Ordoñez 1997

INFORMATION TO USERS
This manuscript has been reproduced from the microfilm master. UMI
films the text directly from the original or copy submitted. Thus, some
thesis and dissertation copies are in typewriter face, while others may be
from any type o f computer printer.
The quality o f this reproduction is dependent upon the quality o f the

copy submitted. Broken or indistinct print, colored or poor quality
illustrations and photographs, print bleedthrough, substandard margins,
and improper alignment can adversely affect reproduction.
In the unlikely event that the author did not send UMI a complete
manuscript and there are missing pages, these will be noted. Also, if
unauthorized copyright material had to be removed, a note will indicate
the deletion.
Oversize materials (e.g., maps, drawings, charts) are reproduced by

sectioning the original, beginning at the upper left-hand comer and
continuing from left to right in equal sections with small overlaps. Each
original is also photographed in one exposure and is included in reduced
form at the back o f the book.
Photographs included in the original manuscript have been reproduced

xerographically in this copy. Higher quality 6” x 9” black and white
photographic prints are available for any photographs or illustrations
appearing in this copy for an additional charge. Contact UMI directly to
order.
UMI
A Bell & Howell Information Company
300 North Zeeb Road, Ann Arbor M I 48106-1346 USA
313/761-4700 800/521-0600
Reproduced with permission of the copyright owner. Further reproduction prohibited without permission.
/ '
Word Order and Clause Structure in

Spanish and Other Romance Languages
by
Francisco Ordonez
h dissertation submited to the Graduate Faculty in

Linguistics in partial fulfillment of the requirements
for the dregree of Doctor of philosophy, The City
University of New York.
1997
UMI Number: 9807978
Copyright 1997 by
Ordonez, Francisco
All rights reserved.
U M I Microform 9807978
Copyright 1997, by UM I Company. A ll rights reserved.
This microform edition is protected against unauthorized

copying under Title 17, United States Code.
UMI
300 North Zeeb Road
Ann Arbor, M I 48103
© 1997
Francisco Ordon ez
All Rights Reserved
This manuscript has been read and accepted for the Graduate Faculty in Linguistics in
satisfaction of the dissertation requirement for the degree of Doctor of Philosophy.
Date Richard Kayne

Chair of Examining Committee
Date 1 Charles Cairns

Executive Officer
Gita Martohardjono
Juan Uriagereka
The City University of New York
Abstract
Word Order end Clause Structure in Spanish and

Other Romance Languages
by
Francisco Orddnez
Adviser: Professor Richard S. Kayne
This dissertation explores various aspects of word order
and clausal structure in the Romance languages, with

special emphasis on Spanish. The different aspects are
looked at in light of the highly constrained theory of word
order proposed in the antisymmetry approach of Kayne
(1994) . This theory makes unavailable certain widely-
assumed mechanisms of analysis such as right adjunction and

multiple adjunction to the same head.
The first part (Chapters 1 and 2) explores new analyses

of postverbal subjects in Romance, which had been assumed
to be right adjoined. The alternation V S 0/V 0 S shows
that there are certain asymmetries between these orders
that are inexplicable under a right-adjunction analysis.
This same alternation is the same found in verb-final

scrambling languages between S 0 V and 0 S V. I propose
V
therefore that V 0 S is the output of the movement of

objects to the left of the subject. Comparison between
Spanish and the other Romance languages, such as Catalan,
Italian, and French leads to the further elaboration on the
analysis proposed. It is argued that the derivation of

VOS also involves movement of the whole TP to a position
above the subject.
The other aspects of clausal structure examined

include interrogatives and the position of preverbal
subjects. In Chapter 4, It is shown that the peculiar
restrictions on the distribution of subjects in

interrogatives cannot be explained by a obligatory overt
verb movement toC°. This last proposal is also
incompatible with the assumptions made in Kayne (1994).
The alternative proposed involves complementizer movement

instead. This alternative makes an important assumption:
pre-verbal subjects have to be left dislocated.
In the final chapter I link the obligatory left-
dislocated nature of subjects to the rich inflectional
system of the language. If we take person agreement to be
a clitic argument that can receive case and be assigned a

0-role, the need for an exclusive position for subjects
disappears. Evidence for the clitic nature of person

agreement is well motivated in the syntax as well as in the
morphology.
Acknowledgment s
I would like to start by thanking my dissertation

committee: Richard Kayne, Juan Uriagereka and Gita
Martohardjono.
One of the best decisions I have made was to come to New

York to study with Richard Kayne. I thank him for my
formation as a syntactician, for communicating his love of

the intricacies of language, for showing me how details are
very important to get to a more rich analysis, and for
teaching me that the most difficult problems lead to the
most interesting solutions. I hope my dissertation can

communicate this way of working in syntax. Richard Kayne
has also been supportive and caring on a personal level
during all these years of learning. I thank him very much.
Juan Uriagereka has always been encouraging and
enthusiastic. He would go for the daring analysis and
inspire me to pursue them. He has also supported me on

many levels, in more ways than I could pay him back. Gita
was a late addition to my committee, and I thank her for

the help and interest she put into this enterprise.
To my professors at CUNY, I would like to thank Charles
Cairns, Bob Fiengo, and Janet Fodor for all I have learned
with them. Carme Picallo was visiting CUNY for a year. I
thank her for written comments on various parts of this

dissertation and for being a good friend.
vii
Many colleagues have directy or indirecly contributed to
this thesis. I thank Elena Herburger for her frienship and
for letting me ask about her German and also about her
Caracas dialect, Antxon Olarrea for interchange of papers
and good comunication over the internet. In CUNY I would
like to thank my friends Sheila Meltzer, Sharon Utakis, and
Ivy Sichel for sharing with me good and not so good moments
during all these years. I deeply thank Marivi Blasco and
Josd Antonio Mendez for their friendship and for being

available whenever I urged for an intuition in Spanish.
Judy Bernstein was a good friend here in New York and in

Boston and Arhonto Terzi shared with me my Mediterranian
views about so many aspects of life. I also thank Nino

Gulli for his judgements and Italian as well as letting me
practice the language, Daniel Chapuis for last minute calls
about doubts in French, and Viviane D6prez for help over
the internet..
Outside of the U.S, I owe thanks to my Mexican "cuates"

Esthela Trevino y Jose Lema for our linguistic discussions
and for making it possible for me to discover the richness

south of the Rio Bravo.
To my professors in the Universitat Autdnoma de
Barcelona, I thank Gemma Rigau, Maria Lluisa Hernanz, and.

Jos6 Maria Brucart for making so appealing the switch from
the world of literature to the world of generative
linguistics. Gemma has also been a good correspondent
viii
whenever I was afraid my Catalan was mixing with my
Spanish. I deeply thank Maria Lluisa for shipping me to

Italy, despite the fact that my knowledge of Italian was
zero. In Girona, I would like to thank Avel.lina Suner and
Laura Ripoll for being a wonderful hosts, and for their
friendship. For Catalan and Spanish intuitions as well as

linguistics discussions and friendship, I thank Francesc
Roca, Joan Rafel i Xavier Villalba.
To my professors in during the year I spent in Venice, I
thank Guglielmo Cinque, Giuseppe Longobardi, and Michael

Kenstowicz. I also thank Anna Cardinaletti for
discussions about post-verbal subjects and Cecilia Poletto
for sending me some of her work on clitics in the Norh of
Italy. Additionally, I would like to thank Maria Teresa
Guasti and Carlo Cecchetto for all the contact we had

through the internet.
During all these years, I had the opportunity to develop
my professional side, besides working on my studies. In

Rutgers, I thank the late Robert Jeffers and Jane Grimshaw.
I thank Cynthia Pyle for her friendship and for my
Renaissance experience with Piero della Francesca. I also

thank Marlene Gottlieb for letting me experience teaching
Spanish at Lehman College.
At a more personal level, I thank my parents Jose and
Francisca for their love across the Atlantic. They always
wondered how come it was taking me so long to finish
studying when they hardly had the opportunity to do it. To
my siblings: Raimunda, Pepin and Ricardo for being there
every other Sunday when I called home.
To my other family in New York, I thank Bob and Ruth

for help and support during all these years since I have
been with Michael. This dissertation owes a lot to Michael

for all the love and support I have received from him.
Michael is also responsible for all the good editing (for
all theerrors I am the only one to blame) . He always
complains that I had never edited his papers written in

Spanish when we were in Barcelona. I do not know how I
will return him this favor. Michael! What about starting
another Ph.D in Spain or Mexico?
X
A B S TR A C T .......................................................................................................................................
A C K N O W L E D G M E N T S .............................................................................................................. vi
1 IN T R O D U C T IO N .........................................................................................................................1
1.1 F ramew ork : antisym m etry ........................................................................................................ 3
1.2 Background : HIERARCHICAL STRUCTURE AND UNEAR ORDER..................................................... 4
13 Som e I nadequacies of the Sym m etric V ie w ............................................................................. 8
13.4 The asymmetry o f specifiers..................................................................................................... 8
1 3 3 Asymmetries in the Agreement patterns................................................................................. 10
13.6 Asymmetries on the directionality o f Head Movement........................................................... 11
1.4 Spec H e a d C o m plem entas U niversalorder . ......................................................................... 12
13 T he form ulation o f the LCA..................................................................................................... 14
1.6 LCA AND rrs CONSEQUENCES........................................................................................................ 22
2 THE V S O /VOS A LTE R N A TIO N IN S P A N IS H ................................................ .28
2.1 INTRODUCTION............................................................................................................................... 28
22 THE V S O AND V O S ORDER IN SPANISH................................................................................. 29
2 2 J The V S O order................................................................................................................... 30
2 2 2 The V O S order................................................................................................................... 33
2.3 The analysis o f V S O and V O S.......................................................................................... 35
2.4. THE ASYMMETRIES........................................................................................................................40
2.4.1 Bin d in g ...................................................................................................................................... 40
2.42 Post-verbal wh-elements.........................................................................................................52
2.43 Interpretation o f indefinites..................................................................................................... 58
2.4.4 The distribution o f post-verbal subject pronouns...................................................................64
1. Doubling with indirect objects................................................................................................... 67
2.5 CONCLUSION...................................................................................................................................74
CHAPTER 3 ................................................................................................................................... 7 7
L IG H T P R E D IC A TE RAISING AND PO ST-VER B A L SUBJECTS IN R O M A N C E 77

3.1 INTRODUCTION............................................................................................................................... 77
32 DISTRIBUTION OF POST-VERBAL SUBJECTS IN ITALIAN, CATALAN AND FRENCH...........................78
32.1 The V O S order.................................................................................................................. 78
3 2 2 V S DP (object)....................................................................................................................80
3 2 3 V S PP (complement).......................................................................................................... 81
32.4 V S A d j................................................................................................................................ 84
3 2 3 V S Adv (de-adjectival)......................................................................................................... 85
32.6 V S INF................................................................................................................................ 86
3.2.7 V S CP.................................................................................................................................. 87
33 Distribution of post- verbal subjects in Spa n ish ...................................................................90
3.4 Post-verbal subjects in N eutral Phrase. ............................................................................... 92
33 Post- verbal subjects in Focus Phrase..................................................................................... 95
33.1 Scrambling............................................................................................................................. 96
3 3 2 Light Predicate Raising (LPR)................................................................................................98
3.6 Proposal: LPR w ith Post-verbal subjects........................................................................... 102
3.7 Consequences for nonarguments.......................................................................................... 105
3.8 Consequences for arguments..................................................................................................110
xi
3J8.1 LPR and C-commanding o f arguments............................. 115
3.9 Spanish V O S o rd e r................................................................................................................. 117
3.10 Comparing LPR 70 th e r ig h t adjunction a lte rn a tiv e .....................................................121
3.11 Conclusion ................................................................................................................................ 127
CHAPTER 4 .................................................................................................................................. 128
T H E IN V E R S IO N C O N S TR U C TIO N IN IN TE R R O G A T IV E S IN SPA N ISH AND

C A T A L A N ........................................................................................................................................ 128
4.1 In tro d u ctio n ............................................................................................................................... 128
4.2 Antisymm etry and the la n d in g site of c u n c s .................................................................... 130
4 3 The position o f th e p o st-verb al subjects in in te rro g a tiv e s ............................................. 135
4 3 J Auxiliaries and Vpp........................................ 236
4 3 2 Catalan and the position o f subjects............................... 238
4 3 3 Floating quantifiers.............. 239
4.4 V-TO-C AND THE “FREE INVERSION “ CONSTRUCTION.................................................................... 142
4.5 PIEDMONTESE................................................................................................................................. 144
4.6 THE OBLIGATORINESS OF INVERSION IN INTERROGATIVES IN SPANISH AND CATALAN............... 145
4.6.1 A -Minimality.......................................... 245
4.62 A-bar minimality......................................... 147
4.63 Head movement. Movement o f the complementizer..................... 251
4.7 Conclusion ..................................................................................................................................158
4.8 A ppendix .......................................................................................................................................159
CHAPTER 5 ..................................................................................................................................164
LE FT DISLO CATED SUBJECTS AND PRO-DROP......................................................164

5.1 INTRODUCTION.......................................................................................................................... 164
5 2 Empirical evidence.................................................................................................................... 168
5.2.1 Ellipsis.............................................. 268
5 2 2 Quantifier extraction........................................ 172
5 2 3 The scope o f quantifiers in preverbal position......................... 279
5 3 Dislocated subjects. Previous accounts............................................................................. 185
5 3 2 Hon Polysynthetic Approaches.................................. 186
2 3 2 Polysynthetic approaches.................................... 288
5.4 Person Agreement as a c l it ic .................................................................................................. 194
5.42 Morphological evidence..................................... 208
5.5 Co n clusio n .................................................................................................................................212
REFERENCES.............................................................................................................................. 214
1
Chapter 1
Introduction
This dissertation explores a number of issues involving

the clausal structure of Romance with special emphasis on
Spanish. The most important theoretical foundation of the
study is Kayne's (1994) antisymmetry proposal, and the
relation of hierarchy to linear order found there directly
informs Chapters 2 to 4. The analyses in these chapters
are not only made possible by the theoretical tools of
Kayne (1994) . The results also provide support for that

proposal. Specifically, the study shows that analyses
compatible with antisymmetry provide superior empirical
coverage to traditional ones that make use of mechanisms*

barred by Kayne's proposal.
The topics of the chapters break down as follows: In
Chapter 2, I present and defend new analyses coherenent
with the principles of antisymmetry for two marked orders

in Spanish, VSO and VOS. I conclude that the
alternation between these two orders can be explained by
obtaining V O S from V S 0 by moving the object to the
left. Chapter 3 explores the distribution of postverbal
subjects comparatively. In this chapter I focus on the

relatively looser restrictions on the distribution of these
subjects in Spanish than in Catalan, French, and Italian.

I propose a hypothesis in which focused subjects end up
post-verbally by moving the TP to the left in a way
parallel to Light Predicate Raising proposed in Larson
(1988) . The topic of Chapter 4 is the analysis of

interrogatives and restrictions on word order. As
predicted by Kayne (1995), I show that there is no movement
of the verb to C. Chapter 5 looks at interesting questions
concerning sentences containing pre-verbal subjects. In
this final chapter, I explore the idea that pre-verbal

subjects are in a left dislocated position. Following up
on work by Taraldsen (1992), I link the left-dislocated
nature of these subjects to the relatively rich
inflectional system of many Romance languages. Unlike the

rest of the problems explored this study, the structures
examined here do not directly impinge on the theoretical
issues elucidated in Kayne (1994) . Nevertheless, the
3
analysis of sentences with preverbal subjects are crucial
in understanding Romance clausal structure. Furthermore,

they follow naturally from the questions explored
previously.
1.1 Framework: antisymmetry
As discussed in the previous section, the most important

theoretical foundation of this study is the antisymmetry
proposal made by Kayne (1994). The core of antisymmetry is

the unification of the two fundamental dimensions of
syntax: hierarchical structure and linear ordering of the
constituents of a sentence. As a consequence, antisymmetry

is more restrictive them previous approaches to syntactic
analysis in terms of the theoretical apparatus allowed.
Evidently, this reduction in available mechanisms is a
potentially important advance because it makes for a more
highly restictive syntactic theory. Nevertheless, it poses
a challenge for the field. Those analyses of empirical
facts that had depended until now upon theoretical
apparatus barred by antisymmetry must be replaced if the

proposal is accepted. Most relevant for this study, a
number of these now eliminated mechanisms have been widely

assumed in syntactic analyses of Romance until now. At the
same time, the plausibility of the theory depends upon the
superiority of the alternative analyses. In the end, the
success of the new theory rests in good part the success of
these new explanations as much as the explanations depend
on the theory. Because of this interdependence, it is

important to have a clear idea of the theoretical issues
involved.
1.2 Background: hierarchical structure and linear

order
Syntacticians have longdistinguished between

hierarchical relations and linear order. In generative
grammar, hierarchical realtions have traditionally been

encoded under the X'-schema as in (1) .:
(1)
XP
Spec
A
/ \
*1
Head CbmpL
Under this schema, the hierarchical arrangement of
phrases corresponds to one where every phrase is made up of
a head and a complement, which together make an
intermediate constituent. This constituent and the

specifier form a maximal projection XP. This pattern has
been taken to apply to lexical categories (i.e.,
adjectives, nouns, prepositions, adverbs, and verbs)
(Chomsky 1970, Jackendoff 1977, Speas 1990). Recently, it
1Chomsky (1995) departs from the tenets of X'-theory. He

proposes a derivational model in which Phrase Structure is
derived by the more primitive operation of Merge.
5
has been extended to complementizers, inflection, and

agreement (Pollock 1989, Chomsky 1986).
The other important notion, the order in which

constituents appear, has been considered to be a product of
language-specific grammars. This less universal approach
follows from contrasts such as those between (2) and (3) .

(2) a. from Tokyo
b. Sandra hit M ary
(3) a. Tokyo kara (Japanese)

Tokio from
b. Sandra-ga M ary -o but - ta. (Japanese)

Sandra Mary hit
In these examples, languages such as English and
Japanese appear to be mirror images of each other, English

being head initial and Japanese head final. Given such
variations, it has traditionally been assumed that the
hierarchical arrangement between head and complement can

have two symmetrical realizations in linear terms. That
is, the head may precede as in English, or it can follow
the complement as in Japanese:

(4)
English Japanese
H ead Compl Compl Head

from Tokyo Tokyo Kara
hit Mary Mary-o but-ta
6
Thus, linear order and hierarchical structure have been
taken to be disconnected. To appreciate the radical nature
of antisymmetry, it is worth noting that separate

treatments of constituent order and hierarchical structure
have been a constant in generative linguistics. Although
the specific formalizations of both facets of phrase

structure have changed considerably over time, there have
been few if any attempts at unification.
For example, in its early stages Transformational

Grammar encoded both relations through different notations
in phrase structure rules. A phrase structure rule such as
(5) states that an NP dominates a determiner and a noun,
the hierarchal relation being indicated by the arrow. In
the same rule the precedence relations were indicated by

the ~ symbol. So' the determiner precedes the noun.
(5) NP —> Dei * N
This over-all separation of hierarchy and linear order
was preserved in the development of nontrans format ional

approaches such as GPSG and HPSG. Gazdar, Klein, Pullum,
and Sag (1985) and Pollard and Sag (1994) propose Immediate
Dominance rules (ID), which express hierarchical relations,
and linear precedence statements (LP), to account for

order. The same rule (5) would be expressed by the two
rules in (6) and (7) . The symbol >" exclusively
signifies dominance, whereas "<" indicates precedence.

(6) NP —> Dei, N (ID Rule)
(7) Det < N (LP Statement)
The GB framework preserved the same separation through
the X' schema presented above. This format expressed that
a maximal projection (XP) dominates an intermediate

projection (X') and a XP (a Spec) (Rule 8a) . It also
expressed the fact that X' dominates a head and a

complement (Rule 8b)
(8) a. XP—» XP(Spec) . X ’
b.X ’—» X°. XP (Complement)
In this approach, hierarchy has been explored with an
eye to universals, while variation in order has been
accounted for by parameters (Chomsky 1986, pag 91 fn 1) .
For example the difference between English and Japanese

with respect to the order of complements and heads was
expressed by the Head Parameter. The Head Parameter states
that heads precede their complements in English, but heads
follow their complement in Japanese. Travis (1989) takes a
further step and proposes that linearity is not only
determined by the Head Parameter. She extends concept of
linearity to other modules of grammar. On this view, case
assignment as well as theta role assignment are sensitive
to what it is called directionality. Different settings
for the directionality determine different typologies of

languages.
8
1.3 Soma Inadequacies of the Symmetric View.
Evidently, in all these approaches, hierarchical
relations and linearity are dissociated. Under these views

symmetric orders for the same hierarchical relations are
crucially not excluded. Thus, we expect the following
possible four realizations of the Spec Head Compl template
in the different grammars.

(9)
XP XP
Spec
Spec
Head Head
Compl Compl
XP XP
Spec
Spec
Compl Compl
Head Head
1.3.4 The asym m etry o f specifiers.
Kayne (1994) shows that this way of conceiving UG is too

permissive; natural languages are less symmetrical than
this pattern predicts. One asymmetry involves the position
of the specifiers in phrase structure. Because complements
may appear either before or after the head, it might be
expected that the specifier would be capable of a similar

mirror-image pattern. However, in human language,
specifiers typically appear initially; the predominant

orders are Spec-Head-Compl and Spec-Compl-Head.
For example, subjects, which are in the specifier
position of a verb phrase, precede the verb and complement
far more often than they follow them. Languages with a
predominantly subject-verb-object (S V 0} order—such as

English—and subject-object-verb (S 0 V) order—such as in
Japanese—are quite common. However, languages with a
predominant order of object-verb-subject (0 V S) and verb-
object-subject (V 0 S)—are, on the other hand, "excessively
rare" as pointed out by Greenberg (1966) in his survey of

genetically unrelated languages.
Another example of the asymmetry of specifiers is Spec
CP. Since Chomsky (1986), it has been assumed that Spec CP
is the landing site for Wh-elements. In languages like
English or Spanish Spec CP is at the left edge of the

clause. However a mirror image language, which
consistently moves its wh-element to a final position is

hardly attested.2
2 The exception to this claim is would be Khashi as

mentioned in Kayne (1994), who cites Ultan (1978) and Bach
(1971) as sources.
10
1.3.5 Asymmetries in the Agreem ent patterns.
Many languages, including Spanish, permit permutations
in constituent order, yet in these cases the different
orderings have different morphosyntactic consequences. For

example, some languages like Arabic alternate S V O and
V S O orders. In the S V 0 order the verb and the subject
must agree, but in the V S O order this agreement may fail
to appear. There is no language that presents the opposite
pattern.
The facts point to the conclusion that the failure of

agreement is possible when the verb C-commands the DP
subject (as in the V S O order) . If subjects in the V S O
order were able to be in the same specifier of the verb,

instantiating a case of specifier to the right, no
asymmetry would obtain between V S O and S V 0 orders.
Yet another asymmetry in the agreement system is found
with adpositions. Post-positions allow agreement between
the NP and P; however, this relation is not typical of
prepositions. If agreement is taken to be a Spec Head
relation, the complements of a postposition are in an
specifier position, while this is not the case with a

preposition, contrary to what one might expect given the
directionality parameter.
In conclusion the asymmetries in agreement patterns
overwhelmingly show that specifiers have to be thought of

as being to the left of the head.
11
1.3.6 Asym m etries on the directionality o f Head Movement.
Another set of asymmetries involve movement. The V2
phenomenon typical of many Germanic languages has been
described as consisting of head movement of the verb to a
C° position. Under this view, verb movement takes place
from right to left. Thus, the complement IP, where the

verb moves from, is to the right of C°
(9) [cpWh [e Vi [1p t i lll
Yet the mirror image of the Germanic V-2 does not seem
to be attested in natural languages. That is, there is no
known language in which a verb precedes the final
constituent of a sentence in root clauses. This fact
implies that the opposite configuration with a Complement
IP preceding the Head C cannot be a possible configuration

in natural languages.
(10) HC t „] V°c] Wh cl
Another case in which head movement seems to be leftward
is shown by the patterns of number agreement in languages
that present the VSO and S V 0 alternation. These
patterns indicate that the verb must C-command the subject
in the V S O order, but it cannot do so in the S V 0 order.
Thus, there can be no rightward head movement of the verb

past the subject to yield the S V 0 order.
The facts presented in this section indicate that head
movement is always to the left. Assuming that head

movement takes place from the head of the complement to a
hierarchically more prominent head, we conclude that heads
must precede the complement where the head movement begins.

In conclusion, the phenomena reviewed in this section
indicate that there is great tendency in natural languages
for specifiers to appear to the left of their respective

head and complement. It was also shown that complements
follow their head in the cases of head movement. These
simple facts indicate that the symmetric view advocated by
theories in which hierarchical structure and linear order
are independent leaves us with a underconstrained theory
for UG.
1.4 Spec Head Complement as Universal order.
Given the asymmetries shown in the previous sections,
Kayne (1994) advances a new and radically restrictive
schema. In spite of superficial appearances, the relation

between linear order and hierarchical structure is rigid.
The only order in human language is the one where the
specifier precedes the head, and the head precedes the

complement as in (11):
13
(11)
Spec
X Head
HmA
Compl
According to this view, the different orders we see on
the surface between take English and Japanese as discussed
in (2) and (3), are product of movement. While in English

the object remains in complement position, it gets
displaced to the left becoming a specifier in Japanese.

(12)
English Japanese
Head _ , Tnirvn rr Compl

Compl lokyo Kara
fr°m Tokyo
Note that according to this hypothesis, these English
and Japanese examples do not merely differ with respect to
linear order; they also differ with respect to hierarchical
structure. That means that the object in Japanese is not

only to the left of the preposition, but it is also more
external in hierarchical terms. Thus, one of the outcomes
of this hypothesis is that every difference in linear order

in the same language or across languages must reflect a
difference in hierarchical structure.
Kayne (1994) attempts to go beyond the idea that the

universal order is specifier-head-complement by positing a
direct mapping between hierarchical structure and linear
order:
14
It Is legitimate and necessary to ask why the human language faculty displays
the particular linear ordering that it does. Why do heads always precede
complements and why do specifiers and adjoined phrases always precede heads?
I provide a partial answer to this question, starting from the assumption that
there exists a mapping between hierarchical structure and the observed linear
order that is rigid. (Kayne, 1994, p. xiv)
He refers to this mapping as the Linear Correspondence
Axiom (LCA) . The function of this axiom is to map the

different hierarchical relations established between
specifier, head, and complement into the corresponding
linear order. The formulation of the LCA signifies a major
shift in our conceptions of how language is designed, and

its consequences are therefore far reaching.
1.5 The formulation of the LCA.
Assuming that the universal order is Spec-head-Compl,
there is a redundancy between hierarchical relations and

linear order. In a given projection, less embedded
elements always precede more embedded ones. Thus

specifiers (least embedded) precede heads and complements.
Complements (most embedded) are preceded by heads and
specifiers.
Thus, there is a need to find a hierarchical relation
with similar properties of linear order. Such a relation
must be antisymmetric, transitive and total. Kayne
15
postulates that such a hierarchical relation is the notion
of asymmetric C-command. C-command is defined as follows:3

(13) C-command: A C-commands B, if all categories dominating A dominate B and A excludes
B.
The relation between hierarchy and linearity is mediated
by a trivial mapping from non terminal nodes, which are the

ones relevant for hierarchical relations, to the
corresponding terminal nodes, the relevant ones for
linearization. This mapping is called d(X) . Kayne
postulates that for any given non terminal node A which
enters into asymmetric C-command with other non terminal B,
there will be a map into precedence of the terminals
dominated by a non terminal node A with respect to the

terminals dominated by the non terminal B .4 The mapping
must necessarily cover all the non terminal nodes in a
given phrase marker. The principle that governs this
mapping is called the Linear Correspondence Axiom (LCA) .

(14) ...Take A to be the maximal such set; that is, A contains all pairs of nonterminals such that
the first asymmetrically C-commands the second.(...) for a given phrase marker P, with T
the set of terminals and A as just given) (Kanye 1994,)
3 The only difference between the notions of linearity

(e.g. precedence) and asymmetric C-command is that
asymmetric C-command is only partially total. Thus, there
might be nodes in which the asymmetric C-command relation
is not established. However, given binary branching, Kayne
observes that it is partially total in the sense that for a
given node in the tree, it covers all the nodes that
asymmetrically C-command that node.
4 This is only a partial answer as to why the order has to
be Spec Head Complement. Nothing bars the possibility that
the mapping from asymmetric C-command into subsequence,
instead of precedence. This second option would yield the
Compl Head Spec order. Kayne takes the first possibility
to be the right one (see his discussion in 4.3)
16
This specific formulation of the LCA has important
consequence for the whole conception of phrase structure.

For the purposes of this dissertation I will just review
the consequences of antisymmetry for the way of conceiving

specifiers and adjunctions.
In order to see how specifiers and adjunctions fit into
this whole approach, we have to discuss the phrasal status
of X'. Consider a tree like (15) in which X' is taken to
be an intermediate projection and YP is in the specifier
position. The tree should not be admissible by the LCA

given the fact that it leads to mutual asymmetric C-command
between YP and X° and between X ' and Y°.
(15)
XP
Y° xw
y x
Therefore, trees with Specifiers must be reconsidered.
Kayne proposes a solution in which X' is taken to be a

segment in the sense of May (1985) and Chomsky (1986) . If
we assume that C-command is only relevant for categories*,
the tree under (15) is allowed. The X' is a segment and it

is unable to C-command any structure inside the specifier.
17
The category YP can C-command X°. The revised definition
of C-command is the following as in (16) :
(16) X C-commands Y ifFX and Y are categories and X excludes Y and every category that
dominates X dominates Y.
This solution has a number of important consequences.
For one thing, the status of intermediate category is

eliminated. Phrase structure only considers heads (non
terminals immediately dominating a terminal), maximal
categories (non terminals dominating another non terminal),
and segments of maximal projections and heads. As a
consequence, the formal differences between specifiers and

adjuncts are eliminated.5
The assumptions thus far also eliminate the possibility
of multiple adjunction to the same projection. Consider
the following tree in which two maximal projections ZP and
YP are adjoin to the same projection XP. ZP asymmetrically
C-command Y° and YP asymmetrically C-commands Z°. We obtain
5 Chomsky's (1995) bare phrase structure system does not

have these consequences. Chomsky dissociates the notions
of maximal/minimal and projected/non projected as in
Muysken (1982) . For Chomsky an intermediate X' projection
is not maximal and not minimal and is therefore invisible
for the computational system. Thus X' is not able to C-
command into its sister specifier. Since Chomsky's system
maintains the notion of intermediate projection and
separate segments, adjuncts and specifiers are formally
distinguished.
18
a situation of mutual asymmetric C-command between ZP and

YP. The trees are not admissible. 6
ZP
YP XP
y X
The situation of multiple heads is slightly different.
In a tree like (18), corresponding to the adjunction of two
heads to the same projection, we obtain that Zc and Yc C-

command each other mutually since they are dominated by the
same maximal projections. There is no asymmetric C-
6 Cinque (1996) considers a new definition of C-command

that would allow multiple specifiers. The definition
proposed is the following:
"X C-commands Y iff X and Y are categories and X excludes Y
and every segment that dominates X, dominates Y"
In that case the upper specifier ZP in the tree in (17) is
dominated by one more segment of XP than the lower
Specifier YP, thus creating asymmetry. This, however, has
some undesirable consequences. It does not allow a head
adjoined to another head to C-command out, since segements
are barriers for C-command. Thus, a head X° adjoined to
another head Y° cannot C-command out its trace since the
upper segment of Y° dominates X° but not the trace of X°.
Equally, an XP adjoined to YP in a successive way will not
be able to C-command its trace. Barbiers (1995) proposes
an alternative in which C-command includes the linear
notion of left "branching". In this alternative multiple
specifiers are allowed and head movement is allowed by the
general principles of semantic interpretation that he
proposes.
command, and no order between the heads Z° and Y° is
obtained.
(1 8 )
X°
However, we still should permit cases of adjunction of
one head to another head. Given the definition of C-
command, we can obtain this result by taking advantage the

notion of exclusion. Thus in (19) , X° does not C-command Yc
since X° does not exclude Y°, even if they are dominated by

the same categories. However, Y° C-commands X° since it
excludes Xa. This asymmetry in the exclusion relation,
allows Y° to C-command X° asymmetricaly. Therefore, head to
head movement is still allowed in the system.

(1 9 )
XP
Let us summarize the different aspects of the definition
of C-command that can lead to antisymmetry in a given tree:

Non Adjunctions
A. Non terminal nodes A and B that exclude each

other. Asymmetry is induced by the fact that B is
20
dominated by one more projection than those that
dominate A. In that cases A asymmetrically C-
commands B. This is a simple case of a tree
containing a head and a complement.

Adjunctions
B. When Category A adjoins to category B, A excludes

category B, while B does not exclude category A.
By definition asymmetry between the two categories
is obtained. Thus A asymmetricaly C-commands B.

C. Asymmetry is also obtained between category A —
adjoining to category B— and the lower segment of
category B. By definition segments cannot C-
command, thus A can C-command into the non

terminals dominated by B, while a segment of B
cannot C-command into the non terminals dominated
by A.
As we have seen so far multiple adjunctions to the same

head, or multiple adjunction to the same maximal projection
is banned. However, more than one adjunction is still

allowed in an interesting way. Adjunction structures are
permitted in the system because the adjoining category
excludes the projection it is adjoining to (the adjoinee)
(Paragraph B). Moreover, adjunction creates segments which

by definition cannot C-command (Paragraph C) . Thus, we can
always create antisymmetry by adjoining to a category, as
21
far as this category has not been already adjoined to. One
example of this type of adjunction is given in the

following tree: YP asymmetrically C-commands ZP since it
excludes it. ZP does not C-command YP because it does not
exclude it and the lower segment of ZP does not C-command
Y°. This is called successive adjunction as opposed to

multiple adjunction.
(20) XP
ZP
ZP
YP
The same procedure of succesive adjunction can be shown
to work with adjunction of one head to another head as in
(21) . Z° C-commands Y° and X° since it excludes both Y° and
X°. However, neither X° nor Y° exclude Z°. And neither
lower segment Y° and X° C-command Z° by definition.
(21) XP
X°
22
To conclude this section, I would like to summarize some

of the most important consequences of adopting the
antisymmetry proposal of Kayne (1994) for specifiers and

adjunctions:
a) The difference between specifiers and adjunctions

collapses.
b) Specifiers and head categories C-command out of

their category they are adjoined to.
c) There is no possibility of multiple adjunction to
the same head or to the same maximal projection.
Thus, there can be no multiple specifiers in the
language.
d) It is nevertheless possible to have "successive"

adjunction.7
1.6 LCA and its consequences
The adoption of the LCA and it consequences for the
conceptualization of phrase structure puts into question

many analysis assumed until now in the Romance languages.
For example, linguists have long taken the availability

of right adjunction or specifier to the right for granted,
and they have used these mechanisms in various analyses.
However, the use of such apparatus obviously conflicts with
7 See Barbiers who exploits this possibility for the

analysis of extraposition in Dutch.
23
the claim that specifiers and adjuncts are to the left of
the constituent [head-complement] .8
One construction in which right adjunction has been used

is subject inversion, a phenomenon common in a number of
Romance languages, as illustrated in (22) to (24).
(22) Ayer compitf eldiccionario un chico.(Spanish)
yesterdaybought the dictionary a kid
A kid bought the dictionary yesterday.
(23) Ahir vacomprar el dicdonari un noi.(Catalan)

Yesterday bought the dictionary a kid
(24) Ieri hacomprato il dizionario un ragazzo.(Italian)
Yesterday has bought the dictionary a kid
Since the earliest analyses of subject inversion in

Romance, it has been assumed that subjects are moved to a
post-verbal position in which they appear adjoined to the
right edge of the VP. This was the position taken by Kayne
(1972) in his analysis of stylistic inversion, Kayne and
Pollock (1978), and Rizzi (1982). It was subsequently
adopted by Suher (1994) and Torrego (1984) for Spanish, by
Motapayane (1991) for Romanian, and by SolA (1992) and

Bonet (1988) for Catalan.9 The rethinking of this analysis
for subject inversion will be the major topic of discussion
of Chapters 2 and 3.
With respect to the new analysis of adjunction,

antisymmetry crucially eliminates multiple adjunction to
8Recall that Greenberg (1966) pointed out that that V O S

orders are quite rare.
9 Giorgi and Longobardi (1991) also adopt the position that
subjects in nominal are to the right in Romance
24
the same maximal projection and multiple adjunction to the
same head. Multiple adjunctions to the same head have been

assumed in certain analysis of clitic constructions in the
framework of Pollock's (1989) inflectional projections.

One example is Kayne (1991) . We can distinguish two sub
cases: (i) adjunction of V and clitic to the same
inflectional projection and (ii) adjunction of two clitics
to the same inflectional projection. For the first case,
Kayne (1991) suggested that in French or Italian verb and

clitics might both directly adjoin to the same inflectional
projection in finite clauses:
(25)
a vi
This multiple adjunction to an inflecional projection

has also been taken as a possibility for the cases of
combination of clitics (see Kayne 1989 for the cases of
clitic climbing) :
(26) Glielo dard. (Italian)
to him/her it w ill give
I w ill give it to him/her
(27) Selodar*. (Spanish)
to him/her I w ill give
These analyses, obviously, are not compatible with the

antisymmetry approach. Therefore, Kayne (1994) considers a
new analysis in which proclitics do not adjoin to the same
inflectional projection than the verb. They are taken to
25
be adjoined to a higher inflectional projection X° as in

(28) :
(28)
CL
This analysis have various consequences. For the
Romance languages, it implies that the verb and a proclitic
do not form a complex head before Spell Out. Consequently,
a verb and a proclitic cannot be subject to head movement,
contrary to what it had been assumed in earlier approaches.
The elimination of this movement is particularly
important for the analysis of interrogatives. As we

observe in examples in (29) and (30), in interrogatives
subjects must appear obligatorily in post-verbal position
in many Romance languages.
(29) a. *Cosa lui mangia? (Italian)

What he eats
What does he eat
b. Cosa mangia lui?

What eats he
(30)a. * iQu6 61 come? (Spanish)

What he eats
What does he eat
b. iQu6 come 61?
What eats he
One common approach to all these cases has been the idea
that verb has moved to a higher position, C° as in Germanic
26
languages. This type of approach was adopted by Kayne
(1983) for subject clitic inversion in French, by Rizzi
(1991) for Italian, by Ambar (1988) for Portuguese, and by
Poleto (1993) for northern Italian dialects. However,

there is a certain clash between all these approaches and
the idea that the verb and clitic do notform a
constituent. As we observe in the following examples
proclitics must precede the verb in interrogatives

(31)a. <;Qu6 le dijo Juan?
what cl-said Juan
What did Juan say to him/her
b. *iQu6 dijo Ie Juan?

what said-cl Juan
(32) a. Cosa gli dice Gianni?

What cl- says Gianni
What does Giannit says to him/them?
b. *Cosa dice gli Gianni?
What says-cl Gianni
Thus, from the perspective of the antisymmetry approach
the obligatory inversion effects cannot only be due to the
fact that verb has moved higher in these constructions.

This will be topic of our discussion in Chapter 4.
The final goal of this study is to be able to arrive at

a better understanding of the various word order
permutations in the Romance languages within the highly
restrictive approach of antisymmetry. Investigations of
this type are necessary in our attempt to attain a better

conceptualization of syntactic knowledge and phrase
27
structure in general and of word order permutation in

particular.
In the end, the main aim of contemporary formal
linguistics is to develop a theory of the human language
faculty, which is general enough to capture the universal
features of language, and flexible enough to account for
the variability that is in fact observed among specific

languages. The variability manifested by the Romance
languages offers an ideal domain of investigation towards
the accomplishment of the above aim.
28
Chapter 2
The V S O /V O S alternation in Spanish
2 .1 Introduction
This chapter deals primarily with the alternation V S 0
and V O S in languages that allow it such as Spanish.
Traditionally the VOS order has been understood as

involving adjunction of the subject to an inflection
projection to the right or as a case in which the Specifier
of the subject projects to the right (Rizzi 1982, Torrego
1984, Suner 1994, SolA 1992) . As I mentioned in the
introduction, such hypotheses are incompatible with Kayne's

antisymmetry proposal. In this chapter, I will propose an
alternative in which the V O S order is generated by the
movement of the objects to the left of the subject.
Reproduced with permission of the copyright owner. Further reproduction prohibited w ithout permission.
29
Empirical support for this hypothesis comes from certain

syntactic asymmetries between V S 0 and V O S . Some of
these asymmetries reflect the fact that the object

C-commands the subject in the V O S order but not in the
V S 0 order. In other cases, the asymmetries show that
certain types of objects cannot move to the left and yield
the V O S order. This is the result predicted by the

constrained nature of scrambling. Furthermore, there is a
parallel between these alternations in Spanish and ones

described in scrambling languages (e. g. German) with the
S 0 V and 0 S V alternations. Finally, this hypothesis
fits neatly with Kayne's antisymmetry proposal since right
adjunction is eliminated in favor of an already attested
mechanism of scrambling.
I will start by discussing the focus properties of V S 0

and V 0 S in Spanish, leaving aside the discussion of the
VOS orders in Italian and Catalan to Chapter 3.
2.2 The V S O end V O S order in Spanish.
Spanish is an S V 0 language which nevertheless allows

its subjects to appear post-verbally before or after
objects. The V S 0 order is attested in declaratives as

well as questions, as in (1) and (2):
(1) £A quidn Ie prestd Juan el diccionario?
to whom cl-lem Juan the dictionary
Who did Juan lend the dictionary to?
30
(2) Espero que tc devuelva Juan ellibro.

I hope that cl-you-retum Juan the book
I hope Juan returns th e book id you.
The V O S order is also attested in questions and

declaratives, as in (3) and (4 ) .
(3) iA qui£n le presuS el diccionario Juan?
to whom cl-lent the dictionary Juan.
Who did Juan lend the dictionary to?
(4) Espero que tedevuelva ellibro Juan.

I hope that cl-you-retum the book Juan.
I hope that Juan returns the book to you.
2.2.1 The V S O order.
Pragmatically, the V S O and V O S orders differ with
respect to their interaction with focus.10 Zubizarreta

(1995) points out that the V S O order can be associated
with a number of different focus structures. The specific
structure depends on which element receives main sentence
stress. If the object receives the main sentence stress,

the assertion could be the object and the subject together
as in (5a). If the subject receives main sentence stress,
the object gets downstressed. In this case the subject will

be the only assertion, as in (5b).11
(5) a. /A au ifn Icnrestd Jaaa «l diccmnario?

to whom cl-lent Juan the dictionary
b. A quign le prestd Juan el diccionario
to whom cl-lent Juan the dictionary
10 For focus I understand all possible material that might be part of

the assertion, as opposed to that material which is already
presupposed.
11 I indicate main sentence stress in bold typeface and possible
assertion in underline typeface.
31
To better see the discourse properties of the V S O

order I will use the question-answer test. For a question
such as (6)— in which the subject is introduced in the
discourse—an answer with a V S 0 order as in (7a) will be

considered inappropriate. Only a response such as (7b),
with the S V 0 order, is possible. It can be concluded

that subjects in the V S O order must not be presuposed
(6) i Qu6 comprd Juan Ayo?

What bought Juan yesterday
(7) a. Answer #Ayer compnS Juan un libra.

Yesterday bought Juan a book.
b. Answer Ayer, Juan compn5 un libro.

Yesterday, Juan bought a book.
On the other hand, the V S O order can be an answer to a
"what happened" type question as in (8) . This fact is
evidence that the subject in the V S O must be included in

the assertion.
(8) a.' iQu6 pastS ayer?
What happened yesterday
b. Answer Ayer gand Juan la lo te rfa .12

Yesterday won Juan the lottery.
1212 In the context of an answer to a question the V S O and

V O S orders seems to require an initial XP before the
verb. So the order XP V S 0 sounds more natural than the
V S O order. The V S O order might also be preceded by the
conjunction "que":
(0 iQud pasd?
what happened
(U) answer # compnS Juan un perro

bought Juan a dog.
que compid Juan un perro.
that bought Juan a dog.
32
Note that the subject need not receive main sentence
stress, and therefore it need not be understood as the only

focus in this V S O orders.
Further support for this last claim can be found in an

examination of Bare Argument Ellipsis (in the sense of
Reinhart 1991) . One important property of this

construction is that the element appearing on the other
side of the conjunction—the remnant—can only be associated
with the DP that bears main sentence stress in the main

sentences (what Reinhart labels as the "correlate").
Observe the following contrast:
(9) a. The King ordered the killing, not the queen.
b. # The King ordered the killing, not the banquet
(10) a. The King ordered the k illin g , not the banquet
b. # The King ordered the k illin g , not the Queen.
In Spanish the V S O order can be associated with two
types of remnants in bare arguments ellipsis cases. Either
the remnant has the subject as the correlate as in (11a) or

the object as the correlate as in (11b).
(11) a. Ayer meregald tuhermano el disco. no tu hermana.
yesterday cl-gave your brother the record, not your sister.
b.No me regakS tu hermano el <tism sino la Iibreta.
Not cl-gave your brother the record, but the notebook.
Interestingly, with determiner less DP objects in the

VSO order, subjects cannot be correlates:
(12) a. En nuestra opinidn, no nos dio tu hermano alegria, sino tristeza.
In our opinidn, not cl-gave your brother happiness, but unhappiness.
b. #En nuestra opinidn, no nos dio tu hermano alegria, sino tu u'a.

In our opinion, not cl-gave your brother happiness, but your aunt.
33
(13) a Hntucasa, no da tuhermana besos, sino tortazos
In your bouse, not gives your sister losses, but slaps.
b. #En tu casa, no da tu heimana besos, sino tu padre.
In your house, not gives your sister kisses, but your father.
This last fact clearly shows, that whereas subjects in
VSO are part of the assertion—as in (7a)—sometimes they
cannot have main sentence stress and be narrowly focused.

To conclude, we have seen that in Spanish the VSO
order can be associated with two informational structures:
In the first the subject is part of the assertion but does
not bear main sentence stress (what I will call a neutral
post-verbal subject) . In the second—and more marked case—

the subject bears main sentence stress and is also part of
the assertion, while the object that follows is
downs tressed.
2.2.2 The V O S order.
In the V O S order, the subject receives main sentence
stress, and it can be the only understood focus of the

sentence.
(14) £,Aqui£n leprestri el diccionario Juan?
to whom cl-lent (he dictionary Juan
This can be shown because V O S order can only be an
answer to a question about the subject, as in (15) . It

cannot be the answer to a question like "what happened?" as
in in (16) .
(15) a. i Qui&i gand la Ioteria ayer?
Who won the lottery yesterday
b. answer Ayer gand la Ioteria Juan.

Yesterday won the lottery Juan
34
(16) a. i Qu6 pasd ayer?
what happened yesterday
answer #que ayer gand la Ioteria Juan.13

that yesterday won the lottery Juan.
The narrow focused, nature of the subject can also be

shown by the bare argument ellipsis construction discussed
in Reinhart (1991) . Example (17b) is marginal with the
object as the correlate.14
(17) a. No me envid un telegrama tu madre, sino tu hermana.
Not cl-sent a telegram your mother, but your sister
b. #No me envid un telegrama tu madre, sino una carta.

Not sent a telegram your mother, but a letter.
Thus, we conclude that in the V O S order in Spanish,
contrary to the V S O order, the subject must receive the

main sentence stress an be narrowly focused.
Finally, parallel to the special nature of the V O S

order in Spanish, Webelhuth (1992, p. 165 fn. 2) has
pointed out that in German, an S 0 V language, the 0 S V
order is possible only when the S is heavily focused.
13 It is important not to confuse the V O S order discussed

here with a V 0 S pattern in which the subject is
downstressed and there is an intonational break before the
subject. Under those circumstances, the answer in (16b)
becomes aprorpriate in a context in which Juan is
presupposed in the discourse.
(Q a. iQu6 pasd ayer ?
what happend yesterday?
b. answer que ayer gano la Ioteria // Juan.

that yesterday won the lottery Juan
5.Again, the sentence is perfect if the subject is right
addojined. This is not the reading that is relevant for
this discussion.
(i) # No escribid un libro # Juan, sino un poema
Not wrote a book # Juan, but a poem.
35
(18) weil das Buch niemand gelesen hat. (from Webelhuth 1992)
because the book nobody read has.
Because nobody has read the book.
Thus there seems to be a parallelism between Spanish

VOS and German 0 S V with respect to the restriction on
focus in both languages.
2.3 The analysis of V S O and V O S .
In view of the proposals about verb movement (Emonds
1978, Pollock 1989) and the VP internal subject hypothesis
(Koopman & Sportiche 1991) the V S O order is derived by

head movement of the verb past the position of the subject.
Suner (1994), in fact, gives independent arguments to think
that verbs in Spanish always move to the highest

inflectional projection of the verb (FP1), and the subject
stays in situ. The core idea of this analysis, which is
that V S 0 is produced by head movement of the verb past

the position of the subject, will be accepted here15.
FP2
VP
comi
Juan
O
ei diccionario
For the V O S order, Suner (1994) and Torrego (1984)
follow Rizzi's (1982) analysis of subject inversion in
15 This analysis will be enriched in chapter 3.
36
Italian and propose that this order is obtained by having

the subject right-adjoined to the VP as in (20) .
(20) The Right Adjunction Hypothesis
VP
Juan
V ei diccionario
Theoretically, there was little reason to question this

analysis before two subsequent advances in linguistic
theory: the proliferation of inflectional projections
(Pollock 1989) and the subject VP internal hypothesis

(Kitagawa 1986, Koopman & Sportiche 1991). After all,
before the VP internal hypothesis subjects were considered
to be base generated in the Spec of INFL. Thus, they could
only end up post-verbally by movement of the subject to the
right, leaving a dummy empty category in Spec of INFL.
Yet even after these advances made alternatives
available, linguists have continued to assume variations of

this Right Adjunction Hypothesis. One representative
version is the idea that post-verbal subjects in Romance
are in Spec VP, which branches to the right. This proposal
has been accepted, for example, by Bonet (1988) and Sol&
(1992) for Catalan and by Friedemann (1995) for French. It

is illustrated in (21).
(21) [vptv ]SU]
37
The proposal that I make here departs from these views.
I claim that the VOS order must be created by the
movement of the object to the Spec of a higher inflectional
projection to the left of the base position of the subject.

The landing site for this movement of the object would be
to the right of the final position of the verb. This

movement should be thought of essentially as scrambling of
a object to the left, and for this reason I refer to it as
"the Scrambling Hypothesis."16
(22) The Scrambling Hypothesis
FP2
VP
el diccionarii
Juan
These two hypotheses differ crucially in the
hierarchical representation of the V O S order. In the

Right Adjunction Hypothesis, the subject adjoined to the VP
C-commands the object in its base-generated position. The
Scrambling Hypothesis posits, on the contrary, that the
16. As we discussed in chapter 1 the distinction between

adjunction to a maximal projection and specifier of a
maximal projection collapses in Kayne (1994). I will
assume that this scrambling is movement to a mixed position
as discussed by Webelhuth (1992) . As we will see in the
discussion of wh-in situ there are good reasons to think
that scrambling is A'movement. However, see chapter 3
sections 3.8.1 and 3.9 for some motivation that this
movement is L-related. For the debate about the status of
scrambling with respect to the A-A' dichotomy see Van
Riemsdjik and Corver (1994).
38
scrambled object in Spec of FP2 C-commands the subject in

Spec of VP.17
Thus, with the Right Adjunction hypothesis there are no

hierarchical differences between the V O S and V S 0. In
both orders the subject C-commands the object. The only
differences lie in the linear ordering of the object and
subject, and so there should be no syntactic asymmetries in

hierarchical terms. From a scrambling perspective, on the
other hand, the orderings are indeed asymmetrical: The

subject C-commands the object in the V S 0 order while the
object C-commands subject in V 0 S order. Therefore, the

Scrambling Hypothesis can accommodate the syntactic
asymmetries shown in subsequent sections.
Interestingly, the alternations between subject and
object in post-verbal position are similar to equivalent
cases that have been described for pre-verbal positions in
17 There is a third possibility to obtain the V O S order,

one in which the object is scrambled to the left and the
subject is right adjoined to the VP. Observe, that under
this alternative the object C-commands the subject, which
is nevertheless right adjoined:
(i) („, object* [w [m» tj t* ] subject^
However, simple economy considerations block the
possibility of this derivation. Given that that the basic
order is S V 0 (see Suher 1994), there are two alternatives
to derive the (V) 0 S with the same number of steps in the
derivation: the Right Adjunction Hypothesis and the
Scrambling Hypothesis. The alternative suggested in (i)
involves the union of those two derivations (right
adjunction and scrambling together). Therefore it involves
one more step and consequently economy will bar it.
Observe that the derivation in (i) also violates the ban
against crossing dependecies (see Pesetsky 1987) .
39
languages such as German, Hindi, and Korean. In these

languages a scrambling analysis is widely assumed. The
differences between Spanish, on the one hand, and Korean,
German, and Hindi, on the other, can be reduced to a

difference in the syntax of verb movement. Verbs end up to
the right of the subject before Spell Out in Spanish but
not in German, Hindi, or Korean. In any case, the

Scrambling Hypothesis predicts that there will be
constraints on V 0 S in Spanish that parallel those on

scrambling found in languages such as German, Korean, and
Hindi.
Before continuing, it may be useful to take an overview
of the different syntactic asymmetries: Section 2.4.1 deals

with the asymmetries between V O S and V S 0 with respect
to binding (quantifier binding in 2.4.1.1, Principle C
effects in 2.4.1.2, and certain reconstruction effects with
subject binders also found in other scrambling languages
(Section 2.4.1.3). Section 2.4.2 examines asymmetries in
the distribution of post-verbal Wh-words in situ. Section
2.4.3 looks at how the interpretation of indefinite objects

is affected by the scrambling of these elements to the
left. Section 2.4.4 deals with constraints on the

distribution of weak subject pronouns. Finally, section
2.4.5 consists of a discussion of scrambling of the

indirect object (10), which has effects on the appearance
of dative clitic doubling.
40
2.4. Tha asymmetries
2.4.1 Binding
2.4.1 A Quantifier binding
The f i r s t a s y m m e try i s fo u n d i n th e d o m a in o f q u a n t i f i e r
b in d in g . It is a s ta n d a r d c la im in th e lit e r a t u r e about
a n a p h o ra th a t in o rd e r to have a p ro n o u n in t e r p r e t e d as a
v a r ia b le b o u n d b y a q u a n t if ie r , t h a t p ro n o u n m u s t b e i n th e
C -com m and d o m a in of th a t q u a n t if ie r ( R e in h a r t, 1983, p.
122, am ong o th e r s ) . T h is c o n d it io n is v io la t e d in th e
e x a m p le s in (2 3 ) w h e re th e s u b je c t p o s s e s s iv e p ro n o u n is
m eant to be u n d e rs to o d as c o r e fe r e n t w it h th e o b je c t
q u a n t if ie r :
(23) a. *Sui amigo leregald un libro [acadaniitojj parasu cumpleafios
his friend cl-bought a book [for each boy] for his binhday
b. *Su i madre IepresenuS (a) [cada nifloJi al director
hisi mother cl-introduced [each boy)] (DO) to the director GO)
As e x p e c te d , th e same c o n s t r a in t h o ld s w it h th e p o s t
v e r b a l s u b je c t s when th e y p re c e d e th e o b je c t q u a n t i f i e r as
in (24) :
(24) a. *&Qu£ le regakS su i amigo [a cada nrfioli
what cl-bought [his friend] (S) [for each boy] GO)
What did his friend buy for each boy for his birthday?
b. *Este libro seloregaUS sui amigo [acadaniflo]j
this book d-gave [his friend] (S) [for each boy] (10)
This book, his friend gave it for each boy.
c. *£Aqui£n IepresenuS sujmadre (a)[cadanifio],?
To whom cl-introduced his mother (S) each boy (DO).
Who did his mother introduce each boy to?
41
d. *Aqui piesentd su imadre (a) [cada nifioJi?
here introduced his mother (S) each boy (DO)
Here, his mother introduced each boy.
In the analysis of the V S 0 order represented in (19),

the possessive pronoun in the subject in Spec of VP is not
C-commanded by the object, and therefore pronominal binding

cannot be established.18
However, when the object quantifier precedes the
subject, the bound interpretation becomes available in all

examples. 19,20
18 As discussed above, there are two prosodic structures

associated with the V S 0 order, a more neutral one- in
which the object receives the main sentence stress and a
marked one in which the subject receives main sentence
stress. While there seems to be consensus that the WCO
clearly obtains under the more neutral structure, some
people find that the cases of WCO improve under the more
marked focus structure in which the subject receives main
sentence stress. There is no agreement between the
speakers I consulted; however, although some seem to get
better, they are not judged as fully grammatical as the
ones in (25).
(ij V. Aqui piesentd SU; MADRE [a cada nifio];.
Here introduced his mother every child
here his mother introduced every child
Observe that the same speakers that allow the marked (i)
also allow tend to allow also (ii):
(iij ?? SU; MADRE piesentd [a cada nifloJi
His mother introduced every child
This problem suggests that certain discourse factors may
override the condition on WCO for somespeakers.See
Zubizarreta (1995) for an account intermsbindingand
discourse.
19 An important issue arises with respect to the
impossibility to C-command out of the dative PP. The
possibility of an NP binding outside its PP hasbeen taken
as aproblem for the C-command approach for binding (see
Reinthart 1983). Recently, Fergusson (1997) proposes a
solution to this puzzle in terms of feature checking at LF.
The relevant features of the NP subcategorized by P would
move out of P and would be taken to a position from which
they would be able to C-command outside.
Reproduced with permission of the copyright owner. Further reproduction prohibited w ithout permission.
42
(25) a. iQu6 leregald [acadaniffoi] su i amigo ?
what cl-bought [for each boy] GO) [his fhend] [S)
b. Este libro se lo regald [a cada nifSo]£ su i amigo
This book cl-gave (far each boy] GO) [his friend] (S)
<;Aqui£n le presentd (a) [cada nilSoli su imadre.?

To whom cl-introduced [each boy] (DO) his mother (S).
d Aquf presentd (a) [cada nifloJi su imadre?
Here introduced [each boy] (DO) his mother (S)
The right adjunction analysis of the VOS as
represented in (26) is incapable of rendering an easy
20 The examples containing negative quantifiers do not seem

to work as well:
(0 *? iQudno leregald [anadieh su* amigo ?
Whatnot cl-gave to nobody his friend
Whad did not give his friend to anybody?
(u) *? £Forqu6 no merecomendd anadie-, su director detesis-,?
why not cl-recommended to nobody his dissertation supervisor
Why did not his dissertation supervisor recommend anybody to me?
However, with more complex negative quantifiers, the
sentences improve, though I do not find them completely
grammatical:
(iii) ?? <,Por qud no me recomendd a ningiin estudiante, suj director de tesis ?
Why not cl-recommend any suident his thesis supervisor
Why didn’t recommend histhesis supervisor any student?
It is interesting to point out that similareffects are
found in English with passives and withcertain double PP
complements. Observe the following contrasts:
(iv) a. This book was given to every boy; by hiSj father,
b. * This book was given to{ no boy by his, father
(v) a. I talked [about every boyj with hiSj counselor
b. *? I talked [about no boy]{ with hist counselor

I think that part of the explanation for these puzzling
facts has to be found in the special properties of negative
quantifiers with respect to focus. If we suppose that
negative quantifiers must get main sentence stress then
that leads to the downstressing of the X P that follows.
This downstressing could be taken to show that that XP must
be necessarily right dislocated. I leave the topic for
further research, noting only that the parallelism between
the Spanish and English facts seems to suggest a uniform
solution.
43
account of these examples in (25) . Under this structural
analysis, the subject right adjoined to the VP would not be
C-commanded by the object quantifier, just as in the V S 0

order.
(26) The Right Adjunction Hypothesis:
-VP
VP,
su i amigo
V [a cada nifioli
Under the scrambling proposal, on the other hand, the
examples are analyzed as in (27) . The object quantifier

(10 or DO) in the Spec of FP2 C-commands the subject in
Spec VP and quantifier binding can be established.
Scrambling creates a new binding possibility which was not
available in either the S V 0 or the V S 0 orders.21
21 I assume in agreement with Frank, Lee, and Rambow (1992)

view that binding theory is a constraint on non-operator
positions.
44
(27) Tree of the Scrambling Hypothesis
FP2
VP
[acadanifiok
su i amigi
Parallel asymmetries with respect to the possibilities

of pronominal binding between the object quantifier binder
and subject bindee have been found as in German, Korean,
and Hindi. These languages differ from Spanish in that the

alternations occur in pre-verbal position. The overt or
covert possesives in the (a) cases of (28)-(30) cannot be
bound because, with their subject object quantifier order,
the C-command condition is not met. However, in the (b)

cases with the order object quantifier subject order, the
binding can be established:
German (From Frank, Lee & Rainbow 1992)22
(28) a. *Ich glaube, daB [seini Vater] jedemidasBild gezeigt hat

I think that [his father (S)] everyone-GO) the picture shown has
I think that his father has shown everyone the picture
22 See Muller and Sternefeld (1994) disagree with the idea

that scrambling has any improvement effect on the binding
facts contrary to Haider (1988), Fanselow (1990),
Grewendorf (1988), Frey (1989), Moltmann (1990), and
Santorini (1990). They do still find some WCO effects in
sentences like in (i):
(0 *?DaB jeden seine Mutter mog
that everybody-ACC his mother-NOM liked
45
b Ich glaube, daB [jedemiJj [sein jVater] dasBildgezeigthat
everyone GO) [his father- (S)] the picture shown has
Korean (From Frank, Lee & Rarobow 1992)
(29) a. *[proi apeci]-ka mwukwu-eykeyi-na yongton-ul cwunta

pro-GEN father (S) everyone-DAT-UQ GO) money-ACC (DO) gives
His father gives everyone money
b [mwukwu-eykeyi-nalj [proj apeci-]Nom yongton-ul cwunta

everyone-DAT-UQ GO) pro-GEN father (S) money-ACC (DO) gives
Hindi (From Mahajan 1992)
(30) a. *[unkii i bahin]sab-koj pyaar Kartii thii

their sister (S) everyone (DO) love do-imp-f be-pst-f
Their sister loved everyone
b [sab-koiJ [unkii [ bahin] pyarr kartii thii

everyone (DO) their sister (S) love do-imp-f be-pst-f
In these languages, all the (b) examples have been

analyzed as cases of scrambling of the object to the left
of the subject. Mahajan (1991) took the facts in (30) to
indicate that there is movement of objects to an A position

(AGRO P) , since scrambling feeds pronominal binding in all
these cases.
2.4.1.2 Principle C effects
Similar asymmetries with respect to scrambling are found
in the domain of Principle C of binding theory. In (31)
the referential expression Eva embedded in the subject
position can be coindexed with the 10 pronoun a ella
without triggering any violation of Principle C since there
is no C-command.23 These same facts hold in post-verbal
23Strong pronouns must be doubled in Spanish. What counts

for binding is the doubled XP. See Varela (1988) for a
justification of this view.
46
position with the order V S 0 in (32) . However, with the
VOS order the co-indexing yields an ungrammatical result

as in (33) .24
Spanish
(31) Los hermanos de Evai Ie compraron el libro a ella;

Eva’s brother (S) bought the book for her (IO)
(32) a. iQu6 Ie compraron los hermanos de Eva; aellai?
What cl-bought Eva’s brothers (S) for her (IO)
What did Eva’s brothers bought for her?
b. El libro, se lo compraron los hermanos de Evai a ellai
the book cl-bought Eva’s brothers (S) for her (IO)
The book, Eva’s brothers bought for her.
(33) a. *(,Qu6 Ie compraron aellai los hermanos de Evai?

What cl-bought for her (IO ) Eva’s brothers (S)
b. *E1 libro, se Io compraron a ellai (IO ) I°s hermanos de Evai (S)

The book cl-bought for her Eva’s brothers
From a scrambling perspective, the ungrammaticality of
(33) with the V 0 S is explained since the 10 has been pre

posed to a C-commanding position with respect to the
following subject, and consequently it is able to trigger a
Principle C effect.25
24 It is important to distinguish the case in (33) from one

where there is a pause between the 10 and the subject as in
(i). With right dislocation of the subject binding can be
established:
(0 &Qu61ecompraron aellai #los hermanos de Evai?
What cl-boughtfor her (IO ) Eva’s brothers (S)
25The deviance of (33) cannot be an effect of backward

anaphora. Backward anaphora is perfectly possible with the
V O S order as in (i).
(0 Ie compraron a suihijo [los directores],?
What cl-bought for his son(IO) the directors (S)?
What did the directors buy for his son?
47
Parallel contrasts to the ones found in (31-33) are
reflected in German and Korean. I just report here the

German examples given by Frank, Lee, and Rambow (1992):
(34) a.., daB [die Briider vom Hansi] ihmi das Bild gezeigt haben
That the brothers of Hans(S) him(IO) the picture shown have
that the brothers of Hans have shown him the picture
b. *..dafi ihmi [die Brfider vom Hans] { das Bild gezeigt haben
That Him (IO) the brothers of Hans(S) the picture shown have
2.4.1.3 Reconstruction effects.
The parallelism between Spanish and the languages where
a scrambling analysis is assumed holds even for cases where
this type of movement shows reconstruction effects. If the

relationship between binder and bindee is reversed— so that
subject is a quantifier binder and the object contains the
possessive pronoun—no asymmetry is established i.e.:
Binding is possible with both orderings as can be seen in
(35) and ( 3 6 ) .
(35) a. iQud le regald [cada niifoli a su[ amigo ?
what cl-bought each boy (S) for his friend (IO )
b. lQu6 Ie regald a su i amigo [cada nifio] i?
what cl-bought for his friend (IO ) each boy (S)
(36) a. Aquf besd [cadanifiali [asuiam iga]
here kissed [each girt] (S) [her friend] (DO)
b. Aquf besd [a su i amiga] [cada nifia];
here kissed [her friend] (DO) [each girl] (S)
A similar lack of asymmetry is found in equivalent cases

in German and Korean:
German (From Frank, Lee, and Rambow 1992)
(37) a. Ich glaube, daB jeder; [seinem i Vaier] die Bilder gezeigt hat.
I think, that everyone-(S) his father-(IO) the pictures shown has
I think that everyone has shown his father the pictures
48
b. Ich glaube, da8 [seinem j Vaier] jeder j die Bilder gezeigt hat.
his father(IO) everyone(S) die pictures shown has
Korean (Prom Frank, Lee & Rambow 1992)
(38) a. [Nwukwuna-kaJi [pro ichinkwuj-eykey komin-ul thelenohnunta.

everyone(S) pro-Gen &iend(IO) problem- tell
Everyone tells his friend problems
b. [pro ichinkwuleykey [Nwukwuna-ka]i komin-ul thelenohnunta.

proGen &iend(IO) Everyone(S) problem- tell
Reconstruction effects are also found with anaphors: the
subject binds the object anaphora—de si m i smas (about
themselves )—whether it precedes or follows the subject
binder. The same effects are found in Hindi with
reciprocals; the scrambling of the object reciprocal does

not destroy the anaphoric relations as shown in (40).
Spanish
(39) a. <;Cudndo hablaron [tus hermanas] j [de si mismas] j ?
When talked your sisters (S) about themselves.
b. i£udndo hablaron [de si mismas] j [tus hermanas] j?
When talked about themselves your sisters (S)
Hindi (Mahajan 1992)
(40) a. [raamOrsiitaa] ek duusre-ko pasand Karte hEn

[Ram and Sita] j (S) each other j (DO) like
Ram and Sita like each other.
b. ek duusre-ko [raamOrsiitaa] pasand Karte hEn

each other] (DO) [Ram and Sita] j (S) like
All these facts add a puzzle to our previous proposal.
Why is it the case that we are not obtaining any

reconstruction effects for all these cases? The first
person to notice any kind of generalization was Santorini

(1990) . She observed that only when the subject is the
binder in all these relations is reconstruction required.
49
Under the Scrambling Hypothesis all the cases that do

not show asymmetry are accommodated in terms of
reconstruction. This in itself is an importantreason for
thinking that the V S O and V O S orders aremediated by
the creation of a chain of the movement of the object to
the left. In Chomsky's (1995, chapter 3) understanding of
reconstruction as an option left by the copy and deletion

theory of movement, these cases would appear as in (41).26
The tail of Chain j, formed by the scrambled object has
been deleted at PF. Observe, that the C-command
requirement is established between the subject quantifier
and the tail of the chain formed by the scrambled object.

(41) Aquf besd [a su i amigalj [cada nifiajj [a su j amigalj
Here kissed [to her friend] [each girl]
As noted in . the previous sections, the relevant
configuration for the computation of bound anaphora and
Principle C is the C-commanding relationship between the

head of the scrambling chain and the subject. Thus, no
reconstruction effects were obtained. The examples are

repeated here as (42) and (43).
(42) Este libro se Io regald [a cada nifioK su i amigo [a cada nino]i
(43) iQud Ie compraron a ellai los hermanos de Evai a ellai •
Therefore there is a need to explain why in some cases

reconstruction effects are obtained.27 To do so, I will*
adopt a derivational approach for the computation of

26 The deleted phonetic material is in bold.
27 I leave aside the question of reconstruction effects
with respect to A-movement.
50
pronominal binding and Principle C effects23. Under this
approach, any point of the derivation in which the relevant

relation of C-command is obtained will trigger pronominal
binding and Principle C effects.
Note that the idea that Principle C has to be computed
derivationally is not a new; it was already proposed by
Lebeaux (1988) and Heycock (1995). Given this approach, in
the previous example (43), the movement of the indirect
object "a ella" enters into C-command with "los hermanos de

Eva" when it is scrambled. At that point, Principle C is
triggered, rendering the sentence ungrammatical. Thus, no
reconstruction effects are obtained.
Similarly, Santorini and Lee (1994) have proposed a

condition on bound interpretation of pronouns that is also
derivational in essence. According to their condition,
binding between a quantifier and a pronoun obtains any time
a quantifier or link of a chain of a quantifier C-commands

a pronoun or a link of a chain containing a pronoun.29 From
that perspective, reconstruction effects depend upon

whether the relevant C-command relation is obtained at any
28 As pointed out to me by Juan Uriagereka, this approach

is incompatible with the minimalist idea that binding is
all computed at LF. However, in an alternative framework
with multiple Spell Outs as in Uriagereka (forthcoming),
the derivational approach could be accomodated.
29 Given that binding had to be understood as restrictions
on nonoperator elements, we automatically exclude the
possibility of Wh in Spec CP binding into a pronoun in
subject position. Thus, the WCO effects are still
maintained in the examples with interrogatives.
51
stage, be it earlier or final, in the derivation. Thus,
pronominal binding for the scrambling of quantifiers can be
satisfied at a later stage of the derivation as we saw in

(25),(28b-29b-30b) and (41) . Specifically, C-command is
obtained between the head of the chain and the subject.
On the other hand, in examples (35b), (36b) (37b), (38b)

and (42) the C-command relationship was already established
at the starting point of the derivation. At this point the
pronoun of the object was already C-commanded by the

subject quantifier.30 Further movement of the object
containing the bound pronoun is irrelevant since the
condition on bound anaphora has already been satisfied. To
conclude, from a derivational perspective, variation

between examples without reconstruction as in sections
2.4.1.1 and 2.4.1.2 and examples with reconstruction as in
section 2.4.1.3 reduces to a single difference: the stage
of derivation in which binding conditions are satisfied.
In conclusion, we observe striking parallelisms with

respect to binding possibilities when the object precedes
the subject in post-verbal position in Spanish and the

equivalent pre-verbal position in German, Korean, and
30 Santorini and Lee's (1994) formulation is more complex

than what I have expressed in the main text. The principle
requires that pronominal binding be satisfied in a certain
domain defined by the existence of a subject Agreement
(what they call binding domain) . In this domain all theta
roles are satisfied (what they call argument domain) . See
Santorini and Lee (1994) for a more explicit formulation of
their principle.
52
Hindi. For the latter cases a rule is widely assumed that
preposes the object over the subject to a higher position.
It is natural and empirically plausible to extend this

analysis to Spanish.
2.4.2 Post-verbal wh-elements.
Another asymmetry between the two orderings is found in
the relative distribution of post-verbal wh elements. The
post-verbal wh-subject must precede the post-verbal wh-
object as can be shown by the contrasts in (44) and (45):

goat-verbal S-Hh-slemeat and .Object.wh-elementa
(44) a t,Qu6 le comprd quidn aquidn?31
what cl-bought who (S) for whom GO)
b. *?&Qu£ le comprd aquiln quidn?
what cl-bought for whom GO) who (S)
(45) a. £Qu6 dijo quidn dequifn?
what said who(S) about who (O)
b. *?i,Qu6 dijo dequidn qui£n?
what said about who (O) who (S)
The same contrast holds with the respective ordering of

post-verbal wh-DO and wh-IO in ( 46) :
(46) a. tQui&i le comprd qu£ aquidn?
Who cl-bought what to whom
b. *?{,Qui£n Ie comprd aquidn
who cl-bought to whom what
31 For all these examples we abstract away from the echo

interpretation.
53
Under the scrambling proposal, the post-verbal sequence
of wh-elements is the result of scrambling the wh-object to
the left of the wh-subject in the examples (44b-45b) and

the movement of the wh-IO to the left of the wh-DO in
(46b).
The distribution of the post-verbal wh-elements in the

(b) configurations seems analogous to a standard case of
violation of the superiority condition. In these cases, a
wh-element has been moved overtly across another

hierarchically superior C-commanding wh-element. However,
it is difficult for a superiority approach reconcile the
ungrammaticality of the (b) cases with the fact that such a
"superiority effect" does not arise between a wh-word in

Spec CP and the post-verbal wh-word (Jaeggli 1982) . A
object-wh can be in CP crossing over a hierarchically
superior subject in (47b):
(47) a. tQ uiln comprd qu£?
Who bought what
b. £Qu£ comprd quidn?

What bought who
It is thus worth considering that the ungrammaticality

of (44b-46b) may not be reducible to the superiority
phenomenon, but to some constraint on the scrambling of wh-
element in situ. It has been noticed for German that wh-
phrases, unlike nonwh-elements, resist scrambling.32 This
32Hindi and Korean do allow scrambling of wh-words. See

Muller & Sternefeld(1995) for a possible parametrical
explanation for the differences between these languages.
54
constraint was pointed out by Fanselow (1990) and Muller
and Sternefeld (1995) as can be shown by thecontrast
between (48b)-with the scrambling of the Wh-Object was -and

(48c)-with the scrambling of the NP das Auto33:
(48) a. Wie hat der Fritz wasrepariert? (Mailer and Sternefeld 1994)
how has Fritz what fixed?
How has Fritz repaired what?
b. *?Wie hat was der Fritz repariert?

how has what Fritz fixed?.
c. Wie hat das Auto der Fritz repariert?
how has the car Fritz fixed?
More examples, similar to the German ones, which show

the ban on scrambling wh-XP in Spanish are given in (49)
which contrasts with (50) :
(49) a iQuidn regald estas flores paraquidn?
who bought these (lowers for who
b iQud dijo comprarle Luis diariamente aquien?
what said to buy Luis everyday to whom
(50) a. *?iQuidn regald paraquidn estas flores ?34

whocl-bought for who these flowers.
Who told whom what happened.
b. *?cQu£ dijo comprarle aqu& i Luis diariamente?

What said to buy to whom Luis everyday.
In order to cope with this ban on the scrambling of wh-
words in situ, Muller and Sternefeld (1993) propose that
the grammar should distinguish between different types of
33 Similar constraint has been pointed out for the

topicalization of wh-elements in Lasnik and Uriagereka
(1988) .
34This order is felicitous with a right dislocation of the
NP that follows as in (i) . This is why the cases involving
more than one wh-word in situ (44-45b) in post-verbal
position are clearer since a right dislocation of a wh-word
in situ is completely impossible.
(i) b tQuidn regald para quidn # estas flores ?
who cl-bought for who # these flowers.
55
A' movement (scrambling, wh-movement and topicalization) in
terms of their landing site.35 They show empirical evidence
that these different processes obey different locality
conditions in the same language. It also supports the idea
that the grammar should have a principle of Unambiguous

Binding that avoids any interaction between these
conditions. This principle would avoid one type of A'
movement (scrambling) feeding another type of movement (Wh-

movement) ,36
35 There are two landing sites for the cases discussed:

Spec of functional projection for scrambled elements and
Spec of CP for wh-movement. We leave topicalization aside.
36 If we assume wh-elements to be indefinites, it is
tempting to think that the ban on the scrambling of wh-
elements is due to the ban on scrambling of nonspecific
indefinites as predicted by the mapping hypothesis of
Diesing (1992) . However, the idea is problematic: on the
one hand, Heim (1987), who assumed that Wh elements are
decomposed into an operator and an indefinite or
existential component, states that the existential
component is always semantically interpreted in its base
position. Thus, at LF the indefinite part would always be
interpreted inside the VP domain in all the cases of
scrambling, rendering the mapping hypothesis irrelevant.
On the other hand, we still might predict that we would be
able to find cases in which the scrambling of wh-word would
be good under a certain specific interpretation. See Heim
(1987) for specific interpretation of sentences containing
a single Wh-elements. However, such sentences are judged *
ungrammatical. Finally, that the restriction on
specificity for scrambled indefinites does not go in tandem
with restriction on scrambling of wh-words is shown by
Hindi, which allows scrambling of wh-elementes (Mahajan
1992), but restricts scrambling of indefinites to the ones
with specific interpretation (Mahajan 1991).
(51) PRINCIPLE OF UNAMBIGUOUS BINDING
a variable that is alpha-bound must be beta-free in the domain of the domain of the head of the
chain (where alpha and beta refer to different types o f positions)
(Mailer & Sternefeld 1993)
In theungrammatical cases (examples 44b-45b) the wh-

word IO a quiSn has been moved to Spec of FP2 the landing
site for scrambled elements before Spell Out. At LF a
qui&n would have to move to Spec CP in order to yield an

operator variable configuration.37 However, at this point
there is a violation of the principleof unambiguous

binding at LF. The original trace of a qui€n in the VP
shell is simultaneously linked to the trace in the
scrambling site Spec of FP2 (t') and the wh-word adjoined
to CP at LF:38
(52)
CP
FP1
FP2
Aquien
VP
Scrambling in he
overt syntax
at LF
37Given Kayne's proposal on antisymmetry (1994) the wh-word

moves at LF adjoins to the one already in Spec CP at Spell
Out.
38We adopt Kayne's definition of C-command. Under this
definition "a quien" C-commands outside the CP in (53).
57
This principle has the additional advantage of

explaining why Spanish has WCO effects with wh-words in
Spec CP as in (53):
(53) a. *? £A qui£n{ vio su j madre?
To whom (DO) saw his mother (S)
Who did his mother see?
b. [VP [A quiSni ItFPi vio[pP2 ti [VP su i madre[ q IDII
In section 2.3.1.1, I showed that scrambling of an DO or

10 can override WCO effects (example 25) . In this case,
this overiding would be possible if overt wh-movement in

Spanish used an intermediate scrambling position as an
escape hatch toward its final landing site in Spec CP.
This potential movement is represented in (52) . In the
first movement to Spec FP2 the Wh-word a qui£n would

C-command the subject in Spec VP su m a d r e and binding could
be established with the possessive pronoun. For the
unambiguous binding approach, however, this could not be a
possibility since scrambling can never feed wh-movement.

In sum, there is good reason to think that the
restricted distribution of post-verbal Wh-words in situ in
Spanish is due to some ban on the scrambling of these

elements. This fact was already noticed for German by
Fanselow (1990) and Muller and Sternefeld (1993) . It is

possible to adopt Muller and Sternefeld's approach of
unambiguous binding which supplies a theoretical way to
deal with the problem. Moreover, the unambiguous binding
condition seems to be independently motivated in Spanish in
58
order to explain the existence of WCO effects with overt wh

extraction.39
2.4.3 Interpretation of indefinites
The objective of this section is to support the claim

that the order verb object-subject is obtained by
scrambling of theobject to the left. As has been pointed
out recently in the literature, scrambling has effects on
the specificity and related scopal properties of
indefinites which are moved by scrambling. For example,
Kural (1992) shows that scrambling can affect the scopal
possibilities of the different quantifiers in Turkish. In
sentence (54a), the subject quantifier "three" is naturally

understood as taking wide scope over the 10 "every car."
However, in sentence (54b),where the 10 is scrambled, the
judgments are reversed, and it is naturally understood with
a wide scope interpretation for the 10:

(54) a. [Oclrisi] [herarafaaya] binmis (Turkish. Kural 1992)
Three person (S) every car(dat) get-in-pst-agr
Three persons got in every car.
b. [Herarabaya] [Qcltisi] binmis
every car(dat) tree person(S) get-in-pst-agr.
Three persons got in every car.
39 The ban of scrambling of wh-words in situ could be

looked from an economy approach as in Epstein (1992) .
Under this view, movement of the wh-word from the scrambled
position would be blocked by a more economical derivation
in which the wh-element in its basic position moves in one
swoop to Spec CP at LF. See Muller and Sternefeld (1995)
for comparison between both approaches and for advantages
of unambiguous binding versus the economy approach.
59
Similarly, an example of the restriction of scrambling
to specific indefinites is pointed out by Diesing (1992)

for German. A subject indefinite inside the VP is
naturally interpreted as nonspecific—or nonpresuppositional
in Diesing's (1992) terminology. If the subject is
scrambled, (i.e., when it appears to the left of the

sentential adverb "indeed"), the interpretation obtained is
specific-or in her terms presuppositional. Thus, example
(55a) simply asserts the existenceof two cellists in a
hotel nonspecifically. In (55b) the twocellists are
already presupposed in the discourse:

(55) a. Weil jadoch zwei Ceilisten in diesem Hotel abgestiegen sind. (From
Diesing 1992)
Since indeed two Cellists in this hotel have-taken- rooms
b. Weil zwei Ceilisten ja doch in diesem Hotel abgestiegen sind.

Since two Cellists indeed in this hotel have-taken-tooms
Given these asymmetries in the scopal and specific
interpretation of indefinites, it is predicted that,
following the Scrambling Hypothesis, similar effects should
be found in Spanish. In fact, the asymmetry can be clearly
seen in cases where the subject is a universal quantifier
and the object is an indefinite. Examples (56) with the

order V S (universal quantifier) 0 (indefinite) are
ambiguous. The indefinite can be interpreted as specific,
a reading that we can express as having an existential

quantifier taking scope over the universal as in (57a)
which corresponds to example (56c) . Another natural
reading is the one where the indefinite is not specific,
60
which we represent as the universal having scope over the

existential (57b).40
(56) a. iQa6 Ie tegalaron todos los estudiantes a un profesor?
what cl-gave all the students (S) to a professor (IO )
b. £Dequ£ informd cadaagente aunespia?
of what informed each agent (S) to a spy.(DO)
What did each agent inform a spy about?
c. Estos Iibros, selosdieron todos los estudiantes a un profesor.

These books, cl-cl-gave all the students (S) to a professor (IO )
(57) a. 3y Vx (student (x) a professor (y)) (give-a-book (x, y))
b Vx 3y (student (x) a professor (y)) (give-a-book (x, y))
However, the previous ambiguity disappears when the

object with the indefinite precedes the subject (58) .
These examples force the specific reading of the indefinite
(see also footnote 40):

(58) a iQu6 Ie tegalaron a un profesor todos los estudiantes?:
what cl-gave to a professor (IO ) all the students (S)
b. <,Dequ6 informd aunespia cadaagente?
of what informed to a spy (DO) each agent (S)
c Estos libros, selosdieron a un profesor todos los estudiantes

These books, cl-cl-gave to a professor (IO) all the students (S)
This same contrast is also found with an indefinite 10
and universal quantifier DO. Sentence (59a) with the order
V DO 10 is also ambiguous between the two readings.
40There is, actually, a third reading in which the

indefinite has narrow scope and nevertheless is understood
as notionally or pragmatically specific. This reading
arises when the indefinite refers to different individuals
out of an already presupposed set. For example in (56b),
that reading would be obtained when we know the spies
involved in the discourse and know that each agent informed
one spy about something -i.e. we have a distributional
reading. For a discussion of this narrow reading of a
specific indefinite see Diesing (1992), and for a general
nonscopal approach to the interpretation of specific
indefinites see En<p (1991) .
61
However, the V IO DO order in (59b) forces the reading
where the indefinite 10 takes scope over the universal

quantifier in the object. This similarity in the
alternation DO 10 with the alternation S IO suggests a
uniform treatment of both cases:
(59) a. Los profesores Ie dieron libro a un estudiante
Professors cl-gave every book (DO) to a student (IO )
b. Los profesores Ie dieron a un estudiante cada libro
Professors cl-gave to a student (IO) every book (DO)
The restriction on specificity of XPs for certain

orderings is further attested with inherently nonspecific
XPs . Such is the case of NPs with the postposed
determiner cualquiera. The prediction that these NPs
could not appear in the scrambled orders V 10 DO and V 10 S

is attested in the contrasts between (60) and (61) :41
41 It has been pointed out to me that the constraints in

examples in (61) might be related to focus. Presumably,
the post-posing of determiner "cualquiera" makes the DP
more emphatic. Parallel effects are found with the post-
posing o f the demonstrative "este" with specific DP's as in
(ii) .
(0 Este nillo.
This boy.
(ii) El nifio este.

The boy this
If focus, and not specificity, were the only factor to
explain the marginality of (61), it might be expected that
examples involving the scrambling of "el nino este" might
be as marginal as the ones involving "una secretaria
cualquiera". However, these examples seem to improve as in
(iii). This, again, suggests that specificificty, and not
only focus, is the crucial factor in explaining the
marginality of the previous examples involving postposed
"cualquiera".
(iii) El director le pasard al nifio este el libro de matemdticas.
The director ci-w ill give to the boy this the book of mathematics.
62
(60) a. El director lepasarS este manuscrito aunasecretaria cualquiera.

The director cl-will give this manuscript (DO) to secretary whichevei(IO)
b. El libro, selodail elprofesor aunestudiantecualquiera.
The book, cl - w ill give the teacher (S) to a student whichever.(IO)
(61) a. ??EI director Iepasarl aunasecretaria cualquiera este manuscrito.
The director cl-w ill give to a secretary whichever (IO ) this manuscript (DO)
b. ??E1 libro, selodard a un estudiante cualquiera el profesor.
The book, cl-will give to a student whichever (IO ) theteacher(S)
The examples with V IO DO and V IO S in (61) each
contrast with those of (62) . The examples in (62) force
the specific reading for the 10 by introducing a partitive

in the sequence with the indefinite determiner
cualquiera:42
(62) a. El director Ie pasara a cualquiera de estas secretarias el manuscrito.
The director will give to any of these secretaries (IO ) the manuscript (DO).
b. El libro, selodari a cualquiera de estos estudiantes el profesor.

The book, cl-will give to any of these students GO) the professor (S).
There has been a common assumption that the indefinite
loses its nonspecific interpretation when it moves out of
its basic position in scrambling structures. In Diesing's
view, for example, this difference on the restriction on
interpretability for scrambled indefinites could be

described under the "mapping hypothesis" under a DRT model.
Diesing postulates that trees in LF are mapped into logical
representation where material inside and outside the VP
corresponds to different kinds of quantification. The
material remaining in the VP maps into the nuclear scope of

the quantification. In this domain all nonspecific
42See Eng (1991) for a proposal that relates specificity to

partitivity.
63
readings of indefinites are obtained by existential
closure.43 The material outside the VP maps into the

restriction clause of the quantification. All
quantificational elements, including indefinites with a

specific reading,44 must appear outside of the VP at LF to
be able to map into the restrictor. One of the empirical

facts that this hypothesis tries to accommodate is that
indefinites already taken outside the VP domain at Spell
Out can only have a specific reading. This is what, in
fact, we have seen for the scrambling example (55b) for
German and (58) and (59b), for Spanish as well.45
43 Diesing (1992) following and modifying ideas by Heim

(1982) and Kamp (1981) supposes that Indefinites with
nonspecific readings introduce a free variable which is
bound by an implicit existential operator in the nuclear
scope.
44Diesing assumes that indefinites with a specific reading
are Quantificational elements which undergo QR.
Indefinites that are in the VP as in examples (56) QR under
the specific reading.
45 Another well-known explatation for the restriction on
specificity has been given by De Hoop (1992) . She divides
indefinites into strong (presuppositional) and weak
(nonpresuppositional), which are assigned two different
types of cases (weak and strong) . Weak case can only be
assigned in its deep structure position (inside the VP)
while strong case is freer. Thus, scrambling is limited to
strong indefinites.
Finally, Reinhart (1995) postulates that the restrictions
on the specificity on indefinites derives from the fact
that the scrambled element is downstressed and therefore
already presupposed in the context under normal
circumstances. This downstressing is consequence of the
rule of assignment of sentential stress. Following
Cinque(1993), she proposes that sentential stress is
assigned to the most embedded element in the structure. In
scrambling structures the downstressed element is the one
that precedes the XP that has sentential stress (see
section 2.2.2 for the V O S order in Spanish).
Consequently, Reinhart's proposal has to assume that the
64
In sum, in Spanish the specific reading is forced for
the object in the V O S order but not the V S O order.

This result is obtained under the standard assumption that
scrambling is only possible if the scrambled element has a
specific interpretation. The effects in Spanish thus match
those found in other languages in which a scrambling

analysis is widely accepted. Again, this fact corroborates
the idea that the first order is obtained from the second
by scrambling of the object to the left.
2.4.4 The distribution of post-verbal subject pronouns.
The fourth asymmetry is found with the distribution of
post-verbal subject pronouns. Unlike fully referential

NP's, unstressed post-verbal subject pronouns are
restricted to the V S 0 order as can be seen by the
contrast in (63) and (64) :46
(63) a. lQ u 6 lescomprd 61 a sushermanos?
whatcl-bought he(S) forhissiblings(IO.).
b. ??<,Qu6lescomprd asushermanos 61?
whatcl-bought forhissiblingsGO) he(S)?
scrambled indefinites are in a more hierarchically

prominent position than the focus element that follows.
This fits very nicely with our proposal that the object is
C-commands the subject in the V O S order.
46 Pronouns which are morphologically complex like nos-
otros, vos-otros do not trigger the effects observed by the
mono-morphemic ones (i) . Under Cardinaletti and Starke's
(1994) approach, these pronouns pattern with what they call
Strong pronouns, which are not subject to the
distributional restriction that the monomorphemic ones are.
(0 b. iQu6 lescomprasteis asushermanos vosotros?
whatcl-bought forhissiblingsG-O) you-plural(S}
What didyou buy forhissiblings?
65
(64) a. iQu6 lecompraronellos a61 ?
what cl-bought they(S)forhim (IO) ?
b. 7?iQu6 lecompraron a61 ellos?
what cl-bought forhim (IO.) they(S)?
The ungrammatical examples with the subject pronoun
following the complement contrast with the example where
the post-verbal subject in the V O S order is a full NP.
Observe the contrast between (63b-64b) and the examples in

(65) .
(65). a. iQu6Iecompraron a61 Iosestudiantes?
whatcl-bought forhim ((IO) thestudents
b. tQuglecotnpnS a 61 el profesor?
whatcl-bought forhim GO) theprofessor(S)
In order to obtain the V O S order in these cases, the
pronouns must be heavily stressed, and an intonational
break must appear before the pronoun as in (66) . It might

also appear in this order when coordinated with another NP
(67) :
(66) a. tQu£ lescompnS a sushermanos //EL?
what bought forhissiblingsGO) HE (S).
b. £Qu6 lecomprarona61//ELLOS?
whatcl-boughtforhim //THEY
(67) iQuS lecompidasushermanos 61y suhermana?
whatboughtforhissiblingsGO) heand hissister(S).
The data in (63) and (64) suggest that post-verbal

subject pronouns in Spanish must necessarily be scrambled
to the left. It has been noted in the literature that the
distribution of unstressed pronouns can be different from

the distribution of full NPs or stressed pronouns.
For example, in Scandinavian languages there is a
process referred to as "object shift", which distinguishes
66
between object pronouns, lexical NPs , and stressed
pronouns. This process obligatorily scrambles unstressed

pronouns to the left of the negation in Icelandic (Johnson
1991) .
(68) a. *,a5 J6n Keypti ekki hann
that John bought not it
b. ,a5 Jdn Keypti hann ekki
That John bought it not
Similarly, subject pronouns also have a peculiar
distribution, which distinguishes them from full NP's.

German allows scrambling of a complement over a subject
full NP (69a) . However, the scrambling of a complement
over an unstressed subject pronoun is much more degraded

(69b). Finally, in Swedish unstressed subject pronouns
must necessarily appear scrambled to the left of sentential
adverbs (70):
(69) a. ..,da£ dem Hans der Automat ein Fahrschein gegeben hat (From CardinaJetti & Starke 1994a)
that to Hans (IO ) the machine (S) a ticket has given
that the machine has given a ticket to Hans.
b. ...,*daB dem Hans er ein Fahrschein gegeben hat
That to Hans (IO ) it(S) a ticket has given
(70) a. —att mdjligen Anna inte har sett boken.

that possibly Anna not has seen book-the
b. ...*att mOjligen hon inte har sett boken.
...that possibly she not has seen book-the
In conclusion, the facts reviewed in this section point
to the conclusion that the constrained distribution of

pronouns in post-verbal position can be explained by the
Scrambling Hypothesis.
67
2.4.5 Doubling with indirect objects.
The fifth asymmetry concerns clitic doubling. Spanish

allows clitic doubling with indirect objects. Some full NP
indirect objects are optionally doubled in the

configuration V DO 10:
Order DO TO:
(71) a Eldirector (les)entregd lasnotas aunos estudiantes.

thedixector(S) dat-cl-gave thegrades(DO) tosome students(IO)
b. Eldirector (le)dio un libro aMaria,
thedirector(S)- dat-cl-gave abook(DO) toMaria(IO)
Hernanz and Brucart (1987, pag 267) observe that
doubling of the indirect object in these situations becomes
necessary when the order of the objects is reversed.47 The

examples in (72) are deviant without the clitic:48
OrderIO DO:
(72) a. Eldirector *?(les)entregd alosestudiantes lasnotas.
thedirectors) dat-cl-gave tothesuidents(IO) thegiades(DO)
47 Zubizarreta(1995) mentions the same intuition (chapter

3, p. 21) . Seealso Uriagereka (1988) for a treatment of
thesecases as counterparts of Dative Shift of English.
48 There is some dialectal variation with respect to the
different possibilities of clitic doubling in Spanish. In
some dialects the clitics are almost always required, in
which case the discussion of this section does not reflect
the grammar of those speakers. For those speakers in which
the clitic is optional, special attention should be paid to
intonation. The judgments reported in (72) have to be read
with sentence stress on the final object, the structure
that corresponds to the intonational properties of
scrambling as in (i). The intonation with the right
dislocation of the DO as in (ii) is not relevant for the
discussion.
(0 a. El director les entregd a los estudiantes las notas.
(ii) a. El director entregd a los estudiante^ # las notas.
68
b. El director *?(Ie)dio a Maria un libro.
the director(S) dat-cl-gave to Maria(IO) a book(DO)
We find that doubling is also required in the V 10 S

order, but not in the V S 10:
SIP,oiden
(73) a. Esta es la asignatura que (les) ensefiaba el profesor a varios estudiantes.
this is the subject that dat-(cl) -taught the professor (S) to some students (IO )
b. Esta es la sefial que Ges)hizo el alcalde a varios hombres.
this is the sign that dat<cl) -made the mayor (S) for some men (IO )
IQ S Oaten
(74) a. Estaes laasignatura que *? (les) ensefia a varios alumnos el profesor.49
this is the subject that dat-ci-teaches to some students GO) the professor (S)
b. Estaeslasefialque *?Qes)hizo a varios hombres el alcalde,
this is the sign that dat-d-made to some men GO) the mayor (S)
These contrasts are clearly problematic for the right-
adjunction hypothesis. Under that proposal, there should
be no contrast for the dative arguments inthe V DO 10 and
V 10 DO orders on the one hand and the V S 10 and V 10 S
orders on the other with respect to clitic doubling. In

both cases the dative is in situ. With thescrambling
proposal, however, there is no incompatibility. Scrambling
therefore must be triggering the clitic doubling, but there
is still is a need to account for these facts. To do so,
it is helpful to employ Torrego's (1992) hypothesis that
clitics head their own DP projection.50 In the case of
doubling, the double XP is realized in the Spec of the DP
49Again, these cases should not be confused with the ones

where we have the same structure but there is a right
dislocation of the final subject.
50 I will put aside DO, wich do not admit doubling in
general. From our perspective, they cannot be subsumed
under representation (75) .
69
where it enters into Spec head agreement with the clitic D

counterpart.51 This DP would be embedded in the lower
larsonian VP shell.52 Example (75) corresponds to (7la)

with the overt clitic dative before Spell Out:
(75)
VP
VP
el profesor y
DO
las notas
PP
a los estudiantes
As a natural extension of the DP hypothesis for clitics,
we also have cases where the D heading the DP is an empty
51This matching is the responsible of the identification of

a "pro" when there is no overt counterpart as in (i):
(i) les entregd las notas pro
to them-cl I gave the grades.
I gave the grades to them
Finally, the matching does not involve the feature ±
specific. It has been shown by Suner (1988) that indirect
object clitics can double nonspecific DP. See example
(73), in which the indefinite can be read as nonspecific
when it is doubled.
52The overt D clitic moves by Spell out to some
inflectional projection. (Kayne 1992) .
70
category. This case corresponds to the counterpart of (71)

with no overt clitic:53
(76) VP
VP
el profesor y
DO
las notas
PP
a los estudiantes q
In general, DPs headed by an empty D have a more
restrictive distribution than lexically realized ones.
Romance DPs headed by an empty head are not permitted in

pre-verbal position.54 They are only allowed in a head-
object relation with the verb as seen by the contrast in
(77) with a passive and (78) with an unaccusative verb:55
53Contrary to overt D's, we assume that empty D's do not

cliticize overtly. If we think that cliticization before
Spell Out is motivated by the need of a clitic to be
checked in an inflectional projection, empty D's are
excluded because they do not have phi-features to be
checked.
54 Only cases of focalization in pre-verbal position seem
to allow such empty D's (Contreras 1985) .
55There are interpretative difference between empty D in
the cases of clitic doubling in (76) and the empty clitic D
in in DP's in (77)-(78) . In this chapter, we are just
concerned about the syntactic distribution of these empty
elements, which seems constrained by the same syntactic
principles. Another element which has a similar syntactic
pattern is empty complementizer. Many linguists have
proposed that empty complementizers are constrained by
lexical government, as well, see Stowell (1980).
(77) a. *? [0 Peiroleo] fueencontrado.
oil (S) was found.
b. Fue encontrado [0 Petroleo] (example from Bosque 1992)

was found oil (S).
(78) a. *?Aquf [0 turistas] Uegan de todas partes.

Here tourists (S) come from everywhere
b. Aquf Uegan [0 turistas] de todas panes
here arrive tourists from everywhere
Longobardi (1994) proposes that the appearance of the
empty determiner is constrained by a lexical government
requirement. As has been pointed out several times in the
literature (Chomsky 1981, Rizzi 1982, among others), this
requirement is not met in the pre-verbal position in (77-

78) . Following this idea, I propose the following
principle for the licensing of 0:56
(79) A DP headed by 0 must be contained in the projection of the head which assigns a theta role
to it 57
56 This proposal is actually very similar to proposal on

the distribution of D 0 by Uriagereka (1988) and
Uriagereka (1996).
57Being in a theta position plays an important role for the
licensing of 0. Empty determiners are not possible for
subjects of small clauses as. pointed out by Contreras
(1985) and Longobardi (1994) . In these cases, the verb in
the matrix clause does not assign a theta role to the Spec
position of the small clause. From this fact, it is
concluded that lexical government is not enough to explain
the distribution of empty D's.
(0 * Consideravo studend intelligent! (from Longobardi 1994)
I considered students intelligent
We could alternatively reformulate (79) in terms of 0-
government in the sense of Chomsky (1986).
(79’) 0 must be 0 -governed.
0 - government: a 0 -governs 3 iff a is a zero-level category that 0-marks 3. and a, 3 are sisters.
(Chomsky 1986)
72
This requirement is met in (77b) and (78b) with the only
argument in the object position, but not when it is moved

to the pre-verbal position in (77a) and (78a). In example
(71) with a dative DP, 0 is also included in the V, which
is responsible for the assignment of the goal theta role to
the whole dative DP. However, this configuration is

destroyed when the DP 10 is scrambled to the left to Spec
FP2. Scrambled XPs end up in a nonthematically selected
projections. This has been taken to be a crucial point in

the explanation for the constraint on extraction out of
scrambled XPs in German (Diesing 1992, Grewendorf & Sabel

1994). In (80), which represents sentence (72a), the DP
dative is in the Specifier of FP2, which is not

thematically selected by V. Consequently the empty
determiner in representation (80) is excluded by principle
in (79) ,58
FPl
FP2
VP
entregd
PP
a los estudiantes DO...
(♦0 ) las notas
58 0—government is not met in (80) either. V and the

dative DP in Spec FP2 are not sisters in the strict sense.
73
Another instance of the requirement in (79) comes from

the distribution of generic bare plural IOs. As pointed
out by Fernandez Soriano (1989) and Jaeggli (1982), these
elements are incompatible with an overt dative clitic.
They are only possible with the empty D counterpart.59

(81) a. No (*Ies) des tusllaves a personas desconocidas.
don’t cl-give your keys to strangers.
b. Luis no (*les) envia sus cartas a niflos indefensos.
Luis not cl-sent his letters to defenseless children.
Consequently, these datives would not be able to

scrambled to the left, as shown below: 60
(82) a *?No des a personas desconocidas tus Haves.
don’t cl-give to strangers your keys.
b. *?Luis no envia a niflos indefensos sus cartas.

Luis not cl-sent to defenseless children his letters
In conclusion, we have seen that empty and overt D's in

a dative XP alternate when these element are in-situ. The
conditions which license empty D's are destroyed once the
DP is scrambled to a higher inflectional projection. As

has been assumed for German (Diesing 1992 and Grewendorf &
Sabel 1994), scrambled XPs end up in nonthematically
selected positions.
59 I have no account for why bare plural IO are impossible

with the overt clitic. Presumably, the same factors that
make the presence of a determiner incompatible with the
generic interpretation for plurals DP in languages like
English, may also explain the incompatibility of the
doubling clitic with the 10 generic bare plural in Spanish.
60 Generics, like specifics indefinites are equally
classified as "presuppositional" in Diesing's sense. They
both, map into the restrictor. Therefore the Mapping
Hypothesis as stated in Diesing (1992), cannot be an
explanation for the marginality of (82).
74
2.5 Conclusion.
In this chapter I have shown certain asymmetries between

objects and subjects in the orders VOS and V S 0 in
Spanish. The asymmetries in quantifier binding and
Principle C between the two orders were explained by having
the hierarchical relation between subjects and objects

reversed, implying that objects C-command subjects only in
the V O S order. This relation could only be obtained by
moving the 0 to the left to some higher inflectional

projection and having the subject in a lower position. The
ban on wh-objects in situ in the V O S order was reduced to

a ban on scrambling wh-in situ. The obligatory specific
interpretation of the indefinite object in V O S order was
seen as a consequence of the effects of scrambling on
indefinites. The restriction of subject pronouns to the

V S 0 order was explained by their semi-clitic nature which
forces them to scramble to the left. Finally, the

impossibility of having an empty counterpart of the dative
clitic in the V 10 S order was reduced to the fact that the
conditions for licensing this empty category are not met in
the landing site for scrambled XPs . In every case support
for a scrambling account is found.

This conclusion implies that the alternation S 0 and 0 S
found in verb final languages like Korean, Hindi and German
needs to be extended to languages where this alternation is

obtained post-verbally such as Spanish. In principle,
75
there is little reason why the same basic alternation
should have two different analyses in U G (scrambling of the

object to the left in verb final languages and adjunction
of the subject to the right in nonverb final languages) .
The introduction of functional projections in the syntactic
trees (Pollock 1989) and the hypothesis of verb movement

(Emonds 1978) , made scrambling a plausible analysis for
nonverb final languages. It can therefore provide this
uniform account. From the particular perspective of
Spanish, it has also been shown that this analysis is

empirically superior to the alternative Right Adjunction
Hypothesis.
Finally, our analysis, contrary to the free inversion

hypothesis, conforms with and supports the line of research
put forward by Kayne (1994), which derives linear order
from hierarchical structure. This isdone by what he calls
the "Linear Correspondence Axiom" (LCA) which maps
asymmetric C-command into linear precedence. Since the

mapping is done into linear precedence, no right-adjunction
is generated in syntactic trees.

For the scrambling proposal, the subject asymmetrically
C-commands the object in the representation in (19) and the
object C-commands the subject in the representation in.
(22) . LCA correctly yields the twoorderings V S 0 and
VOS. However, the Right Adjunction Hypothesis
represented in (20) is incompatible with such a reduction
of linear ordering from hierarchical relations. According
to this hypothesis, the subject asymmetrically C-commands
the object in both V S 0 and V O S . It is then a logical
result to eliminate right adjunction of subjects from UG,

in favor of an independently established phenomenon of
scrambling.
77
Chapter 3
Light Predicate Raising and post-verbal

subjects in Romance
3 .1 Introduction.
In Chapter 2 I showed that Spanish allows both the V S 0
and V O S orders, and I proposed that the derived V O S
order is obtained by movement of the object to the left.
In this chapter, I address the question of the status of

post-verbal subjects in other Romance languages such as
Catalan, French, and Italian. This comparison shows that
Spanish allows a wider distribution possibilities for post-"

verbal subjects than these other languages. In order to
understand the more restricted nature of post-verbal
78
subjects in these languages, I consider an approach in
which the TP has moved over the subject in a process akin
to Light Predicate Raising (LPR) . The subjects are then in
a focus position. This analysis of post-verbal subjects in
Catalan, French and Italian is then extended to Spanish

VO S orders I analyzed in Chapter 2.
3.2 Distribution of Post-verbal subjects in

Italian, Catalan and French.
As we examined in the previous chapter, Spanish seems to
allow post-verbal subjects in the V S 0 or V O S order.
However, Italian and Catalan or French are not as flexible

with respect to the different possible positions of post
verbal subjects. I will first examine the distribution of
post-verbal subjects of these languages in the following
sections.
3.2.1 The V O S order.
The V O S is perfectly possible in Catalan with a
focused subject as in (1) .

(1) Demi comprank unUibre la Magda.
Tomorrow w ill buy abook Magda
However, this kind of sentence appears to be more

delicate in Italian. For instance Rizzi (1991) judges a
parallel sentence such as (2) as marginal.61
61It is important not to confuse this V O S sequence with

one in which the S is marginated or right dislocated.
Those sentences are characterized by a clear intonational
79
(2) ?Harisolto ii problema Gianni (Rizzi 1991)
Has solved the problem Gianni
Nevertheless, Calabrese (1990) has pointed out that the
order is plausible in some presentational contexts with an

indefinite object as in (3) .
(3) Ti ha dipinto unquadro Sandro, (from Calabrese 1990)
cl-has painted a portray Sandro
Similarly, Delfitto and Pinto (1992) also report that

the V O S order is plausible with an indefinite subject as
in (4) and (5).
(4) Le ha cantato una canzone Paolo, (from Calabrese 1990)
cl-has sung a song Paolo.
(5) Ha recensito il libro un professore. (from Delfitto & Pinto 1992)

Has summarized the book a professor
Therefore, I will assume that the VOS order is

possible in principle in Italian
French, on the other hand, is more restrictive. The

VOS order is very marginal as reported in Friedemann
(1995) .
(6) *A qui donnera le livre ton ami? (finom Friedemann 1995)
Who w ill give the book your friend.
Nevertheless, the most important differences between
Spanish, on the one hand, and Italian, Catalan, and French,
on the other, is found in the possible elements that can
follow the postverbal subject. In these languages, the
order V S 0 seems to be quite restricted. Here we should
break before the final subject such as in (i) (see

Calabrese 1990).
(i) Harisolto il problema// Gianni
has solved the problem Gianni
80
distinguish different subcases: V s DP, V S PP, V S Adj.,

V S Adv., V S INF, and V S CP.
3.2.2 V S DP (object)
It is generally reported that Catalan does not allow the

V S DP (object) order, and this is judgment is reflected in
(7a), (8a), and (9a).62 The (b) cases with the VOS
cases, on the contrary, are all grammatical.

(7) a. ???Avui fari en Joan el dinar, (from Soli 1992)
Today will make Joan lunch.
b. Avui farh el dinar en Joan.
Today will make lunch Joan.
(8) a. * Vacdrrer en Llufs la Maratd. (from Bonet 1988)

Ran Llufs the marathon.
b. Vacdrrer laMaratd en Llufs.
Ran the marathon Llufs
(9) a. •Quan va fer I’Anna el caffe? (from Vallduvf 19

When made Anna the coffee.
b. Quan va fer el caffe 1’Anna ?
When made the coffee Anna
However, there seems to be a dialectal split in this
respect. The negative judgments are much less robust in
some central and Southwestern varieties63.
62 Again, I leave aside emarginated or right dislocated

subjects in my discussions of Catalan and Italian.
63 I thank Francesc Roca from Sabadell (Central Catalan)
and Merc£ Pujol from Lleida (Western Catalan) for judgments
in this respect. It remains to be seen whether there is a
new tendency in Catalan towards admitting these orders.
Such a tendency would presumably be the result of the
influence of Spanish, owing to the large number of
bilingual speakers (See Ruaix 1994 pp. 12-14).
81
Arguably, in spite of the dialectal variation, all
varieties seem to coincide in not allowing the V S 0 order

with determinerless DPs as shown in the following examples:
(10) *Quan va discutir el professor linguistics?
When discussed the professor linguistics
(11) *A la Maria, lidonari enJoan pastissos.
To Mary, w ill give Joan pastries.
Italian seems clearly more restrictive than Catalan.

The order V S O is very difficult in all varieties
consulted.
(12) ‘ Scrisse Gianni la Iettera. (Grom Zubizarreta 1994)
Wrote Gianni the letter
(13) *Hamangiato spesso Gianni gli spaghetti
Has eaten often Gianni the spaghetti
Obviously, the sentences are ungrammatical with

determinerless DPs:
(14) *Ieri, hannomangiato i ragazzi pisellL
Yesterday, have eaten the boys peas.
The same kind of restriction is also encountered in

French:
(15) *Quand adcrit Jean la lettre? (from Zubizarreta 1994)
When has written Jean the letter
(16) *Ou t6parc Marie sa voiture? (from Friedemann 199S)

Where repaiis Marie his car
(17) *A qui donnera ton ami le Iivre ? (from Friedemann 1995)

to who gave your friend the book
3.2.3 V S PP (complement)
The V S PP order is slightly better, but still marginal

in these languages. Observe the contrast betwen V S PP and
V PP S in (18), (19), and (20). Note that thoseCatalan
82
speakers, who accept V S 0 with full DP objects, find the

(a) sentences grammatical:
(18) a. ??Afur parlava en Joan de la universitat
yesterday spoke Joan of the university
b. Ahir parlava de la universitat en Joan,
yesterday spoke of the university Joan
(19) a. ??No pensa La Magda en les consequ&ncies.
Not thinks Magda about the consequences
b. No pensa en les conseqfi&ncies La Magda.
Not thinks about the consequences Magda
(20) a. ??Ahir va discutir el professor sobre lingOistica?

Yesterday discussed the professor about linguistics.
b. Ahir va discutir sobre Iingfiistica el professor?

Yesterday discussed about linguistics the professor.
SolA (1992) points out animprovement with IO PPs.64

This improvement is reported by all speakers.
(21) ?Aquest Uibre el va regaiar en Joana la Maria (from Solh 1992)
This book gave Joan to Maria
Finally, sentences with temporal and locative PPs are
grammatical:
(22) Qu6 feia en Joan al 1968?
what <Sd Joan in 1968
(23) El Uibre, l’ha posat en Joan al prestatge. (from Soli 1992)

The book, put Joan on the shelve.
For Italian, V PP S orders are alsodifficult as shown

in (24a), (25a), and (26a), which contrast with their (b)
counterparts:65
64 In some varieties of spoken Catalan there is doubling of

IO. For those speakers the S V IO without clitic is
difficult.
65These sentences should be read with heavy stress on the
final subject. It is interesting to point out that the V S
PP in relative clauses these sentences seem to improve.
(i) ?I giomo in cui ha lin'gaio Piero con sua soreUa_
the day in which has fought Piero with his sister
83
(24) a. ??Ha litigato Piero con sua sorella.
fought Piero with his daughter
b. Ha litigato con sua sorella Piero.

Has fought with his daughter Piero
(25) a. ??Non si & lamentato Piero del mal di testa,
Not complained Piero about his headache
b. Non si k Iamentato del mal di testa Piero,
Not complained about his headache Piero
II
(26) a. mi ha parlato Gianni del suo professore.

cl-talk Gianni about his ptoffesor
b. Non mi ha parlato del suo professoreGianni .
Not cl-taik about his professor Gianni
However, the sentences seem to improve with prepositional

IOs:
(27) Questo regaio, I’haspedito Giuliana al suo nipote
This gift cl-has given Giuliana to her nephew
(28) n regaio que haspedito Giuliana al suo nipote
the gift that has sent Giuliana to her nephew
Furthermore, sentences with locative and temporal PPs are

grammatical:
(29) Dove eta Gianni nel 1968?
where was Gianni in 1968?
(30) Ti- ha visto Gianni nella tua casa

cl-saw Gianni in your house.
The French judgments are somewhat similar. Sentences
with complement PPs are not as bad as with DOs, but they
are still awkward:
(31) n Le jour ou parlera Jean k Marie. (Kayne 1986)
The day when w ill speak Jean to Marie
(32) ??Quand changera cette fille d’avis? (Kayne 1972)
When w ill change this girl of opinion.
(33) ??Quand 6crira ton frfcre k sapetite amie? (Kayne 1972)
When w ill write your brother to his friend.
(ii) ?n giomo in cui si k lamentaro Piero del mal di testa

the day in which complained Piero of a headache.
84
IOs in French can appear after the post-verbal subject,
and they render the sentence completely acceptable:

(34) Lecadeau qu’a envoy6 Jean &Marie, (from Kayne 1986)
The gift that has sent Jean to Marie.
(35) La fille dequi s’est plaint Jean k Marie, (from Kayne 1986)
The girls to whom complained Jean to Marie.
(36) Qu’a dit Jean aujardinier?

What said Jean to the gardener
(37) Que promet un bon mari k sa nouvelle Spouse?
What promises a good husband to his new wife?
Other locative PP or temporal PP allow the V S PP order:

(38) Que fait ton fr6re dans la vie? (from Kayne 1972)
What makes your brother with his life
(39) Ou estalld votreami pour trouver la paix? (from Kayne 1972)
Where has gone your brother for finding peace
(40) A quoi s’intdressait cette personne en 1968? (from Kayne 1972)
What was interested this person in 1968
3.2.4 V S Adj.
In the previous sections, there was some variation in
the possible positions of post-verbal subjects in the
different languages. However, all thevarieties explored
seem to coincide in not allow adjectives, presumably part

of a small clause, to follow the post-verbal subject. Thus
we find the following contrastsbetween V S Adj and

V Adj S.
(41) a. * Perqud no sembla en Joan intelligent? (Catalan)
Why not seems Joan intelligent?
b. Perqufc no sembla intelligent en Joan?

Why not seems intelligent Joan
(42) a. *Esfa enJoan veil (Catalan)
becomes Joan old
b. Es fa veil en Joan.
becomes old Joan
85
(43) a. *?Les ha vistes en Joan borratxes. (Catalan)
them has seen en Joan drunk
b. Les ha vistes bonatxes en Joan .

them has seen drunk en Joan
(44) a. *Perch£ non sembrava Gianni intelligente? (Italian)
why did not seem Gianni intelligent.
b. Perchd non sembrava intelligente Gianni? (Italian)

why did not seem intelligent Gianni .
(45) a. *S i sono sentiti tutti i soldati scoperti. (Italian).
Have felt all the soldiers discovered
All the soldiers felt discoverd
b. Si sono sentiti scoperti tutti i soldati . (Italian).

Have felt discovered all the soldiers
(46) a. * E rimasta Maria delusa. (Italian)

Has remainedMaria disappointed
b. E rimasta delusa Maria.
Has remained disappointed Maria
(47) a. ?*Ieri, li ha visti Rita ubriachi. (Italian)
Yesterday cl-has seen Ritadrunk
Yesterday, Rita has seen them drunk
b. Ieri, li ha visti ubriachi Rita. (Italian)
Yesterday cl-has seen drank Rita
(48) *Quand deviendra cecomddien cÎfebre? (French, from Kayne 1972)
When w ill become this comedian famous.
3.2.5 V S Adv (de-adjectival).
The same ungraramaticality is found with adverbials

derived from adjectives in all varieties of Catalan,
Italian, and French in the V S Adv order:

(49) a. *No parla elteugermh clar. (Catalan)
Not speaks your brother clear
b. No parla clar el teu germh. (Catalan)

Not speaks dear your brother
(50) a. *Tteballa en Joan dur. (Catalan)

work Joan hard
b. Treballa chr en Joan. (Catalan)
work hard Joan
86
(51) a. *Sempre juga la Magda bruL (Catalan)
Always plays Magda dirty
b. Sempre juga brut la Magda.

Always plays dirty Magda
(52) a. *Lavora Rita dosso, nel suo libro. (Italian).
works Rita hard in his book
b. Lavcra dosso Rita, nel suo libro.

works hanl Rita in her book
(53) a. *? A New York, cammina Gianni veloce. (Italian)

In New York walks Gianni fast
b. A New York, cammina veloce Gianni .(Italian)
In New York walks fast Gianni
(54) a. *Quand arisqud cet duidiant gros? (French)

When has risked this student big.
When has this student risked a lot?
b. TQuand arisqud gros cet&udiant ? (French)
When has risked this student big.
3.2.6 V S INF
Subjects followed by infinitives of modal verbs are all

ungrammatical in Catalan, French, and Italian, as shown in
the following contrasts between the subject following or

preceding:
(55) a. *Eldia que volia en Joan menjar. (Catalan)
The day in which wanted Joan to eat?
b. Eldia que volia menjar en Joan. (Catalan)
The day in which wanted to eat Joan?
(56) a. * No podria la Magda menjar? (Catalan)

Not could Magda to eat
No podria menjar la Magda? (Catalan)
Not could to eat Magda
(57) a. *D giomo in cui voleva questo ragazzo mangiare. (Italian)
The day in which wanted this boy to eat?
b. II giomo in cui voleva mangiare questo ragazzo.
The day in which wanted to eat this boy?
(58) a. *Non poteva Gianni mangiare (Italian)
No couldn’t Gianni to eat.
87
b. Non poteva mangiarc Gianni.
No couldn’t to eat Gianni
(59) a. *Que voulait cejeune homme manger? (from Kayne 1972)

What wanted this young man to eat.
b. Que voulait manger cejeune homme ?

What wanted to eat this young man.
(60) a. *Que pouvait la mbre hue?

What could your mother to make
b. Que pouvait faire tambre ?

What could to make your mother
However, sentences in Catalan and Italian become less
deviant if we change to nonrestructuring verbs or verbs
that might be preceded by the complementizer di/de.

(61) ?A nosaltres enshaprombs en Pere de ponar-nos a la platja.
To us , has promised Pere to take us to the beach.
(62) ?Quel giomo in cui ci ha promesso Gianni di portarci in spiaggia.

The day in which has promised Gianni to take us to the beach
In French, however, the sentences are still quite poor:

(63) ‘ Pourquoi t’a demandb Jean deparler?
Why has asked Jean to read this letter?
(64) ?*Quand aessayb Jean dele lire?
When has tried Jean to read it
3.2.7 V S CP
With finite clause sentences there is an improvement in

general.
(65) Qub deia en Joan que volksferdesptbs? (Catalan)
what said Joan that you wanted to do afterwards?
(66) Ahir pensava la Magda que podries venir? (Catalan)
When thought Magda that you could come?
(67) ?Cosa diceva Piero chetu avevi fatto? (Italian)
What said Piero that you had done?
(68) La personaa cui diceva Pietro che ilprofessare eravenuto (Italian)

the person to who said Piero that the professor had come.
(69) A qui adit Jean que lejardinierbtait venu ? (French, from Friedemann 1995)
To Whom had saidJean that the gardener had come?
(70) ?Avec qui a pr&endu Marie que sortirait Jean ? (from Kayne and Pollock 1978)
With whom has claimed Marie that w ill go out Jean?
The following table summarizes the results we have

collected in the previous sections:
89
Key:*=impossible order/ ?=slightly marginal order/
??=marqinal or very marginal order/ O.K=possible order
Orders Catalan French Italian
V S DP ?? (north and * ??
(object) central)
OK southwest
V S DP * * *
determinerless
V S PP ?? (North) ?? ??
(complement) OK/? central
V S PP (10) ? to OK O.K ? to O.K

(doubling
interferes)
V S PP 0. K 0 K O.K.
(locative)
V S Adi. * * k
V S Adv * * k
de-adiectival
Modal S Inf. * * k
V S (di +INF) ? (better than * ?
Modals)
V S CP OK ? to OK ? to OK
In sum, the three languages differ in many details.

Note, however, that they coincide in not allowing the
V S DO (determinerless), V S Adv (de-adjectival) , V S Adj.,
and Modal S infinitive.
90
3.3 Distribution ot post-verbal subjects in
Spanish
S p a n is h , c o n tr a r y to C a ta la n , F re n c h and I t a lia n a llo w s
a w id e r d is t r ib u t io n o f s u b je c t s . It a llo w s th e V S DO
o rd e r q u it e f r e e ly as show n in C h a p te r 2. S u b je c ts ca n
a p p e a r b e tw e e n th e v e r b a n d a d e te r m in e r le s s o b j e c t : 66
(72) iCu£ndo comprd used manzanas?
When bought you apples
(73) Ayer compnS tuhermana manzanas.
Yesterday bought your sister apples
S u b je c ts can even b re a k id io m c h u n k s 67:

(74) £P0rqu6 no Ieda tuhermana coba?
Why not cl-give your sister “coba”
Why doesn’t your sister kiss up to him?
(75) Nocreo que tenga usted tablas parahaceresto.
I do not think that have you tables to do that.
I do not think that you have experience to do that.
S p a n is h a ls o a llo w s s u b je c t s to a p p e a r b e fo r e a d je c t iv e s
p a r t o f a s m a ll c la u s e s : 68
66 T h e re i s c e r t a in s e n s i t i v i t y t o th e h e a v in e s s o f th e
s u b je c t . S e n te n c e s s t a r t t o d e g ra d e w hen th e s u b je c t i s
made h e a v ie r , e s p e c ia lly w it h r e l a t i v e c la u s e s :
(0 ?£Cudndo comprd el hermano de Luis manzanas?
When bought the brother o f Luis apples
(ii) *?£Cudndo comprd el chico que vino manzanas?
when bought the boy that came apples
67 The sam e o b s e r v a tio n s a b o u t h e a v in e s s a p p ly t o th e s e
e x a m p le s a s w e ll:
(i) ??Nocreo que tenga el hermano de Luis tablas parahacereso
Not think that had Luis’ brother tables to do that
(ii) *? Nocreoque tenga lachicaque vino tablas parahacereso
Not think that had the g irl that came tables to do that
68 A g a in s p e c ia l a t t e n t io n m u s t b e p a id t o in t o n a t io n
p a tte r n s . The a d je c t iv e m u s t b e fo c u s e d .
(0 A mi nome parecen los nifios inteligenteg sino esuipidos.
To me not cl-seem these guys intelligent, but stupid.
These guys don’t seem intelligent, but stupid.
91
(76) Se hace usted viejo.
become you old.
(77) En el hospital, siguen las so(dados borrachos.

In the hospital, continue the soldiers drink
(78) Pot qu£ no parecen los niflos inteligentes?
Why not seem the boys intelligent
(79) Los vio Rita borrachos.
cl-saw Rita drunk
Rita saw them drunk.
They are also allowed preceding de-adjectival

adverbials:
(80) Jugar Juan limpio a las cartas es una contradiccidn. (from Suiter 1994)
Play John clean the cards is a contradiction.
Juan playing cards properly is a contradiction.
(81) Camina usted lento.
Walk you slow
You walk slowly.
(82) Esteafio trabaja Paco duro en su tesis.

This year works Paco hard on his dissertation
They are also allowed to intervene with modals and

nonmodals:
(83) Con quidn pocfcd Cecilia ir a Jackson Heights?
With whom w ill be ableCecilia to go to Jackson Heights.
(84) Qu6 viene Juan de hacer aqui? (from Torrego 1984)
What comes John of doing here.
(85) No quiere usted salir con este tiempo tan bueno?
Not want you to go out with this weather so good?
(86) Durante dase, no nos permiten los profesores codllear con nuestras compafleras.
During class, not cl allow the professors talk with our partners.
And finally, with CPs sentences are perfect as would be

expected:
(87) £Qu6 piensa Juan que habta decidido la familia? (adapted from Torrego 1984)
What thinks Jua that had decided the family
The facts examined raise many questions. I will explore

the following two in the upcoming sections:
92
a) How can we explain the difference between Spanish,
on the one hand, and Catalan, Italian and French, on

the other?
b) How can we account for the Italian, Catalan and

French facts from an antisymmetric point of view?
c) What explains the restrictions on what can follow a

post-verbal subject in Catalan, French and Italian?
3.4 Post-verbal subjects in Neutral Phrase.
As we saw in Chapter 2 (section 2.2.1), post-verbal
subjects in Spanish can end up in the V S 0 order without

main sentence stress. Thus, subjects can be neutral in
Spanish, apart of being focused as in the V O S order. In
Catalan and Italian, on the contrary, post-verbal subjects

of nonunaccusative verbs necessarily get main sentence
stress,69 and they must be focused.70 Thus, these languages
represent a subset of the possibilities allowed in Spanish.
In other words, Catalan, Italian, and French lack the
derivation that yields the possibility of having the
69 French is more delicate in this respect. The pragmatics

of stylistic inversion are complicated given that French is
not pro-drop, and the inversion needs a trigger such as wh-
movement or the subjunctive mood. Focus, however, seems to
play a crucial role in the subjunctive constructions
studied by Hampers-Manhe (1997) . For the purposes of this
chapter I assume French like Italian and Catalan lacks the
extra inflectional projection responsible for the extra
distributional possibilities of Spanish.
70 See SolA (1992) for such a claim for Catalan and
Calabrese (1990) for similar claim in Italian.
93
subject in neutral position in Spanish. The only

derivation left out in these languages must be what is
responsible for all the restrictions in the distribution of
subjects.
The way to work out this intuition is to propose that
Spanish has an extra inflectional projection I call Neutral

Phrase. This neutral phrase is below TP and above VP.
Thus, subjects move to this NeutP,71 and Verbs move past
this position to T° by head movement yielding V S 0. The

analysis is similar to that analysis proposed for V S 0
languages as in McCloskey (1996) .72

(88)
TP
NeutP
VP
Since, V S 0 is just a product of head movement to T°

over NeutP, there will not be any restriction as to the
elements that follow the post-verbal subject. In this way,
determinerless DPs, PPs, objects of idiom chunks,

infinitives, small clauses, de-adjectival adverbials, and
71 I borrow this term from Szabolscsi (1996).

72 However, the details cannot be the same. It is clear
that heaviness plays a role for the V S 0 order as we saw
in footnotes (66) and (67) in this chapter. It looks like
very heavy DPs are not able to move to this Neutral Phrase
position.
94
modal verbs, will all be allowed. These results are all
trivially derived from a head movement analysis

perspective.
One interesting prediction that this head movement to T°
analysis makes is that in constructions with a series of

verbs, the subject might only be able to follow the finite
tensed verb. Recall that I have proposed that this special
NeutP is immediately below TP. Therefore, only the
sequence V S INF INF, but not the V INF S INF, is allowed.
This prediction is confirmed by the following contrasts.73

(89) a. Nos pidid Juan voLver a recogerlo.
To us asked Juan to return to pick him up.
Juan asked us to pick him up again.
b. *?Nos pidid volver Juan a recogerlo.

To us asked to return Juan to pick him up
(90) a. Nos permite usted poder ayudarlo?
to us permitted you to be able to help you.
b ?*Nos permite poder usted ayudarlo?
To us permitted to be able you to help her.
(91) a No quiso usted poder subir.

Not wanted you to be able to go up
You didn’t want to be able to make it
b. ?*No quiso poder usted subir.

Not wanted to be able you to go up
In conclusion, I have proposed that subjects in Spanish

in post-verbal position can be in the Spec of aNeutral
position. I have also proposed that verbsmove past this
position to tense in a parallel fashion to what has been

proposed for V S 0 languages. It is therefore not
73 I will leave to section 3.9 the explanation for the

ungrammaticality of the (b) examples.
95
surprising that no restriction applies as to what follows
the subject in this neutral position. Finally, we have

assumed that Catalan, French, and Italian lack this extra
inflectional projection and therefore a derivation in the

same fashion is impossible
3.5 Post-verbal subjects in Focus Phrase.
I have pointed out that post-verbal subjects have a very

restricted distribution in Catalan, French, and Italian in
Section 3.2. I have linked these restrictions to the
special status of post-verbal subjects with respect to

focus in Italian and Catalan.74
To account for these patterns, it is possible to
consider various alternatives compatible with antisymmetry.

The first possibility is that complements in Catalan,
French, and Italian must scramble to the left of the

subject, which stays in the VP. Nevertheless, as we will
see, this scrambling is not sufficient. Instead, the
proposal I put forward requires another step. It involves
movement of the subject to a focus position followed by
movement of the TP past this position. Let us examine the

two alternatives in the following sections.
74 For French, see footnote 65.
96
3.5.1 Scrambling.
With this first alternative, I propose that Catalan
French and Italian move their arguments past the post
verbal subject because they need to check their features

overtly in a projection above VP. The subject would remain
in situ in Spec of VP and receive focus by default (see

Cinque 1993 and Reinhart 1995)
(92)
TP
AgrO
VP
OB
SU
For the V INF S order in (55b-60b) and the V Adj S order
shown in (41b-44b) one could consider a possibility in
which VP INF and Adj must scramble to the left of the
subject. The nature of this new projection could be
thought in similar lines to Koster's (1989) Predicate

Phrase.
Koster (1993) adopting the antisymmetry proposal of
Kayne (1994) proposes that Dutch is an V S 0 language. In
order to accommodate the overt S 0 V order, he claims that
objects move overtly to a higher AgrO projection in order
to get case and that the verb always remains lower. Dutch
also presents the S PP V order as well as the S Adj V and
97
the V SP V order. These facts lead Koster to propose the
existence of what he calls Predicate Phrase (PredP). Non
finate Vps and adjectives would move for checking reasons

to this projection which, by hypothesis, lies between AgrO
and VP.
(93)
AgrO
S
PredP
VP
We could also propose that arguments (PPs or DPs) and

predicates (Adjectives and Non Finate VPs) always move to a
higher projection above Spec of VP while subjects stay in
Spec of VP (see D§prez (1988) . The difference between
Dutch, on the one hand, and Catalan, French and Italian, on
the other, would be that the verb always moves higher than
AgrO P in these Romance languages while it remains lower
than PredP in Dutch. The Romance configuration is
illustrated in (94):
98
(94)
AgrO
PredP
DP
VP
PP
VP
Adj
There various reasons to make us depart from this simple

proposal. The parallelism between Dutch and Catalan,
French, and Italian breaks down in a number of respects.
Most importantly, the Romance V O S order corresponds to a

very marked prosodic construction while the Dutch S 0 V
construction corresponds to the less marked focus pattern.
On the other hand, scrambling has been always considered

to be an optional operation. Our conception about Spanish
in Chapter 2 fits well into this assumption since it allows
both VOS and v S 0. Why should this operation be
obligatory for Catalan, French and Italian? How is it
linked to the fact that subjects are necessarily focused in
Italian and Catalan? This is something that a simple
obligatory scrambling alternative seems unable to easily

answer.
3.5.2 Light Predicate Raising (LPR).
I will develop an alternative that is in accordance with
the antisymmetry proposal. The gist of thisidea has its
origins in the heavy NP shift constructions discussed in
99
Larson (1988). In the VP shell framework Larson gives a
new alternative for the traditional cases of heavy NP-shift
of sentences like (95) . Under this new hypothesis the verb

and the 10 reanalyze into a head unit in (96a) . This head
unit moves to the upper VP shell yielding the desired order
V 10 Heavy NP in (96b)75.
(95) Max sent to me the longest letter anyone had ever seen
(96a) (96b)
VP
VP
To me
the longest letter y T

anvone
sent the longest letter
anyone
Light Predicate Raising
Recently, Den Dikken (1995), following the lines of
Kayne's (1994) theory, proposes an antisymmetric analysis
of the Heavy NP-Shift in English. This analysis is

proposed in response to a heavy NP-shift analysis proposed
by Kayne (1994), according to which cases of Heavy NP Shift
75 Larson's (1988) analysis is not compatible with the

antisymmetry proposal of Kayne. This analysis is based on
the idea that the verb and the complement form a head.
This would be a case of an XP adjoined to an X°. These
structures are not permited in Kayne (1994) pag. 18.
100
could be treated as simple cases of scrambling of XP over

the heavy N P .
In Kayne's original analysis, in example (97a)— analyzed

in (97b)— since 1939v ould have been scrambled over the heavy
NP the treasure said to have been bu r i e d on that i s l a n d .

(97) a. Johan has secretly possessed since 1939 [the treasure said to have been buried on that
island)
b Johan has secretly possessed [ [since 1939] [ X° [the treasure... buried on that island]
[Y°[[e]]]
One of the problems, pointed out by Den Dikken is that
this analysis cannot easily cope with more complex cases

like (98) :
(98) Johan has expected to find [since 1939] [the treasure... buried on that island]
From an scrambling perspective it is implausible that

the adverb since 1939 modifying the main clause could end
up betwen the infinitive and the object of the lower clause
as shown in (99):
(99) John has expected to find[ [since 1939]; X° [ the treasure... buried in that island] ^ Y°
Den Dikken proposes an alternative to this scrambling

proposal in which the heavy object has moved out of the
clause containing the verb and infinitive. This movement

will take the heavy object to Spec of AgrO position. The
adverbial modifying the matrix clause would originate in
the upper clause.
(100) ...[pp] [fp since 1939]]...[AcnP [ the treasure said to have been buried on that island ] [ ^
[[sp« [vp has expected to Find t,]])]_
101
Finally, the whole VP is moved to the specifier of the
adverbial in a parallel fashion to proposals by Barbiers

(1995) about intraposition.
(101) ...[pp [sp^ [ has expected to find tj ] ] [n> since 1939]]— [ the treasure said to have
been buried on that island ] U Ill-
One advantage of this analysis is that maintains

Larson's idea that these cases of heavy NP shift involve
movement of a verbal projection to the left, instead of

movement of the heavy object to the right. Another
important feature is that it takes the heavy object to have
moved out of the VP to an AgrO projection, previous to the

raising of the VP. This analysis is parallel to Den Besten
and Webelhuth's (1989) description of remnant

topicalization. Namely, there is movement of a VP
projection containing a trace of an element that has

scrambled out previously:
(102) [ t; Gelesen] hat Hans das Bucli; nicht (from Den Besten & Webelhuth 1989)
lead has Hans the boolc not
To summarize, there are alternatives to simple
scrambling, which involve the movement of VPs (Light
Predicate Raising LPR). These alternatives have been
adopted for the analysis of Heavy NP-shift as in Den Dikken

(1995) .
102
3.6 Proposal: LPR with Post-varbal subjects
In this section I will propose an analysis of Catalan,
French, and Italian following proposals by Den Diken and

Larson.
Catalan, French, and Italian, contrary to Spanish, lack

a Neutral Phrase position between TP and vp. Thus,
subjects cannot end up in a post-verbal position by simple

head movement of the verb to a higher position past NeutP
as we saw for Spanish. Instead, subjects when focused must
move overtly to Spec of a FocP.76 I assume that that FocP

is above TP.77 This first step is represented in (104) for
the Catalan sentence (103).
(103) Volia venir en Joan. (Catalan)
Wanted to come Joan
TP
VP
volia
venir
76 I will assume that French follows this pattern. I will -

take that French subjects move to a nonneutral position.
This position will have similar properties of focus in
Catalan and Italian.
77 Focus P could also be between TP and VP, I do not think
there is much at stake on this point. Focus could be above
CP for this purpose.
103
This first step is followed by the light predicate
raising of the TP to a Spec of a projection X° above the FP

as in (105) :78
(105)
XP
En Joaiii
volia
venir
78 For French this LPR of the TP is only possible when TP

contains a Wh-element. From this perspective, wh-elements
are the only XP able to pied-pipe the whole projection to
this higher projection X above the subject, before they
can procede to move to Spec CP.
104
One obvious question one might ask is the reason for the
movement of this TP above the DP in Focus position. This
is a common problematic feature of all LPR analyses.

However,there are various possible answers: one of them is
that some sort of Extended Projection Principle requires
the movement of TP above Focus. Another possible answer is

based on Kayne's class lectures. The core of the idea is
that when Juan moves to Spec of FocP there is a mismatch
(anti-agreement) between the DP in Spec of FocP and the
Focus head. As a consequence of this mismatch the Foe head
moves a notch up to X°. However this movement is not
sufficient; X° needs to satisfy its categorial features in a

Spec-Head agreement configuration. This requirement to
enter into Spec Head agreement in X° triggers the
obligatoriness of the second movement of the TP. This is
represented in the following tree:
(106)
FocP
TP
105
Thus, the core of this proposal, consist in linking and
triggering the two XP movements (the movement of the DP

Joan and of the TP volia venir ) through the existence of
head movement from thefocus phrase to the higher

inflectional projection
To summarize this section, I have proposed that subjects

in Catalan, French, and Italian move overtly to a special
postion I call FocP. This movement is followed by LPR of
the TP to the Spec of a Higher X° projection. The
consequences of this analysis are obvious: First of all,
we capture the fact that subjects must be focalized in

order to trigger this kind of light predicate raising in
Catalan and Italian. As a consequence of this LPR, the TP
moves to a higher projection, with the result that is
subject always to the left of TP.
3.7 Consequences for nonarguments
The movement of the TP focus discussed above can explain
some of the data given in the first section. In this

section I examine the sequences V + Nonarguments (e.g.
V Inf, V Adj, and V Adv) . I will proceed with the the
consequences of verbs + arguments (e.g. V DO, V PP, and

V 10) in section 3.8.
In the introductory section I discussed some sharp
contrasts. These concerned the possible insertion of a
106
focused subject in Catalan, and Italian between the verb
and adjective in a small clause, verb de-adjectival
adverbial, and verb and modal and infinitive. I will repeat

some of the contrasts in Catalan:
(41) a. * Per qu6 no sembla enJoan intelligent?
Why not seems Joan intelligent
b. Perqud no sembla intelligent enJoan?

Why not seems intelligent Joan
(49) a. *No parla el tea germh clar.
Not speaks your brother clear
b. No parla clar elteugermh.

Not speaks clear your brother
(56) a. * No podria la Magda menjar?

b. No podria menjar la Magda?
Not could to eat Magda
Given the proposal above we can derive all the (b) cases
as we showed in (104); the issue here is to derive the
ungrammaticality of (a) examples. From the perspective of
the LPR, the derivation would have to proceed as follows

for a sentence like (41a) . Joan moves to FocP position in
(107):
107
(107) FocP
En Joai){
TP
VP
sembla
Clause
tj intelligent
The second would involve movement of the small clause to
a projection below focus and above TP as in (108) yielding
the sequence En Joan i n t e l .ligentj [sembla tj ]. Following

Szabolsci (1996), I will call this projection LP.79
(108) FocP
LP
TP
Small Clause
VP
sembla
tj intelligent
79 Nothing bars a derivation in which the Small Clause

moves previously to the movement of Joan to FocP.
Cyclicity would probably decide in favor of this last
derivation.
108
The third step in the derivation would require final LPR
of the TP above FocP yielding the final [Sembla t}J en
Joan i n t e l .ligent
XP
TP
FocP
VP
sembla
LP
♦
tj intelligent
4
This gives an explanation of the ungrammaticality of
(41a) . What is problematic appears to be the extra
movement of the small clause [e i n t e l .lxgent] to the Spec

of LP. Let us suppose that, the movement to Spec of LP is
as restricted as scrambling. Work on Germanic (e.g.

Hoekstra 1984, Koster 1987, Zwart 1995) makes clear that
small clauses are highly constrained with respect to their
movement possibilities. For example, they cannot be

extraposed in Dutch.
(110) a.. dat Jan Marie intelligent vindt
that John Mary intelligent find
109
b. *.dat Jan Marie vindt intelligent
that John Maty finds intelligent (from Zwart 1995)
Muller (1995) also points out that adjectivals that form
a part of a small caluse are difficult to scramble in

German:
(111) ??daBderFritz [ * ,stolzauf sein Kind ]j nichtt- gewesen isL (from MQller 1996)
That Fritz proud of his child not been is
It is plausible to think that such a ban against moving
small clauses is responsible for the ungrammaticality of
example (41a). Similar kinds of constraints must be behind
the moving of de-adjectival adverbials— as in (49a)—and bare
infinitves to this LP positions—as in (56a)—before the LPR

of the TP.
It is interesting to point out that the improvement of
the V di/de INF order in (61) and (62) in Catalan and
Italian recalls the following Dutch contrast between bare
infinitives and te + Inf studied in Reuland (1981) with
respect to the possibility of following a verb:

(112) *dat Ik tra d e (Max een verhall Vertellen.] (from Reuland 1981)
that I heard Max a story tell
(113) dat hij Kees vroeg [het boek telezen.]

that he Kees asked the book to read.
Finally, we saw that CPs could follow postverbal
subjects in Catalan, French, and Italian in (65)-(70). As
a matter of fact, CPs must follow the verb in Dutch as
pointed out by Koster (114):

(114) dat hij dacht dat zij kwan. (from Koster 1993)
that he thought that she came
110
The crosslingustic similarities observed in these
examples require a uniform analysis. Thus movement of
Romance infinitives to LP is blocked for the same reasons
that Dutch could not have bare infinitives following the

verb in (112) .
In conclusion, we observe that the LPR analysis can

accommodate straightforwardly that the expected orders in
Catalan is V Adj S, V Adv S, and V INF S. From our
perspective the V S Adj, V S Adv, and V S INF require an
extra step in the derivation, which consists of moving Adj,

Adv, and INF to the left previous to the LPR of the entire
TP. I proposed that the nature of this movement is highly
constrained. I have also pointed out that Movement of
adjectival Small Clauses and Bare Infinitives is similarly

constrained in Dutch. It is plausible to think that the
same analysis in terms of movement should be extended to
Catalan, French, and Italian.
3.8 Consequences for arguments
The consequences of this hypothesis for complements are

more complex. As discussed above, in many varieties, the
insertion of the subject between the verb andobject is
quite difficult as in the Catalan sentences (7) and (19)

repeated below.
(7) ???Avui Taraen Joan el dinar (from Soli 1992)
Today w ill make Joan lunch.
I ll
(19) ??No pensa La Magda a les conseqiidncies
Not thinks Magda about the consequences
From the perspective of the analysis proposed in this

chapter, the marginality of these examples implies that
movement to LP seems quite restricted.80 Observe, however,
two important facts that seem to corroborate the line of
thought exposed in our analysis. First of all, in the less

restrictive varieties of Catalan we found a gradation
between determinerless DPs after the subject and regular
DPs. This contrast is even felt by the most restrictive

speakers:
80 There is a possible way to explain the marginality of

moving arguments to LP by invoking Muller's (1996)
generalization about the restrictions on remnant movement
in German. Muller observed that remnant movement was
banned whenever the kind of movement the remnant undergoes
is of the same type that the antecedent of the remnant had
undergone. He explains this generalization in terms of
Unambiguous Domination. His principle explains why
remnants in German cannot be scrambled.
(I) *DaB [ tj zu Iesen] keiner [das buch], versucht hat
that to read nobody the book tried has.
Let's assume that the movement of XP to LP and LPR of the
TP are basically the same kind of movement. This way, we
would have a formal explanation of the difficulty of moving
to LP in all derivations in Italian and restrictive
dialects of Catalan. The only exception is 10 PPs as we
will see. Observe that if the movement to LP were of a
different kind, the derivation would be allowed under
Muller's (1996) generalization. I think this is indeed
what we find with emarginazione or Right Dislocation of the
XP that may follow the subject in Focus. Let's suppose
that right dislocation or emarginazione are movement (Kayne
(1994) and they can be characterize as different from LPR
of the TP. By Muller's observation, emarginazione or right
dislocation would not interfere with LPR.
(0 No hi pensa la Magda # a les conseqii&ncies (Catalan)
Not cl-thinks Magda about the consequences
(ii) Non si Iamenta Rita # del mal di testa. (Italian)
Not complains Rita, about the headache
112
(115) *Volia enJoan permis. (Catalan, all dialects)
Wanted Joan permission.
(116) ??Volia en Joan el permis. (O.K for Some Catalan varieties)
wanted en Joan the permit.
I would attribute this gradation to the fact that
movement to LP would take the determinerless DP to a
position not governed by the verb. As discussed in Chapter

2, bare DPs cannot be scrambled in general in Spanish. It
is plausible to extend this restriction to Catalan. From
this perspective, it is understandable why there is a

gradation between determinerless DPs and normal DPs in
these sentences.
The second interesting fact comes from French and the
possibility of allowing prepositional IOs to follow the

post-verbal subject, contrary to object DPs:
(117) Lecarirau qu’a envoyd Jean & Marie, (from Kayne 1986)
The gift that has sent Jean to Marie.
(118) *Ou ngpare Marie sa voituie? (from Friedemann 1995)

Where repairs Marie his car.
In the perspective defended here, this means that Step 2
involving movement of 10 PPs to LP previous to LPR of TP is
exceptionally allowed, contrary to DO.8182 Interestingly,
81 The same contrast is found with constructions with

extraposed prepostional datives in Dutch after the verb.
Koster (1975) gives the following example with a PP 10:
Q) Omdat hij het boek geft. aan Norval (from Koster 1975)
because he the book gave to Noval
However,a direct object NP after the verb is ungrammatical:
(ii) *dat hij las het boek (from Koster 1993)
that he read the book
82 The contrasts recalls Cinque's (1990) discussion of
Italian clitic left dislocations and is parallel to the
discussion of topicalization in French in (119)-(120). We
113
it is important to notice that the 10 seem to be
exceptional in other constructions. For instance, an 10
can be topicalized without a clitic, while a DO cannot:

(119) A Jean, j ’ai donnd un livre.
To John, I gave a book.
(120) *Le livre, j ’ai achetd.
The book, I bought
This contrast parallels another contrast pointed out by

Kayne (1975) with respect to pre-posing of 10 PPs in front
of a VP.
a. Je ne veux ni k lui ni&elle due la veritd.
I not want nor to him nor her to say the truth.
b. * Je ne veux ni lui ni elle revoir demain
I not want nor him nor her see tomorrow
(122) J’aurais , k ces gar?ons-&, permis de fumerune ciganette. (from Kayne 1975)
I would to these boys peimitted to smoke a cigarette
(123) *J’aurais, Jean, invitd klasoirde
I would Jean invited to the evening.
Finally, another construction that shows similar
contrasts between DOs and IOs is found with subdeletion
with le as studied in Kayne (1984b). Observe the contrast

between (124) and (125):83
(124) *(H sera photographic beaucoup d’enfants.) el il le sera beaucoup de parents aussi.
it/there will-be photographed many boys, and it w ill be many parents also.
(125) (Ce livre-ri sera oflert k Marie), I’autre le sera &Jeanne,

this book w ill be offered to Marie, the other w ill be to Jean
could adopt his explanation for these contrasts by

proposing the empty category left by the movement of the DO
to LP requires a clitic to identify it, while this is not
the case with PPs moving to LP.
83 I thank Richard Kayne for pointing this out to me.
114
The analysis of these clauses might arguably be
considered to the same cases of VP subdeletion in English.

(126) Mary spoke to John, but she didn’t to Bill.
Kayne (1994) proposes that these cases of subdeletion

should involve an inner topicalization of to John above the
VP, followed by deletion of the VP. If the analysis of the
French cases (124)(125)is the same, they would also involve
leftward movement of DO, 10 followed by deletion. From the
LPR analysis the contrast between the DO and the 10 PP in
subject inversion constructions (e.g. 117 and 118) should
be treated in the same fashion as the same contrast in the
other constructions discussed above.84 These include

topicalization (e. g. 119 and 120), leftwards movement
above the VP (e.g. 121a and 121b) , and le subdeletion in
French in (124 and 125) . In this way, this analysis of
leftwards movement of DOs or prepostional 10s to LP, below
the position of the post-verbal subject, allows a uniform
treatment of all these apparently different constructions.

In conclusion, theanalysis presented allows us to
understand the deviance of determinerless DPs after focus
subjects. Additionally, it permits a uniform account of the
DO/10 contrast in French with respect to four
84 On this perspective, it could also be explained why

Small Clauses are also bad in this sort of subdeletion.
These cases would involve leftwards movement of small
clauses, which is not permited as we saw in the previous
section.
Q) ‘ Marie seraiendue hereuse, Jeanne le sera malhereuse
Marie w ill be made happy, Jeanne w ill be unhappy
115
const xructions: Subject inversion, topicalization
possibilities, movement above VP, and le subdeletion.
3.8.1 LPR and C*commanding of arguments
The analysis proposed so far presupposes that DP

arguments move with the VP to the left of the subject and
yield the [VO] S order. Thus, one might expect that
objects would not be able to C-command the subject under
such a configuration. This prediction is the opposite of
what we had shown for the V O S order in Spanish in Chapter
2. However, the prediction is not borne out: Objects are

indeed able to C-command the subject in the V O S order in
Catalan and Italian. Thus, a quantifier object can bind a
postverbal subject in the V O S in (127a) in Italian.
There is a clear -contrast between (127a) and (127b): 85

(127) a. Questo libro, I’haregalato [aogniragazzojj sua^ madre/ la propriaj madre.
This book, cl-has given to every boy his mother/ his own mother.
b. *SUj madre/la propriaj madre haregalato un libro [a ogni ragazzo ];.
his mother/his own mother has given a book to every boy
Catalan also allows objects to C-command the subject in
the V O S order as shown in (128):

(128) a. Qufe (li)varegalar acadanen elseuamic eldiadeNadal/
What (d)-gave to each boy his friend during Christmas day
85 I thank Nino Gulli for providing me with the Italian

data. Careful attention should be paid to intonation. The
most normal intonation with downstressing of the subject in
the V O S order is not the one relevant in these cases.
116
With respect to principle C effects, we find similar
results. The argument lei C-commands the post-verbal

subject [i fratelli di Eva] in the V O S order:86
(129) *Ieri, hanno visto [i fratelli di Eva],-, nonle sue sorelle. (Italian)
yesterday have seen her Eva's brothers.
I conclude that previous to the LPR of the TP, objects

have moved independently to a position above the focus
subject position. I take this position to be of the same
nature of AgrO, an L-related position in which objects and

other complements check their features:87 In this sense
this position differs from the LP position below focus,
which we take to be non L-related. Let us see how the

derivation would proceed for a simple sentence like:
(130) [vp En Joan [v-escriur4 [ la carta]]]
Joan w ill write the letter
Step 1: movement of the subject to focus:

(131) [rkp En Joanj (ip tj [v-escriur& [ la carta]]]
Step 2: movement of the object to AgrO, above FocP:

032) l^ la carta; [*«,. En Joanj tj [v.escriiirh [ t,]]]]
Step 3: Light Predicate Raising of TP:

(133) (Tp tj [v> escriuri ( t, ]]], U^jla carta; [p^pEnJoanj t,]]]
86 This cases cannot simply be reduced to problems with

respect to backward' anaphora as the following example
shows:
(I) ?Non hanno visto leij i suoij fratelli, non le sue sorelle.
Not have seen her her brothers, not her sisters.
87 There is another possible way to integrate this facts
into the general picture if the Formal Features of the
object are raised out of the raised TP at LF to a position
from which they can C-command the subject.
117
The derivation as proposed explains the C-coiranand facts

at the same time it is compatible with the LPR analysis I
have proposed in this chapter
3.9 Spanish V O S order.
The analysis proposed above captures the fact that

subjects are in focus when they appear post-verbally in
Catalan and Italian. We also observed in Chapter 2 that
the V O S order of Spanish is characterized by having
obligatory focus on the subject. Consequently, the
analysis proposed above can be extended to Spanish as well.
The V O S order proceeds by the subject moving to focus,
objects moving above subject to a position in which they

can C-command the subject, and finally, LPR of the TP
remnant. Through this comparative perspective, the
analysis for Spanish gets considerably enriched.
In Section 3.6, I proposed a different derivation for
the V S 0 order. According to this analysis, subjects move
to NeutP, and the verb raises above it by head movement.

Thus, we obtain a configuration in which verb can end up to
the left of the subject in two different ways in Spanish:

by head movement when subject is in NeutP or by LPR of a
remnant TP when the subject is focused.88
88 I do not have any deep explanation for why derivation

should proceed in different ways when the subject is
focused or neutral. I leave this question for further
investigation.
118
The analysis using focus of the subject and LPR of the
TP has the advantage of explaining why there are

restrictions as what can follow the post-verbal subject.
This is important because of the existence of constraints
shared by all the languages examined. In none, for
instance, could determinerless DPs, INFs, adjectives, or

de-adjectival adverbials follow a focused subject. Given
the existence of an alternative derivation in Spanish in
which the subjects are in NeutP, these restrictions are not
easily noticed. However, there are two contexts in which
Spanish does indeed show these constraints. We have
already seen one of them with examples such as (89) to (91)

involving a sequence of infinitives.89 Compare the
ungrammaticallity of (134b) with acceptability of (134c) :

(134) a. Ayer. nos permitid Juan poder hacerio.
Yesterday To us permitted Juan to able to make it
b. *? Ayer, Nos permitid poder Juan hacerio.

Yesterday, To us permitted to be able Juan to make it
c. Nos permitid poder hacerio Juan, no Maria
To us asked to be able to makeit Juan, not Mary
I take the contrast between (134b) and (134c) to be the
same as the one found in Catalan in the following sentences
repeated below:
(56) a. * No podria la Magda menjar? (Catalan)
89 The sequence V INF S CP becomes much better. This

recalls the fact that in V S CP was the most acceptable in
Catalan, Italian and French
(i) Queman saber los estudiantes sipodrian ir a Brooklyn Heights.
Wanted to know the students wether they could go to Brooklyn Heights
119
b. No podria menjar la Magda?
Notcoold to eat Magda
Thus, the derivation of the ungrammatical Spanish (134b)

case would proceed as follows:
Step 1: subject moves focus
(135) [ftdJuaiij [n. tj Nos permitid [poder [hacerio]]]
Step 2: Movement of [ hacerio] to LP

(136) (tfepTuanj [[(jjiacerlo, [vp q Nos permitid [jpoder t, IB
Step 3: LPR of the remnant TP:

(137) [-rp tj Nos perm itid [poder t, ]]. [R„1Juanj [ljhacerIo, t . ]]
As in the cases of Catalan and Italian the problem is

found in Step 2, with the movement of the infinitive to LP.
I concluded in previous sections that scrambling of
predicates and bare infinitives to this LP position is

impossible, and that the obstacle lies in the constrained
nature of this movement.90 Observe thatif Step 2 is
eliminated the output after the LPR of thewhole TP is the
grammatical sentence (134c). I show the final stage in the

derivation below without step 2:
(138) [-rp tj Nos perm itid [poder [hacerio]]], tp^pJuanj t,
Other cases that show the restriction on moving to LP
previous to LPR involves small clauses. Observe the

following contrasts:
90 The alternative derivation with the subject in NeutP

cannot yield the required order of sentence (134b) either,
since, as I argued, this would proceed by movement of the
verb to tense leaving the post-verbal subject after the
finite verb.
120
(139) a. Porentonces consideraban tushermanas brillantes alosestudiantes.91
By then considered your sisters (S) brilliant the students (O)
b. ‘ Porentonces consideraban alosestudiantes. tus hermanas brillantes.

By then considered the students(0) your sisters (S) brilliant
c. TPorentonces consideraban alosestudiantes. brillantes tushermanas.92
By then considered thestudents(0) brilliant your sisters (S)
(140) a. Violahijadelprofesor borrachos alosestudiantes.
saw the professor’s daughters) drank the students(0)
b. *V io alosestudiantes la hija del profesor borrachos.
saw the students (O) the professor’s daughters) drunk .
c. Vio a los estudiantes borrachos la hija del profesor
saw the studentsfO) drank the professor’s daughter(S)
It is clear that sentence (139b) and (140b) are the most

difficult. The representation of (140b) is in (141) with
the object in AGRO , TP above AgrO, subject in FocP and
crucially the small clause brillantes in LP:
(141) h r <>consideraban]]] Lôa los estudiantes [Focûs hermanas [lp( tj brillantes]]]
On this analysis the degradation of all the (b) cases is

due to the same reason the following simple Catalan cases
are degraded. All these cases involve an illicit step,
movement of small clause to LP, just as in the Catalan

examples (41) and (42) repeated below:93
91 Other plausible permutations are:

(0 Porentonces consideraban tushermanas alosestudiantes brillantes.
By then considered your sisters (S) the students (O) brilliant
(ii). Porentonces consideraban alosestudiantes brillantes tushermanas.

By then considered the students (O) brilliant your sisters (S)
92 The sentence requires heavy emphasis on tus hermanas.
93 There are even some cases that recall the French
difference between DO and 10. The sentences involved are
fairly heavy, but a distinction is noticed. The sentence
with the DO at the end is definitively worse:
(i) ?Ayer le escribid la carta mianigo asuabuela.
Yesterday wrote the letter(DO) mifriend(S) to his grand-mother (IO )
121
(41) a. •Pfcrqufc no sembla en Joan [ intelligent]?
Why not seems Joan intelligent?
(42) a. •E sfa en Joan veil.
becomes Joan oid
To conclude,I have extended the results of previous
section about Catalan, French, and Italian to the Spanish

VOS order. This extension leadsto a welcome result
given the fact that some of therestrictions on what
follows the focused subject can be seen in Spanish as well.
The sentences had to be made more complex given the fact
that otherwise an alternative derivation with subjects in

NeutP could interfere in evaluating them.
3.10 Comparing LPR to the right adjunction

alternative.
Earlier analyses, (e.g. Bonet 1988 and Soli 1992— for
Catalan—and Friedemann 1995— for French) have proposed that
the Spec of VP branches to the right. As discussed above,
the main support for this hypothesis is that complements

frequently precede post-verbal subjects in Catalan or
Italian as in (142) . In order to account for complements
that follow the subject, advocates of this approach have
had to propose a rule of extraposition to the right as in

(143) and (144) :
(142) g V O ] S]
(ii) ??Ayer le escribid asuabuela mi amigo la carta

Yesterday wrote to his grandmother(IO) my friend (S) the letter (DO)
For French, such an extraposition analysis is not so
obvious since V O S order is not good in general as shown
in (145) :
(145) *A qui donnera le livre ton ami? (from Friedemann 1995)
Who w ill give the book your friend.
Nonetheless, Friedemann takes the so-called "crossing
effects" proposed by Kayne (1986)—based in turn on a

generalization by Korzen (1983)—to be an indirect argument
in favor of the right adjunction approach in that language.
The approach assumes the two points mentioned above: ie.,
that subjects are adjoined to the right, and any element

following the subject in French must have been extraposed
as (143) (144) . Given the existence of this extraposition,
any movement of Wh-words originating to the right of the
trace of the extraposed element would create a crossing

effect:
(146) [[ V t t ] S] PP
^ I I f ^ extraposition
In this way, it is possible to explain the marginality
and ungrammaticality of the following sentences:.

(147) ??Lejour ou [[a& rit t vpJJeanyp] t k Marie t ?
the day when has written ■ Jean to Marie
t 1 H
123
(148) *A qui domera t i ton ami le livre?
to whom w ill give your friend the book
t L -l 1
(149) ??Ou a tfl6phon£ Marie&Jean ? (from Friedemann 1995)
where has phoned Marie to Jean
t
(150) * Dans quelle flection a vot£ votre ami [pour Nixon] ?
In which election has voted your friend for Nixon
t _____________________
J
However, these crossing effects are not triggered when
the Wh-element is closer to the verb than the trace left by
the element extraposed to the right, thus the

grammaticality of the following sentences:
(151) Lecadeau qu’a envoy6 t t Jean & Marie. (From Kayne 1986)
the gif^ that has sent ^ ^ Jean to yarie
(152) Qu’a dit t Jean au Jardinier (from Friedemann 1995)

what has said Jean to the gardener
J L
(153) La fille dequi s’est plaint Jean & Marie (from Kayne 1986)
the girl of whom complained Jean to Marie
t J L J
On this point, it is worth noting that parallel effects
are found in Italian and Catalan:
(154) a. Gianni conosceva la ragazza della quale si &Iamentato Francesco con Maria.
Gianni knew the girl of whom cl-compIained Francesco with Maria. (from
Zubizaneta 1994)
b. ??Gianni conosceva la ragazza con la quale si t lamentato Francesco di Maria
Gianni knew the girl about whom cl-compiained Francesco of Maria
(155) a. El tema del qual parlava el teu pare al president.
the topicof which spoke your father to the president
b. ??La persona a la qual parlava el teu pare de polftica.
the person to whom spoke your father about politics
124
Spanish contrasts with the above languages as pointed
out by Zubizarreta (1994):

(156) a Eltema delcual hablaba tu padre con el presidents.
the topicof whichspoke your father with the president
b. La persona con Iacual habiaba tu padre depolftica.

the person with whom spoke your father about politics
Still, there are various problems with a crossing
perspective. For one thing, this approach predicts that
the effects should be the same in all the French cases no
matter which element is crossed over. Thus, there should
be no difference between sentences with the extraction of

an extraposed PP as (149) or an extraposed DO as in (148) .
However cases with DO are clearly degraded. Also, the same
effect should be found with extraposed CPs. A sentence

like (157) would exemplify such a case with a CP extraposed
crossing over an 10.94 However, these cases are perfectly
grammatical according to Friedemann (1995) .

(157) Aqui adit t tJean [<, quelejardinierdtaitvenu]?(fromFriedemann 1995)
To whom has said Jean that the gardener had come
t u ----
Finally, and most importantly, if we can show that the
grammatical contrasts are found without wh-movement, then a
crossing approach becomes unlikely. This sort of contrast

is not testable in French since inversion requires wh-
movement. Such case are found in Catalan and Italian. For
example, as discussed in the introduction to this Chapter,
94 Friedemann notes this problem and proposes CP objects

are generated in a more external position than DP objects.
125
the V S 0 order independent of wh-movement is difficult in
Italian and many varieties of Catalan.

(8) "Vactiner en Liars la Manub. (Catalan, from Bonet 1988)
Ran Uius the marathon
(13) *Ha mangiato spesso Gianni gli spaghetti. (Italian)

Has eaten often Gianni the spaghetti
Furthermore, saw that in many Catalan varieties and

Italian, the sentences with PPs are not very good.
However, as in French, the sentences are better than the
previous ones with DOs:

(20) a. ??Ahir va discutir el professor sobre linguistica?
yesterday discussed the professor about linguistics.
(24) a. 7?Ha Iitigato Piero con sua soreHa-
Has fought Piero with his daughter
However, 10 PPs seem to be exceptional and can appear

after the focused subject without difficulties:
(21) ?Aquest llibie elvaregalar en Joan a la Maria (from Sol£ 1992)
This book gave Joan to Maria
(28) □ regalo que ha spedito Giuliana al suo nipote
the gift that has sent Giuliana to her nephew
Finally examples with CPs are substantially better as
seen in (66):
(66) Ahir pensava la Magda que podries venir. (Catalan)
Yesterday thought Magda that you could come.
(68) La persona a cui diceva Piero che il professore era venuto (Italian)
the person to who said Piero that the professor had come.
The parallelisms in the restrictions on what is allowed
after the post-verbal subjects in cases without wh-movement
internal to Catalan and Italian and in comparison with
French deserve a common explanation. Yet the idea that
crossing a wh-word is what is determining the
126
ungrammatical ity of examples (147-150)) in French cannot
sustained in the Catalan and Italian examples above.
The third problematic issue comes from comparison with

Spanish. Why should the effect not appear in (156)?.
Recall that Spanish was exceptional in that normal

declarative clauses allow V S 0, V S PP, V S Adj. I have
attributed this exceptionality to the fact that there is an
alternative derivation in which the subject in NeutP.95

In conclusion, in this section I have reviewed the major
arguments that linguists have proposed for having subjects
located on the right edge of the VP. As we have been able
to see, the reasons are insufficient. The fact that
complements generally precede the subjects as noted by Sola

(1992) and Bonet (1988), could also be explained from the
light Predicate Raising alternative sketched above. The
crossing effects advocated by Friedemann are also an
insufficient reason. This solution was unlikely for a
number of reasons: 1) it could not explain easily why PPs
and DOs have different effects, and 2) the proposal for
French cannot be extended to languages that show similar
95 Recall that there are complex contexts in which subjects

could not be in NeutP in Spanish, but in FocP. In those
contexts we might expect similar effects to the ones
pointed out by Korzen (1983). The sentences are heavy;
however, (ii) is worse than (i)
6) Eldlaenque hablaba depolftica tu padre con Juan.
The day in which spoke about politics your father with Juan
(ii) ??E1d£aen que hablaba con Juan tu padre de polftica.
The day in which spoke with Juan your father about politics
127
restrictions such as Catalan and Italian. As we saw,

those effects show up independently of Wh-movement.
3.11 Conclusion
In this chapter I have proposed that the wider

distribution of post-verbal subjects in Spanish is due to
the existence of an extra inflectional projection called
NeutP. I have proposed that the verb passes this position
by head movement yielding the V S 0 order. I have also

proposed that the other languages examined, French,
Catalan, and Italian lack this possible derivation.
Subjects are necessarily moved to a special position that I
have identified with Focus P in Catalan and Italian. This

movement of the subject to focus is followed by LPR of the
entire TP to a position above this focus projection. All
XPs that follow the subject in focus must scramble out of
the TP previous to the LPR. It has been pointed out that
this operation is tightly constrained. This analysis
permits us to make a parallelism between the restrictions
on this extra leftward step and other leftward operations

in the same language or across languages. Finally, the
analysis is extended to Spanish V O S cases, with the
subsequent consequence of enriching the account proposed in

Chapter 2. Indirect evidence of the LPR analysis in
Spanish was seen with certain complex structures with the
VOS XP(Adj), V INF S INF, as well as V 10 S DO.
128
Chapter 4
The inversion construction in interrogatives

in Spanish and Catalan
4.1 Introduction
In this chapter, I switch to an examination of the
structure of interrogatives with special attention to
Spanish and Catalan. In Spanish and Catalan, as
illustrated in examples (la) and (2a), there is an
obligatory subject-verb inversion in interrogatives.

(1) a £A quite visitd Juan ? (Spanish)
Whom visited Juan
b. *£A quite Juan visiuS?
Whom Juan visited
129
(2) a- Qui veu en Joan ? (Catalan)
whom sees Joan
b. *Qui en Joan veu?

whom Joan sees
Some linguists (e.g. Torrego, 1984) have taken this
inversion to be the output of a movement of the verb to a
position to the left of the subject. Accepting this

account, Rizzi (1991) explains the obligatoriness of the
movement in terms of the wh-criterion,96 a well-formedness
principle at LF that requires a Spec-head configuration
between a wh-word and a head with wh-features:

(3) Wh-criterion. (from Rizzi 1991)
a. A wh-operator must be in a Spec-head configuration with an x [+wh]
b. A X [+wh] must be in a Spec-head configuradon with a wh-operator.
Rizzi postulates that the carrier of the wh-feature in

(la) and (2a) is the verbal inflection, which has moved to
C overtly in order to enter into a Spec head agreement with
the Wh-XP. Such a configuration is not obtained in (lb)
and (2b) since the subject in Spec IP is placed between the
wh-word and the inflectional head.
Nevertheless, overt V-to-C is problematic on several
levels.97 Theoretically, it is incompatible with Kayne's

(1994) antisymmetry proposal, particularly in its
consequences for the position of clitics. Furthermore, it
leads to the erroneous conclusion that postv^rbal subjects
96Rizzi discusses mainly English, Italian, and French.

However, his proposal is valid for Spanish and Catalan.
97 Suner (1994) is the first to deny the existence of overt
V-to-C in Spanish. I will not repeat her compelling
arguments in this chapter but in Chapter 5.
130
could occupy the position of Spec of IP. In addition, it
creates a nonuniform account of inversion phenomena.
Finally, and crosslinguistically, this approach assumes a
similarity between the Catalan and Spanish inversion

phenomena and the V2 and Sub-Aux inversion of Germanic.
This parallelism, I will show, cannot account for
differences in the distribution of clitics and the possible

placements of subjects after the auxiliaries in questions
between both groups of languages.
The alternative proposed here relies on the basic idea

that pre-verbal subjects are in the Spec of ToPP whose head
Top0 conflicts in features with the wh-requirements of the

Wh-word.
4.2 Antisymmetry and the landing site of clitics
As we saw in Chapter 1, Kayne (1994) concludes that
pro-clitics in Romance cannot be adjoined to the same
inflectional projection as the verb in sentences such as
(4) and (5) :

(4) Juan le escribid. (Spanish)
Juan to-him wrote
(5) En Joan li ddna el llibrc. (Catalan)
Joan to-him gives the book.
The argument is presented in (6) :
131
a. incompatible with LCA b compatible with LCA
The Linear Correspondence Axiom (LCA) bars multiple

adjunction to the same head such as the one shown in (6a)
in which the pro-clitic is adjoined to the same head as the
verb. Therefore, pro-clitics must be adjoined to an

inflectional projection above the landing position of the
verb, as represented in (6b),98
98Another possibility compatible with the LCA is that the

clitic adjoins itself to the verb ,which is already
adjoined to the inflectional projection. However, this
possibility is unlikely given that pronominal elements are
excluded from appearing within words (see Kayne, 1994,
p.42) .
132
Since right adjunctions are also banned in the

antisymmetry approach, there are two possible resolutions
when the verb has to move to a position above the clitics,
such as C°. In the first alternative, the verb could move
directly over the position of the clitics as illustrated in
(7a) .99 In the second, it could move left adjoining to the
clitic itself, and move along with it to C°, as shown in

(7b) mo. in either case, the order obtained is always verb-
clitic.
(7)
a.
V +lo
t
From this perspective, the postulation of movement of
the verb to C° in interrogatives in Romance would predict
the V—cl order. However, the contrasts in (8) and (9) for
Spanish and Catalan show that the prediction is not borne
out. The sequence V-cl for interrogatives is

ungrammatical.
"This possibility is entertained by Terzi (1996). The

movement of the verb over the place holder would not
constitute a violation of the HMC under the shortest
movement approach proposed by Ferguson and Groat (1995).
100This would be a case of incorporation as proposed by
Cinque and Beninci. (1993).
133
(8) a. *£Qu6 escribid le?
What wrote to her/him
b. £Qu£leescribhS?
What to him/her wrote
(9) a. *Qu&dtinesIi?
what give-for her/him
b. Qu&liddnes?
What to him/her give
In this respect, Spanish and Catalan clearly differ from

the cases of the V-2 in Germanic. The V-2phenomenon has
been analyzed as involving the movement ofthe verbto C.
In clear contrast to the Spanish or Catalan examples, the

Germanic object clitics necessarily follow the verb in
second position as shown in the contrasts in (10a) and
(10b) :
(10) a. Gestem hat sich der Hans ein Buch gekauft. (from Kayne 1995, pg. 43)
Yesterday has cl the Hans a book bought.
Yesterday Hans has bought a book for himself
b *Gestem sich hat der Hans ein Buch gekauft
Finally, the antisymmetry account leads naturally to the

idea that verbs have moved further in the order V-cl than
in the order cl-V. This allows a simple characterization
of the difference between imperatives/nonfinites versus
finite verbs in Spanish and Catalan. For instance Rivero
(1994) and Rivero and Terzi (1995) propose that verbs in

imperatives move to C to get their illocutionary force.
Thus, it is not surprising that V-cl order obtains in these
cases:
(11) a ddna-li (Catalan)
give it to him/her
b *Iiddna
134
(12) a cdmpralo (Spanish)
buy it
b. *k) compra
Similarly, Kayne (1991) proposes that infinitives in
Italian (and by extension Spanish and Catalan) move to a
projection higher than the one where the clitics are

adjoined.101
(13) a. Comprarlo (Spanish)
To buy it
*L o comprar
(14) b donar-li (Catalan)
To give to him
*Ii donar
This differentiation between infinitives and imperatives
on the one hand, and finite verbs, on the other, is
problematic for the overt V-to-C analysis in
interrogatives. This approach leads to the conclusion that

the clitic and verb have moved together as a unit in
interrogatives. Thus, it inevitably leaves the contrasts

between V-cl and cl-V entirely unmotivated in terms of verb
movement. Unlike an antisymmetry based account, it cannot
elucidate why the properties of the verb (e.g. tense) play
a crucial role in motivating one order or the other in the
same language. For example, it might be expected that the

verb should allow the cl-V in imperatives in some Romance
languages. However, such a possibility is barred in all
101Kayne's (1991) proposal would need to be modified since

it involved multiple adjunction of the verb and the clitic
to the same inflectional projection.
135
Romance languages with a specific imperative morphology as
pointed out by Rooryck (1992).
In conclusion, V-to-C in interrogatives in Spanish and
Catalan is incompatible with the idea that pro-clitics and

verbs cannot be adjoined to the same inflectional
projection and the general ban on right adjunctions.

Additionally, an analysis that denies V-to-C in finite
interrogatives in Romance accounts straightforwardly for
the difference between Germanic V-2 and Romance

interrogatives. In particular, it shows why clitics have
to follow the verb in Germanic but not Romance. Finally,
it also gives simple characterization the distribution of
the clitics in the different tenses in terms of verb

movement.
4.3 The position of ths post-verbal subjects in

interrogatives
Another problematic issue for the overt V-to-C approach
can be found in the position of subjects in interrogatives.

As we discussed in Chapter 3, Spanish and Catalan allow
their subjects to appear post-verbally in normal
declaratives. Subjects can appear below TP in NeutP, or
they can appear in focus position after the LPR of the TP:
The existence of overt V-to-C adds a new possible
derivation in which the subject might end up post-verbally
in Spanish and Catalan. However, it would need to be shown
136
that verbs have moved by head movement overtly above Spec
IP and that none of the other possible derivations are at
hand in interrogatives.
As I will show, there are important reasons to doubt
that subjects occupy Spec of IP in interrogatives. If this
conclusion is correct, that removes one of the most
compelling reasons for the overt movement of the verb to a
higher C, namely the nonvacuous movement of the verb over
Spec of IP position.
4.3.1 Auxiliaries and Vpp.
On Rizzi's account, an auxiliary, in moving to C, leaves

the Vpp in some projection internal to the IP. This leads
to the prediction that subjects in Spec of IP might be
placed between the auxiliary and the Vpp. However,
examples (15)-(17) show that this prediction is not borne

out.
(15) a. *Qui va la Magda veine? (Catalan)
who aux- Magda see
who did Magda see
b. Qui va veure la Magda?
Who aux-see Magda
(16) a *Qui havia la Magda vist? (Catalan)

who had Magda seen
b. Qui havia vist la Magda?

Who had seen Magda
(17) a. *£ A quite habia la madre de Juan visto? (Spanish)

Who had John’s mother seen?
b. £A quite habia visto la madre de Juan ? (Spanish)

who had seen John’s mother
137
In this respect, there is a clear cross-linguistic
difference between Spanish and Catalan, on the one hand,

and English and German on the other. In the Germanic
languages, the subject is necessarily placed between the
auxiliary and the Vpp, as in (18) and (19) :

(18) a. Who has John’s mother seen?
b. •Who has seen John’s mother?
(19) a. Was hat dein Bruder gekauft?

What has your brother bought?
b. •Was hat gekauft dein Bruder?

What has bought your brother
In order to explain this parametric contrast, while

maintaining the V-to-C approach, one might conclude, that
contrary to German and English, the auxiliary and Vpp do
not form a breakable unit in Romance because the vpp

incorporates to the auxiliary. However, there is good
reason to be doubtful about such an alternative. For
instance, Suher (1988a) has shown that an incorporation

approach would be implausible for Spanish, given the fact
that certain tenses in some dialects allow the insertion of
material between the auxiliary and the Vpp.102 On the other
102 In my dialect the aux-vpp unit is breakable with

infinitives and with certain subjunctive tenses. Observe
that even in these two cases the complementizer is filled
by de in (i) and by si in (ii) . Therefore, it is unlikely
that the auxiliary is in C.
(0 De haberk) yo sabido, no te habria dicho nada.
of having-it I known, I wouldn’t have told you anything
Had I known, I would not have told you anything
(ii) ?Si Ie hubiese yo hecho caso, no tendria ningtin problema.
If I to him/her had paid attention I would not have problems.
The length of the intervening subject is also an important
138
hand, Rizzi (1991) mentions that the auxiliary-vpp sequence
is breakable in Italian with adverbials. Nevertheless, the
subject cannot intervene between the auxiliary and the Vpp.

(20) a. *Che cosa ha il direttore detto? (From Rizzi 1991)
What has the director said
b. Che cosa ha detto il direttore?

What has said the director
In conclusion, if Aux moves to C with compound tenses

the subject cannot be in Spec of IP as would be expected
given the English and German examples.
4.3.2 Catalan and the position of subjects.
More evidence against the idea that the subject occupies
the Spec of IP in interrogatives is given by Catalan in

examples of a verb with one DP object and with modal verbs.
The overt movement to C approach predicts that subjects in

Spec of IP should precede any other object complement in a
lower inflectional projection in Catalan. However, this is
factor that limits the possibility of breaking the verb and

auxiliary. There is a clear contrast between the above
examples and the following ones.
(iii) *De haberio la chica que visitaste el otro dia sabido, no tendrias problemas ahora mismo.
O f having the girl that you visited the other day known, you would had no problems
(iv) *Si le hubiese la chica que visitaste el otro dia hecho caso.
If tod the girl that you visited the other day paid attention
In this respect the construction differs clearly from the
English subj-Aux inversion.
Finally, it is important to note that some of the
examples Suner (1988) mentions, in which a tense Aux Vpp
can be broken, come from the Caracas dialect. This variety
belongs to the Caribbean dialect area which also allows the
insertion of the subject between the verb and the Wh-word
(Thanks to E. Herburger, a native speaker of this dialect
for the relevant information) .
139
o b v io u s ly n o t th e case, as we saw i n C h a p te r 3 . I re p e a t
h e re som e o f th e c o n tr a s ts under (2 1 ) . The s u b je c t m u st
n e c e s s a r ily f o llo w th e o b je c t . 103

(21) a. ??Aqui donaA la Magda eIUibre?VSO
whom w ill give Magda the book.
Who w ill Magda give the book to?
b. Aqui denari el Uibre la Magda? VO S
whom w ill give the book Magda.
The sam e p ro b le m is a tte s te d w it h m ore c o m p le x v e rb a l
s tr u c tu r e s in v o lv in g m o d a ls . If th e m o d a l h a d m oved t o C,
th e s u b je c t s h o u ld a p p e a r in S pec o f IP b e tw e e n th e m o d al
in C , a n d th e in f in it iv e in an em bedded c la u s e . H o w e ve r,
as we saw i n C h a p te r 3 th e c o n tr a s ts in (2 2 ) and (2 3 ) show
t h a t th e s e s e q u e n c e s a r e u n g r a m m a tic a l.
(22) a *A qui vol la leva germana [donar aquest Uibre]? Modal-SU-INF
to who wants your sister give this book?
Who does your sister want to give this book to?
b. Aqui vol [donar aquest Uibre] la tevagermana? Modal-INF-SU
to who wants give this book your sister
(23) a. *A qui pot la leva germana [demanaraquest Uibre]?Modal-SU-INF
to who can your sister ask this book
Who can your sister borrow this book from?
b. A qui pot [demanaraquest Uibre] la leva germana? Modal-INF-SU
who can ask this book your sister
4.3.3 Floating quantifiers
F lo a t in g q u a n t if ie r s a re lic e n s e d in b o th S p a n is h and
C a ta la n a s show n i n (24)-(26)
(24) Aquellos turistas vienen todos de Francia (Spanish)
these tourists come all from France.
(25) Mis vecinos recibieron ambos una carta de recomendacidn (Spanish)
my neighbors received both a letter o f recommendation
103T he sam e a rg u m e n t a p p lie s fo r I t a lia n .
140
(26) Aquests turistes v6nen tots de Franca (Catalan)
these tourists come all from France
Sportiche (1988) analyzed these cases of floating

quantifiers as instances where the NP subject and the
floating quantifier originate as a syntactic unit in Spec
of VP. In a later stage of the derivation the subject
moves to some inflectional projection leaving the floating

quantifier stranded as represented in (27) :
(27) [q >Aquests turistesi [i*v6nen [vp tots t j de Franca] ]](Cataian)
these tourists come all from France
If the verb had moved to C in interrogatives, the FQ

might still be stranded with the subject in Spec of IP.
However, the sequence V Sub FQ is ungrammatical:104
(28) * [cp D’on v6nen [jp aquests turistes [yp tots till ?(Catalan)
from where come those tourists all
Where do all those tourists come from?
(29) * [cp &De ddnde vienen [fp estos turistas [VP todos ]]]? (Spanish)
from where come those tourist all
Where do all those tourists come from?
(30) * (CP cQu6 recibieron [iptus vecinos [yp ambos 1] ]? (Spanish)

what received your neighbors both
What did your both neighbors receive?
To finish, all these constraints on the placement of
subjects point to the conclusion that subjects cannot
appear in Spec of IP in interrogatives in Spanish and

Catalan. If V-to-C existed in these languages, it would
always have to take place vacuously.
104 The argument against subjects in Spec IP in

interrogatives, based on the distribution of FQs is equally
valid if FQs are treated as predicative adverbials.
141
In order explain v^jxy subjects cannot appear in Spec of
IP, Rizzi (1991), following Rizzi and Roberts (1989),
postulates that V-to-C in interrogatives destroys the Spec-
head configuration necessary for case assignment to the
subject in Spec of IP. Consequently, Rizzi claims that
nominative case is alternatively assigned by tense to the

right in sentences like (la) and (2a) .
This solution is problematic in various respects. First
of all, we are forced to conclude that expletive or
argumental pro, which by hypothesis is in Spec of IP, does
not receive case in interrogatives. At the same time, this

idea clashes with the derivational approach advocated in
the minimalist program (Chomsky 1995). From the minimalist

perspective, once a configuration for case is obtained at a
certain stage of the derivation, it cannot be undone in
later steps.
Finally, this solution is too strong given the existence
of Aux-to-Comp constructions. Rizzi (1982) postulated that
these constructions involve movement of the Aux-to-C. The
auxiliary in C would assign case to the subject in Spec of

IP. This at least implies that it is possible for the verb
in C to assign nominative case in certain instances.105
105 The fact that the subject can appear between the Aux and
the Vpp suggest, that contrary to finite verbs, infinitives
and gerunds move further up. This seems to be corroborated
since, as we mentioned in Section 4.2, clitics follow
gerunds and infinitves.
142
(31) [Avendo [Mario t [acceuaio di aiuiarcL ]]]
I________________ I
Having Mario accepted to help us
4.4V-to-C and tha "free invars ion " construction.
Spanish and Catalan allow subjects to appear post-
verbally in declarative sentences in what it is called

"free inversion" as we saw in Chapter 2 and 3. I will
repeat some examples for convenience:
(32) Van venir aJguns estudiants (Catalan)
came some students.
(33) Vinieron varios estudiantes (Spanish)
came some students
The obvious question is whether there should be a common
analysis of this type of inversion and the one resulting

from interrogatives. The answer must be affirmative since
both constructions show exactly the same constraints. For
instance, subjects are not allowed between the auxiliary
and the Vpp as shown in examples (34).

A ik -SJU-V versus Aua-.V-SU
(34) a. * La habia lamadredeJuan visto. (Spanish) Aux-SU-V
cl- had Juan’s mother seen.
Juan’s mother had seen her.
b. La habta visto la tnadre de Juan. Aux-V-SU

cl-had seen Juan’s mother
Q) Avendolo Gianni comprato

Having Gianni bought it
Having Gianni bought it
143
Also, the subject cannot appear between the object and
the verb in the Catalan as we discussed in the previous
chapter. Here I repeat (35) :

Catalan V S O versus V O S and Modal-SU-INF versus Modal-INF-SU
(35) a * Em donard la Magda el Uibre. V S O

cl-to me w ill give Magda the book.
Magda w ill give the book to me
b. Em donaii el Uibre la Magda. V O S

cl-to me wiU give the book Magda.
Additionally, we also saw that the subject cannot appear

between the infinitive and the modal verb in Catalan as in
(36) :
(36) a. *VoI la teva germana demanaraquest Uibre Modal-SU-INF
wants your sister to order this book
Your sister wants to order this book
b. Vol demanar aquest Uibre la teva germana. Modal-INF-SU

wants to order this book your sister.
Finally the sequence V S FQ is not allowed either.

(37) a. *V6nen de’Franca aquests turistes tots. (Catalan) V-SU-FQ
come from France these tourists all
A ll these tourists come from France.
However, the possibility of extending the overt V-to-C
to the free inversion construction is problematic in two
respects. From a theoretical point of view, it is not a

clear what the trigger for V-to-C movement in these cases
could be since there is no obvious criterion to be
satisfied. Even more importantly, the free inversion

construction can appear in an embedded sentences with a
filled overt complementizer as in (38) :

(38) M ’handit que vindran alguns estudiants(Catalan)
to me have told that wiU come some students.
They told me that some students wUl come
144
As we have shown in Chapter 3, a simple head movement
analysis cannot render an account of all the restrictions
we find in the post-verbal subject position in Catalan,

French and Italian.
However, the impossibility of extending the overt V-to-C
analysis to the cases of free inversion poses problems to
this approach overall. On the one hand, it is clearly
unsatisfactory to have two analyses for constructions that
are otherwise alike. For instance in the interrogative
construction V-to-C would apply vacuously, and the free

inversion construction V-to-C would not apply at all. On
the other hand, it shows the insufficiency of the
explanation based on the loss of case of subjects in Spec
of IP in the interrogatives. The explanation for the
constraints could not be extended to the free inversion
construction. There must be a common explanation for the
contrasts, and this account evidently cannot not rely on

overt V-to-C.
4.5 Pi«daontas«
Another problem with overt V-to-C is encountered in

certain languages with double-filled Comp such as in
Piedmontese as shown in (39).
(39) a. Cosa ch'a fa la mama d' Giuanin tuti i di?
What that-cl makes the mother of John Everyday
What does John’s mother make every day?
145
b. *Cosa che la mama d’ Giuanin a fa tuti i di?
what that the mother of John cl-makes everyday
Given the presence of an overt complementizer, there

should not be any inversion effect since the verb would not
have moved to C°. Nevertheless, we do find these
inversions.106 These facts show, that an overt V-to-C
approach is not sufficient to explain the obligatory

inversion in languages with overt complementizers. Some
extra mechanism must be responsible for these effects.
4.6 The obligatoriness o f inversion in

Interrogatives in Spanish and Catalan.
Once the overt V-to-C approach is eliminated, the
explanation for the ungrammaticality in (lb) and (2b)
clearly needs to be approached in terms of what blocks pre

verbal subjects from appearing in interrogative contexts:
(lb ) b. *A quiln Juan visitd?
Who Juan visited
(2b) b. *Qui enJoan veu?
who Joan sees
There are various possible ways to account for this

prohibition.
4.6.1 A -Minimality
Suner (1994) explores the possibility that the
ungrammaticality of (lb) and (2b) could be due to the fact
106 I thank Luigi Burzio for this information. There are

many other Northern Italian dialects that seem to show this
pattern.
146
that subjects in Spec of IP block the movement of a wh-
word. She proposes a specific principle of argumental

agreement, which expresses the idea that an argumental wh-
word in Spec of CP maintains a long distance agreement

relation with the inflection of the verb. Such an
argumental agreement relationship cannot be interrupted by
any other argumental DP between the highest verb in I and
the wh-word. Therefore, the ungrammaticality of (lb) and
(2b) is accounted for because the subject in Spec of IP
counts as an argumental element blocking this relationship.

(40) Argumental Agreement Licensing (From Suiter 1994)
a. Argumental wh-phrases must be licensed through symmetric Rag-agreement between a

(=SpecQ and P(=c)
b. p arg agrees with y (=V) only if P and y are arg-marked and no other Arg-marked
element is closer to y,
Sentences with post-verbal subjects as (41) do not have

the same effect since the element in Spec of IP is a non
argumental pro, which by (40) does not count as a possible

blocking element:
(41) tAquidn (IP proexpl visitd Juan] ?
Who Juan visited
As in the case of overt V-to-C, however, this solution
encounters problems when the status of argumental pro is

considered in sentences like (42) . If argumental pro
occupies Spec of IP as in (43a) , it should have the same

blocking effect as a argumental lexical DP and a sentence
147
like (42) should be ungrammatical107. Alternatively, the
analysis to be adopted would involve a nonargumental pro in
Spec of IP co-indexed to an argumental pro presumably in
Spec of VP as in (43b) . However, such an alternative

cannot be adequate either since argumental pro in Spec of
VP would fail to be identified by Agr S, which is too far

away.
(42) Aquidn visitd?
Who visited-3p.s?t?
(43) a A qui£n pro+arg visitd
b. Aquidn pro.arg visit6 tvPPro+arg
Moreover, this approach also turns out to be problematic

from a cross-linguistic point of view. It is not clear why
subjects in Spec of IP in English or German do not have the
same blocking effects given the grammaticality of run-of-

the-mill English sentences like (44) .
(44) What cfid he buy?
4.6.2 A>bar minimality
A variety of factors lead to the conclusion that
preverbal subjects in Spanish and Catalan must have a

different distribution than pro and DP subjects in English.
107 the distribution of "pro" is also a problem to those

theories that propose that wh-words land in Spec IP while
the subject remains in a lower position (Spec VP) (Goodall
1991, Fontana 1993). Under those proposals "pro" would
have to remain post-verbal in interrogative contexts,
making it impossible for it to be identified under spec-
head agreement. If we eliminate the possibility for pro to
appear in Spec IP, there is no much justification for Spec
IP altogether as I will discuss in Chapter 5.
148
They include (i) elimination of overt V-to-C (ii) the
different behaviors of lexical subjects and pro, and (iii)

the difference of Spanish and Catalan with respect to
English with respect to lexical subjects.
The question then is what that distribution is. One
hypothesis is that pre-verbal lexical subjects in Spanish
and Catalan are dislocated to a more peripheral position.
This notion is not new to this study. It has been proposed
for Spanish by Contreras (1991), Zubizarreta (1994) , Uribe

Etxebarria (1995), for Rumanian by Dobrovie-Sorin (1991),
and for Catalan by SolA (1992) and Rigau (1988) and for
Greek by Alexiadou and Anagnostopoulou (1996) . I will
explore this idea with some of the results discussed in

Chapter 5.
In my hypothesis, DPs appear displaced to a more
peripheral position in Spanish. Following antisymmetry and

the ban on multiple adjunction, I will assume that
preverbal lexical DPs are in the Spec of a topic projection

(see also Rizzi 1995). However, let's assume provisionally
that pro occupies Spec IP.

TopP
Juan
pro
vino....
came
149
Since Lexical subjects and pro occupy different
positions, the fact that there is no intervention effect of

the wh-word when there is no preverbal lexical DP is less
surprising. Also, the contrast between Spanish and Catalan
with respect to English becomes understandable if preverbal
subjects in English are not topicalized, but ,contrary to

Spanish, have to appear in Spec IP. We will see this in
more detail in Chapter 5.
The idea that subjects occupy a more external position

in Spanish than in English could be corroborated from the
behavior of complementizerless dependent clauses in both
languages. It is well-known that complementizers can be
dropped in English clauses like (46) .
(46) Jane regrets Melissa is not home.
Certain dialects of Spanish also allow this kind of
complementizer deletion (see Torrego 1982). What is
important is that, contrary to English, Spanish pre-verbal

subjects are not allowed in complementizerless dependent
clauses as shown in (47b).
(47) a. Lamento no estdconienta Carmen. (Spanish)(from Torrego 1982)
I regret is not happy Carmen
I regret Carmen is not happy
b. ‘ Lamento Carmen no est£ conten(a (Spanish) (from Torrego1982)
I regret Carmen is not happy
It is tempting to think that the ungrammaticality of

(48a) in Spanish is parallel to the ungrammaticality of a
sentence of English with a peripheral adjunct as in (48b) .
150
(48) a. ‘ Lamento [Top P Carmenj [IP pro j no estd contenia]]
b. * Jane regrets [Top P yesterday [q> Melissa went to her house 1]
By the analysis given before, the adjunct y esterday in

English and the pre-verbal subject in Spanish would occupy
more peripheral positions, which could not be licensed in
this type of construction. Thus, preverbal subjects in

English correspond to pro in Spanish as lexical subjects in
Spanish correspond to adjuncts in English.
If subjects in Spanish and Catalan are in an A'
position, it would be possible to understand the

ungrammaticality of (lb) and (2b) in terms of A'
minimality. Unfortunately, taking an A' minimality
approach alone is too simple. From this perspective,

topics in an A' position would block the movement of the
wh-word to Spec of CP. Nevertheless, Torrego (1984) has
shown that the extraction of a wh-word over a pre-verbal

subject is available in embedded declaratives where long
distance wh-movement has taken place as in (49a).108 The

same possibility seems to be available with other
nonsubject topics in embedded declaratives such as in
108 Torrego (1984) claims there is a distinction between

having a sentence with two embeddings and one with three
embeddings. The one with three embeddings requires
inversion of the lowest clause. In my dialect I do not
find appreciable difference between both clauses. They are
both fairly heavy because of the number of embeddings.
(il) iQu6 piensas que Juan habia dicho que la revista habia publicado?
what think-2ps that Juan had said that the journal had published
(ii) iQu6 piensas que Juan habia dicho que habia publicado la revista?
what think-2ps that Juan had said that had published the journal
151
(50a) . If a subject in topic position were blocking the
extraction of a wh-element, there should not be any

contrast between (a) and (b) in the two pairs of sentences:
(49) a. Qu£ pensabas que [Top P la revista [q > habia publicado]] (Spanish)
What you thought that the magazine had published
What did you think that the magazine had published?
b. *Qu6 [Top P la revista [jp habia publicado ]]? (Spanish)

What the magazine had published
What did the magazine publish?
(50) a. Qud imaginabas que [Top a ti [ipte iban a regalar? ]]S)p

What you thought that to you they were going to give?
What did you think that they were going to give to you?
b. *Qu£ [Top p a ti (ip te iban a regalar]]? (Spanish)
What for you they were going to give.
What were they going to give to you?
Thus, it is promising to think that lexical subjects are
in a topic position in Spanish and Catalan. This

explanation would account for why lexical subjects, but not
pro, have a blocking effect in interrogatives. However,
the simple idea that topics would create an A' minimality
effect in interrogatives seems too strong given the facts
from long distance extraction out of embedded declaratives.
4.6.3 Head movement. Movement o f the complementizer.
The contrasts in (49) and (50) is also indicative of one

other syntactic effect: The blocking of the lexical DP
subject only occurs in the sentence where the wh-word
lands. Therefore, the principle that explains the contrast

in (1) and (2) must be very local. I will maintain, with
Rizzi (1991) , that this local principle is an agreement
152
relationship. However, contrary to Rizzi, I propose that
the agreement is with the complementizer rather than the
verb. The agreement relationship involves an empty

complementizer in Spanish and it involves the auxiliary do
in English as in (51) and (52) :
(51) [cp W h a t^ didt^,, |jp he I [yp buy t ID?
(52) ICPAqui&Urt C0^, tip [I’visltd]]]]? (Spanish)
From our perspective topics interfere with this

agreement relationship. We saw that coding this
interference in term of A' minimality was too simple.

However, there is an interesting difference between
interrogatives and embedded declaratives with respect to
the complementizers. Embedded declaratives have an overt

que whereas interrogatives have an empty complementizer.109.
I will start by deriving the grammaticality of embedded
declaratives with a pre-verbal subject as in (49a) or an
embedded topic as in (50a). I will follow the basic lines

of Watanabe (1992) and Browning (1996) and their ideas
about CP recursion These authors propose that embedded
topicalizations involve a CP recursion structure la

Larson." The overt complementizer starts in the lower CP
shell and enter in a Spec-head relation with a topic XP.
109 Rizzi (1991) postulates z complementizer for French

sentences like (i) or English sentences like (ii) :
CO Je nesais pas [qui 0 eile a rencontrg
I not know who she has met
(ii) I wonder [what 0 you saw
153
This is shown in (54), which represents the first stage in
the creation of the complex CP of a sentence like (53):

(53) Robin met the man that Leslie said that for all intents and purposes was the mayor of the
city.
(54) said [ for all intents and purposes [ that was the mayor of the city.
CP C°
Watanabe proposes that the complementizer that targets
its own projection and creates another CP shell. The
motivation for this head movement is found in the fact that
the verb say requires a declarative complementizer (see

Cheng 1991) .
(55) said [q> that for all intents and purposes [c l [p
f 1
Browning, following Chomsky (1995, Chapter 3) observes
that this head movement can explain why the topic for all
intents and purposes does not count as a blocker for A'
minimality effects. The head movement of that to the upper
CP shell extends the minimal domain of C°. Thus the two CP
specifiers are equidistant with respect to a wh-word to be
moved, given Chomsky's relevant definition of equidistance
in (56). Wh-movement can proceed as shown in (57).

(56) If a j) are in the same minimal domain, they are equidistant from y.
(57) said that; [&, for all intents and purposes L ^ [p

a a C"
I am going to adopt this approach for Spanish embedded
topics in sentences like (49a) and (50a) . The first step

is shown in the structure in (58); la revista isin the
154
specifier of the complementizer que, which is in the head

of Top0:. :
(58)
op P
pensaba
la revista
qtt
The verb p e n s a r "think" requires a CP with declarative
force. Top P is part of the CP Field (Rizzi 1995), but has

no declarative force. Que in Top° moves a notch up and
creates a new specifier CP as in (59):
<59) / \ cp
pensaba
-wh
Top P
la revista
Top
The head movement of the complementizer will extend the
domain. The upper Spec of CP and the lower Spec of Top
would be equidistant to any wh-word in IP. This analysis

predicts why no A ' -Mininality (or MLC) is created when the
wh-words move cyclically in (49a) and (50a).
The extension of this perspective lead us to think that
in matrix clauses there is movement of a complementizer
155
when a topic XP pre-verbal as in (49b) and (50b) . This
complex CP can only be created by recursion through head
movement, more or less in the same fashion as VP shells are
created. This leads us to aderivation similar to the one

seen previously for the embedded cases. However, there is
a difference between embedded declarative CPs and matrix
question CPs: their specification with respect to features.

I am assuming that Top0 is always specified as [-wh] . In
the cases of embedded declarative in (49a) and (50a) this

[-wh] specification does not create a problem since traces
left by wh-elements are [-wh] anyway, as argued by Lasnik
and Saito (1992):
(60) Qu6 pensabas [ t.„h que..,, [xon p la revista t [ip habia publicado t U
t It * > I
However, with a matrix clause, the empty complementizer
starts in the topic head, and it would be specified as
[-wh] by Spec-Head agreement. It moves a notch up to

create an Spec for the Wh-word. This head movement
guarantees that la revista in Spec Top0 will not block wh-
movement. However, there is a mismatch between the empty

complementizer, which started in Top0 and the Wh-word in
Spec CP as represented in (61). This mismatch in features
explains the ungrammaticality of (49b) and (50b) :
156
(61)
CP
Qu6
+Wh
la revista
Top
IP
In conclusion, the matrix embedded asymmetry in the
Spanish examples reduces to a difference between Wh-traces

being [-wh] versus wh-words being [+wh]. The same mismatch
of features found in matrix questions with the 0

complementizer in Spanish is found with the overt
complementizer che in Piedmontese. As in Spanish, I assume
subjects are in Spec Top0. The complementizer when

appearing with a pre-verbal subject would start off as the
head of topic and because of the agreement with subject in
topic will be [-wh]. In this configuration, this
complementizer would raise to create another specifier for

the wh-word. That in turn causes a mismatch configuration
between the wh-word in the upper CP and the complementizer

che:
(62) a. [o>*Cosa [i>afa
Importantly, when all these complementizers are not part

of a topic, they would acquire their features with direct
157
agreement with the wh-feature of the XP moved to Spec of
CP. Here no problem for mismatching would occur:

(63) [cpCosa(+wh) ch’a(+wh) [9 fa la mama d’ Giuanin
(64) [ Qu6(+wh) 0 (+wh) [ q> habia publicado la revista]]?
In previous sections, I have proposed that subjects in

pre-verbal position in Spanish (48a) have the same
distribution as pre-verbal adjuncts in English, as in
(48b) . Thus we predict rightly that these adjuncts block
wh-movement in the same manner as pre-verbal subjects do in

Spanish. The DO moves from I to Top0 acquiring a [-wh]
specification which would create mismatching problem when
moving raising to the higher CP as shown in (65):

(65) * [cpWhat (+wh) did(-wh) [TopPyesterday TOP0 [IP Peter buy
t_________ It I
This approach also predicts that there should not be any
mismatch between matrix and embedded questions, whenever a
topic intervenes between IP and CP. This topic would
always be specified as [-wh] and would block the matching
of features with the higher CP.110
110 Italian subjunctive is a potential problem. Contrary to

matrix clauses preverbal subjects can appear pre-verbally
producing no intervention effects for some speakers (see
Giorgi and Pianesi 1996):
(i) Gianni si domanda che cosa Mario abbia fana (Giorgi and Pianesi 1996)
Gianni wonder what Mario had-3ps done
However, the idea that subjects in subjunctive clause are
not in topic position seems to be confirmed by the fact
that the fact can also appear in complementizerless
clauses:
(ii) Credeva tu fossianivato in tempo(Giorgio and Pianesi 1996)
I thought you had arrived on time
158
(66) *1 wonder [& who(+wh) o(-wh) [T_ yesterday t [ you saw?
(67) *No sabia [q, qug(+wh)

t________I
0(-wh) [T_ la revista t [ habia publicado
To conclude this section,

t I
I h a v e e x t e n d e d an approach
due to Browning (1996), Watanabe (1992) about CP recursion

to the cases involving preverbal subjects in Spanish. I
have assumed that subjects are in Topic and that the [-wh]
feature value of Top0 conflicts with the [+wh] feature of
the wh-word, creating a mismatch. This conflict is
obviated in embedded declarative contexts in which the
trace of the wh-element is specified as [-wh] .
4.7 Conclusion.
It has been shown that movement of the verb in Spanish

and Catalan in overt syntax conflicts in important ways
with the antisymmetry theory and its implications for the
landing sites of clitics. The difference between Catalan

and Spanish with respect to the Germanic languages examined
here in this respect can only be understood if the former
languages lack overt V-to-C. It has also been shown that
post-verbal subjects in interrogatives are not in Spec of

IP. These post-verbal subjects in interrogatives behave
like post-verbal subjects in the so-called free inversion
Recently Giorgi and Pianesi have pointed out the there is a
correlation of the acceptability of (i) with the
acceptability of (ii) for Italian speakers. This confirms
the correlation previously made in the text inversion
interrogatives and insertion of the pre-verbal subject in
complementizerless clauses.
159
construction. It has been pointed out that the overt V-to-
C approach is unable to give a uniform account of both

constructions.
Once overt V-to-C is eliminated, however, a puzzle
appears: Why do empty subjects in Spanish and Catalan not
have the same blocking effects as overt subjects?

Similarly, why do DP subjects in English, contrary to
Catalan and Spanish, allow such extraction? The answer
appears to be that lexical subjects in Catalan and Spanish
must occupy a topic position. Finally, to explain why

there is obligatory inversion in interrogatives, I have
adopted a CP recursion approach. I have proposed that the

mismatch between the complementizer which starts in Top0 and
the wh-word in CP is the responsible for the
ungrammaticality of sentences with topics such as (lb) and
(2b) . As we predict, this effect obtains in matrix and
embedded questions (67). We have also seen that embedded
topics in embedded declaratives do not produce
ungrammaticality. This is .as predicted by the idea that
the trace left by a wh-element in Spec CP and the

complementizer are both specified as [-wh].
4.8 Appendix
Finally, some remarks need to be made concerning the

fact that certain wh-words do not trigger obligatory
160
inversion in Spanish. Some authors (e.g. Torrego 1984 and
Suner 1994) have proposed that non argumental wh-words do

not require inversion, while argumental wh-words do.
I think that the distinctions the do not cut so much
across the argumental/nonargumental difference as the one

between complex and noncomplex wh-elements as shown in the
following contrasts:
(68) a. * £Cu£ndo tus hermanas se fueron?
When your sisters left
b. Ên qu6 momento tus hermanas se fueron?

In which moment your sisters left
(69)a. *£Cdmo tuhermana secayd delacama?
How your sister fell from bed?
b. <;De qud manera tuhermana secayd delacama?

In which way your sister fell from bed.
(70) a. * <;Ddnde tus amigos se divotciaron?

Where your friends got divorce.
b. Ênqudlugar tus amigos se divotciaron?
In which place your friends got divorced.
The same problem cuts across argumental wh-questions.
Speakers find an improvement with complex wh-words compared
to simple ones (see chapter 5, section 5.2.2.2):111

(71) a. *£Aquidn tuhermana visitd?
Who your sister visited
b. £A cudl de estas chicas tuhermana (la) habia visitado en Sicilia?

Which of the girls your sister had visited in Sicily
c. <;Acuil de las chicas que vinieron tuhermana (la) habia visitado?
Which of the girls that came your sister had visited
xllCalabrese (1982) reports a similar phenomenon in Italian,

as illustrated in (i)
6) QuaJIe delle ragazze che abbiamo incontrato, Mario ha conosciuto in Sicilia?
Which of the girls that we just saw, Mario has already met in Sicily? (From
Calabrese 1985)
L61
Observe, by contrast, that in English overt V-to-C in
questions is triggered no matter how complex the wh-

element, as in (72).
(72) a. Which of the girls that we had met has Mario fallen in love with?
b. * Which of the girls that we had met Mario has fallen in love with?
This crosslinguistic difference constitutes another

argument against a uniform treatment of Spanish/Catalan
inversion constructions and the V-2/Aux-NP inversion of

Germanic.
Instead, I would like to adopt the idea that complex wh-
word in the preceding examples in Spanish are not in Spec

CP but are left dislocated. Therefore, since there is no
[+wh] agreement involved, it becomes less surprising that

preverbal subjects can intervene between the complex wh-
word and the verb. As pointed out by Rizzi (1995), complex
negative quantifiers in Italian can also be left dislocated
contrary to simple ones as shown in the following contrast:

(73) a Nessuno di questi ragazzi io conosco veramente bene (From Rizzi 1995)
No one of these boys cl-1 know very w ell
b. ’ Nessuno l’ho visto.

No one cl-have seen
In Rizzi's view, the reasons for this contrast have to

do with the possibility of the quantifier in the Spec of
the complex DP moving at LF and binding a variable inside
that DP as in (74) . Even if the whole DP is left
dislocated, the quantifier in its Spec can still form the

operator-variable configuration needed.
(74) Nessuno [ec di questi ragazzi] TOP k> conosco veramente bene.
162
The similarity between these cases requires a uniform
treatment. In other words, if the behavior of simple
versus complex negative quantifiers is the same as that of

simple versus complex question words, these contrasts are
likely to have the same explanation. I would like then to
suggest that the complex wh-*constituents are left

dislocated and that the quantifier part is moved at LF in
order to obtain the needed quantificational interpretation.
They, thus, function in a similar fashion to the negative

cases.112.
112 An important question that arises is the fact that

resumptive clitics do not seem to be required in the
examples with complex quantifiers even if they are left
dislocated. The same applies for complex negative
quantifiers at least in Spanish:
(0 A ninguno de estos chicos ui conoces realmente.
It is still conceivable that these quantifiers in the Spec
of the left-dislocated DP are capable of licensing the
variable internal to the IP and therefore accounting for
the absence of the clitic. It has been noticed by Cinque
(1991) that certain quantifiers can be left-dislocated and
nevertheless license a variable
(ii) Qualcosa £ai6. (Cinque 1991)
163
(75) £Acu£l [ec dc las chlcas] TOP tu hennana la habfe visitado antes?
Which o f these girls your sister has seen before.
In conclusion, the complexity of the wh-words seems to
be an important factor that determines the possibility of

inversion in Spanish. Following Rizzi (1995), I have
proposed an alternative where complex wh-words are left

dislocated.
164
Chapter 5
Left dislocated subjects and Pro-Drop*
5 .1 INTRODUCTION.
In the last chapter I proposed an analysis of

interrogatives that does not involve verb movement to C.
This conclusion relied on the idea that subjects in Spanish

are in a topic-like position. In this chapter I would like
to further motivate this assumption. I will concentrate in
Spanish, but the results evidently apply to Italian and
Catalan as well.
It has been assumed that in Romance preverbal subjects

such as those in (1) occupy a functional projection Spec of
IP, in which case and Agreement are satisfied (Rizzi 1990,

Motapayane 1988, Cardinaletti 1996, Belletti 1990). On the
* This chapter is based on work in conjunction with Esthela

Trevino (see Orddnez & Trevino 1997).
165
other hand, preverbal objects in (2) and (3) are taken to
occupy a more external position corresponding to that of a

topic.
(1) Juan Ie dio las Uaves a Pedro.
Juan cl gave the keys to Pedro.
(2) Las Uaves se las dio Juan a Pedro.

The keys cl-cl gave Juan to Pedro.
0) A Pedro ledio Juan las Uaves.

To Pedro cl gave Juan the keys.
Thus, sentences with preverbal subjects receive the
analysis in (4) in which subjects are in Spec of IP,113
while preverbal DOs and IOs receive a different analysis as

in (5) . Here, the IP projection is always present either
occupied by pro as indicated in (5a) or by a lexical DP
subject as indicated in (5b).
(4) [IP Juan Ie dio las Uaves]
(5) a. XPDO/IO [IP pro V (subject)]

Las llaves / a Pedro [pro Ie dio (Juan)]
The keys / to Pedro cl gave Juan
b. XPDO/IO [IP [subject] V]

Las Uaves/aPedro Juan Ie dio.
The keys / to Pedro Juan cl gave.
The standard analysis therefore leads to a dual
characterization depending on whether preverbal subjects or
preverbal objects are involved. The evidence to be
introduced in this final chapter challenges this dual

analysis in two important ways. On the one hand, the
113 Given an articulated theory of inflectional projections

as Pollock (1989), this projection would be denominated
AgrS. See Belletti (1990). In most of the chapter I will
continue using the term IP except when the denomination
AgrS becomes relevant.
166
assumption that there is a null pro in Spec IP with a
preverbal DO or 10 in (5b) conflicts with some facts of
ellipsis and extraction of quantificational elements (e.g.

negative quantifiers and wh-expressions as we saw in
Chapter 4). The conclusion that pro cannot be postulated
in the preverbal subject position leads naturally, in turn,

to the elimination of the idea that there is a preverbal
Spec projection exclusive for subjects altogether. The
second kind of evidence in favor of our hypothesis comes
from the fact that overt preverbal subjects share certain
important similarities with preverbal DOs and IOs regarding

constraints on quantificational interpretation. These
similarities call for a uniform analysis for all preverbal
arguments.
These admitedly radical reconceptualizations require
certain fundamental theoretical changes. For example, I
propose the elimination of Spec IP as a projection

exclusive for subjects. Instead, I analyse preverbal DOs,
IOs, and subjects as occupying the same topic position as

in (6) :
(6) [tcp XP(subject/DO/IO) Top] V.
Also, as a consequence of this analysis, the theoretical
status of preverbal arguments must be revised. With

regards to subjects, the main question to be addressed is
how case and agreement properties are satisfied. In view
of the fact that subjects may appear in other positions, as
167
examples in (7) and (8) illustrate, I also sketch an
alternative characterization to the so-called pro-drop

instances like those in (9) .
(7) Le dio las Haves tu hermano.
Q gave the keys your brother
'Your brother gave him/her the keys'
(8) Le dio tu hermano las Uaves.

Cl gave your brother the keys
(9) Le entregd las Uaves a Pedro.

C l gave the keys to Pedro.
The chapter is organized as follows. In section 5.2, I
describe the empirical data that support the conclusion
that preverbal overt subjects and objects have the same
syntactic distribution. Given such evidence, I reconsider
the theoretical status of preverbal subjects in Section 5.3

and revise some of the alternatives proposed in the
literature by other researchers. In Section 5.4 I reject
the idea that there is such an inflectional projection
called AgrS in which case and agreement are satisfied for
subjects in a preverbal position. Instead, I adopt

Taraldsen's (1992) position that agreement, specifically
person agreement, should be considered a clitic and an
argument of the verb. Evidence that Agr is a clitic is
given by the striking similarities between the relation of

doubling XPs and clitics, on the one hand and agreements
and subjects, on the other, with respect to certain

mismatches in person specification. Evidence that Agr is
an argument is given by the fact that it is crucial for the
168
computation of A binding in these cases of mismatches.
Some interesting morphological evidence is discussed at the

end in section 5.4.1.
5.2 Empirical evidence
5.2.1 Ellipsis
Spanish exhibits ellipsis phenomena which differ in
several ways from English VP ellipsis. Brucart (1987)
shows that certain discourse polarity particles such as si
('yes'), no ('not1), tambi&n ('too'), and tampoco
('neither') license ellipsis. The elements that can stand

as remnants of this kind of ellipsis include not only
preverbal subjects as in (10), but also preverbal direct
and indirect objects, as in (11) and (12).114

(10) El le dio unos libros a Pia y Pepe tambidn [le dio unos libros a P u l
He cl gave some books to Pia and Pepe too [cl gave some books to P£a]
(11) Unos libros Ie dio Juan a Pfa y unos cuadros tambi£n [le dio Juan a Pu]
Some books cl gave Juan to Pia and some paintings too [cl gave Juan to Pia]
(12) A Pfa le dio Juan unos libros y a Sara tambidn [le dio Juan unos libros]
To Pia cl gave Juan some books and to Sara too [cl gave Juan some books]
Furthermore, all the remnants of this type of ellipsis
can be easily subordinated, as shown in (13) and (14):

(13) Juan le dio unos libros a Pia y me parece que Pepe tambidn [le dio unos libros a Pia]
Juan cl gave some books to Pfa and cl seems that Ptepe too [cl gave some books to Pia]
(14) A Pia le dio Juan unos libros y me parece que a Sara tambiln [le dio Juan unos
libros]
To Pfa cl gave Juan some books and it cl seems that to Sara too [gave Juan some books]
114 The elided material is indicated in bold face enclosed

in square brackets.
169
The possibility of subordinating these remnants
indicates that ellipsis of this sort involves a copy of the
inflectional projection that follows the discourse polarity

items. Thus, an analysis involving the movement of the
correlate, for example, as proposed in Reinhart (1991) for
bare argument ellipsis, or in Larson (1990) for cases of

gapping, is immediately excluded since those two types of
constructions do not allow subordination of the remnant.
The parallelism between DO, 10, and subject remnants

remains even in those cases in which ellipsis is licensed
in island contexts. Examples include extraposed clauses

such as (15) and (16)— instances of a weak island contexts—
and relative clauses—strong islands-as in (17) and (18) :

(15) Maria no sabe ruso pero es posible que Luis si [sepa ruso] (Brucart 1987)
Maria not knows Russian but it is possible that Luis yes [knows Russian]
'M aria doesn't know Russian, but it is possible that Luis does'
(16) A ti te admitirdn en Harvard pero es probable que a tu amiga no [la adm itirin en
Harvard]
To you cl w ill admit in Harvard but is probable that to your friend not [cl w ill admit in
Harvaid]
'You w ill be admitted to Harvard, but it is probable that your friend w ill not'
(17) Luis no sabe traducir pero yo conozco a una alumna que si [sabe traducfr]
Luis not knows to translate but I know a student that yes [knows to translate]
'Luis doesn’t know how to translate, but I know a student who does'
(18) Hay gente a la que le puedes decir bromas y hay gente a la que no [le puedes decir
bromas]
There Is people to whom cl can tell jokes and there is people to whom not [cl can tell
jokes]
'There are people who you can tell jokes to, and there are people who you cannot'
Under the dual hypothesis represented in (5), in which

preverbal subjects occupy a more internal position than
preverbal DOs and IOs, it would not be easy to capture the
170
parallelism shown by all remnants in the former examples.

From that perspective, two different types of remnants must
be postulated, as shown in (19). Thus, there would be one
constituent remnant with preverbal subjects, the subject
itself (19a), and a two constituent remnants for preverbal

objects and pro (19b).115
(19) a. [SU no/tambi&i/tampocQ/sf]
b. [DO/IO [pro no/tambten/tampoco/sfl
Assuming an analysis such as (19b), we would make the
prediction that it could be possible to substitute pro in
Spec IP, for an overt subject. Nevertheless, (20b) below,

which could be derived from (20a) by assuming that the
projection in bold is subject to ellipsis, shows that this
prediction is incorrect. It is clear that remnants

containing a preverbal 10 do not admit a preverbal overt
subject.116
(20) a. A ti los policias te van a detener, pero me parece que a Maria el detective no la va a
d etm er.
To you the police cl w ill detain, but it seems that to Maria the detective no.
115This is especially clear for the case of the polarity

item "no". Observe that overt subjects obligatorily
precede "no" as shown in the contrast in (i) and (ii) .
Therefore if pro has the same distribution as an overt
subject, it will always have to appear be preverbally with
the polarity item "no" with all DO and 10 remnants.
0) * No Juan vino,
not Juan came.
(ii) Juan no vino

Juan not came
116 There is clear contrasts between (20b) and (ii) below:
Q) A ti los policias te van a detener, pero me parece que a Maria no.
To you, the policement they w ill arrest, but it looks lite to Mania no.
171
b. ??A ti los policias te van a detener, pero me parece que a Maria el detective no [la va
detener]
To you the police cl w ill detain, but it looks like to Maria the detective no.
Additional cases with other discourse polarity items

like tambi&n, as in (21), confirm this fact.
(21) a. *?A Marfa, los niflos Ie dieron un libro y a Pfa, Pedro tamb&n [le dieron un
lib ro ].
To Marfa, the irids cl gave a book and to Pia, Pedro too.
b. *?A Marfa, Juan le dio un libro y me han dicho que a Tomds, Tito tambidn [le dio
un lib ro ].
To Marfa, Juan cl gave a book and cl have told that to Tomds, Tito too
Therefore, we must conclude that at least the assumption

that pro and overt subjects have the same distribution is
untenable. By the same token, the behavior of all cases
presented thus far can be only accounted for if overt
preverbal subjects, DOs, and IOs occupy the same syntactic
position. From this perspective, all of the examples (10)-

(18), represent instances of one constituent remnant,
unlike (20a), (21a), and (21b) which constitute cases of two
constituent remnants.
Interestingly, examples such as these fall into the same
category as those studied by Williams (1977) and Sag (1976)
where a more external XP binds into the elided material

predicated of a subject. Some examples include the
following:117
(22) * The police, I called yesterday but the firefighters, you didn’t
(23) *What did Harry take a picture of?.*What did B ill ? (from Sag 1976)
(24) *John, who B ill saw, and who Bob did too. (from Williams 1977)
117 Judgements vary in this respect. Contrary to Sag's or

William's judgement some speakers find some of these
sentences not that deviant.
172
The degraded nature of the two constituent remnants can
be explained, in part, by the fact that no predication or

binding relation from an external element can be
established into an ellided constituent. Williams and Sag
have accounted for these cases by a condition of identity

of predication on the copied material.
5.2.2 Quantifier extraction.
5.2.2.1 Negative quantifiers.
In Spanish, negative quantifiers can appear in preverbal

position. Example (25) shows a subject negative
quantifier, and (26) and (27) show a DO and an 10 negative

quantifier respectively.
(25) Nadie le debe la renta a Maria.
Nobody (S) owe-3ps the rent to Maria
(26) Nada le debe Juan a sus amigos.
Nothing (DO) cl owe-3ps Juan (S) to his friends GO)
(27) A nadie le debe Juan la renta.
To nobody GO) cl ow-2p.s Juan (S) the rent.
Following the hypothesis illustrated in (5), it could be

argued that while a subject negative quantifier surfaces in
Spec IP, DO and 10 ones arise in a more peripheral

position; one likely to be identified as a focus site.118
If this is correct, the representation of (26) and (27) is
that in (28) and (29) :
(28) A nadie [IP pro (Ie) debe Guan) la renta]
118 As we will see later on, it is not necessary for

preverbal IOs with doubling to be focused when they appear
preverbally (see footnote 121). However, this issue is
irrelevant for the main point of section.
173
(29) Nada [ IP pro le debe (Juan) a sus amigos]
Again, from this analysis we may infer that pro may be

substituted, by an overt subject without inducing a deviant
outcome. Yet, this is plainly wrong as sentences like (30)

and (31) demonstrate.
(30) * Nada Juan Ie debe a sus amigos.
Nothing (DO) Juan (S) cl owe-3ps to his friends
(31) *A nadie Juan Ie debe la renta.
To nobody GO) Juan (S) cl owe-3ps the tent
A similar contrast is observed in contexts with
nonnegative quantifiers:
(32) a. *?A cualquiera tu madre pone contento.
To anybody (DO) your mother (S) make-3ps happy
b. A cualquiera pones contento.

To anybody (10) make-2ps happy
(33) a ?*A todo el mundo Iaprensahaaceptado
To everybody GO) the press (S) have-3ps accepted
b. A todo el mundo ha aceptado la prensa.

To everybody have-3ps accepted the press
Contrary to what we assumed in describing (28) and (29) ,
Laka (1990) contends that the contrast between (25) to (27)

and (30) to (31), shows that the verb has to move above
Spec IP. She argues this is necessary so that a Spec-head

relationship with the negative quantifier can be
established. Nonetheless, certain types of evidence lead
us to question the appropriateness of such an approach.

For example, Spanish, unlike Italian and French, allows
frequency adverbs to precede the verb in declarative
clauses, as it is shown in (34) :
(34) Juan siempre me engafia.
Juan always cl cheat-3ps
174
If, in fact, the verb moves higher up in contexts with
preverbal negative quantifiers, we would expect frequency
adverbs to follow the verb in these contexts. Examples

(35)-(37) indicate that such an expectation is not met; a
preverbal negative quantifier and a negative preverbal

frequency adverb render a grammatical outcome:
(35) Nadie nunca te vio tan triste
Nobody ever cl saw-3ps so sad
'Nobody ever saw you so sad'
(36) A nadie nunca le das iasgracias.

To nobody ever say-2ps thank you
'You never say thank you to anybody’
(37) A nadie ya saludas por las maflanas.
To nobody you greet in the morning.
In addition, the verb movement approach is incompatible

with the idea that Romance clitics cannot be head-adjoined
to the verb or the inflectional projection where the verb
is located as proposed in Kayne (1994) . I discussed
evidence in favor of this prohibition in Chapter 4. If
clitics always precede the verb but are not adjoined to it,
there is no possibility for a Spec-head relation to obtain

between the verb and the preverbal negative quantifier in
(38) .
(38) A nadie le debe la renta.
To nobody owe-3ps the rent.
To nobody, he owes the rent.
If, as I argue, the verb movement hypothesis is not

viable, some other mechanism must be at play in order to
explain the apparent blocking effect exerted by preverbal
overt subjects, DOs, IOs. Otherwise the contrast between
175
cases like (30) and (31) versus those like (25) to (27)
remains unaccounted for. Observe that the contrast is

especially problematic if pro occupies the same structural
position of overt subject as proposed in (5) .
The conclusion I draw from the preceding evidence is
that the position preverbal subjects surface in is not
exclusive of subjects. Instead that position is parallel

to a typical left dislocation. In fact, the
ungrammaticality of (30) and (31) , repeated below as (39a
and b), can be explained in terms of A' relativized
minimality.
(39) a. *A nadie[Juan[Iedebelarenta.
b. ’Nadie [aJuan [ledebelarenta..
We take the subject Juan in (39a), and the 10 a Juan in

(39b) to be in an A' position. Thus, movement of the
negative quantifier produces a blocking effect when it
moves over it. This situation does not arise in (26) and
(27) since, under our view, no intruding material, overt or
covert, intervenes when a negative quantifier is preposed.
5.2.2.2 Extraction o f wh-elements.
As we saw in Chapter 4, the distribution of wh-elements

and the availability preverbal subjects seem to support the
conclusion that overt subjects do not have the same

distribution as nonovert ones. As we saw before, the
obligatory inversion exhibited in contexts like (40) has
led linguists to two different conclusions.
176
(40) a.*iQu& Pedro cotnprd en el mercado?
What Pedro bought in the market?
b.*Â qui6n Susana le dio el paraguas?

To whom Susana cl gave the umbrella?
For example Lema (1989) and Goodall (1991) argue that
the subject cannot occupy Spec IP because the wh-trace

already occupies that position. Alternatively, Rizzi
(1991) proposes that V must move to a higher position
(higher than IP), such as C, to enter into a checking
relation.
Yet as we also saw in Chapter 4, the contrasts offered

by examples (41a) versus (41b) , (42a) versus (42b), and
(43a) versus (43b) below, show that an obligatory inversion
effect is not always forced:
(41) a. ^Qu6 tipo de literatura Octavio Paz nos sugiere que debamos leer?
What type of literature Octavio Paz cl suggests that we should read?
b *i,Qu& Octavio Paz nos sugiere que debemos leer?
What Octavio Paz cl suggests that we should read?
(42) a. £Qu6 tipo de libros a nosostros nos sugiere Paz que debemos leer?
What type of books to us cl suggests Paz that we should read?
b *Qu6 a nosotros nos sugiere Paz que debemos leer?
What to us cl suggests Paz that we should read?
(43) a. A ver, ^de qu£ manera a ti te podnan criticar tus enemigos?
See, in what way to you could criticize your enemies?
See, in what way could your enemies criticize you
b. *A ver, &ctimo a ti te podrian criticar tus enemigos?

See, how to you could criticize your enemies?
See, how could your enemies criticize you.
The conclusion given there was that the inversion effect
vanishes when the wh-constituent is a complex one, such as
Qu& tipo de literatura, ('what type of literature’), and De

qu6 manera, ('in what way’), but it surfaces when faced
with a bare wh-constituent like gu£ ('what'), or cdmo,
177
('how'). Interestingly, when inversion is not required,
subjects, DOs, and IOs pattern equally with respect to the
possibility of appearing between the verb and the Wh-

element as shown in (42b) and (43b).
Again as discussed in Chapter 4, another context where

no obligatory inversion is required, even when a bare wh-
is involved, is found in Spanish in cases of long wh-
extraction:
(44) a. te dijo Maitina que el abogado no le habia dado a tiempo?
lQ u£
What cl said Maitina that the lawyer not cl had given on time?
b. <;Qti£ dijiste que tus padres te iban a regalar?

What you said that your parent were going to give you
(45) a. iA. quiln te dijo Martina que el citatono ya se lo habian dado?
To whom cl said Martina that the subpoena already cl they had given?
b. £Qu6 pensabas que a ti te iban a regalar?

What you said that for you they were going to give you?
In the examples in (44), the wh-element has been

extracted over a subordinate preverbal subject, and in (45)
it has been extracted over a preverbal object. What we
concluded from these data was that the obligatory inversion
effects must crucially depend on the syntactic nature of
the moved wh-element. When this effect is controlled, we

observe that preverbal subjects, DOs, and IOs do pattern
the same; they are all allowed between the wh-word and the
verb.
Finally, as in the case of negative quantifers, Suner
(1994) has pointed out the possibility of insertion of
178
adverbials between the wh-word and the verb as in (46) and

(47) :119
(46) ^Con quidn nunca piensas tu hablar? (From Sufier 1994)
With who never think-2ps you to speak
Who don’t you ever think to speak
(47) quidn siempre le dices tus secietos?
to whom always cl-say-2ps your secrets.
Who do you always tell your secrets
These facts again, show that a V-to-C is not adequate

for the inversion effect in Spanish. Nevertheless, it is
not clear why pro should have a different behavior than
overt subjects given the hypothesis illustrated in (5) .

Thus, we seem to run into the same paradox for the cases of
questions as we did for negative quantifiers as the
contrast in (48) shows:
(48) a. *iQ*d Pedro compnJ en el mercado?
What Pedro bought-3p.s in the market.
b. &Qu6 pro comprd en el mercado?

What pro bought in he market
In conclusion, the internal structure of wh questions in

Spanish also leads to unexpected asymmetries—as in (48)— if
pro and preverbal overt subjects have the same
distribution. We have seen when no inversion is required,
preverbal DOs, IOs, and subjects can appear between the
Wh-element and the verb as in (41a), (42a), and (43a). If
more structure is involved with preverbal DOs and IOs than

with preverbal subjects, as proposed in (5) , one might
119 See also the same point with quantifiers in the previous
section.
179
expect some contrasts depending on whether subjects or DOs

and IOs intervene.
5.2.3 The scope o f quantifiers in preverbal position
In the previous section the kind of evidence introduced

argued against the idea that there is more structure with a
preverbal DO or 10 than with a preverbal subject. It also
showed that given the postulation of subject pro, overt

subjects and pro must have a different distribution. These
two conclusions indicate that the standard hypothesis
represented in (5) is inadequate. Thus, until now I have
shown that pro and lexically realized subjects have
different distributions. In this section, I demonstrate

that lexically realized subjects and IOs share the same
distribution. In this way I provide more evidence for the

uniform hypothesis showing that the restrictions on
quantificational interpretation of preverbal DOs and IOs is
parallel to that one of preverbal subjects.
The special properties of subject quantifiers in
preverbal position have been taken as evidence for the
appearance of these elements in Spec IP, a nonproperly
governed position. For example, Rizzi (1982)—following
Kayne (1984a)—reports that elements like nessuno cannot be

licensed as NPI by a negation in the matrix when they
appear in the preverbal subject position of a finite
embedded clause. They thus cannot have clause wide scope
180
interpretation. They can only be licensed as an NPI with
respect to the matrix negation when they appear

postverbally as shown in the following contrast:
(49) Non credo che sia anestato nessuno.
I do not think that be arrested nobody
I don't think anybody will be arrested
(50) Non credo che nessuno sia anestato.

* I don't think that nobody w ill be arrested (NPI)
I think that nobody w ill be arrested.
Rizzi (1982) gives an explanation for this contrast in
terms of the ECP. He assumes that there is movement of the

NPI to target the licensing negation at LF. Consequently,
the movement of the preverbal subject nessuno at LF would
not be possible in (49) because Spec IP is not a properly

governed position. However, the movement would be
perfectly licensed from the postverbal position in (50)
where the NPI nessuno is governed by the inflection.120
120The adaptation of the nes su no facts into Spanish is

rather controversial. Not all speakers reject the wide
scope interpretation in (i). For those speakers there is
no contrast between (i) and (ii).
(0 No creo que nadie Uegri.
I don’t think that nobody came
(ii) No creo que llegd nadie.
I think that came nobody.
Interestingly, the speakers that do not allow wide scope
interpretation for (i) do not allow it either for other
preverbal 10 negative elements. For those speakers, wide
scope interpretation is easier with the b) cases thanwith
the a) cases in iii) and iv) respectively.
(iii) a. No creo que le gustti a nadie.
I do not think Chat cl-liked to nobody
I do not think that anybody liked it.
b. No creo que a nadie le gustd.
I do not think that to nobody cl -liked
I do not think that anybody liked it
181
Jaeggli (1987) adopts this hypothesis for Spanish and
observes that the same contrast holds in the distribution
of wh-elements in situ. Compare (51) with (52) :

(51) a. *Qu6 dijiste que qui&i comprti el otro dia?
What did you say that who bought the other day.
b. Qufc dijiste que compid qufti el otro dia?
Who did you say that bought who the other day
(52) a. •Quien piensa que quidn Io gand?
Who thinks that who won it
b. Qui€n piensa que Io gand quidn?
Who thinks that won it who
The contrast between (51) and (52) is parallel to the
one shown for Italian; the subject wh-element cannot appear
in the Spec IP. Jaeggli (1985) suggests that the
ungrammaticality of (51a) and (52a) has to berelated to
the inability of the subject wh-word to move at LF from a
nonproperly-governed position like Spec IP. However,

similar contrasts hold for other preverbal wh-elements like
IOs or DOs. Observe the following contrasts when the 10 is
preverbal or postverbal as in (53a and b) . As shown in

(53c), a non-wh-IO can appear preverbally in the embedded
clause, minimally contrasting with (53b).
(53) a. £Qui£n czees que va a saludiar a qui£n?
Who do you think w ill greet who
b. * Q uiln crees que a qui£n va a saJudar?
Who do you think that to whom w ill greet
(iv) a. No creo que se Io enviase a nadie.

I do not think that cl-cl sent to nobody
I do not think that they sent it to anybody
b. no creo que a nadie se Io enviase
I do not think that to nobody cl-cl sent
I do not think that they sent it to anybody
182
c. iQuidn crees que a ti te va a saiudar?
who do you think that to you w ill greet
We believe that the explanation provided for the pairs
(51) and (52) should be the same as the one given for (53)
with preverbal IOs. Thus, an explanation in terms of the
ECP and the crucial idea that Spec IP is a non properly
governed position falls short of giving a complete

understanding of the entire paradigm.
By contrast, Uribe-Etxebarria (1992) provides the basis

for such a uniform explanation. She observes that the
interpretation of preverbal universal quantifier subjects
is more restricted than that of postverbal universal
quantifier subjects in Spanish. For example, preverbal
subject quantifiers in an embedded clause cannot take wide

scope over an extracted wh-object. Thus, in (54) the
reading in which cada senador loves a different person is
not possible. The most salient reading is the one in which
cada senador loves the same person (narrow scope of the

universal quantifier) .
(54) Âquifo dices que cadasenador amaba?
Who did you say that every senator loved?
However, when the subject quantifier is in a postverbal
position, the ambiguity reappears. Sentence (55) can have

both a wide scope and a narrow scope reading.
(55) Â qui6n dices que amaba cada senador?
Who did you say that loved every senator
Who did you say that every senator loved
183
On this point, English significantly differs from

Spanish. May (1985) claims that a preverbal quantifier can
take wide scope over a extracted wh-element:
(56) What do you think everyone bought? (from May 1985)
From this contrast, it can be concluded that the nature
of the position occupied by the subject in Spanish has to
be characterized in a rather different way from English.
Uribe-Etxebarria (1992) proposes that the mismatches can

be understood if subjects are in an A' position in
Spanish.121 She makes the interesting parallel between this
121 The idea that preverbal subjects like preverbal DOs and
IOs are dislocated to an A ' position is usually challenged
by the following contrast from Rizzi (1986b) :
(i) *Nessuno, I’ho visto.
Nobody, I have seen him.
(ii) Nessuno &venuto
Nobody came.
Rizzi (1986b) takes (ii) to show that negative quantifiers
cannot be left dislocated in an A' position. Thus, if
subjects are in an A' position in (i) it is hard to
understand the contrast in grammaticality between the two
examples. We take this contrast just to mean that object
negative quantifiers are incompatible with the accusative
clitic. Thus, the contrast in (i) and (ii) is reminiscent
of the ban on the doubling object negative quantifiers with
accusative clitic (see Dobrovie Sorin 1990 for Rumanian
and Suiter 1988 for Rioplatense Spanish) . Observe that an
IO negative quantifier in preverbal position, which
requires doubling in all dialects of Spanish, behaves
similarly to a preverbal subject negative quantifier. For
instance, it does not need to be focused in order to appear
preverbally as shown in the parallelism between (iii) and
(iv) .
(iii) Dicen que a nadie Ie tienes miedo.
they said that to nobody you fear
They said that you don’t fear anybody.
(iv) Dicen que nadie se airevid

they said that nobody dared
184
position and the topicalization position of English. For
example, a wide scope interpretation of a topicalized
quantifier in English is not possible as shown in the

contrast in (57a) and (57b) . Similarly, no wh-element
seems to be allowed when it is topicalized as shown by the
ungrammaticality of (58b).
(57) a. Someone thinks that every problem, Mary solved, (from Lasnik and
Uriagereka.1988).
b. Someone thinks that Mary solved every problem
(58) a. Who thinks that I like who?

b. *Who thinks that who, I like
Therefore, this type of approach leads to the treatment

of preverbal subjects as having the same quantificational
constraints of a topic-like element. This is feasible if

the preverbal DOs and IOs, on the one hand, and preverbal
subjects, on the other, are characterized in the same
fashion.
The conclusion arrived at from these facts is that no
further movement is licensed when a quantifier has been

already moved to an A' peripheral position before Spell
Out. The facts are subsumed under the Unambiguous Binding
approach of Muller and Sternefeld (1993). Namely, subjects
in a left dislocated position (one type of A' movement) are
There is a difference between the former cases and the one

with direct object negative quantifiers, which cannot be
doubled. The DO must be necessarily focused to apppear
preverbally.
(iv) Dicen que NADA compraste en la tienda
They said that nothing you bought in the store.
185
followed by movement of the quantifier at LF (another type
of A' movement) . This yields a violation of the principle
of Unambiguous Binding in (51) of Section 2.4.2.122
In conclusion, it has been shown that preverbal subject
quantifiers in Spanish behave in many cases in a similar

fashion to preverbal object quantifiers. This similarity
can be seen in the fact that we find the same contrasts
with direct objects in wh-questions such as in (51) and
(52) and with preverbal IOs as in (53) . It has also been
shown that Spanish preverbal subjects differ from their

English counterparts with respect to wide scope
interpretations of a quantifier. This cross-linguistic

difference confirms our suspicions regarding the assumption
that the preverbal subject occupies the Spec IP in Spanish.
5.3 Dislocated subjects. Previous accounts.
If, as we have been showing, preverbal subjects parallel
preverbal objects in their behavior, we have to give an
answer as to what prevents subjects from appearing in Spec
IP. We will see that the answer is crucially related to a
122 Epstein (1992) gives a theoretical motivation for this

ban in terms of economy of derivation. For a quantifier in
a topic position to have wide scope, it is necessary to
move further at LF in a second step. This derivation in
two steps is blocked by a derivation in one step where the
quantifier and wh-element are not moved to the topic
position before Spell Out, but are moved in one swoop at LF
from a postverbal position. For advantadges of an
Unambiguous Binding Approach see Muller & Sternefeld
(1995) .
186
broad question of what allows Spanish to have the V S 0 as
well as the V O S orders.
There are two lines of inquiry that have attenpted give

an answer to these questions. In this section I review
them and point out some of their shortcomings. I propose a
solution which in some respects combines some important

conclusions from these two approaches.
5.3.1 Non Polysynthetic Approaches.
One type of answer to the question posed above is the
idea that case and agreement are satisfied postverbally,
and therefore movement to a preverbal position is triggered

by discourse reasons as in any left dislocation. This kind
of approach has been proposed by Contreras (1991) and

Zubizarreta (1994).
Contreras (1991), for example, suggests that economy
principles prevent IP from projecting a specifier position.
In his analysis, subjects in Spanish are assigned case and

agreement under C-command by inflection, which is taken to
be lexical and, as such, is able to L-mark a postverbal (C-
commanded) subject. Given this perspective, there is no
justification for the projection of a specifier position of
IP. Preverbal subjects are generated as adjuncts, and case
and agreement are presumably assigned postverbally to a

null pro when no overt subject appears.
Similarly, Zubizarreta (1994), assuming a richer system
of inflectional projections, proposed that subjects are

assigned case and agreement prior to the movement of the
verb to the highest inflectional projection. The highest
Spec of the projection where the verb ends up is a position
related to what she calls 'neutral topics' and movement of

the subject to that position is purely optional.
(61)
AGRS
In important respects, these approaches claim that

Spanish is the mirror image of English. While in English,
subject NPs satisfy their agreement and case properties in

a Spec position above the position of the verb, in Spanish
subject NPs satisfy these properties in Spec positions
before the final target movement of the verb at Spell Out.

In this respect, Spanish is thought to share important
commonalties with V S 0 languages like Irish or Arabic.
188
However, there are many aspects of Spanish that do not seem
to fit into this perspective. From a discursive point of

view, the Spanish V S 0 order has special pragmatic
properties that do not seem to occur in Irish or Arabic.
From a syntactic point of view, there seem to exist quite

important differences between standard V S 0 languages and
Spanish. For instance, it is typical of this type of
language for there to be adjacency requirements between the
verb and the subject that follows. However, such
constraints does not seem to apply in a language like

Spanish, which, for example, also allows the V O S order as
we have shown in Chapters 2 and 3.123

(62) Ayercomprd ellibroJuan.
YesterdayboughtthebookJuan.
2.3.2 Polysynthetic approaches
Another type of approach claims that Spanish is
nonconfigurational or polysynthetic with respect to the
distribution of subjects. This seems to be corroborated by
the existence of a rich inflectional paradigm, the
possibility of pro-drop, and the different possible
positions of the subject (free word order).
This intuition has been put to work in different ways by
various linguists (see Jelinek, 1984) . Baker (1996), in

his study of polysynthetic languages claims that lexical DP
123 A detailed explanation of the syntactic differences

between Spanish and V S 0 languages goes well beyond the
scope of this chapter.
189
arguments are always associated with pro. In his view,
pro's are the real argument and lexical DPs are adjoined to
a more peripheral position as in (62). This proposal is
adopted by Barbosa (1996) in her study of the distribution
of preverbal subjects in European Portuguese. A pro would
be in a Spec IP position, while preverbal subjects are

always left dislocated or adjoined.
Top
VP
There are two important questions that arise from

adopting this proposal: (1) Why could a lexical DP not
occupy the argumental position? (2) What kind of syntactic

relation is established between pro and the corresponding
lexical'DP?
Question 1 is related to the fact that pro and lexical

DPs do not seem to have the same distribution as we have
seen in Section 1. The ideathat lexical subjects and pro
must have different distributions has been also argued by

Rigau (1988) for Catalan and Cardinaletti (1996) for
Italian (see also Bonet 1988 and SolA 1992) . For example,
Rigau shows that in Catalan the behavior of pro parallels
that of clitics and not that of strong pronouns. She

proposes that pro appears in Spec IP, and a strong pronoun
190
appears adjoined to IP. A similar intuition is found in
Cardinaletti who suggests that in Italian the behavior of
pro runs on a par with some lexical pronouns but not with
others. Italian seems to have strong pronouns like lui
('he ') and loro ('they1) and weak pronouns like egli
('he'), and esso ('it'), and pro seemingly behaves like

weak pronouns. Cardinaletti1s conclusion is that pro must
move to the specifier of a functional projection Agr2,
whereas lexical subjects may only appear in the specifier

of a higher Agrl. If these arguments are correct, they
would build a strong case against considering lexical DP's
and pro as exhibiting the same distributional properties.
The question not yet answered by these approaches is why

pro and a lexical subject must have a different
distribution. Baker (1996) proposes an answer based on the
idea that rich inflections absorbs case:

(64) An agreement morpheme adjoined (o a head X receives that head’s case at S-
structure/PF.(From Baker 199S, page 86)
Thus, the reason why lexical DPs cannot occupy the

argumental position is due to the lack of case. This
reasoning, however, crucially implies that pro does not
have case previous to Spell Out. Therefore, the case
filter has to formulate as a condition on S-structure that
applies only to lexical XPs:
(65) The Case Filter
* NP without case if NP has phonetic features and is in an argument position.
However, a formulation of the case filter in these terms

is dubious given the visibility condition, which requires
an XP to be in a case position in order to be visible for
theta role assignment (see Lasnik & Uriagereka 1988 and
Chomsky 1995, chapter 1) . On the other hand, Rizzi (1982

Chapter 4, 1986a) has also shown that pro is licensed only
if it is in the context of assignment of case. This is
especially crucial for the instances of pro in the so-

called Aux-to-Comp constructions in which Rizzi argues that
expletive pro is licensed by the auxiliary in C, which

exceptionally assigns nominative case to it:
(66) Ritengo esser pro nevicato anche sotto I mille metri.
I believed [to have snowed even below 1000 meters]
Following these reasonings, new premises need to be
added to case theory. One must assume that even if pro
does not get morphological case at PF because it is

absorbed by the agreement, it must still receive it at LF.
Thus Baker (1996) adopts a conjoined theory of the case
filter that applies to lexical XPs at PF and applies to pro
at LF. In this way, agreements are deleted at LF and case
is assigned to pro at that level. Notice, that it still
difficult to understand how the lexical DP associated with
pro gets its morphologic case before Spell Out, given that
case has been absorbed by the corresponding agreement.
192
This issue is particularly problematic if the DP and pro
form a chain as is claimed by Baker (1996) (see below).

The second question that a polysyntesis type approach
has to confront is how to define the syntactic relation
established between pro and a lexical DP. It is obvious

that the relation cannot be pure binding because that would
lead to a principle B violation since pro would be bound by

the lexical DP related to it:
(67) Juan{ pro-, vino.
Baker assumes that the XP and pro are connected by a
chain of the type proposed for CLLD by Cinque. This kind

of chain, however, has a rather special status. It is not
derived by movement, but it is computed

representationally .124 There must be a matching of indexes
between the adjoined XP elements and pro.125 Baker (1996)

formulates the chain condition in the following way:
(68) The chain condition
X and Y may constitute a chain only if:

(0 X C-commands Y.
(ii) X and Y are co-indexed.
(iii) There is no barrier containing Y but not X
(iv) X and Y are no distinct in morphosyntactic features (Le. category, person, number, gender,
case, etc..)
124 If it were derived by movement, there would be no

reason for the existence of pro in argumental position in
the first place. The chain has to be always computed
representationally.
125 The nature of the algorithm that determines how this
matching of indexes takes place is not explicitly discussed
by Cinque (1990) nor by Baker (1996).
193
As the chain condition is formulated, it leaves open the
possibility that the two elements involved could be either

a pronominal and a lexical XP or two lexical XPs with the
same index. As we mentioned earlier, binding plays no role

in the computation of this kind of chain. However, as
proposed by Baker, the only two elements involved in this

kind of chain are a lexical XP in an adjoined position and
a empty pronominal in the argument position. Why should
this be so? It remains unclear why the existence of this
special kind of chain (lexical XP, pro) relies on the

already stipulated property that pro can only occupy
argumental positions. There does not seem to be any

obvious answer. Thus, Baker (1996) proposes the adjunct
licensing condition below:
(69) The adjunct Licensing Condition (from Baker 1995)
An argument-type phrase XP generated in adjoined position is licensed if and only if it forms a

chain with a unique null pronominal in an argument position.
In conclusion, we have seen that there are two possible

alternatives to account for why preverbal subjects are
dislocated. The first type of approach relies on the idea
that subjects satisfy case and agreement before the verb

moves to a higher inflectional projection in a similar
fashion as has been described for V s 0 languages. This
proposal encounters problems in that Spanish does not seem

to fit very well into the class of verb-initial languages
from either a discursive or a syntactic point of view. The
second alternative, adopts the idea that Spanish shares the
194
property of being polysynthetic with respect to the
syntactic distribution of subjects. In this respect,

Baker's polysynthesis proposal, which requires pro's to be
in the argumental position and any other XP to be adjoined,
seems promising. However, if we adopt this hypothesis we
have to adopt nonobvious assumptions like a conjoined
formulation of the case filter. Even in this case, it is
unclear how an adjoined lexical DP receives case. On the
contrary, we saw that the chain relation between the
dislocated XP and p r o had a special status. The chain is

not the product of movement but takes place by a mechanism
of co-indexing. Finally, it is unclear why the existence
of this special type of chain is contingent on the fact
that an empty pronominal is in an argumental position.
5.4 Person Agreement as a clitic
The reason preverbal subjects behave like preverbal DOs

and IOs is, it would seem, related to the fact that the
language allows free variation with respect to the position

of the subject. By the same token, it has to be also
related to the possibility of pro-drop. In this respect it
is feasible to group Spanish with the polysynthetic

languages.
However, the best way to derive these properties is

different from Baker's polysynthesis parameter. Instead,
195
it makes sense to adopt the proposal made by Taraldsen
(1992) that agreement is a clitic and, as such, it not only

absorbs case, but it also receives theta role assigned to
the subject.126
The idea that Agr is an argument of the verb seems

confirmed by how binding is crucially determined by it.
Hurtado (n.d.), Jelinek (1984), Olarrea (1994) and Torrego
(1996) notice that Spanish allows certain plural DPs to
agree with lp, 2p, as well as 3p in the plural verbal

paradigm.
(70) a. Los estudiamestenemosmalamemoria.
The studentshave-lpp badmemory.
b. Los estudiamesten&smalamemoria
Thestudentshave-2pp badmemory.
c. Los estudiamestienenmalamemoria.
The studentshave-3pp badmemory.
Crucially, there is a change in patterns with respect to
binding depending on whether the plural DP is associated or

not with such agreeing elements. For instance, a plural DP
object such as los estudiantes, not associated with any
agreement or clitic, cannot be coindexed with either the
lpp pronoun in the adjunct clause in either (71a) and

(71b) . The DP los estudiantes takes the default 3pp
pronoun or an agreement feature as shown in (72) :
(71) a. *Acusaron [alosestudiantes],despudsdequesepeleasenconnosotros,.
They accused [the students], after they fought with us,.
126 The theory outlined here differs from Taraldsen's in

several respects. For example, Spanish agreement does not
head an independent functional projection, but it is a pure
argument which enters into Spec-head agreement with a
doubling XP.
196
b. * Hablaronde [los esmdiantes]; despuds deque sepeleasen con nosotros:.
They talked about [the students]: after they fought with u%.
(72) a. Acusaran [a los estudiantes]: despuds deque sepeleasen con eilos:.

They accused [the students]: after they fought with them^
b. Hablaronde [los estudiantes]: despuds deque sepeleasen con eilos:.
They talked about the students: after they fought with them;.
These facts show that a plural DP binds a 3pp pronoun or
agreement by default. However, Olarrea (1994) points out
that when the plural DP is associated with lpp or 2pp
subject agreement, the binding effects are the opposite.
The plural DP associated with lpp subject agreement can

only be bound with lpp in the adjunct clause as in (73b) ,
but not with a 3pp as in (73a) .
(73) a. *[Los estudiames]: salimos de la reunion despuds de que los; acusaran.
The students; leave-lp.ers pL the meeting after they accused them:.
b. [Los estudiantes]: salimos de la reunidn despuds de que nos; acusaran.

The students leave-lp.ers pi the meeting after they accused us:
Therefore, the examples above point to the conclusion

that the appearance of agreement on the verb is crucial in
evaluating what the possible antecedent for a pronoun can
be. This could only be shown in a language that allows

different agreement possibilities for the same DP as shown
above. When the DP appears not to be associated with any

agreement element, it can only bind a default 3p pronoun.
When the DP is associated with an agreement morpheme it can
only bind a pronoun whose person specification matches the
agreement associated with the DP binder.127
127 The conclusion is rather puzzling for those theories

that claim that pro and subject DPs might have partially
the same distribution (eg Chomsky 1982 and Rizzi 1986a) .
From that perspective, it is harder to express how the
197
The same kind of paradigm can be observed in the domain
of standard clitic doubling. A non doubled DP cannot be

the binder of a lpp or 2 pp pronoun in an adjunct clause.
It can only be coindexed with a 3pp pronoun in (72b) .
However, when the DP is doubled by a lpp pronoun, the
judgments reverse, and co-indexing can only take place with

a lpp pronoun. We obtain the following minimal pair in
(74), which parallel the case we have seen for the the
subject agreement paradigm in (73) .

(74) a. * Acusaran [a los estudiantes]; despuds de que hablasen de nosotros;
They accused [the students]; after they talked about us-,
b. Nos acusaran [a los estudiantes]; despuds de que hablasen de nosotros-,.

cl -Ip . pi accused [the students]; after they talked about us-,
The facts point to the same conclusion with respect to

the paradigm of clitic doubling. The presence of the
clitic crucially determines the coreference possibilities
of the DP associated with it. All these data raises two
questions: (i) How can we express the parallelism between
clitic doubling and subject agreement with respect to their

co-reference patterns? and (ii) How can we integrate a DP
and Agr or clitic to explain why they are crucial in
determining coreference with any other pronoun in the

sentence?
We might start answering the first question by taking
seriously the idea that person agreement is a clitic. The

facts discussed above not only showed that there was a
different agreement changes affect the binding

possibilities of these DP.
198
parallelism between clitic doubling and agreement; but also
that agreement was crucial in the determination of A-
binding. Such a result leads us to think that person
agreement itself should be considered an argument. We
propose that the relation between the Agr and the DP is the
same kind of relation established between a DP and a clitic
as we saw above: they are both instances of "clitic
doubling".128 In doing so, we adopt a specific proposal
about clitic doubling made by Torrego (in progress)
Uriagereka (1995a). Torrego has proposed that the clitics

(and in our proposal, Agr) head a DP in the spirit of
Postal (1974) . This DP is also integrated by the doubling
DP, which merges with the head to form what we can call a
big DP. 129
Doubling DP cl agr Doubling DP clacc/dat
128 There is an obvious question to be answered. If person

agreement is a clitic, why should it be obligatory present
on the verb. We think that part of the answer has to be
encoded in the parameter that determines non
configurationality with respect to the subject argument
(see Jelinek (1984) and the discussion below) .
Observe, that the same obligatoriness of the presence of
the clitic is obtained with 10 in certain Spanish dialects.
I will not touch here on the interesting question of why
this is not the case for DO.
129The analysis is not exactly the ones proposed in Torrego
or Uriagereka (1994). They propose that there is a
complement pro of the D. Observe, that given antisymmetry,
the clitic agreement projects to an intermediate X' which
the DP attaches to in order to be a specifier.
199
The virtue of this proposal is that it integrates the
relation between the doubling element and the clitic (or

Agr) into a core notion such as Spec-head relation.130 In
this way the big DP would inherit the phi-features of the
both elements that integrate it: the clitic and the DP.
that doubles it.131 Thus, gender is determined by the DP as
shown in the following example:

(76) Las estudiantes creemos que nosotras/*nosotros tenemos razdn.
The students-fem believe-lpp that we-fem/we-mas are right
The Big DP occupies the corresponding subject argumental

position for cl Agr, as the DO doubled element occupies the
corresponding argument position for object clitic.
In this way, we obtain a way to express why agreement is

crucial in the determination of the binding possibilities
when associated with plural DPs in Spanish. The clitic Agr

will be the head of the Big DP, and therefore it will
determine the person specification on the index of the big
DP. This is the element that enters in the computation for
binding purposes. The Big DP will be only able to be co
indexed with another element that shares the same person
specification already determined by the agreement clitic as
130 Sportiche also claims that the relation of the doubling

DP and the clitic is one of Spec-head at LF. Earlier
theories of doubling either treated the relation of the
doubled and the clitic as one of representational chain
Jaeggli (1982), Borer (1983). Baker's proposals follow the
second way of reasoning in dealing with the problem of
polysynthesis. Some of the problems with that second type
of approach have been pointed out in the previous section.
131 We assume with Harris (1995) and Torrego (1996) that
there is no 3 person feature.
200
the paradigm in (73) showed. When the DP is not doubled by
any clitic, no person specification is explicit in the
reference of the DP. The DP would only be able to be co

indexed with a 3 person by default as shown in the contrast
in (71) and (72) .
As we mentioned before, the agreement in features
between the doubled DP and the clitic agreement is resolved
internal to the big DP, which is in an argumental position

in D-structures. Since the clitic is the head of the DP,
it will be also the element to absorb the nominative case
from the tense specification. They will then be
transmitted by Spec-head to the doubled DP. Following

Alexiadou and Anagnostopoulou (1996) we could also say that
clitic agreement gets case by incorporating to T°. Tense
assigns its case to the big DP by being in a very local
relation with the upper Spec of the VP in which the subject
argument is situated.132
(77)
Nominative
Case
DP Q agr
Clitic agreement would move to a functional projection

external to T as it has been proposed for any kinds of
132For proposals that nominative case is assigned in this

configuration see Koopman and Sportiche (1991) and Kitagawa
(1986) .
201
clitics.133 For example, Uriagereka (1995) proposes an F
position unrelated to tense to which clitics adjoin in
Galician and Portuguese. Here we will adopt the proposals

made by Kayne (1994) in this respect and assume that
agreement clitics like other clitics in tensed clauses
adjoin to an abstract functional head devoid of features
(see also Taraldsen 1992) ,134 Subsequent to the adjunction
of the agreement clitic to the abstract functional head we
call F, the verb with tense would adjoin to it as has been
already proposed for other cases of enclisis with

imperatives and infinitives as proposed by Kayne (1994),
and BenincA and Cinque (1990)
133From our perspective, person agreement is not functional

projection in the language, contrary to tense. (see
Chomsky 1995 and Rouveret 1991 for a discussion of the
special status of Agr in the collection of inflectional
projections
134 As we have discussed before The reasons that drive Kayne
to this conclusion are two: In the first place if clitic
was able to adjoin to a head with features as tense and the
verb subsequently adjoined to tense, a case of multiple
adjunction to the same head. Secondly, if the verb
adjoined to the clitic, already adjoined to tense, the
sequence V CL Tense would be obtained. Kayne (1994) point
out that pronouns never appear within words. It is
feasible to extend his idea to this case and propose that
the sequence V CL Tense is a case of a pronoun within a
word. By treating person agreement as a pronomial clitic,
we therefore offer a formal explanation of why the sequence
Verb-agreement-tense is not attested in the Spanish verbal
paradigm. Observe that such a conclusion is not
immediately available if agreement is treated as a
functional projection as in Pollock (1989) and Belletti
(1990).
202
(7 8 )
FP
TP
V+T CL agr p
VP
The motivation for this last incorporation of the verb
into the clitic could understood under Guasti (1994) and

Rizzi and Roberts's (1989) extension of morphological
subcategorization into the syntax. For instance, Guasti

(1994) in her discussion of causative verbs extends
proposals of Selkirk about subcategorizaion in morphology.
Causative verbs as fare come with a morphological

subcategorization frame that indicates that they require
the incorporation of an infinitive as shown in (79) :
203
(7 9 )
InfP
[inf--
lavorare
A
t
Let us assume that clitic agreement has a morphological
subcategorization frame requiring a verbal host as in (80):

(80) Clv [— VI
This requirement would immediately explain why verbs
appear incoroporated to the left of the clitic agreement.
From this perspective, there is a crucial dictinction

between clitic agreement and object clitics: the latter
lack a subcategor'ization for verbs.135
Now that we have laid out our hypothesis, we can revisit
some of the major conclusions with respect to the empirical
data discussed in section 5.2. The data showed, on the one
135 This perspective of agreement as a clitic poses the

question of what the status of subject clitics are. I
propose that subject clitics are like clitic agreement, but
they lack a verbal subcategorization frame and therefore do
not need a verbal host. Thus, we predict that there is a
complementary distribution between agreement (clitic
agreement in this study) and subject clitics. This
complementary nature is obtained in some Northern Italian
dialects (see Poletto 1993). However, what happens with
cases in which both agreement and subject clitics co-
appear? I am lead to think these are cases of clitic
"tripling" with a structure as in (i). I leave for futher
research what the consequences of this possibility are. I
would assume that French falls in this category :
(0 [[NPcUlcIÎ
204
hand, that perverbal subjects and preverbal objects have
similar syntactic behavior, and, on the other, that the

distribution of lexical subjects seems to differ from the
distribution of pro and silent subjects.
With respect to the first issue, I take the position

that the clitic is the only element that can satisfy 6—role
assignment and case for the subject argument. Thus, we
follow line of thought established by Jelinek (1984) that

in some languages the real arguments are the agreement,
which she also takes to be clitics. This defining property
is encoded in the parameter which distinguishes non
polysynthetic languages from polysynthetic ones.136 As

mentioned before, there are good reasons to think that
Spanish can be classified into the first group with respect
to the subject argument. Thus, we start to understand why

Spanish does not pattern like a canonical S V 0 (English)
or V S 0 (Irish and Arabic) as implied by Contreras (1991)
and Zubizarreta's (1994) approach.

From our perspective the DP is the doubling element of a
clitic argument, and it is completely optional. Its
presence or absence is immaterial to the discharge of the

theta role or the assignment of nominative case to
subjects. Lexical DPs are licensed by being in a Spec-head
136 Baker (1996) encodes the parameter in a different

fashion. He proposes that arguments in these languages
have to be co-indexed to a morpheme (see his 1996
Morphological Visibility Condition p. 17), even though the
morpheme is not the real argument.
205
relation with the clitic agreement inside the big DP, and
they will inherit all their properties in that way. The
position of the doubled DP is motivated purely by pragmatic
reasons since it plays no role for case. In Spanish
subjects would have the option of moving to NeutP or to

FocP as proposed in Chapter 3.137
We are assuming that there is no inflectional projection
Agr and that verbs in Spanish move beyond T° in order to

incorporate to the subject clitic agreement. So any
movement of the doubling DP SU past the V+T+ClatJ, must be
driven by considerations of semantic-discourse
interpretation such as is typical for the case of left
dislocations of DOs and IOs. Thus, we are able to
understand why preverbal subjects and preverbal DOs and IOs

behave in the same fashion. Parallel to the analysis of
preverbal subjects, DOs and IOs are also associated with a
clitic, which in the same fashion is responsible for the
satisfaction of case and theta role assignment.138
Our data also showed that sentences with silent subjects
pattern differently from sentences in which a preverbal
overt subject appears. Given the postulation of pro as a

subject argument, the conclusion was that pro must have a
different distribution from overt subjects. However, from
137 This leads to the conclusion that subjects are always

moved out of the VP.
138 Thus, our perspective leads us to think that CLLD of DO
and 10 are instances of clitic doubling as has been
suggested by Kayne (1995) .
206
our perspective there is never any pro argument in the Spec
of an inflecional projection139. The subject argument is

always the agreement clitic that appears enclitic on the
verb. Obviously, sentences with a preposed doubling DP
subject differ from sentences in which no doubled DP is

preposed, just as sentences with a dislocated object differ
from sentences in which the object is not left dislocated.
Specifically, the preposing adds a new layer of structure
in preverbal position: a topic phrase. In this sense, we

adopt Rizzi's (1995) proposals that left dislocated
elements must involve a topic head as in (81a) . When no
preposing takes place, no TopP appears preverbally as in

(81b):
(81) a. Top P[SU/DO/IO Top] V+T+CLag^ [[tj tj
b. V+T+Cl agr; [VP DP V --
Finally, our proposal compares in interesting ways with
Baker's polysynthesis proposal. Baker proposes that three
important elements are involved in explaining the property
of polysynthesis: pro, lexical DPs, and agreement. The
element pro is licensed by being in Spec-head with
agreement. The lexical DP is licensed by being in a chain

relation with pro. Thus, the polysynthesis parameter
relies on the important idea that rich agreement absorbs

case. This leaves pro being the only possible element to
139 Uriagereka (forthcoming, b) proposes also the

elimination of pro in favor of feature checking in Basque.
His proposal also predicts that when no lexical DP appears
preverbally no specifier of IP is involved.
207
appear in argumental position, given the assumption that
pro does not require case. Thus, case is the factor that
explains the different distribution of lexical XPs and
pro's.
From our perspective there are two elements involved in
explaining non-configurationality: the Agr itself and the
doubling DP. The parameter in question also involves the

special properties of agreement. Agreement morphemes not
only absorb case, but they can also bear a 0-role and are,
therefore, the real argument of the verb. This fact can be

expressed if agreement is thought to be a pronominal
clitic. The question of the different distribution of Agr
and lexical DPs is understood since they have different X'

status: the Agr is a head and must end up in a head
position by Spell Out, while the lexical DP is a maximal

projection and must always be in an specifier position.
Since there is no pro licensed by an inflectional
projection, the question of the different distributions
does not arise.
(82)
XP
Spec
DP
a.*
208
Finally, in the Polysynthesis Hypothesis, the lexical DP
is licensed by being co-indexed to pro given the
postulation of a chain with the following properties:

a) X C-command Y (from Baker 1995)
b) X and y are co-indexed
c) There is no barrier con-containing Y but not X
Also, a condition should be added that postulates that
lexical DP can only be licensed by being co-indexed to a

pro (see the adjunct condition above) .
Given Torrego's account of clitic doubling, the DP is

licensed by being in a Spec-head relation with the
agreement clitic as shown above. Thus conditions a) b) c)
of the chain condition in the polysynthesis parameter
reduce to the standard notion of Spec-head agreement under

our doubling hypothesis. On the other hand, we assume
doubling can only take place when two elements with the
same referential index differ in X' status: one is a head
and the other element is an XP in Spec. Therefore, an XP
will be only able to be related to a referential head

element, in our case a clitic pronoun.
5.4.1 M orphological evidence
In this section, I will give some interesting evidence
that points to the fact that subject agreement has the same
morphological structure of an object clitic. This evidence
209
can be observed by examining -mos. a lpp agreement.
morpheme, as in (83):
(83) hablamos
we speak
hablibamos
we spoke
Harris (1995) in his discussion of object clitics in the
framwork of Distributive Morphology proposes that clitics

share the same internal morphological structure of
nominals. It is interesting to point out -mos shares some
properties with those object clitics.

(84) m-o-s
The m- of m-o - s is arguably the same root that we see in

other lpp pronoun clitics as me. Thus, m- is taken to be
an indicator of 1 person clitic:
(85) m-e
lps
Harris, claims that the object pronouns nos, os, and los
belong to the same word class many ordinary words ending in
o, such as p a l o , ajo, lado , etc. The morpheme o is the
default word marker of what he calls the I class nominals:

(86) n-o-s o-s l-o-s
lpp 2p 3pp ACC
(87) pal-o
It is natural to assume that the o of mos is also the

default word marker of class I nominals:
Finally, the -s is the same marker for plurality we find

in the nominal/clitic system. Evidence that -s is plural,
and not an arbitrary consonant for AgrS, comes from
210
sentences such as (88a) in which the -s of mos disappears

when it amalgamates with the object pronoun nos. There is
no phonological explanation for this phenomenon as shown by
the contrast between (88b) and (89) . The phonological
structure of los and nos is the same is the same, however -

s only must disappear with nos:
(88) a. Va-mo(*s)-nos
Go- 1 p.p. SU- 1 p p OB
Let’s go
b. Uev6mo(*s)-nos este libro

take-lpp this book
Let’s take this book
(89) Llev6mo*(s)-Ios
Let’s take them
Llevemos dos
I think that the apparent anomaly of (88a) can be taken
as a case in which the pronominal roots mo and no form a

clitic cluster, and -s is directly adjoined to the whole as
in (90) 140.
(90) Va- [ [MO] [NOl sj
toot- [ [1 p SU ]- [1 p OB] -Plural]
Harris (1995) has proposed this kind of analysis for
many dialects of Spanish in which the plurality marker -s
is used. He proposes that certain anomalies on the

distribution of the -s in clitic cluster can be understood
if plural -s adjoins as a whole to the cluster. For
example, in varieties of Caribbean Spanish the -s, which
140 Some further morphological condition must explain why

this double adjunction is possible when the two clitics
agree in person specification. However, as we saw in (89),
this is not possible with combination lpp and 3pp in
Standard Peninsular Spanish.
211
clearly indicates the plurality of n os appears following
the singular object clitic lo. Harris (1995) takes the

fact to indicate that s adjoins to the whole cluster as in
(91) . A similar phenomenon appears in Mexican Spanish with
the combination of double objects. In Spanish the

combination of double object clitic in 3 person yields
reflexive/impersonal se instead of le, for the dative.

However, se, contrary to le, cannot indicate plurality.
Nevertheless, in Mexican Spanish a plurality marker for the

dative is added to the whole cluster when se combines with
the object clitic. Harris proposes the analysis in (92) .
He calls the cases of (91) and (92) "parasitic plurals".
It is plausible to think that the same mechanism is
involved in (90) with the MO-NO sequences.

(91) iQuc si NOS ley«Scl cuento? Si, [[NO] [LO] S] ley6 (from Harris 1995)
Did he read us-DAT the story-ACC? Yes, he read it ACC
(92) El sombrero [ [SE] [LO] S] quit£ a los hombres (from Harris 1995)
I tood the men’s-Dat hat-ACC off (them)
Finally, in certain Andalusian dialects and in Judeo-

Spanish, the lpp object clitic is identical to that of
subject agreement, as in (93) . While, it is not natural to
have imperatives with 1 plural in these varieties, speakers

still found that mos subject and mos object in combination
in imperatives are inconcivable as in (95) . However, this

is not the case with other elites as in (96) . The
impossibility of (95) recalls the typical ban against
sequences of the same clitic in combination (Bonet 1991).
212
(93) mos vamos (Andalusian, Judeo-Spanish)
lpp go-lpp
we go
(95) *Vamo(s)mos (Andalusian. Judeo-S)

go 1 ppS-lppO.
(96) ?Comamoslo
Hai-lpp-3ps.
Let’s eat
5.5 Conclusion
In this chapter, I have shown that there is no exclusive
position for preverbal subjects in finite clauses in

Spanish. Evidence against such a position was drawn from
data on ellipsis, extraction of quantifiers, wh-questions,
and finally the restrictions on quantifier interpretation
in this position. The facts either showed that we could

not assume the existence of pro in preverbal position or
that there were striking similarities between preverbal
subjects, DOs, and IOs with respect to quant ificational
interpretation. Thus, the standard hypothesis in (5) was

rejected.
To solve these puzzles, I have proposed the elimination
of AgrS as a functional projection. Rather, we have

proposed that person agreement should be considered an
argument of the verb. Thus, we adopted the position that
AgrS is a clitic, which absorbs theta role and case (see
Jelinek 1984 and Taraldsen 1992). Independent evidence for
such a position comes from the similarities between clitic
213
doubling constructions and agreement. We saw that person

agreement, like a clitic in a doubling construction,
crucially determines the binding possibilities. This could
be shown in those cases in which different person
agreements can be associated with the same plural DP. It
also leads to the proposal that the relation between DP and

agreement is one of clitic doubling.
As a consequence of this analysis, we eliminate the

notion that pro is an argument of the verb in favor of
thinking that person agreement is the real argument. Thus
the empirical argument that pro seems to show a different
distribution from overt DPs reduces to the fact that a

clitic has a different distribution from a DP. Under
Torrego's analysis of doubling, we can accommodate the
different distribution of both elements since one of them

is a head (the clitic agreement) and the other is an XP.
Finally, we have shown there is some morphological evidence
to think that subject agreement has the same morphological
structure of an object clitic. This was mainly illustrated

with the lpp clitic mos.
214
REFERENCES
Ambar, Manuela (1988) Para uma Sintaxe da Inversao Sujeito
Verbo em Portugues, Unpublished ms., Universidade de
Lisboa.
Alexiadou, Artemis & Elena Anagnostopoulou (1996a) "SVO and
EPP in Null Subject Languages and Germanic,"
Unpublished ms, FAS Berlin, University of
Tilburg/UCLA.
Alexiadou, Artemis & Elena Anagnostopoulou (1996b) "On the
Common Formal Properties of Scrambling and Clitic
Doubling," Unpublished ms., ZAS, Berlin and University
of Tilburg, Holland.
Baker, Mark (1996) The Polysynthesis Parameter, Oxford
University Press, New York.
Bach, Eammon S. (1971) "Questions,"Linguistic Inquiry , 2,
153-166.
Barbiers, Zjef (1995) The Syntax o f Interpretation, HIL
Publications, The Hague, Holland.
Barbosa, Pilar (1996) "Clitic Placement in European
Portuguese and the position of subjects," A. Halpern,
A. Zwicky (eds.) Approac hi n g Second: Second Position
Clitics an d R e l a t e d Phenomena, CSLI, Stanford.
BenincA, Paola and Guglielmo Cinque (1993) "Su Alcune
Differenze Fra Enclisi e Proclisi," in Omaggio a
Gianfranco Folena, 2313-2336. Editoriale Programma,
Padova.
Belletti, Adriana (1990) Generalized Verb Movement.
Rosenberg & Sellier, Turin.
Den Besten, Hans and Gerth Webelhuth (1987) "Adjunction and
Remnant Topicalization in the Germanic SOV
languages,"paper presented at the GLOW, Venice.
Den Besten, Hans and Gerth Webelhuth (1989) "Stranding," in
Grewendorf and W. Sternefeld (eds) Scrambling and
Barriers, John Benjamins, Amsterdam pp 77-92.
Bonet, EulAlia (1988) "Post-verbal Subjects in Catalan,"
Unpublished m s ., MIT, Cambridge, MA.
Bonet, EulAlia (1991) M or ph o lo g y After Syntax, Unpublished
Ph.D Dissertation, MIT, Cambridge, MA.
Bosque, Ignacio (1990) Las Categorias Gramaticales,
Editorial Sintesis, Madrid.
Browning, Maggie A. (1996) "CP Recursion And That-t
Effects," L in guistic Inquiry, 27, pp 237-257.
215
Brucart, Jos6 Maria, (1987) La Elisidn en Espahol,

Publicacions de la Universitat Autdnoma de Barcelona
Bellaterra.
Calabrese, Andrea (1982) "Alcune Ipotesi Sulla Struttura
Informazionale Della Frase in Italiano e sul Rapporto
con La Struttura Fonologica," Rivista di Grammatica
G e n e r a t i v a 7 3-38.
Calabrese, Andrea (1990) Some Informal Remarks on F o c us and
Logical Structures In Italian , Unpublished ms,
Harvard University, Cambridge, MA.
Cardinaletti, Anna, (1996) "Subjects and Clause Structure,"
University o f Venice Working Papers in Linguistics,
Vol.6, n.1,1996.
Cardinaletti, Anna and Starke, Michal (1994) Deficient
P ron o un s: a View from G e rm a ni c ms University of Venice
and University of Geneva.
Cheng, Lisa (1991) On the Typology of Wh-Questions,
Unpublished Ph.D. MIT,Cambridge, MA.
Chomsky, Noam (1970) "Remarks on Nominalizations," in R.
Jacobs and P. Rosembaum (eds) Readings in English
Transformational Grammar, Ginn, Waltham, MA, 184-221.
Chomsky, Noam (1981) Lectures on Government a n d Binding,
Foris, Dordrecht.
Chomsky, Noam, (1982) Some Concepts and Consequences o f the
Theory of Government an d binding. MIT Press,
Cambridge.
Chomsky, Noam (1986) B a r ri er s , MIT Press, Cambridge, MA.
Chomsky, Noam (1993) "A Minimalist Program for Linguistic
Theory," in K. • Hale and S. Keyser (eds.) V i e w from
B u i l d i n g 20, MIT Press, Cambridge, MA, pp. 1-52.
Chomsky, Noam (1995) The M i n imalist Program, MIT Press,
Cambridge, MA.
Cinque, Guglielmo (1990) Types o f A '- De p endencies, MIT
Press, Cambridge, MA.
Cinque, Guglielmo (1993) "A Null Theory of Phrase and
Compound Stress," Li ng u is ti c Inquiry, 24, 239-298.
Cinque, Guglielmo, (1996) "The 'Antisymmetric' Program:
Theoretical and Typological Implications," University
of Venice Working Papers in Linguistics, 5, 1-21.
Contreras, Heles (1985) "Spanish Bare NPs and the ECP," in
I. Bordelois et al. (eds.) Generative Studies in
S pa n i s h Syntax, Foris, Dordrecht, pp. 25-51.
Contreras , Heles (1991) "On the Position of Subjects," in
S. Rothstein (ed.) Perspectives on phrase structure:
216
Heads and Licensing, S y n ta x and Semantics, 25.

Academic Press, San Diego, pp. 61-79.
Corver, Norvert and Henk Van Riemsdijk (1994) Studies on
S c r a m b l i n g , Mouton de Gruytier, Berlin.
Delfitto, Dennis & Manuela Pinto (1992) "How Free is 'Free
Inversion'?, Recherches de Linguistique Romane et
Frang a is e d'Utrecht XI, Utrecht University, Utrecht.
Diesing, Molly (1992) Indefinites, MIT Press, Cambridge,
MA.
Den Dikken, Marcel (1995) "Extraposition as Intraposition,
and the Syntax of English Tag Questions," Unpublished
ms, Vrije Universiteit Amsterdam.
D6prez, Viviane (1992) "On the Typology of Syntactic
Positions and the Nature of Chains," Unpublished Ph.D.
MIT, Cambridge.
Dobrovie-Sorin, Carmen (1990) "Clitic Doubling, Wh-
Movement, and Quantification in Romanian," Linguistic
Inq ui ry 21: 351-398.
Dovrobie-Sorin, Carmen: 1991. The syntax of Romanian:
Comparative studies in Romance, ms., CNRS, Paris.
Emonds, Joe (1978) "The Verbal Complex V"-V in French,"
Lin g uistic Inquiry, 9, 151-176.
Eng, Miirvet (1991) "The Semantics of Specificity,"
L i n g ui s ti c Inquiry, 22, 1-27.
Epstein, Samuel (1992) "Derivational Constraints on A'
chain Formation," Linguistic Inquiry, 23 235-258.
Fanselow, Gilbert (1990) "Scrambling as NP Movement," in G.
Grewendorf and W. Sternefeld (eds.) Scrambling and
Barriers, John Benjamins Publishing Company,
Amsterdam, pp. 113-143.
Fergusson, Scott (1997) "Deducing the Invisibility of PP
nodes from Case Checking and Full Interpretation," in
GLOW Nesletter 38, pp 22-23.
Fergusson Scott and Eric Groat (1995) "Defining Shortest
Move," Unpublished ms, Harvard University.
Fernandez Soriano, Olga (1989) Reccidn y Ligamiento en
Espahol (Aspectos del Par&metro Nulo, Unpublished Ph.D.
Universidad Aut6noma de Madrid.
Fontana, Josep Maria (1993) "A residual A-Bar Position in
Spanish," in Duncan et A1 (eds) The Proceedings o f
the T w e l f West Coast Conference in Linguistics. CSLI,
Stanford.
Frank, Robert, Young-Suk Lee, and Owen Rambow (1992)
"Scrambling as a Non-Operator Movement and the Special
217
Status of Subjects," in Proceedings o f the Third

Leiden Conference for Junior Linguists, 3 ppl35-154.
•
Frey, W. (1990) Syntaktische Bedingungen fur die

Interpretation, Unpublished Ph.D. University of
Stuttgart.
Friedemann, Marc-Ariel (1995) Sujets Syntaxiques:
Positions, Inverstion et pro. Unpublished Ph.D.
University de Geneve.
Gazdar, Gerald, Ewen Klein, Geoffrey.K Pullum and Ivan. A.
Sag (1985) G eneralized Phrase Structure Grammar,
Harvard University Press, Cambridge, MA.
Giorgi, Alessandra and G. Longobardi (1991) The Syntax o f
Noun Phrases, Cambridge University Press, Cambridge,
England.
Giorgi, Alessandra and Fabio Pianesi (1996) "Verb Movement
and Syncretic Categories,"Probus 8, 137-160.
Goodall, Grant (1991) "On the Status of Spec IP"in
Proceedings of WCCFL, 10, CSLI, Stanford.
Greenberg, Josef (1966) "Some Universals of Language with
Particular Reference to the Order of Meaningful
Elements"in J. Greenberg (ed) Universals o f Language,
, MIT Press, Cambridge, 73-113.
Grewendorf, G. (1988) Asp e kt e der deutschen Syntax. Eine
Rektions-Bindungs-Analyse, Gunter Narr Verlag,
Tubingen.
Grewendorf, Gunther and Joachim Sabel (1994) "Long
Scrambling and Incorporation," Linguistic Inquiry 25,
263-309.
Haider, Hubert (1988) "0-tracking Systems — Evidence from
German," in L. Marcicz and P. Muysken (eds.)
Con figurational ity, Foris, Dordrecht 185-206.
Harris, James (1995) "The Morphology of Spanish Clitics,"
in H6ctor Campos & Paula Kemchinsky (eds) Evolution
and Revolution in Linguistic Theory, Georgetown
University Press, Washington D.C.
Heim, Irene R. (1982) The Semantics o f Definite and
Indefinite Noun Phrases, Unpublished Ph.D. UMASS,
Amherst.
Heim, Irene R. (1987) "Where Does the Definiteness
Restriction Apply? Evidence from the Definiteness of
Variables," in E. Reuland and A. Ter. Meulen (eds.)
The Representation o f (In) definiteness.
218
Hernanz, Maria Lluisa and Jos§ Maria Brucart (1987) La

Sintaxis, Editorial Critica, Barcelona.
de Hoop, Hellen (1992) Case Configuration and NP
Interpretation, Ph. D Dissertation, Croningen,
Holland.
Heycock, Caroline (1995) "Asymmetries in Reconstruction,"
L i n g u i s t i c Inquiry , 26, 547-570.
Hoekstra, Teun (1984) Transitivity, Foris Pubications,
Dordrecht, Holland.
Hurtado, Alfredo, ND. "La Hipdtesis de la Discordancia"
Revista Argentina de linguistica.
Jackendoff, Ray (1977) X ' Syntax: A Study of Phr a se
Structure, MIT Press, Cambridge, MA.
Jaeggli, Osvaldo (1982) Topics in Romance Syntax, Foris,
Dordrecht.
Jaeggli, Osvaldo (1987) "ECP Effects at LF in Spanish," in
David Birdsong and Jean-Pierre Montreuil (eds)
Advances in Romance Linguistics, Foris Publications,
Dordrecht.
Jelinek, Eloise, 1984. "Empty Categories, Case, and
Configurationality," Natural Language and L in g uistic
Theory 2, 39-72.
Johnson, Kyle (1991) "Object Positions,"Natural Language
and Lin gu i st i c Theory 9, 577-636.
Kamp, Hans (1981) "A Theory of Truth and Semantic
Representation," in Jeroen Gronendijk et al.(eds.)
Formal M ethods in the Study o f Language, Mathematical
Center, Amsterdam, pp. 277-321.
Hampers-Manhe, Brigitte (1997) "Je Veux Que Parte Paul: A
Neglected Construction, "Unpublished ms Groningen
University.
Kayne, Richard S. (1972) "Subject Inversion in French
Interrogatives"in Jean Casagrande and B. Saciuk (eds)
Generative Studies in Romance Languages, Newbury House
Publishers, Rowley, Massachussetts, pp 70-126.
Kayne, Richard S. (1975) French Syntax, MIT Press,
Cambridge, MA.
Kayne, Richard, S (1983) "Chains, Categories External to S
and French Complex Inversion," Natural Language an d
L in guistic Theory 1, 107-139.
Kayne, Richard, S. (1984a) Connectedness and B in a r y
Branching. Foris, Dordrecht.
Kayne, Richard, S (1984b) "le" Talk Delivered in the Fifth
Groningen Round Table.
219
Kayne, Richard S. (1986) "Connexit6 et Inversion du

Sujet"in M. Ronat and D. Couquaux (eds) La Graimaire
Modulaire, Edition de Minuit, Paris.
Kayne, Richard S (1989) "Null Subjects and Clitic
Climbing, "in Jaeggli, 0 and K. Safir (eds) The Null
Subject Parameter, Kluwer Academic Publishers,
Dordrecht, Holland.
Kayne, Richard S. (1991) "Romance Clitics, Verb movement,
and PRO," Linguistic Inquiry, 22, pp.647-686.
Kayne, Richard S . (1994) The A n ti sy m m e t r y o f Syntax, MIT
Press, Cambridge, MA.
Kayne, Richard S.and Jean-Yves Pollock(1978) "Stylistic
Inversion, Successive Cyclicity, and Move NP in
French, "Linguistic Inquiry, 9, pp 595-621.
Kitagawa, Y (1986) Subjects in Japanese an d English,
unpblished Ph.D. UMASS, Amherts.
Koopman, Hilda and Dominique Sportiche (1991) "The Position
of Subjects," Lingua 85.2/3.
Korzen, H. (1983) "Reflexions Sur 1/ inversion dans les
Propositions Interrogatives en Frangais,"In M.
Herslund, 0 Mordrup & F. Sorensen (eds.; Analyses
Grammat ic a le s du Frangais: £tu d es Pu bl i cs A 1'occasion
du 50®”* Anniversaire de Carl Vikner, Revue Romane,
num6ro special 24, 50-85.
Koster , Jan (1975) "Dutch as an S 0 V language,"
L i n g u i s t i c Analysis, 2, 111-136.
Koster, Jan (1987) Domains and D yn a s t i e s , Foris, Dordrecht,
Holland.
Koster, Jan (1993) "Predicate Incorporation and the Word
order of Dutch"Unpublished ms, University of
Croningen.
Kural, Murat (19920 "Properties of Scrambling in Turkish,"
Unpublished ms. UCLA.
Laka, Itziar (1990) Negation in Syntax: On the Nature o f
Functional Categories and Projections. Unpublished
Ph.D dissertation, MIT, Cambridge.
Larson, Richard (1988) "On the Double Object Construction,"
L i n g u i s t i c Inquiry 19, 335-391.
Larson, Richard (1990) "Double Objects Revisited: Reply to
Jackendoff," Linguistic Inquiry 21, 589-633. .
Lasnik, Howard and Mamoru Saito (1992) Move-a, MIT Press,
Cambridge, MA.
220
Lasnik, Howard and Juan Uriagereka (1988) A Course in GB

Syntax, MIT Press, Cambridge, MA.
Lebeaux, David (1988) L an gu a ge Acquisition and the Fo r m o f
the Grammar. Unpublished Ph.D. dissertation,
University of Massachussetts, Amherts.
Lema, Jos6 (1987) "Pronombres Sujeto y Valor Param6trico en
Frances, "Signos 1, 263-288, Universidad Autdnoma
Metropolitana-Iztapalapa, Mexico.
Lema, Josd (1989) "Category Switching: Configurational
Effects of Verb Movement, "in Proceedings o f N E L S 19:
301-313, University of Massachusetts, Amherst.
Longobardi, Giuseppe(1994, "Reference and Proper Names,"
L in gu i st i c Inquiry, 25, 609-667.
Mahajan, Anoop (1991) "Clitic Doubling, Object Agreement
and Specificity," NELS 21, GLSA, UMASS at Amherts.
Mahajan, Anoop (1992) The A / A - B a r Distinction and Movement
Theory, Unpublished Ph.D. dissertation, MIT.
May, Robert (1985) Logical Form, MIT Press, Cambridge, MA.
Moltmann, Frederike (1990) Scrambling in German a n d the
Specificity Effect, Unpublished ms. MIT, Cambridge.
Montalbetti, Mario, 1984, After Binding. On the
Interpretation of Pronouns. Unpublished Ph.D
Dissertation, MIT.
Motapanyane, Virginia (1991) Theoretical Implications o f
the Complementation in Romanian, Unpublished Ph.D.
Dissertation, University de Geneve.
Muller, Gereon (1996) "A Constraint on Remnant
Movement, "Natural L an g ua ge a n d Linguistic Theory, 14,
pp 355-407.
Muller, Gereon and Wolfgang Sternefeld (1993) "Improper
Movement and Unambiguous Binding," Linguistic Inquiry,
24, 461-508.
Muller, Gereon and Wolfgang Sternefeld (1994) "Scrambling
as A-bar Movement" in N. Corver and H. Van Riemsdijk
(eds) Studies in Scrambling, Mouton de Gruytier,
Berlin.
Muller, Gereon and Wolfgand Sternefeld (1995) "A"-chain
Formation and Economy of Derivation," L in gu i st i c
Inquiry , 28, 480-511.
Muysken, Peter (1982) "Parametrizing the Notion of 'Head' "
Journal of Linguistic Research, 2, 57-75.
Olarrea, Antxon (1994) "Notes on the Optimality of
Agreement" Unpublished ms. University of Washington.
221
Orddnez, Francisco and Esthela Trevino (1997) Left

Dis l ocated Subjects and the Pro-Drop Parameter: A Case
S tudy of Spanish, Unpublished ms. CUNY and
Universidad Autdnoma Metropolitana-Iztapalapa.
Pesetsky, David (1987) "Wh-in situ: Movement and
Unselective Binding," In E. Reuland and A. Ter Meulen
(ed) The Representation of (In)definiteness, MIT Press,
Cambridge, MA.
Poletto, Cecilia (1993) La Sintassi del Soggetto nei
Dialetti Italiani Settentrionali, Unipres, Padova.
Poletto, Cecilia (1994) "Three Kinds of Subject Clitics in
Basso Polesano and the Theory of pro" in A. Belletti
and L. Rizzi (eds) Parameters and functional heads,
Oxford University Press, Oxford.
Poletto, Cecilia ( to appear) "The internal Strucuture of
Agr S and Subject Clitics"in Henk V. Riemsdjik (eds)
Clitics in the Languages o f Europe, Mouton de
Gruytier, Berlin.
Pollard, Carl and I. A. Sag (1994) H e a d Driven Phrase
Structure Grammar, University of Chicago Press,
Chicago.
Pollock, Jean-Yves (1989) "Verb Movement, Universal
Grammar, and the Structure of IP," L i n guistic Inquiry,
20, 365-424.
Postal, Paul (1974) "On So-Called 'Pronouns' in English"in
David A. Reibel and sanford A. Schane, M odern Studies
in English, Prentice-Hall, Inc, Englewood Cliffs, New
Jersey.
Reinhart, Tanya
(1983) Anaphora and Semantic
Interpretation, University of Chicago Press, Chicago.
Reinhart, Tanya (1991) "Elliptic Conjunctions—Non-
Quant if icational LF" in Kasher (eds) The Chomskyan
Turn, Blackwell, Oxford.
Reinhart, Tanya (1995) Interface Strategies, OTS Working
papers, Utrecht.
Reuland, Eric (1981) "On Extraposition of Complement
Clauses" NELS. GLSA, UMASS at Amherts.
Rigau, Gemma (1988) "Strong Pronouns, "Linguistic Inquiry,
19, pp. 503-511.
Rivero, Maria Luisa (1994) "Clause Structure and V-movement
in the language of the Balkans," Natural Language and
L in guistic Theory 31, 301-332.
Rivero, Maria Luisa and A. Terzi (1995) "Imperatives, V-
Movemen and Logical Mood,"The Journal o f Linguistics
31, 301-332.
222
Rizzi, Luigi (1982) Issues in Italian Syntax, Foris,

Dordrecht.
Rizzi, Luigi (1986a) "Null Objects in Italian and the
Theory of pro," Linguistic Inquiry 17, 501-588.
Rizzi, Luigi, (1986b) "On the Status of Subject Clitics in
Romance," O. Jaeggli and C. Silva-Corvaldn (eds)
Studies in Romance Syntax. Foris, Dordrecht, 391-419.
Rizzi, Luigi (1990) Relativized Minimality, MIT Press,
Cambridge, MA.
Rizzi, Luigi (1991) "Residual Verb Second and the Wh-
Criterion," Technical Reports 2, University of
Geneva.
Rizzi, Luigi (1995) "The Fine Structure of The Left
Periphery," Unpublished ms. University of Geneva.
Rizzi, Luigi and Ian Roberts (1989) "Complex Inversion in
French"Probus 1, 1-30.
Roberge, Yves, (1990) The Syntactic Recoverability O f Null
Arguments. Mcgi11-Queen's University Press.
Rooryck, Johan (1992) "Romance Enclitic Ordering and
Universal Grammar," The linguistic Re view 9, 219-250.
Ruaix, Josep (1994) Observacions Prdctiques sobre el Catald
d' Avui, "Ruaix, Moid.
Saito, Mamuro (1992) "Long Distance Scrambling," Journal
of East Asian Linguistics 1, 69-118.
Sag, Ivan, 1976. Deletion and Logical Form. Unpublished
Ph.D dissertation, MIT.
Santorini, Beatrice (1990) "INFL and Scrambling in German,"
Unpublished ms. University of Pennsylvania.
Santorini Beatrice and Young-Suk Lee (1994) "Torwards
Resolving Webelhuth's Paradox: Evidence from German
and Korean," In H. Corver . and H. Van Riemsdijk (ed)
Studies in scrambling, Mouton de Gruytier, Berlin.
Sold, Jaume (1992) Agreement an d Subjects, Unpublished
Ph.D. Universitat Autdnoma de Barcelona, Bellaterra.
Speas, Margaret (1990) Phrase Structure in Natural
Language, Kluwer Academic Publishers, Dordrecht,
Holland.
Sportiche, Dominique (1988) "A Theory of Floating
Quantifiers and its Corollaries for Constituent
Structure," Linguistic Inq ui r y 19, 203-238.
Sportiche, Dominique (1996) "Clitic Constructions," in
Rooryck and Zaring (eds) Phrase Structure and the
Lexicon, Kluwer, Holland, pp 213-276.
223
Sportiche, Dominique: ND. Subject Clitics in French and

Romance: Complex Inversion a n d Clitic Doubling -, ms.
UCLA.
Stowell, Tim (1981) Origins of Phrase Structure,
Unpublished Ph.D.dissertation MIT, Cambridge, MA.
Suner, Margarita: (1988a) "Haber +■ Past Participle,"
L i n g ui st i c I n q u i r y 18, 683-690.
Suner, Margarita (1988b) "The Role of Agreement in Clitic
Doubled Constructions," natural Language and
linguistic Theory, 6, 391-434.
Suher, Margarita (1994) "V-Movement and the Licensing of
Argumental Wh-phrases in Spanish," Natural Language
a n d L i n g u i s t i c Theor y , 12, 335-372.
Szabolcsi, Anna (1996) "Verb and Particle Movement in
Hungarian," Unpublished ms. UCLA.
Taraldsen, Knut T. (1992) "Subject/Verb-Agreement in Celtic
and Romance," Proceedings o f NELS, GLSA, Amherts.
Terzi, Arhonto (1996) "The Linear Correspondence Axiom, the
Adjunction Sites of Clitics" in A.M. Di Sciullo (eds)
Configurations, Cornell Casadilla Press.
Torrego, Esther (1982) "More Effects Of Successive Cyclic
Movement " L i n g u i s t i c Inquiry, 13, 561-565.
Torrego, Esther (1984) "On Inversion in Spanish and Some of
its Effects," Linguistic Inquiry, 15, 103-129.
Torrego, Esther (1992) "Case and Argument Structure,"
Unpublished ms. UMASS, Boston.
Torrego, Esther (1996) "On Quantifier Float In Control
Clauses," L i n g u i s t i c Inquiry 27,111-126.
Torrego, Esther (in progress) "Pronouns and Determiners: A
DP Analysis of Spanish Nominals," ms, University of
Massachusetts, Boston.
Trevino, Esthela, 1993. Minimalidad en las Construcciones
Causativas del Espahol. Ph.D. dissertation, El
Colegio de Mexico.
Travis, Lisa (1989) "Parameters of Phrase Structure," in A.
S. Kroch and M.R. Baltin (eds) Alternative Conceptions
o f Phrase Structure, University of Chicago Press,
Chicago, 263-279.
Ultan, R. (1978) "Some General Characteristics of
Interrogative Systems." in Joseph Greenberg (ed)
Universals o f Human Language. Vol 4 , Syntax, 211-
248.
Uriagereka, Juan (1988) On Government, Unpublished Ph.D.
University of Connecticut, Storss.
224
Uriagereka, Juan (1995a) "Aspects of the Syntax of Clitic

Placement in Western Romance," Lin g ui s ti c Inquiry 26,
79-124.
Uriagereka, Juan, (1995b) "An F Position in Western
Romance," in: Kiss, Katalin (ed) Discourse
Configurational Languages, Oxford University Press,
Oxford.
Uriagereka, Juan (1996) "Determiner Clitic Placement," in
Freidin, Robert (eds) Current Issues in Comparative
Gra m ma r , Kluwer Academic Publishers, Dordrecht,
Holland.
Uriagereka, Juan (forthcoming) "Multiple Spell-Out," in M.
Browning (eds) Working Minimalism, MIT Press,
Cambridge, MA.
Uriagereka, Juan (forthcoming) "Minimal Restrictions on
Basque Movements," Natural Language and Linguistic
Theory.
Uribe-Etxebarria, Myriam (1992) "On The Structural
Positions of the Subject in Spanish, their nature and
their consequences for quantification," J. Lakarra
and J. Ortiz de Urbina (eds) Syntactic theory and
Basque syntax, 447-491, Supplements of the ASJU,
Donostia, San Sebastian.
Uribe Etxebarria, Myriam (1995) "On the Structure of Spec
IP and its Relevance for Scope Asymmetries in Spanish
and Englsih," In Amastae et al. (ed) Contemporary
Research in Romance Linguistics, 355-367, John
Benjamins, Amsterdam, Holland.
Vallduvi, Enric (1990) The Informaliona_ Component,"
Unpublished Ph.D, University of Pennsylvania.
Varela, Ana (1988) Bind i ng in Spanish: A Theoretical and
Experimental Study, Unpublished Ph.D. dissertation,
University of Connecticut.
Watanabe, Akira (1992) "Larsonian CP Recursion, Factive
Complements and Selection"in Proceedings of NELS 23,
vol, 2, 523-537. GLSA, University of Massachusetts,
Amherts.
Webelhuth, Gert (1992) Principles a n d Parameters of
Syntactic Saturation,Oxford University Press, Oxford.
Williams. Edwin, 1977. "Discourse and Logical Form'.
Linguistic Inquiry, 8, 101-139.
Zanuttini, Raffaella (1990) 'Two Types of Negative
Markers," Proceedings of N E L S 20, Amherst: University
of Massachusetts.
225
Zubizarreta, Maria Luisa (1994) "El Orden de Palabras en

Espanol y el Caso Nominative," Violeta Demonte (eds)
Gram&tica del Espanol , Pubicaciones de la Nueva
Revista de Filologia Hisp&nica, Mexico, D.F.
Zubizarreta, Maria Luisa (1995) P r o s o d y , Focus, and Word
Order, Unpublished ms., USC.
Zwart, Jan Wouter (1993) D u t ch Syntax, Croningen
Dissertations in Linguistics, Croningen.

Ordoñez 1997

Uploaded by

Copyright:

Available Formats

Ordoñez 1997

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Ordoñez 1997

Uploaded by

Copyright:

Available Formats

INFORMATION TO USERS

The quality o f this reproduction is dependent upon the quality o f the

Oversize materials (e.g., maps, drawings, charts) are reproduced by

Photographs included in the original manuscript have been reproduced

Word Order and Clause Structure in

h dissertation submited to the Graduate Faculty in

This microform edition is protected against unauthorized

Date Richard Kayne

Date 1 Charles Cairns

The City University of New York

Word Order end Clause Structure in Spanish and

Adviser: Professor Richard S. Kayne

This dissertation explores various aspects of word order

and clausal structure in the Romance languages, with

(1994) . This theory makes unavailable certain widely-

assumed mechanisms of analysis such as right adjunction and

The first part (Chapters 1 and 2) explores new analyses

that are inexplicable under a right-adjunction analysis.

This same alternation is the same found in verb-final

therefore that V 0 S is the output of the movement of

Spanish and the other Romance languages, such as Catalan,

Italian, and French leads to the further elaboration on the

analysis proposed. It is argued that the derivation of

The other aspects of clausal structure examined

restrictions on the distribution of subjects in

verb movement toC°. This last proposal is also

incompatible with the assumptions made in Kayne (1994).

The alternative proposed involves complementizer movement

pre-verbal subjects have to be left dislocated.

In the final chapter I link the obligatory left-

dislocated nature of subjects to the rich inflectional

system of the language. If we take person agreement to be

a clitic argument that can receive case and be assigned a

disappears. Evidence for the clitic nature of person

I would like to start by thanking my dissertation

One of the best decisions I have made was to come to New

formation as a syntactician, for communicating his love of

very important to get to a more rich analysis, and for

teaching me that the most difficult problems lead to the

most interesting solutions. I hope my dissertation can

Juan Uriagereka has always been encouraging and

enthusiastic. He would go for the daring analysis and

inspire me to pursue them. He has also supported me on

was a late addition to my committee, and I thank her for

To my professors at CUNY, I would like to thank Charles

thank her for written comments on various parts of this

Many colleagues have directy or indirecly contributed to

this thesis. I thank Elena Herburger for her frienship and

like to thank my friends Sheila Meltzer, Sharon Utakis, and

Josd Antonio Mendez for their friendship and for being

Judy Bernstein was a good friend here in New York and in

views about so many aspects of life. I also thank Nino

Outside of the U.S, I owe thanks to my Mexican "cuates"

and for making it possible for me to discover the richness

To my professors in the Universitat Autdnoma de

Barcelona, I thank Gemma Rigau, Maria Lluisa Hernanz, and.

linguistics. Gemma has also been a good correspondent

whenever I was afraid my Catalan was mixing with my

Spanish. I deeply thank Maria Lluisa for shipping me to

friendship. For Catalan and Spanish intuitions as well as