG
Criterion Linguistic Review 8, 233251.
Hale, Kenneth, and Samuel J. Keyser (eds.). 1993. The View from Building 20,
Cambridge, Mass.: The MIT Press.
Hanssen, Federico. 1945. Gramtica histrica de la lengua castellana. Buenos Aires:
Librera y Editorial El Ateneo.
Hatcher, Anna Granville. 1956. Theme and Underlying Question. Two Studies of
Spanish Word Order Word 12, Supplement 3.
Hernanz, Maria Luisa, and Jos Mara Brucart. 1987. La sintaxis. Barcelona: Editorial
Crtica.
Higginbotham, James. 1985. On Semantics Linguistic Inquiry 16, 547593.
Howard, Harry. 1993. P, Aective Inversion, and Topicalization in English and
Spanish. Unpublished Ph.D. dissertation. Cornell University, Ithaca, NY.
Hurtado, Alfredo. 1989a. El control mediante clticos Revista Argentina de
Lingstica 5, 1356.
1989b. Las cadenas clticas Revista Argentina de Lingstica 5, 77133.
Iatridou, Sabine. 2000. The Grammatical Ingredients of Counterfactuality
Linguistic Inquiry 31, 231270.
Jackendo, Ray. 1972. Semantic Interpretation in Generative Grammar. Cambridge,
Mass.: The MIT Press.
1977. X-Bar Syntax: A Study of Phrase Structure. Cambridge, Mass.: The MIT
Press.
Jaeggli, Osvaldo. 1982. Topics in Romance Syntax. Dordrecht: Foris Publications.

276

References

Kaisse, Ellen. 1983. The Syntax of Auxiliary Reduction in English Language 59,
93102.
Kany, Charles. 1951. American Spanish Syntax. Chicago: University of Chicago Press.
Kayne, Richard. 1975. French Syntax: The Transformational Cycle. Cambridge, Mass.:
The MIT Press.
1981. Unambiguous Paths in Levels of Syntactic Representation, ed. Robert May
and Jan Koster. Dordrecht: Foris Publications, 143183.
1989. Null Subjects and Clitic Climbing in The Null Subject Parameter, ed.
Osvaldo Jaeggli and Kenneth Sar. Dordrecht: Klewer Academic Publishers,
239261.
1994. The Antisymmetry of Syntax. Cambridge, Mass.: The MIT Press.
Koopman, Hilda, and Dominique Sportiche. 1991. The Position of Subjects Lingua
85, 211258.
Kurian, George (ed.). 1992. Encyclopedia of the Third World. New York: Facts on File.
Laka, Itziar. 1990. Negation in Syntax: On the Nature of Functional Categories and
Projections. Unpublished Ph.D. dissertation. Cambridge, Mass.: MIT.
Lapesa, Rafael. 1968. Sobre los orgenes y evolucin del lesmo, lasmo y losmo in
Festschrift Walther von Wartburg, 2 vols. Tbingen: Max Niemeyer, 1, 523551.
1981. Historia de la lengua espaola. Madrid: Editorial Gredos. (8th edn.)
Larson, Richard K. 1985. Bare-NP Adverbs Linguistic Inquiry 16, 595621.
1988. On the Double Object Construction. Linguistic Inquiry 19, 335391.
Lema, Jos. 1991. Licensing Conditions on Head Movement. Unpublished Ph.D.
dissertation. University of Ottawa.
Levin, Beth. 1993. English Verb Classes and Alternations: A Preliminary Investigation.
Chicago: University of Chicago Press.
Levin, Beth, and M. Rappaport-Hovav. 1994. Unaccusativity: At the Syntax Lexical
Semantics Interface. Cambridge, Mass.: The MIT Press.
Lipski, John M. 1977. Preposed Subjects in Questions: Some Considerations
Hispania 60, 6167.
1994. Latin American Spanish. London and New York: Longman.
Lloyd, Paul M. 1987. From Latin to Spanish, vol. I: Historical Phonology and
Morphology of the Spanish Language. Philadelphia: American Philosophical
Society (Memoirs of the American Philosophical Society; 173).
Lois, Ximena. 1982. Sur LAccusatif Prpositionnel. Masters essay. Universit de
Paris VIII.
1986. Les groups nominaux sans dterminant en espagnol Recherches linguistiques
de Vincennes.
Longobardi, Giuseppe. 1994. Reference and Proper Names Linguistic Inquiry 25,
609665.
Lujn, Marta. 1976. The Analysis of Reexive Inchoatives in Current Studies in
Romance Linguistics, ed. Marta Lujn and Fritz Hensey. Washington, D.C.:
Georgetown University Press, 377387.
1987. Clitic-doubling in Andean Spanish and the Theory of Case Absorption in
Language and Language Use: Studies in Spanish, ed. T. Morgan, J. Lee and B.
Vanpatten. Washington: University Press of America, 109121.
Malln, Enrique. 1989. The Internal Structure of Determiner Phrases. Unpublished
Ph.D. dissertation. Cornell University, Ithaca, NY.

References

277

1992. Subject Topicalization and Inection in Spanish Theoretical Linguistics 18,

179208.
Masullo, Pascual Jos. 1992. Incorporation and Case Theory in Spanish: A
Crosslinguistic Perspective. Unpublished Ph.D. dissertation. University of
Washington, Seattle, Wash.
Mazzola, Michael (ed.). 1994. Issues and Theory in Romance Linguistics: Selected
Papers from the Linguistic Symposium on Romance Languages XXIII.
Washington, D.C.: Georgetown University Press.
Montalbetti, Mario. 1986. How Pro is it? in Studies in Romance Linguistics, ed.
Osvaldo Jaeggli and Carmen Silva-Corvaln. Dordrecht: Foris, 137152.
Moore, John. 1991. Reduced Constructions in Spanish. Unpublished doctoral dissertation. The University of California, Santa Cruz.
Olarrea, Antxon. 1996. Pre- and Postverbal Subject Positions in Spanish: A
Minimalist Account. Unpublished Ph.D. dissertation. University of
Washington, Seattle, Wash.
Ordez, Francisco. 1997. Word Order and Clause Structure in Spanish and Other
Romance Languages. Unpublished Ph.D. dissertation. New York: The City
University of New York.
Otero, Carlos P. 1971. Evolucin y revolucin en romance, 2 vols. Barcelona: Seis Barral.
1975. The Development of the Clitics in Hispano-Romance in Diachronic Studies
in Romance Linguistics, ed. Mario Saltarelli and Dieter Wanner. The Hague:
Mouton, 153175.
Palmer, L. R. 1954. The Latin Language. London: Faber and Faber Limited.
Parodi, Claudia. 1994. On Case and Agreement in Spanish and English DPs in
Mazzola (1994), 403416.
Perlmutter, David. 1971. Deep and Surface Structure Constraints in Syntax. New York:
Holt, Rinehart and Winston.
1978. Impersonal Passives and the Unaccusative Hypothesis Proceedings of the
Fourth Annual Meeting of the Berkeley Linguistics Society, ed. Jen J. Jaeger,
Anthony C. Woodbury, Farrell Ackerman et al. Berkeley: Berkeley Linguistic
Society, 157189.
Plann, Susan. 1980. Relative Clauses in Spanish Without Overt Antecedents and Related
Constructions. Berkeley: University of California Press.
1982. Indirect Questions in Spanish Linguistic Inquiry 13, 297312.
Pollock, Jean-Yves. 1989. Verb Movement, Universal Grammar and the Structure of
IP Linguistic Inquiry 20, 365424.
Quicoli, Carlos. 1976. Conditions on Clitic Movement in Portuguese Linguistic
Analysis 2, 199223.
Raposo, Eduardo. 1984. The Null Object in Portuguese. Paper presented at the 14th
Linguistic Symposium on Romance Languages (University of Southern
California, Los Angeles).
Reichenbach, Hans. 1947. Elements of Symbolic Logic. New York: Macmillan.
Reinhart, Tanya. 1982. Pragmatics and Linguistics: An Analysis of Sentence Topics
distributed by the Indiana University Linguistics Club, Bloomington.
Rigau, Gemma. 1992. Propiedades de FLEX en las construcciones temporales de
innitivo: la legitimacin del sujeto. Paper presented at the Second
Colloquium on Generative Grammar (Vitoria, Basque Country).

278

References

Ritter, Elizabeth. 1991. Two Functional Categories of Noun Phrases: Evidence from
Modern Hebrew in Perspectives on Phrase Structure: Heads and Licensing, ed.
Susan D. Rothstein. New York: Academic Press (Syntax and Semantics 25),
3762.
Rivas, Alberto. 1977. A Theory of Clitics. Unpublished Ph.D. dissertation.
Cambridge, Mass.: MIT.
Rivero, Mara-Luisa. 1978. Topicalization and WH-Movement in Spanish
Linguistic Inquiry 9, 513517.
1980. On Left-dislocation and Topicalization in Spanish Linguistic Inquiry 11,
363393.
1986. Parameters in the Typology of Clitics in Romance and Old Spanish
Language 64, 774807.
1990. Especicidad y existencia in Indicativo y subjuntivo, ed. Ignacio Bosque.
Madrid: Taurus Universitaria, 261279.
1991. Clitic and NP Climbing in Old Spanish in Current Studies in Spanish
Linguistics, ed. Hctor Campos and Fernando Martnez-Gil. Washington,
D.C.: Georgetown University Press, 241284.
1993. Long Head Movement vs. V2 and Null Subjects in Old Romance Lingua 89,
217245.
1994. Clause Structure and V-movement in the Languages of the Balkans Natural
Language and Linguistic Theory 12, 63120.
Rizzi, Luigi. 1982. Issues in Italian Syntax. Dordrecht: Foris Publications.
1996. Residual Verb Second and the Wh-Criterion in Parameters and Functional
Heads: Essays in Comparative Syntax, ed. Adriana Belletti and Luigi Rizzi.
New York and Oxford: Oxford University Press, 6390.
Rizzi, Luigi, and Ian Roberts. 1989. Complex Inversion in French Probus 1, 130.
Roberts, Ian. 1990. Laccord et les auxiliaires dans lhistoire de langlais. Talk
given at the Sminaire de Recherche, University of Geneva (cited in Belletti
(1990)).
Rooryck, Johan. 1992. Romance Enclitic Ordering and Universal Grammar
Linguistic Review 9, 219250.
Rothstein, Susan. 1983. The Syntactic Forms of Predication. Unpublished Ph.D.
dissertation. Cambridge, Mass: MIT. (Circulated by Indiana University
Linguistics Club, Bloomington, 1985.)
Saltarelli, Mario. 1994. Voice in Latin and Romance: On the Representation of
Lexical Subjects in Issues and Theory in Romance Linguistics: Selected Papers
from the Linguistic Symposium on Romance Languages XXIII, ed. Michael L.
Mazzola. Washington, D.C.: Georgetown University Press, 445478.
Snchez, Liliana, and Jos Camacho. 1995. Equative ser in Spanish. Unpublished
MS. Los Angeles: The University of Southern California.
Snchez-Albornoz, Nicols. 1984. The Population of Colonial Spanish America in
Bethell (1984) 336.
Sportiche, Dominique. 1988. A Theory of Floating Quantiers and Its Corollaries for
Constituent Structure Linguistic Inquiry 19, 425449.
Stalnaker, Robert C. 1978. Assertion in Pragmatics, ed. Peter Cole. New York:
Academic Press (Syntax and Semantics; 9), 315332.

References

279

Stowell, Tim. 1981. Origins of Phrase Structure. Unpublished Ph.D. dissertation.

Cambridge, Mass: MIT.
Strozer, Judith. 1976. Clitics in Spanish. Unpublished Ph.D. dissertation. Los
Angeles: The University of California.
Suer, Margarita. 1982. Syntax and Semantics of Presentational Sentence-Types.
Washington, D.C.: Georgetown University Press.
1984. Free Relatives and the Matching Parameter Linguistic Review, 363387.
1986. Lexical Subjects of Innitives in Caribbean Spanish Studies in Romance
Linguistics, ed. Osvaldo Jaeggli and Carmen Silva-Corvaln. Dordrecht: Foris,
189203.
1987. Haber + Past Participle Linguistic Inquiry 18, 683690.
1988. The Role of Agreement in Clitic-doubled Constructions Natural Language
and Linguistic Theory 6, 391434.
1991. Indirect Questions and the Structure of CP: Some Consequences in Current
Studies in Spanish Linguistics, ed. Hctor Campos and Fernando Martnez-Gil.
Washington, D.C.: Georgetown University Press, 283312.
1994. V-Movement and the Licensing of Argumental Wh-Phrases in Spanish
Natural Language and Linguistic Theory 12, 335372.
1995. Negative Elements, Island Eects and Resumptive no Linguistic Review 12,
233273.
Toribio, Almeida Jacqueline. 1993. Parametric Variation in the Licensing of
Nominals. Unpublished Ph.D. dissertation. Ithaca, NY: Cornell University.
1996. Dialectal Variation in the Licensing of Null Referential and Expletive
Subjects in Aspects of Romance Linguistics: Selected Papers from the Linguistic
Symposium on Romance Languages XXIV, ed. Claudia Parodi, Carlos Quicoli,
Mario Saltarelli and Mara Luisa Zubizarreta. Washington, D.C.: Georgetown
University Press, 409432.
Torrego, Esther. 1984. On Inversion in Spanish and Some of its Eects Linguistic
Inquiry 15, 103129.
1989. UnergativeUnaccusative Alternations in Spanish MIT Working Papers in
Linguistics 10, 253272.
1998. The Dependencies of Objects. Cambridge, Mass.: The MIT Press (Linguistic
Inquiry Monograph: 34).
Tortora, Christina M. 1997. The Syntax and Semantics of the Weak Locative.
Unpublished Ph.D. dissertation. University of Delaware.
Tovena, L. M. 1996. An Expletive Negation Which Is Not So Redundant in Zagona
(1996), 263274.
Trevio, Esthela. 1990. Non-Canonical Subjects in Spanish: Evidence from
Causatives and Psych Verbs. Unpublished MS. University of Ottawa.
Uriagereka, Juan. 1992. Extraction Parameters: A Case Study on
Underspecication. Unpublished MS. University of Maryland.
1995. A Focus Position in Western Romance in Discourse Congurational
Languages, ed. Katalin E. Kiss. Oxford: Oxford University Press, 153175.
Uribe-Etxebarra, Miriam. 1994. Interface Licensing Conditions on Negative
Polarity Items: A Theory of Polarity and Tense Interactions. Unpublished
Ph.D. dissertation. Storrs: The University of Connecticut.

280

References

Vendler, Zeno. 1967. Verbs and Times in Linguistics and Philosophy. Ithaca, N.Y.:
Cornell University Press, 97121.
Vincent, Nigel. 1982. The Development of the Auxiliaries HABERE and ESSE in
Romance in Studies in the Romance Verb, ed. Nigel Vincent and Martin Harris.
London and Canberra: Croom Helm, 7196.
Wanner, Dieter. 1987. The Development of Romance Clitic Pronouns: From Latin to Old
Romance. Berlin: Mouton de Gruyter.
Williams, Edwin. 1982. The NP Cycle Linguistic Inquiry 13, 277295.
Wright, Roger (ed.). 1991. Latin and the Romance Languages in the Early Middle Ages.
London and New York: Routledge.
Zagona, Karen. 1988. Verb Phrase Syntax: A Parametric Study of English and Spanish.
Dordrecht: Kluwer Academic Publishers.
(ed.). 1996. Grammatical Theory and Romance Languages. Amsterdam/
Philadelphia: John Benjamins Publishing Company.
Zanuttini, Rafaella. 1990. Two Types of Negative Markers in Proceedings of NELS
20, ed. Juli Carter, Rose-Marie DEchaine, Bill Philip and Tom Sherer.
Amherst: Graduate Linguistic Student Association of the University of
Massachusetts, II, 517530.
Zubizarreta, Mara Luisa. 1987. Levels of Representation in the Lexicon and in the
Syntax. Dordrecht: Foris Publications.
1994. El orden de palabras en espaol y el caso nominativo in Demonte (1994),
2149.
1998. Prosody, Focus, and Word Order. Cambridge, Mass.: The MIT Press (Linguistic
Inquiry Monograph; 33).
Zwicky, Arnold. 1977. On Clitics. Bloomington: Indiana University Linguistics
Club.

Index
A-bar movement, see movement
Abney, Steven 98
Adjective Phrases 82, 88
in exclamatives 53
order in NP 9091, 109111
adjunct clauses 39, 51 n. 54
adjuncts 8
innitival 65
of VP 131135
participial 121
adverbs 34, 35, 37, 66
and negation 45
in exclamatives 53, 54
IP-adverbs 160161, 168, 169
order 114, 196
VP-adverbs 160170
affected arguments 64
ax-hopping 172, 177
Agreement (AgrP) 114115, 193, 253
agreement 3032, 223, 232
absence of 69
Argument 246
dialectal variation 80 n. 7
in participial clauses 63
invariant 31
person disagreement 228
rightward 32
Aissen, Judith 259, 268 nn. 13 & 14
Alatorre, Antonio 3, 4
Alvar, Manuel 2, 4, 72
A-positions 79
Arabic 4
Argentina 2, 68, 69, 70
argument, external see Verb Phrase, external
argument of
argument clauses 40, 5456
argument NPs see NP, argument
argument structure 41, 76
arguments, direct object 16
arguments, internal 127

Arnaiz, Alfredo 51 n. 54
Aspect (AspP) 179180, 181182, 191193, 253
aspect 3338, 48
delimited 35, 36 n. 47, 42, 64
frequentitive 36 n. 47
habitual 36
punctual 35
Asturian 2
auxiliary verbs 8, 166167, 170171, 174183,
194195
aspectual 174, 182
in participial clauses 64
Aymara 1
bare NPs 2122, 152, 153, 153 n. 20, 154
Basque 1, 4
Belletti, Adriana 121, 155, 180 n. 12, 191,
196, 238, 264, 266
Bello, Andrs 63 n. 64
benefactives 137, 144, 145
Binding 259
Bolivia 1, 68, 69, 70, 71
Bonet, Eulalia 16 n. 30
Bordelois, Ivonne 258 n. 10
Borer, Hagit 189
Bosque, Ignacio 21, 23 n. 39, 24 n. 40, 40 n.
50, 46, 70, 199, 199 n. 21
branching, binary 86, 112, 148, 165
Bright, William 1
Brucart, Jos Mara 225, 241, 250, 251
Burzio, Luigi 153 n. 21
Camacho, Jos 48, 71 n. 66, 183 n. 15
Campos, Hctor 39 n. 49, 55, 57 n. 59, 67,
252, 263
cardinals 102103
as speciers of NP 93
Case 814
absorption in passives 139
abstract 8082, 84

281

282

Index

Case (cont.)
Accusative 5, 16, 17, 29, 30, 41, 136, 139,
145 n. 15, 146, 154
Accusative, double 13
and agreement 189
and clitics 184
assignment 81, 128129, 230231
assignment by nouns and adjectives 87 n. 13
assignment to DP 136
Dative 11 n. 20, 14, 16, 17, 29, 30, 56, 142,
145 n. 15
Exceptional (ECM) 81 n. 8
Genitive 10, 71
inherent 87 n. 13
markers 55
Nominative 5, 9, 11 n. 20, 16, 29, 30, 63, 67,
8182, 84, 139, 153
Nominative assignment 129, 236
Objective 9, 11, 8182
Oblique 11
Partitive 155
structural 87 n. 13
transmission 155
visibility 84
Catalan 1, 37, 222 n. 1, 262
causation, direct and indirect 30
C-command 78 n. 5, 81 n. 8, 127
chain 225
A chain 260
Chomsky, Noam 78, 87 n. 13, 128, 170, 172,
173, 204, 230, 247 n. 3, 257, 260
Cinque, Guglielmo 114, 220
clitic se 5, 16 n. 30, 17, 32, 41
aspectual 188
clitics 1519, 184194
and negation 44
base-generation 189190
climbing 267270
Dative 142, 142 n. 11
dialect variation 6870
doubling 51, 56, 59, 136, 138, 187188, 191
in Gerunds 17
in imperatives 17, 52
in participial clauses 64
movement of 184188
order 178
reduplication 69
reexive 41
cognate objects 75 n. 2
Colombia 69, 70, 71, 72
complementizers 234 n. 17, 247
omission 55

complements 7
DP 135141
of adjectives 56
of nouns 56, 8689, 9193
of verbs 131135
complex adjectivals 259262
COMP-trace effects 26
Comrie, Bernard 209, 211, 249 n. 4
conditional de-clauses 66
Conditionals, counterfactual 34
constituent order 78
adverbs 162170
clauses 113114
declarative 4850, 213216
interrogatives 19, 50
NP 9192, 109116
subject 204208, 229239
contrastive contexts 40 n. 50, 43, 49, 72, 223,
227 n. 15, 252
copular sentences 32, 4748
Costa Rica 2
Dative a 147, 149
Dativization 145147
denitive articles, see determiners
demonstratives 93, 100
Demonte, Violeta 14, 137, 143 n. 12, 148 n.
16, 151
Determiner Phrases (DPs) 98101
determiners 2023
argument NPs 94102
denite 5
neuter lo 21, 53
pleonastic 100
direct object 13, 16
discourse context 262
dislocation construction 137138
Dominican Republic 2, 70, 72
DP, direct object 135141
pronominal 136
reexive 136
D-structure 126
economy principles 204, 235
Ecuador 2, 69, 71, 72
El Salvador 71
Emonds, Joseph 114, 165, 185
endocentricity 86
English 1, 2, 49
adverb order 167
auxiliaries 15, 16 n. 27, 167, 197
benefactives 137 n. 7

Index
constituent order 113
Dative-shift 143 n. 12
double genitives 10
double negation 46 n. 53
gerunds 11
interrogatives 51, 234 n. 17, 244, 256
one 21, 92
passives 140 n. 8
pleonastics 31, 80
progressives 38
richness of agreement 232, 237
unaccusative verbs 154
Equatorial Guinea 1
Estar 38, 4748, 178 n. 11
events, ordering 6062, 65
exclamatives 5354
existential sentences 25
Extended Projection Principle (EPP) 124
feature checking 172173, 177, 205, 207208,
216
nite clauses 31, 55, 67
focus 22, 25, 49, 50, 209211, 248254
Fontana, Josep 6
Franco, Jon 188 n. 17, 192
French 6, 17, 29 n. 45, 37, 49, 153, 232, 238,
244, 262, 269
functional categories 8, 116, 183, 190194,
204206, 217219
Galician 1
Galms de Fuentes, Alvaro 3
Garca-Mayo, Pilar 258
German 6
Gerundive Adjunct Clauses 6265
gerunds 17, 29
Giusti, Giuliana 102 n. 18
Goodall, Grant 214, 246 n. 2, 247, 248
Green, John 1
Grimes, Ruth 2
Grimshaw, Jane 159
Guaran 2
Guatemala 2, 71
haber 33, 177, 178 n. 11, 180, 182
hacerlo replacement 132134, 135, 137 n. 7,
142144, 149151, 152, 155156
Haegeman, Lilianne 198, 199
Hanssen, Federico 5, 6, 7, 13
Hatcher, Anna 49
Hernanz, Maria Luisa 225, 250, 251
Higginbotham, James 98 n. 15

283

Hindi 13 n. 23
Honduras 2, 69, 71
Howard, Harry 249
Hurtado, Alfredo 31, 188
Iatridou, Sabine 34 n. 46
Icelandic 6
if-clause 34
imperatives 5253, 264265
clitics in 17, 18
morphology 53 n. 55
inchoatives 42
indirect objects 55, 59
pronominal 143
reexive 143
indirect questions 19 n. 34, 55, 247248
individual-level predicates 4748
INFL 119122, 171174
features of 119, 125, 172173, 181, 192,
205, 232
split INFL hypothesis 196
information structure 49, 209
interrogative forms 10 n. 19, 11
interrogatives 19, 26, 5052, 5657, 72, 93,
226 n. 13, 234 n. 17
see also Wh-movement
intonation 49, 50, 63, 67, 92, 210211, 220,
249
islands 224, 257258
Italian 6, 7, 10, 18, 37, 62 n. 62, 102 n. 18,
153, 232, 236, 238, 244, 262
Italic, Oscan-Umbrian subfamily 3
Jackendoff, Ray 79 n. 6, 112
Jaeggli, Osvaldo 25, 69, 189, 203, 232235,
256
Judeo-espaol 2
Kaisse, Ellen 16 n. 27
Kany, Charles 71
Kayne, Richard 69, 86, 105 n. 20, 113, 186,
236, 255, 268, 269, 270
Kempchinsky, Paula 258
Koopman, Hilda 214, 236
Kurian, George 1, 2
Ladino 2
Lasmo, 17 n. 33
Laka, Itziar 195
Lapesa, Rafael 3, 5, 6, 7
Larson, Richard 74, 148
Latin 3, 5, 16, 17 n. 32

284

Index

Latin America, indigenous languages 1

Lesmo 17 n. 33
Lema, Jos 180 n. 12, 183
Leonese 2
Levin, Beth 152 n. 19
Lipski, John 3, 6872
Lloyd, Paul 4
locative inversion 22
locative PPs 47
Logical Form 100
Lois, Ximena 22, 140 n. 10
Losmo 17 n. 33
Longobardi, Giuseppe 101 n. 17
Lujn, Marta 188
main clauses 4854
Malln, Enrique 114, 214 n. 9
Masullo, Pascual 144 n. 14, 150
Mayan languages 2
M-command 123, 123 n. 2
Mexico 2, 69, 70, 71
modal verbs 4243
Montalbetti, Mario 238 n. 18
mood 3840
subjunctive 38, 39
Moore, John 268 n. 14
morphology, inectional 1415
movement 1920
A-bar 214, 218, 225227, 252, 260261
clitic 267270
clitic left dislocation (CLLD) 228
clitics to Comp 267270
constraints 235, 259
DP to specier of IP 154
Focus 217218, 250254
in NP 100101
in passives 139
NP 82
of subject 130, 165166, 206208
of Topic 213216
of verb 121, 148, 165166, 166167, 168,
171174, 245247, 263267
optionality of 204
rightward 112113
to Specier of IP 128
Wh see Wh-movement
N-ellipsis 9293
Navarro-Aragonese 2
Neg Phrase 253
negation 21, 4347, 7071, 178, 219, 194200,
266, 267

in imperatives 52
multiple 46
Neg-raising 46
n-words (NPIs) 24, 197200, 253
Negative Concord 2324
Negative Polarity Items, see Negation, n-words
Nicaragua 2, 69
nominals, derived 62 n. 62
non-nite clauses 22, 2829, 55, 65, 265266
Noun Phrase, see NP
nouns, picture 88
NP, 85, 8991, 9194
argument 7382
NP-trace 233 n. 16
null objects 262263
null subject languages 190
null subject parameter 26, 229239
null subjects 2426, 124, 269
Olarrea, Antxn 221, 227 n. 15, 228
Old Spanish 6
Ordoez, Francisco 216, 241, 255256
Otero, Carlos 2, 6
Palmer, L. R. 5
Panama 2, 71, 72
Paraguay 2
parameters 206, 238 n. 18
parasitic gaps 257259
Paredes, Liliana 71 n. 66
Parodi, Claudia 102 n. 18
Participial Adjunct Clauses 6265
participial clauses 266
participles 18, 22, 40
morphology 179
partitive constructions 106, 107 n. 23
passives 20, 32, 40, 138139, 144
clitic 42
perfectives 33
Perlmutter, David 16, 154, 259, 268 nn. 13 &
14
Personal a 6, 7, 1314, 58, 59, 84, 136, 140,
140 nn. 9 & 10, 144, 149151, 190, 221
Per 2, 70
Plann, Susan 55 n. 56, 57 n. 59, 241, 247
Pollock, Jean-Yves 171, 196
Portuguese 262
possessives 71, 93, 100
pre-determiners and post-determiners 9597,
102107
Predicate Constituent 3348
predication 83, 123126

Index
predicative DPs 21
predicative NPs 8285, 107109
Prepositional Phrases 59
prepositions 65
covert 74
locative 134135
stranding 52
presentational predicates 4950, 152
presupposed information 49, 212
progressives 18, 179180
pronouns, pleonastic 25
Proto-Romance 5
Puerto Rico 1, 72
purpose clauses 41
Quantied NPs 54
Quantier-oat (Q-oat) 105, 109
Quantier Phrase (QP) 102107
quantiers 23, 93, 96
Quechua, 1, 2
questions, see interrogatives
Quicoli, Carlos 185
Raposo, Eduardo 262
Rappaport-Hovav, M. 152 n. 19
reexives 12, 125
emphatic 136
Reichenbach, Hans 60
Reinhart, Tanya 212 n. 8
relative clauses 29, 5660
non-restrictive 58
restrictive 59
relative pronouns 5960
restructuring predicates 123
Rigau, Gemma 9
Ritter, Elizabeth 102 n. 18
Rivas, Alberto 189
Rivero, Maria Luisa 6, 225, 241, 247, 264,
265
Rizzi, Luigi 25, 203, 232, 233235, 243244,
248, 255, 268 n. 14, 270
Roberts, Ian 238, 247 n. 3
Romanian 244 n. 1
Rooryck, Johan 264
Rothstein, Susan 124 n. 3
Saltarelli, Mario 6, 188
Snchez, Liliana 48, 71 n. 66, 183 n. 15
Snchez-Albornoz, Nicols 4
Scrambling 216, 255256
Secondary Predicates 44, 120
semi-auxiliaries 122, 125, 145 n. 15

285

ser 6, 32, 40, 4748, 183 n. 15

Sicilian 7
small clauses 27, 30, 67 n. 65, 87 n. 13, 88, 90
Spanish
Andean 69
Caribbean 7172
dialects 2, 6872
Leonese 72
Mozrabe 3
Spanish as primary language 1
speech-time 60
Sportiche, Dominique 105 n. 20, 214, 236
S-structure 126
stage-level predicates 4748
Stalnaker, Robert 212 n. 8
Strozer, Judith 16 n. 30, 137, 143 n. 12, 150,
189
subject constituent 2632
Nominative 67
raising 130131
subordinate clauses 46, 5468
Suer, Margarita 49, 51, 69, 72, 167 n. 9, 188
n. 17, 199 n. 21, 241, 245246, 247 n. 3,
248
superiority 256 n. 9
Telicity, see Aspect, delimited
Temporal Adjunct Clauses 6062
Tense 3338
compound 37
Conditional 33
Tense, features of 171172, 173, 177
Future 33, 34
Imperfect past 33, 35, 37
Indicative Past 35
Present indicative 3435
Preterite past 33, 35, 36
sequence of tense 33, 61
simple Present 41
Thematic hierarchy 79 n. 6
Thematic roles see Theta-theory, Theta-roles
Theme and Rheme 49
Theta-binding 98 n. 15
Theta-role assignment (theta marking) 78,
127, 147, 260
compositionality of 79 n. 6
Theta-roles 77, 125
Theta-theory 7580, 126, 164, 165, 230
Topic (see also dislocation construction) 22,
25, 44, 49, 211213
dislocated 220229, 251
movement of 213216

286

Index

Toribio, Almeida J. 72
Torrego, Esther 6, 22, 13 n. 23, 19 n. 34, 140
n. 10, 243, 245
Tortora, Christina 152 n. 19
Tovena, L. M. 62 n. 62
traces 260
transitivity 76
Trevio, Esthela 269
unaccusatives, see verbs, unaccusative
Uriagereka, Juan 191, 254
Uribe-Etxebarra, Myriam 198
Uruguay 2, 70
Vendler, Zeno 35
Venezuela 71
verb movement, see movement
Verb Phrase 118157
distribution 118126
external argument of 126131
Verb Second 6
verbs, activity 36
aspectual 122
atmospheric 31
auxiliary 121, 125
causative 122, 145 n. 15
copular 66
ditransitive 141
intransitive 22, 75, 153
modal 182

of contact 141, 145147

perception 122
restructuring 43
semi-auxiliary 18
stative 34, 35, 36, 53, 65, 134
transitive 30, 75
unaccusative 64, 152156
volitional 122
Vincent, Nigel 6
voice 4042
middle 41
VP-internal subject hypothesis 128, 231
Wanner, Dieter 6
Wh-criterion 242, 243244
Wh-movement 241248
Wh-phrases 50, 51, 250
Williams, Edwin 84 n. 10
word order, see constituent order
Wright, Roger 3
X-theory 93, 94, 174
Zagona, Karen 161 n. 2
Zampini, Mary 252
Zanuttini, Rafaella 23, 198, 199
Zubizarreta, Maria Luisa 14, 150 n. 17,
209213, 214 n. 9, 216, 218, 222, 223,
249
Zwicky, Arnold 16 n. 27