Linguistic Inquiry Why Agree? Why Move?

Monograph Fifty-Four
Unifying Agreement-Based
and Discourse-Configurational
Languages

Shigeru Miyagawa
Why Agree? Why Move?
Linguistic Inquiry Monographs
Samuel Jay Keyser, general editor

A complete list of books published in the Linguistic Inquiry Monographs

series appears at the back of this book.
Why Agree? Why Move?

Unifying Agreement-Based and Discourse-Configurational Languages

Shigeru Miyagawa

The MIT Press

Cambridge, Massachusetts
London, England
6 2010 Massachusetts Institute of Technology

All rights reserved. No part of this book may be reproduced in any form by any
electronic or mechanical means (including photocopying, recording, or informa-
tion storage and retrieval) without permission in writing from the publisher.

For information about special quantity discounts, please e-mail special_sales@

mitpress.mit.edu

This book was set in Times New Roman and Syntax on 3B2 by Asco Typesetters,
Hong Kong.
Printed and bound in the United States of America.

Library of Congress Cataloging-in-Publication Data

Miyagawa, Shigeru.
Why agree? Why move? : unifying agreement-based and discourse-
configurational languages / Shigeru Miyagawa.
p. cm. — (Linguistic inquiry monograph)
Includes bibliographical references and index.
ISBN 978-0-262-01361-1 (hardcover : alk. paper) — ISBN 978-0-262-51355-5
(pbk. : alk. paper)
1. Government-binding theory (Linguistics) 2. Grammar, Comparative and
general—Agreement. I. Title.
P158.2.M59 2010
415—dc22 2009015564

10 9 8 7 6 5 4 3 2 1
Contents

Series Foreword vii

Preface ix
Acknowledgments xiii

1 Why Agree? 1

2 Why Move? 31

3 Unifying A-Movements 59

4 aP, f-Features, and the A/Ā Distinction 93

5 Wh-Questions and Focus 125

6 Concluding Remarks 143

Notes 145
References 157
Author Index 173
Subject Index 177
Series Foreword

We are pleased to present the fifty-fourth in the series Linguistic Inquiry

Monographs. These monographs present new and original research be-
yond the scope of the article. We hope they will benefit our field by bring-
ing to it perspectives that will stimulate further research and insight.
Originally published in limited edition, the Linguistic Inquiry Mono-
graphs are now more widely available. This change is due to the great in-
terest engendered by the series and by the needs of a growing readership.
The editors thank the readers for their support and welcome suggestions
about future directions for the series.

Samuel Jay Keyser

for the Editorial Board
Preface

One of the great mysteries of human language is the existence of move-

ment operations. Why does movement occur? At every stage in the devel-
opment of generative grammar, an attempt has been made to answer this
question. It would not be an overstatement to say that the answer at any
given point characterizes in an essential way the general nature of the
theory at that particular point. In other words, every step in the evolu-
tion of generative grammar has been, in no small measure, an attempt
to construct an answer to this question that is better than the one before.
In Government-Binding Theory (GB), movement operations—in fact, all
operations—are entirely optional, so Move a can move anything any-
where, anytime. Independent universal principles such as the Empty Cat-
egory Principle and Subjacency extract from this overgenerated set of
strings the subset that constitutes the grammatical strings of a particular
language. In this way, the independent principles not only allow the
theory to reach descriptive adequacy—in the ideal, of course—they also
allow it to reach explanatory adequacy in that they ‘‘give a general theory
of linguistic structure of which each [grammar of a particular language] is
an exemplification’’ (Chomsky 1955/1975:77).
In GB, this ‘‘general theory of linguistic structure’’ is the principles-
and-parameters approach, which informs us, among other matters, of
how language acquisition proceeds from the initial state to mastery of a
language. This is a particularly attractive prospect in that we have, in
principle, a clear description of the initial state of Universal Grammar
(UG), and such a description is a principal goal of linguistic theory.
However, there is one problem. These so-called universal principles are
often—perhaps always—a description of the problem. So long as we
depend on such description, we cannot really know the nature of
language—or, more precisely, I-language (Chomsky 1986). This is the
basis for the Minimalist Program (MP), where e¤ort is made to rid the
x Preface

theory of any element that does not have a natural and independent justi-
fication. In an attempt to live up to this ideal, the direction that the theory
has taken—although by no means the only direction possible—is to view
operations as taking place only as a ‘‘last resort’’ (Chomsky 1993).
Since the early 1990s, at least three classes of movement have been dis-
cussed, two of which observe the last-resort nature and a third that does
not. One type of last-resort movement is found in the work on lineariza-
tion initiated by Kayne (1994).1 On the basis of simple assumptions about
hierarchical structure (asymmetric c-command) and its relation to linear
order (precedence), Kayne has argued that ‘‘[l]anguages all have S-H-C
order’’ (Kayne 1994:47), where S stands for specifier, H for head, and C
for complement. This means that SVO (subject-verb-object) is the basic
word order, and SOV and other word orders that do not conform to the
universal order must have arisen by some obligatory movement. For ex-
ample, ‘‘[i]n an OV language . . . the O must necessarily have moved left-
ward past the V into a higher specifier position’’ (Kayne 1994:48). Setting
aside the precise nature of these movements, the theory predicts that they
are obligatory, and further, that they are restricted to cases in which the
output adheres to the ‘‘antisymmetry’’ order of S-H.
The second type of last-resort movement is a kind of movement that
one might call ‘‘EPP-triggered movement’’ (where EPP stands for Ex-
tended Projection Principle). It is this type of movement that I address in
this monograph, taking liberty with the term EPP to refer to a broader
range of movements than just movement of the subject to Spec,TP.
Included in this general type of ‘‘last-resort’’ movement are certain head
movements, which I discuss in conjunction with pro-drop, and move-
ments of the Ā variety such as wh-movement.
The third type of movement is a purely optional movement that has
properties very di¤erent from the last-resort type. Although it does not
adhere to the last-resort nature of the first two kinds of movement, in re-
cent theory it is suggested that optional movement is motivated in that it
allows an interpretation that is otherwise not possible (Fox 2000; see also
Chomsky 2001, Miyagawa 2005a, 2006). I will not discuss optional move-
ment in this monograph.
Another mystery is the occurrence of agreement systems in natural lan-
guage. There are two general questions to ask about agreement. First,
what is the purpose of agreement? On the surface, agreement appears
entirely superfluous in that information in one part of the sentence (e.g.,
plurality of the subject noun phrase) is repeated in another part of the
sentence (e.g., as plural verbal inflection). Moreover, the content of the
Preface xi

agreement system sometimes appears patently random, as, for example,

in the assigning of gender to noun phrases (e.g., Russian assigns feminine
gender to the word lampa ‘lamp’). Second, why do some languages (e.g.,
the Indo-European family) have agreement, while others (e.g., languages
of East Asia) apparently do not?
The goal of this monograph is to try to answer these questions about
movement and agreement. But does it make sense to address, in one
work, these two issues that are often handled as distinct phenomena? As
it turns out, the answer to one depends on the answer to the other. So it is
not only critical that we deal with both—in fact, the two issues at their
core must be made to interact with each other in a meaningful way.
Why agree? Why move? The simple answer, I suggest, is that although
agreement and movement are the result of distinct operations, they work
in tandem to substantially enhance the expressive power of human lan-
guage. Without agreement and movement, human language would be a
shadow of itself for expressing human thought, impoverished to the de-
gree that it would not be able to express such common notions as topic-
comment, subject of a clause, focus, and content questions. Crucially,
‘‘agreement’’ here includes grammatical features of topic and focus found
in what É. Kiss (1995) has called discourse-configurational languages.
This monograph is organized as follows. In chapter 1, I explore the an-
swer to the question ‘‘Why agree?’’ In chapter 2, I take up the question
‘‘Why move?’’ In the rest of the monograph, I look at the consequences
of the analysis presented in the first two chapters. In chapter 3, I suggest
a way to unify two types of A-movement, the so-called EPP movement in
languages such as English and a class of local scrambling in languages
such as Hindi and Japanese. In chapter 4, I examine Kinande and Kilega,
both languages of the Bantu family, which exhibit ‘‘agreement’’ and
‘‘movement’’ that are implemented di¤erently but are consistent with
the proposed analysis. I also look briefly at Finnish, which shares some
elements with Bantu, and I propose a way of distinguishing A- and Ā-
movements based on phase architecture. In chapter 5, I explore issues re-
garding wh-questions. I also take up a related issue, the intervention e¤ect
invoked by certain elements in wh-questions. In chapter 6, I present con-
cluding remarks.
Acknowledgments

This work began in discussions with Noam Chomsky over several years
about how to make a minimalist approach relevant to languages that do
not have f-feature agreement—Japanese, for example. I am grateful to
Noam for these discussions and for many key suggestions that pushed
the project forward at critical junctures. In the fall of 2006, I taught a
graduate seminar at MIT with Norvin Richards where I presented earlier
versions of the material, and Norvin lectured on material that will appear
in Uttering Trees in the same Linguistic Inquiry Monographs series as this
work. Norvin’s comments and questions, and those from the students,
were enormously helpful in solving some daunting problems and in un-
derstanding what needed to be done for this work to see the light of day.
Others in the MIT Department of Linguistics and Philosophy have
generously o¤ered their time and advice; these include Sylvain Brom-
berger, Michel DeGra¤, Kai von Fintel, Edward Flemming, Danny
Fox, Claire Halpert, Irene Heim, Sabine Iatridou, Michael Kenstowicz,
David Pesetsky, and Omer Preminger. I wish also to acknowledge my
colleagues Morris Halle and Jay Keyser for their encouragement. I am
grateful to many people outside of MIT whose comments I benefited
from. There are too many for me to list everyone, but they include Mark
Baker, Cedric Boeckx, Vicki Carstens, Gennaro Chierchia, Yoshio Endo,
Liliane Haegeman, Nobuko Hasegawa, Anders Holmberg, Jim Huang,
Kazuko Inoue, Kyle Johnson, Hisa Kitahara, Jaklin Kornfilt, Jonah Lin,
Hiroki Maezawa, Keiko Murasugi, Luigi Rizzi, Mamoru Saito, Miyuki
Sawada, Hiroyuki Tanaka, Dylan Tsai, and Yukiko Ueda. Two anony-
mous reviewers provided extraordinarily detailed and critical comments
that helped to shape the final version.
Portions of this work were presented at Harvard University, Kanda
University of International Studies, Keio University, MIT, Nagoya Uni-
versity, Nanzan University, National Tsing Hua University, University of
xiv Acknowledgments

Geneva, and University of Michigan. I thank those who attended these

events for numerous suggestions.
I wish to express my appreciation to Pierre Mujombo, the Kinande
consultant whom I worked with, for his remarkable insights and for his
patience. I thank David Hill for creating the index, Eli Laurençot for
assisting with all aspects of preparing the manuscript, and Anne Mark
for a terrific job of copyediting the manuscript.
1 Why Agree?

1.1 Introduction

In this chapter, I explore issues related to agreement in human language.

Why does agreement occur? Why do some languages appear to have it
while others don’t? I will begin by demonstrating the direct relation be-
tween agreement and movement. I will then make a proposal about the
‘‘agreement’’ and ‘‘agreementless’’ languages that minimizes the di¤er-
ence between them: they are identical in all respects except in what shows
up at T or some related head that triggers A-movement. Both typically
exhibit movement triggered by a grammatical feature at T; in agreement
languages this is f-feature agreement, whereas in agreementless languages
it is topic/focus when such a feature occurs, the latter reflecting what É.
Kiss (1995) has called discourse configurationality. I will hold in abey-
ance until chapter 2 a detailed discussion of why movement occurs.

1.2 The Extended Projection Principle

Chomsky (1981) proposed the Extended Projection Principle (EPP) be-

cause of the appearance of the expletive in existential constructions
(There stands a statue in the town center). The agreement is between the
verbal inflection and the postverbal nominal, and the expletive there fills
Spec,TP. The expletive makes it possible for the existential construction
with this ‘‘long-distance agreement’’ to have a subject. The EPP is, in
fact, informally referred to as the requirement that a clause must have a
subject. With the advent of the predicate-internal subject hypothesis (e.g.,
Kuroda 1988, Sportiche 1988; see also Fukui 1986, Kitagawa 1986), the
theory had to account for the movement of the external argument from
Spec,vP to Spec,TP, and it is the EPP that has been invoked to drive this
operation (Chomsky 1995). To make the EPP applicable to T, Chomsky
2 Chapter 1

(1995) argues that T has a D feature that has to be checked, and attract-
ing a DP (e.g., the external argument) to Spec,TP accomplishes this.
Alexiadou and Anagnostopoulou (1998), who argue that this EPP prop-
erty of T is universal, show that in pro-drop languages, rich agreement in
the form of a head that contains a D feature raises to T to check this D
requirement on T, making it unnecessary for a DP to move to Spec,TP.
Given that the EPP requirement can be met by movement of a DP to
Spec,TP (or an agreement element to T) or by merging an expletive on T,
one question that this general approach to the EPP raises is this: if there is
a choice between movement and merger, is one favored over the other?
Chomsky (1995) notes the following pair of examples as evidence that,
when either merger or movement is possible to meet the EPP, merger is
favored over movement:
(1) a. There seems [TP to be a man in the garden].
b. *There seems [TP a man to be in the garden].
The lower clause has a T, hence the EPP requirement, and this require-
ment is filled in di¤erent ways in these two sentences. In the grammatical
(1a), the expletive there is merged on this T to fulfill the EPP requirement,
and subsequently moves to the matrix Spec,TP, where it again fulfills the
EPP requirement, this time of the matrix T. In the ungrammatical (1b),
the EPP requirement of the lower T is met by moving the DP a man to
its specifier. Under the general approach to the EPP in Chomsky 1995,
both options are theoretically available, but the pattern of grammaticality
suggests that merger (as in (1a)) is favored over movement if there is a
choice between the two, which suggests perhaps that merger is a simpler
operation than movement. I will return to this pair later.
A question that arises with the above approach to the EPP concerns
the fact that the EPP always appears to operate in tandem with some other
element, a point noticed by a number of linguists. If we look at a typical
EPP movement, whereby the external argument moves to Spec,TP, we see
that two elements are involved besides the EPP: Case and agreement.
(2) [TP He is eating pizza].
The subject, he, which has undergone movement to Spec,TP, agrees in
Case (nominative) and number (singular) with T. This situation, in which
both Case and agreement identify the target of the EPP movement, is
typical—in fact, so typical that various linguists have proposed that the
EPP should be combined with, or derived from, either Case or agreement.
For example, Bošković (1997, 2002) and Martin (1999) argue that an ex-
pletive must have Case, and this, according to them, makes it possible to
Why Agree? 3

predict its distribution. Behind their studies is the desire to derive the EPP
from Case considerations, something they try to accomplish by assuming
that Case can only be checked in the specifier position of the head respon-
sible for valuing Case (e.g., T) (Boeckx 2000, Epstein and Seely 1999;
see also Koopman 2003, Koopman and Sportiche 1991). In contrast,
Chomsky (2000, 2005, 2007, 2008), Kuroda (1988), Pesetsky and Torrego
(2001), and I (Miyagawa 2005b), among many others, suggest that the
EPP is identified with agreement.
Which is the right answer for the EPP—Case or agreement? Or is the
EPP simply an independent phenomenon, as previously and still widely
assumed (e.g., see discussion in Landau 2007)? Looking only at languages
such as English, which is the language in which the EPP has been most
extensively studied, it is di‰cult to tease apart the di¤erent components
to get at the exact identity of the EPP. To find compelling evidence, we
have to go beyond the familiar languages whose EPP properties have
been investigated.
A number of languages display a phenomenon called ‘‘agreement
asymmetry’’ in which the agreement on the verb di¤ers depending on
whether the subject occurs pre- or postverbally.1 In the Northern Italian
dialects of Trentino (T) and Fiorentino (F), verbs do not agree with post-
verbal subjects; the verb instead has the unmarked neutral form (third
person masculine singular) (Brandi and Cordin 1989:121–122).
(3) a. Gli è venuto delle ragazze. (F)
b. E’ vegnú qualche putela. (T)
is come some girls
‘Some girls have come.’
Full agreement on the verb—agreement in number, in this case—is not
allowed with postverbal subjects, as shown in (4).
(4) a. *Le son venute delle ragazze. (F)
b. *L’è vegnuda qualche putela. (T)
they are come some girls
‘Some girls have come.’
In contrast, full agreement must occur, as in (5a–b), if the subject moves
to preverbal position (presumably Spec,TP) (Brandi and Cordin
1989:113).2
(5) a. La Maria la parla. (F)
b. La Maria la parla. (T)
the Maria she speaks
‘Maria speaks.’
4 Chapter 1

Presumably, the subject is Case-marked in both pre- and postverbal

positions. Consequently, the pattern of grammaticality found in these
Northern Italian dialects clearly shows that the occurrence of agreement
correlates with movement. If agreement occurs, the subject must move to
Spec,TP; but if there is no movement, agreement does not occur.
In certain agreement asymmetries, the asymmetry is between partial
and full agreement. A well-known asymmetry of this type is found in
Arabic: a postverbal subject triggers partial agreement of person and gen-
der as in (6a) (the verb also has the default singular agreement form),
whereas a preverbal subject triggers full agreement of person, gender,
and number as in (6b) (e.g., Bahloul and Harbert 1993, Benmamoun
1992, Fassi Fehri 1993). The following examples are taken from Bahloul
and Harbert 1993:15.
(6) a. Qadim-a (/*qadim-uu) al-awlaadu.
came-3ms came-mp the-boys-3mp
‘The boys came.’
b. Al-awlaadu qadim-uu (/*qadim-a) [t].
the-boys-3mp came-3mp came-3ms
‘The boys came.’
What we can deduce from these Arabic data is that, just as we saw with
the two Northern Italian dialects, agreement triggers movement. Unlike
in the Northern Italian dialects, where no agreement emerges if move-
ment does not take place, in Arabic, person and gender agreement
appears when there is no movement. Number agreement cannot occur if
there is no movement, but number agreement, along with person and
gender agreement, must occur if the subject moves to preverbal position.
This clearly shows that number agreement is responsible for movement.
Arabic demonstrates that, although it is correct to associate agreement
with movement, not all agreement forms are equal in this regard. In this
monograph, I abstract away from these interesting but complex issues re-
garding various types of agreement and focus by and large on the general
point that agreement, not Case, triggers movement.
Returning to the pair in (1) from Chomsky 1995, repeated here, we can
see that the ‘‘agreement’’ approach to the EPP provides an alternative ac-
count of these examples.
(7) a. There seems [TP to be a man in the garden].
b. *There seems [TP a man to be in the garden].
Given that the lower TP is nonfinite, it has no agreement; hence, there is
no reason for anything to move to the specifier of this lower T. That is
Why Agree? 5

why (7a) is grammatical; Spec,TP is not filled (the gap should in fact fol-
low to to show that there is no ‘‘EPP’’ within this TP). In contrast, in (7b)
a man has moved to this specifier—an instance of unmotivated move-
ment, hence ungrammatical. On this account, we need not specify that
merger takes precedence over movement, a desirable outcome given the
recent assumptions about these two operations. Bošković (1997, 2002)
o¤ers an alternative to this pair based on Case considerations. If we limit
our data only to English, it is di‰cult to choose between the two
approaches, but the data from agreement asymmetries in Arabic and
Northern Italian indicate that agreement is the correct option.3
One of the achievements of the Minimalist Program (MP) has been to
unify merger and movement under the general operation Merge, where
external Merge covers what used to be the domain of phrase structure
and X-bar theory, and internal Merge takes over what used to be the
domain of movement (Chomsky 2000, 2001, 2005, 2007, 2008). Both are
exactly the same operation, the only di¤erence being what is merged. Ex-
ternal Merge takes an item from the numeration, so that what is being
merged is being merged for the first time. Internal Merge takes something
that is already in the structure and remerges it. An important result of
unifying the two operations is that the theory is able to account for a
key insight from the Standard Theory era by Emonds (1976). Emonds
observed that movement operations lead to structures that are identical
to those produced by phrase structure rules, a phenomenon he called
‘‘structure preservation.’’ Given the pre-MP theory, there was no reason
why structures that result from movement should be identical to those
built by phrase structure rules. On the other hand, Merge, external or in-
ternal, predicts the structure-preserving nature of movement: movement
is simply another instance of Merge. Ideally, then, we want to avoid mak-
ing any qualitative distinction between the two types of Merge, such as a
preference for one over the other when both are possible. We can accom-
plish this by adopting the agreement approach to the EPP. (Case would
work, too, but recall the data from agreement asymmetries.) Of course,
we have more to do to make this work; but both conceptually and empir-
ically, there is ample justification for pursuing this line of reasoning. I
now turn to the question of why agreement occurs in human language.4

1.3 Why Agree?

What is the purpose of agreement? Many linguists have asked this ques-
tion, from a variety of perspectives, but nothing close to a consensus has
emerged. What is particularly striking about agreement is that, on the
6 Chapter 1

face of it, it seems to be entirely superfluous. In its most basic form, the
‘‘agreement rule,’’ let us say, targets information in one position—the in-
formation contained in a nominal such as the subject—and reproduces it
in another position, commonly as some form of an inflectional element on
a verb or some such ‘‘head.’’ In (8a), the verb inflects for singular, where-
as in (8b), the lack of overt inflection indicates plurality—in both cases
reflecting the nature, singular or plural, of the subject.
(8) a. Mary walks.
b. They walk.
This redundant nature of agreement is puzzling. Why should human
language contain a rule that represents information redundantly? There
are other domains of language where information is repeated—for in-
stance, pronominalization—but the repetition is informative.
(9) John thinks that he will win the race.
John and he are repetitive, to the extent that they refer to the same entity
in discourse, but they do not reproduce the same information: John is the
subject of think and he is the subject of win, so these two occurrences pro-
vide distinct information.
The puzzling nature of agreement goes further. In the Russian example
in (10) taken from Corbett 2006:2, not only is the singularity of the sub-
ject redundantly reproduced on the verbal inflection, but its grammatical
gender is as well, and the choice of feminine gender for a lamp is patently
arbitrary—there being nothing inherent about lamps that would make
them feminine.
(10) Lamp-a stoja-l-a vugl-u.
lamp(f)-sg stand-pst-f.sg in.corner-sg.loc
‘The lamp was standing in the corner.’
So, agreement is not only redundant, but sometimes entirely arbitrary in
its content as well.
Given these puzzling properties, what could the purpose of agreement
possibly be? Levin (2001) points to a variety of functional approaches
that appear in the literature, most of which boil down to the idea that
the redundancy helps the addressee accurately comprehend the informa-
tion by repeating it across the expression. Such a proposal faces the di‰-
culty of accounting for the wide variety of agreement systems that exist
in languages, including, most critically, lack of agreement, as in the East
Asian languages. Are East Asian languages simply nonredundant in com-
municating information relative to, say, subject-verb?
Why Agree? 7

Although there is no consensus on the outer boundaries of what consti-

tutes agreement, there is reasonable concurrence that agreement is a form
of covariance between two elements, such as the covariance between the
subject nominal and verbal inflection (Steele 1978:610). Various studies
assume this notion of covariance for agreement (e.g., ‘‘feature sharing’’
in Pesetsky and Torrego 2007). Furthermore, many of these studies de-
scribe covariance as an asymmetric relation, whereby one element, the
goal/controller, in some fashion is deemed the source of the information
for the probe/target (e.g., Anderson 1992, Chomsky 1965, 2001, Gazdar
et al. 1985, Keenan 1974, Pesetsky and Torrego 2007). There are other
approaches, such as the unification-based frameworks (Pollard and Sag
1994, Shieber 1986), that reject the asymmetric, ‘‘copying’’ approach, in-
stead positing that agreement emerges from an accumulation of infor-
mation from a variety of sources in the structure. I do not pursue the
unification approach in this monograph.
I am now ready to begin to answer the question, what precisely is the
purpose of agreement? Up to now we have seen that the formal agree-
ment system is redundant in that it comprises a covariance of two or
more elements, each expressing the same information. It is asymmetric,
in that one of the elements participating in the covariance relation pro-
vides the agreement information. Finally, the semantic content of the
agreement information is apparently not significant, and may even be ar-
bitrary, as in the case of the feminine gender on ‘lamp’ in Russian. We
find all three properties represented in the probe-goal system of Chom-
sky’s work (Chomsky 2000, 2001, 2005, 2007, 2008). A probe (the ‘‘tar-
get’’ of agreement) is an uninterpretable feature by virtue of not having a
full value for its feature; for example, it is unvalued for gender. The goal
(the ‘‘controller’’ of the agreement) provides the value, thereby account-
ing for the covariance and the asymmetric nature of agreement. Finally,
an uninterpretable feature must be deleted once it is valued so that it will
not receive semantic interpretation, a fact that directly reflects the notion
that the actual content of agreement is irrelevant. Although the probe-
goal system captures the essential properties of formal agreement, it
makes the notion of agreement all the more puzzling. Why would the
computational system insert something into the derivation of an expres-
sion only to delete it so completely that nothing remains of it for semantic
interpretation? It seems utterly counterintuitive.
The answer to the true identity of agreement, I suggest, is based on
what is sometimes referred to as ‘‘the duality of semantics’’: the well-
established distinction between lexical and functional heads. Lexical
8 Chapter 1

heads select their complements to create the argument structure of an

expression. In contrast, functional heads, which are commonly merged
higher than the basic argument structure, create an expression structure
that ‘‘consists of the modes of expression of the language’’ (Bresnan
2001:9–10). The functional layer of a clause gives rise to such notions as
topic-comment, subject of a clause, focus, and content questions, among
many other modes of expression. In other words, functional heads sub-
stantially enhance the expressiveness of human language.5
I will call the relations found in the argument structure lexical relations,
for the obvious reason that these relations are defined over a lexical head
and its argument, typically a head-complement relation. The nominal—
and the complement is almost always a nominal, although in certain cases
it is a PP or a CP—bears a particular relation to the lexical head such
that its referent is understood to be a participant in the event or the situ-
ation described by the meaning of the head. I will call the second type
functional relations, since they always involve a relation between a nomi-
nal and a functional head, such as C, T, or v. As noted above, the
purpose of functional relations is to enhance the expressive power of lan-
guage by providing the tools to express such notions as topic-comment,
subject of a clause, focus, and content questions.
We can see the independence of functional relations from lexical rela-
tions in a number of constructions. For example, in Japanese, the reflex-
ive anaphor zibun ‘self ’ is subject oriented. In the following example,
zibun can only take as its antecedent the subject Taroo:
(11) Tarooi -ga Hanakoj -o zibuni= j -no-heya-de sikat-ta.
Taro-nom Hanako-acc self-gen-room-in scold-past
‘Taro scolded Hanako in his/*her room.’
However, under direct passivization, the internal argument, Hanako, may
function as the antecedent of zibun (Kuno 1973), which shows that the
notion ‘‘subject’’ plays a crucial role independent of lexical relations.
(12) Hanakoj -ga Tarooi -ni zibun i=j -no-heya-de sikar-are-ta.
Hanako-nom Taro-by self-gen-room-in scold-pass-past
‘Hanako was scolded by Taro in her/*his room.’
The original external argument, Taroo, no longer the subject of the over-
all expression, cannot function as the antecedent of zibun.
How are the two types of relations, lexical and functional, established
in the linguistic structure? Lexical relations are thematic relations. They
are established by external Merge, in which a lexical head (or v) combines
Why Agree? 9

with its complement in a binary fashion (Chomsky 2001, 2005, 2008,

Kayne 1984). Lexical relations are therefore defined by the binary-
branching structure of sisterhood, itself created by external Merge. What
about functional relations? There is no simple structural way to establish
a relationship between, say, the external argument and T. T does not di-
rectly select the external argument, for example. (I discuss the expletive
construction, which is ostensibly a counterexample, in chapter 2.) In the
literature on this topic, a typical suggestion is that the relation that holds
between a functional head such as T and the nominal with which it agrees
(or assigns Case to) must be established by moving the nominal into
Spec,TP (Koopman 2003, 2005, Koopman and Sportiche 1991). In the
main, I believe that this intuition that agreement emerges as a specifier-
head (Spec-head) relation is correct, although there are exceptions, one
being pro-drop. Nevertheless, I will assume that agreement relations are
established independently of movement, by a process Chomsky calls
Agree (Chomsky 2000, 2001, 2005, 2008). We can thus state the purpose
of agreement as follows:
(13) Purpose of agreement
Agreement occurs to establish a functional relation.
I will capture the Koopman-Sportiche intuition that agreement requires a
Spec-head relation by showing that Agree takes place to establish func-
tional relations. Such a functional relation, which is always nonlocal,
must be transformed into a local relation by moving the goal to the
probe. The purpose of this movement is to keep a record of the func-
tional relation beyond narrow syntax so that semantic interpretation
and information structure can make use of it. This, in e¤ect, is Spec-
head agreement, but implemented as two independently motivated
operations—Agree and Move.6
Pesetsky and Torrego (2006) and Sigur¶sson (2004, 2006) indepen-
dently argue that Agree, or some form of probe-goal relation, exists for
all instances of Merge, external and internal. These proposals in one way
or another blur the distinction between lexical and functional relations.
It is quite possible that something must trigger even external Merge, as
these studies suggest. However, I will distinguish between functional and
lexical heads, as noted earlier, and presume that the kind of Agree rela-
tion I wish to explore here is relevant only to functional heads. After
all, we never see formal agreement inflection reflecting a relation between
a lexical head and its argument; such inflection is found only between
a functional head and some XP. This is the Agree relation I wish to
capture.7
10 Chapter 1

There is another proposal, very di¤erent in nature, that also blurs the
distinction between lexical and functional relations. Chomsky (2007) sug-
gests that the edge feature is responsible for Merge, both external and
internal. Note that the edge feature is independent of Agree. The edge
feature brings us back to something akin to GB in one respect: in princi-
ple, it allows free movement—free internal Merge—to any head. The
grammar simply has to ensure that the movement is motivated. Although
I will not include the edge feature in the narrow syntax, I assume with
others that movement must be justified, and that where agreement is con-
cerned, the justification is that movement brings the goal close to the
probe. In chapter 2, I discuss the notion of closeness in detail and what
precisely it accomplishes.
As a final note on why agreement occurs in human language, the pic-
ture I drew above provides a natural way to think about which categories
count as phases (Chomsky 2001). In recent minimalist work, it is thought
that computation in language, such as the numeration and Merge, occurs
within specific local domains called phases. Once the computational sys-
tem completes its work within one phase, the products of this computa-
tion are sent to PF and semantic interpretation, and the computation
then goes on with its work in the next higher phase. What are these phase
categories? Chomsky (2001, 2005, 2007, 2008) proposes that minimally
they are CP and vP. From the perspective taken here, these two catego-
ries comprise the two principal parallel structures in language: the expres-
sion structure and the argument structure. CP is the complete expression
structure, and vP is the complete argument structure. Chomsky (2001)
uses the notion of completeness as well; in the approach taken here,
phases have a highly specific and concrete underpinning—that is, the
phases comprise the two principal structures that the computational sys-
tem builds to create the expressions of a language.
I have given an explanation for why agreement occurs in human lan-
guage. I now turn to the second question about agreement: why does it
occur in some languages but not in others?

1.4 Agreement, Topic/Focus, and Strong Uniformity

1.4.1 Strong Uniformity

We saw that in subject-verb agreement languages such as English, a sub-
ject moves to Spec,TP if there is agreement inflection on T that agrees
with it. Otherwise, there is no reason for the subject to move, and it stays
in the position where it was externally merged. It was Kuroda (1988) who
Why Agree? 11

proposed that movement is forced under agreement. Further, he proposed

that agreementless languages such as Japanese do not force movement;
instead, any movement that might be observed occurs as an entirely op-
tional operation (see Saito 1989, 1992 for the view of scrambling as a
purely optional operation). I agree with Kuroda that agreement, as
opposed to Case, triggers ‘‘forced’’ movement. But does an agreementless
language such as Japanese involve no forced movement at all? I will
argue that in discourse-configurational languages, of which Japanese is
one, something else that is computationally equivalent to f-feature agree-
ment triggers forced movement.
I will argue that in discourse-configurational languages, topic/focus
establishes functional relations in the same way as f-feature agreement
in agreement languages. I hasten to add that it is not the case, for exam-
ple, that agreement languages do not also have focus, or that discourse-
configurational languages do not have f-features. In fact, we will see
that all languages have both kinds of grammatical features: f-features
and topic/focus features. Much of the monograph will address how par-
ticular languages deal with this uniform set of features. Work by Cinque
(1999) and the cartography linguistics of Rizzi (1997, 2004) and others
hint at this idea that all languages have essentially the same universal
features/structures.
The overall approach that I adopt here rests on the Uniformity Princi-
ple (Chomsky 2001:2).
(14) Uniformity Principle
In the absence of compelling evidence to the contrary, assume
languages to be uniform, with variety restricted to easily detectable
properties of utterances.
My approach is very much in line with that of Sigur¶sson (2003), who
assumes this Uniformity Principle literally for all languages and suggests
the Silence Principle, by which he means that any given language shares
the universal set of features with all other languages but does not pro-
nounce all of them. This is, at least in part, the reason for the di¤erences
among languages. Here, I will adopt an even stronger interpretation of
the Uniformity Principle and assume that, at least for grammatical fea-
tures such as agreement and focus, every language not only shares a uni-
form set of features but also (contra Sigur¶sson) overtly manifests these
features in some fashion. Although this assertion is part of the Unifor-
mity Principle—in fact, if I am right, it is a strong a‰rmation of this
principle—I will give it a name for ease of exposition and call it Strong
Uniformity.
12 Chapter 1

(15) Strong Uniformity

All languages share the same set of grammatical features, and every
language overtly manifests these features.
For example, Japanese, which shows no agreement inflection except in
rare cases such as subject honorification, nonetheless is predicted to have
f-feature agreement in some form, as I will attempt to show. In chapter 2,
I will show that Chinese, another ‘‘agreementless’’ language, in fact evi-
dences person agreement.
This strong interpretation of the Uniformity Principle cannot be right
for all features of a language. After all, languages do vary. In adopting
the strong version at least for grammatical features, I intend to explore
some of the outer bounds of the Uniformity Principle.

1.4.2 Discourse-Configurational Languages

It has long been observed that in many languages a phrase identified as
topic or focus undergoes movement. Such a language is what É. Kiss
(1995) describes as discourse-configurational.
(16) Discourse-configurational languages
a. ‘‘In a topic-prominent language, the topic is, in a way, an
alternative to the subject [in a subject-prominent language] as
the VP-external argument.’’ (É. Kiss 1995:4)
b. ‘‘Focus movement is triggered in some languages but not in
others.’’ (É. Kiss 1995:5)
É. Kiss notes that in many discourse-configurational languages, both
topic and focus are associated with movement, although there are lan-
guages where only one is. In a paper very much related to discourse con-
figurationality, Grewendorf (2005) has argued that the movements in
the German middle field that have typically been characterized as scram-
bling are nothing but topic or focus movement. Although I di¤er from
Grewendorf in not assuming specific projections that host topic and
focus, I will demonstrate that in Japanese both topic and focus trigger
movement just as in German, and that the movement is equivalent to
the movement caused by f-feature agreement—A-movement to Spec,TP
or some related ‘‘A’’ position. In this way, in discourse-configurational
languages, topic/focus has the same role as f-feature agreement: both
establish a functional relation. Here, I will briefly describe the discourse-
configurational nature of Japanese, holding a more extensive discussion
in abeyance until chapter 3.
Why Agree? 13

In Japanese, there are cases where movement is forced, but what forces
it is not f-feature agreement. Rather, these movements are triggered by
topic or focus. The term topic as I am using it refers to the entity the sen-
tence is about. It is close to, but di¤erent from, discourse topic in that, for
example, a topic need not refer to an anchored expression in the conver-
sation; it simply needs to be characterizable as ‘‘what the sentence is
about.’’ A sentence with a topic falls into the class of expressions that
Kuroda (1972–1973) calls ‘‘categorical’’ as opposed to ‘‘thetic,’’ a distinc-
tion he bases on the logical theory of Marty (1918, 1965). Japanese has a
topic construction where the topic, marked by wa, is always the discourse
topic (Kuno 1973). The topic construction that I will discuss is di¤erent:
here, something is moved and is given the property of topic in the broad
sense of ‘‘topic of the sentence.’’ I will argue that topic/focus in Japanese
constitutes a grammatical feature that is computationally equivalent to f-
feature agreement in forcing movement that results in A-chains. The idea
that focus in some languages functions as a grammatical feature that
drives movement has been suggested by a number of linguists (see, e.g.,
Brody 1990, Horvath 1981, 1986, 1995, É. Kiss 1995). Below, I will give
one example of focus in Japanese that results in A-movement; in chapter
3, I will give evidence that these focus movements in Japanese undergo A-
and not Ā-movement.
In Japanese, a wh-phrase can be interpreted as an indeterminate pro-
noun in the context of the universal quantificational particle mo. This
combination of wh-mo is a negative polarity item.
(17) Taroo-ga nani-mo kawa-nakat-ta.
Taro-nom what-mo buy-neg-past
‘Taro didn’t buy anything.’
As is well known, the wh-phrase portion and mo can be separated
(Kuroda 1965, Nishigauchi 1990).
(18) Taroo-ga nani-o kai-mo si-nakat-ta.
Taro-nom what-acc buy-mo do-neg-past
‘Taro didn’t buy anything.’
Here, the wh-phrase as an indeterminate pronoun occurs in object posi-
tion with the accusative case marker -o, and the universal quantificational
particle mo occurs on the verb stem. One distinct property of the indeter-
minate pronoun expression is that it is typically associated with focus—
meaning something like ‘absolutely nothing/no one’. I will assume that
the indeterminate pronoun is associated with the focus feature, which is
14 Chapter 1

licensed by mo; it is this focus feature that gives the indeterminate pro-
noun this ‘‘identificational focus’’ interpretation (but see note 8).
I will make use of Kishimoto’s (2001) analysis of the indeterminate
pronoun. As noted above, the indeterminate pronoun is a wh-phrase that
is interpreted as an indeterminate pronoun in the context of the universal
particle mo. Kishimoto proposes that in order for the wh-phrase to be
interpreted as an indeterminate pronoun, it must be dominated by the
same immediate maximum projection that dominates mo; that is, mo
and the indeterminate pronoun must occupy the same minimal domain.
As part of his analysis, Kishimoto argues that the verb raises to v in Jap-
anese, taking mo with it, as shown in (19).
(19)

In this structure, mo can license any indeterminate pronoun in its local

vP. In Kishimoto’s analysis, the object is assumed to move to Spec,vP.
This is why it is fine to have an object indeterminate pronoun like the
one in (18). As a piece of evidence for his analysis, Kishimoto observes
that an indeterminate pronoun cannot occur in subject position.
(20) *Dare-ga piza-o tabe-mo si-nakat-ta.
who-nom pizza-acc eat-mo do-neg-past
‘Anyone didn’t eat pizza.’
Kishimoto assumes the EPP here and argues that the subject indetermi-
nate pronoun dare ‘who’ raises to Spec,TP to satisfy the EPP requirement
of T; this movement takes it outside the scope of mo, which is on v. This,
then, is a case of forced movement that takes a phrase to Spec,TP. In-
stead of f-feature agreement, what is operative here is focus.8
There is further evidence for this analysis beyond Kishimoto’s data.
First, recall from (18), repeated here, that the indeterminate pronoun is
fine in object position.
(21) Taroo-ga nani-o kai-mo si-nakat-ta.
Taro-nom what-acc buy-mo do-neg-past
‘Taro didn’t buy anything.’
Why Agree? 15

Now observe what happens if we scramble the object to the head of the
sentence.
(22) *Nani-oi Taroo-ga ti kai-mo si-nakat-ta.
what-acc Taro-nom buy-mo do-neg-past
‘Taro didn’t buy anything.’
As shown, if the object indeterminate pronoun is scrambled to the left
of the subject, the sentence becomes ungrammatical. What does this fact
indicate? Kishimoto’s analysis does not predict this ungrammaticality.
The problem is that this kind of scrambling may be A-movement, which
is what we are attempting to analyze, but it may also be Ā-scrambling.
Ā-scrambling allows reconstruction (Mahajan 1990, Saito 1992, Tada
1993), so that in (22), the moved object indeterminate pronoun nani
should in principle be interpretable in its original complement position.
This should lead to a grammatical sentence. We can see the Ā-movement
possibility of local scrambling in (23), where an anaphor has been moved
to the head of the sentence.
(23) Zibun-zisin-oi Taroo-ga ti hihansi-ta.
self-acc Taro-nom criticize-past
‘Self, Taro criticized.’
The fact that the indeterminate pronoun in (22) cannot be so recon-
structed indicates that it has undergone A-movement, which normally
does not reconstruct. In Miyagawa 2001, 2003 (see also Hasegawa 2005,
Kitahara 2002), I argued that the landing site of this kind of A-movement
is Spec,TP. This is what we predict if focus in discourse-configurational
languages like Japanese functions as a grammatical feature that triggers
A-movement. In chapter 3, I will give evidence for the ‘‘A’’ nature of
this movement. There, I will also revise the view that the movement al-
ways takes place to Spec,TP; I will suggest instead that it can sometimes
move to an A-position above the TP, which I will call aP.
Let us return now to the agreement–topic/focus parameter. An imme-
diate problem arises with the idea of such a parameter. Take focus, for
example. Focus and agreement are usually thought to be located on fun-
damentally di¤erent heads. Focus is commonly postulated to occur on
the focus head that is higher than T and in the region of C (e.g., Culicover
and Rochemont 1983, Rizzi 1997), or, in languages such as Hungarian
and Turkish, possibly lower (see É. Kiss 1995 for discussion of various
approaches). In contrast, agreement in (for example) subject-verb agree-
ment is normally construed as being located on T. Although it is not
entirely implausible for two features on fundamentally di¤erent heads to
16 Chapter 1

vary parametrically, the idea of an agreement–topic/focus parameter

would be more plausible if focus and agreement were not found on such
vastly di¤erent heads. There is su‰cient evidence to associate focus with
a head higher than T, so if we are to do anything about ‘‘head parity,’’ we
need to look at agreement. To get right to the point, I suggest, following
Chomsky (2007, 2008), that agreement in (for example) subject-verb
agreement is associated with a head higher than T—namely, with C (see
Boeckx 2003, Carstens 2003, Kornfilt 2000, 2004 for a similar idea).
There are conceptual and empirical reasons for assuming this. Conceptu-
ally, merging the agreement feature on C means that grammatical fea-
tures that are responsible for computations such as movement show up
solely on phase heads—C, v, and possibly D, although I will limit my dis-
cussion largely to C. Given that any operation beyond initial Merge takes
place within phases, it makes sense that the elements triggering these
operations are merged on phase heads, f-feature agreement being one
such element.
There is also empirical evidence for assuming agreement to merge on
C. First, in English, environments where agreement (and Case) is not
assigned, such as the ECM (exceptional-Case-marking) construction, in-
volve a ‘‘bare’’ TP that does not have a CP (Chomsky 2005, 2008). A
simple way to view this is that C provides the agreement, and in its
absence, T by itself cannot bear agreement (or Case). A second piece
of empirical evidence is that agreement actually shows up on C in some
languages. For example, Carstens (2003:393) notes the following West
Flemish examples (from Haegeman 1992):
(24) a. Kpeinzen dan-k (ik) morgen goan.
I.think that-I (I) tomorrow go
‘I think that I’ll go tomorrow.’
b. Kpeinzen da-j (gie) morgen goat.
I.think that-you (you) tomorrow go
‘I think that you’ll go tomorrow.’
c. Kvinden dan die boeken te diere zyn.
I.find that.pl the books too expensive are
‘I find those books too expensive.’
Although a number of linguists have proposed that the complementizer-
subject agreement is an instance of the agreement on T raising to C
(Hoekstra and Marácz 1989, Watanabe 2000, Zwart 1993, 1997), Cars-
tens argues that the agreement originates on C (see Carstens 2003 for ad-
ditional references for and against this idea). One piece of evidence is that
Why Agree? 17

the complementizer must be adjacent to the subject it agrees with. In

(24a–c), the embedded verb also inflects for agreement, which suggests
that the agreement also shows up on T. Under my analysis, which closely
follows Chomsky’s (2007, 2008) proposal, this suggests that the agree-
ment on C may percolate down from C to T, a point I discuss later.
The picture that emerges is that both topic/focus and agreement ini-
tially occur on C, as shown in (25). (I use the generic term f-probe for
the uninterpretable f-feature agreement at C.)
(25) Uniform set of features

Technically, what is merged on C is an uninterpretable feature. For

agreement, it is an uninterpretable agreement feature, the f-probe, that
must be valued by the interpretable feature on the goal (Chomsky 2000,
Pesetsky and Torrego 2006). For focus, what is the nature of the uninter-
pretable feature? Expanding on a line of investigation in Holmberg and
Nikanne 2002, I propose that there is just one feature for topic/focus,
and it is the ‘‘default’’ topic feature, focus. If this feature enters into
agreement with a focused element, it is turned into a þfocus feature.
Details will be given in chapters 3 and 4.
The topic/focus feature is matched with a relevant phrase in the
structure—for example, the thematic subject. In most cases, this category
is raised to Spec,TP because that is the head the probe ultimately ends up
on by inheritance from C. One ostensible exception is long-distance
agreement, where the goal appears to occur lower than the head with the
f-probe. I will take up the issue of long-distance agreement in chapter 2
and show that for many cases, long-distance agreement in fact is not an
exception to the need for agreement to form a Spec-head relation. An-
other exception to the idea that the f-probe and its goal occur in the
Spec,TP region comes from Bantu: we will see instances where the f-
probe occurs on a projection higher than TP, which I call aP, the same
projection I will posit for some of the topic/focus movements in Japanese
(for detailed discussion, see chapter 3).
18 Chapter 1

By initially placing agreement as well as topic/focus on a ‘‘high’’ head

in the C region, we make it plausible for these two features to be the two
polarities of the same parameter. Because agreement and topic/focus con-
stitute the primary grammatical features in the proposed system, we also
thereby isolate all such features on a phase head, C. This is a desirable
outcome: agreement and topic/focus are two major elements of computa-
tion in narrow syntax, and the idea of isolating them initially on C means
simply that we identify the major elements of computation with phase
heads—C, v, and possibly D. Although I will not have much to say about
v and D, I will explore this view in detail with regard to C. Indeed, if
topic/focus and agreement are essentially two sides of the same coin as
far as computation is concerned, we gain a conceptual argument that
agreement should be associated with C. There is su‰cient evidence that
topic/focus is associated with the C domain (see, e.g., Culicover and
Rochemont 1983 for focus). If agreement has the same formal function
as topic/focus—to trigger movement—then it would be conceptually
plausible to locate agreement at C as well.
But do topic/focus and agreement constitute a natural class indepen-
dent of syntactic movement? If so, there is a plausibility argument to add
to our empirical and theoretical arguments. Historical analysis suggests
a relation between topic/focus and agreement. It is widely assumed that
subject agreement morphology historically develops from subject pro-
nouns (e.g., Givón 1976). Givón proposes that the process by which this
happens relates to topicalization. When something is topicalized, it typi-
cally leaves a pronoun in its original position. The idea is that this pro-
noun in the topic construction gets reanalyzed and becomes part of the
verbal morphology. Alternatively, Simpson and Wu (2001) suggest that
the subject pronoun’s reanalysis as agreement morphology has to do
with its being associated with focus. On either account, there is a clear
link historically between agreement and topic/focus, which lends further
credence to my claim that these features are computationally equivalent.
A major point I will illustrate is that in a discourse-configurational lan-
guage, topic/focus, which occurs on C, ultimately shows up on a lower
node such as T, triggering A-movement to this lower node. In certain
cases, the head that ultimately hosts topic/focus may be a head higher
than T and lower than C, as I demonstrate in chapter 3, but for the time
being I will keep the picture simple and assume that in these languages
topic/focus appears on T. This means that the topic/focus feature that
starts out on C is inherited by T, as shown in (26).
Why Agree? 19

(26) The topic/focus feature is inherited by T in a discourse-configurational

language

This agreement–topic/focus parameter boils down to whether the topic/

focus feature is inherited by T. If it is, the structure will represent a dis-
course-configurational language as described by É. Kiss (1995, etc.) and
others. I will discuss this inheritance of topic/focus by T in detail in chap-
ters 3 and 4.
What about the f-probe? In earlier work (Miyagawa 2005b), I assumed
that the f-probe also has the option of staying on C or being inherited by
T. This assumption was based on an analysis of Kinande as described by
Baker (2003). However, in chapter 4, I will present an alternative analysis
of the Kinande facts that shows the f-probe being inherited by a lower
head other than T. What this means is that the f-probe, if it occurs, is ap-
parently always inherited by a lower head regardless of whether the topic/
focus feature is also inherited (see Chomsky 2007, 2008), as shown in
(27).
(27) The f-probe is always inherited by a lower head such as T

What is the reason for this inheritance? There are two questions here:
why inheritance occurs at all, and why f-probes are always inherited by
a lower head. Concerning the first, I will assume the reason Chomsky
gave when he originally proposed feature inheritance (Chomsky 2005,
2008): it enables languages to have A-chains. Without inheritance by T,
20 Chapter 1

all movement would be Ā-movement; that is, all movement would be

operator movement. Language would be deprived of movement for pur-
poses of informational structure, such as creating subject-of relations,
topic-comment relations, and so forth. From the perspective of the
agreement–topic/focus parameter, this means that in some languages,
such as English, A-chains are created on the basis of f-probe inheritance,
but in the discourse-configurational languages, they are created on the
basis of topic/focus.9 A central question of this monograph is, then,
what happens in constructions in a discourse-configurational language
where both the f-probe and topic/focus appear? I take up this question
particularly in chapter 4 when I look at Kinande and Kilega (both Bantu
languages) and at Finnish.
Let us turn to the second question: why f-probes are always inherited
by a lower head. If the f-probe is always inherited in this way, an objec-
tion that one might raise to the overall approach outlined so far is, why
not simply merge the f-probe at T to begin with? That was, in fact, the
assumption prior to the recent works by Chomsky and others. We have
seen conceptual and empirical arguments that favor merging the f-probe
at C. When we look at Bantu and Finnish, we will see cases where the f-
probe is not inherited by T; rather, it is inherited by a head higher than T
but lower than C, which I call aP.
(28)

This type of projection between TP and CP, illustrated in (28), has been
proposed for a variety of languages, including Bantu (Baker 2003), Hun-
garian (É. Kiss 1995), and Romance (Uriagereka 1995). Particularly in
Why Agree? 21

the case of Bantu, the f-probe at a gives rise to interactions between the
f-probe and movement that di¤er sharply from the familiar Indo-
European situation where the f-probe typically picks out the grammati-
cal subject and it is this subject that moves to Spec,TP. Being inherited
by a higher head, the f-probe in Bantu is able to pick out any DP in its
search domain, so that what raises may be the subject, the object, or the
locative (the locative in the Bantu languages we will look at is apparently
a DP), and the raised phrase—the subject, the object, or the locative—is
what enters into agreement with the f-probe. (The following examples
are from Baker 2003:113.)
(29) a. Omukali mo-a-seny-ire olukwi (lw’-omo-mbasa). (SVO)
woman.1 aff-1.s/t-chop-ext wood.11 lk11-loc.18-axe.9
‘The woman chopped wood (with an axe).’
b. Olukwi si-lu-li-seny-a bakali (omo-mbasa). (OVS)
wood.11 neg-11.s-pres-chop-fv women.2 loc.18-axe.9
‘Women do not chop wood (with an axe).’
c. ?Omo-mulongo mw-a-hik-a omukali. (LocVS)
loc.18-village.3 18.s-t-arrive-fv woman
‘At the village arrived a woman.’
In (29a), the verb agrees with the subject ‘woman’; in (29b), it agrees with
the raised object ‘wood’; and in (29c), it agrees with the raised locative ‘at
the village’. We have to assume that for the object or the locative to be in
the local search domain of the f-probe, it must occur at the edge of the
vP (Carstens 2003). In chapter 4, I will draw on Baker’s (2003) work to
show that the agreed-with phrase (subject, object, or locative) occurs
higher than Spec,TP.10
Why is the f-probe that occurs at a in Bantu free to pick out any DP in
its search domain, while a f-probe at T is limited to the grammatical sub-
ject? One possibility is that in ‘‘subject agreement’’ languages such as En-
glish, there is no reason for a DP within vP—typically the object—to
move to Spec,vP, so that such a DP can never find itself in the local do-
main of the f-probe in the higher phase. On this account, the fact that the
f-probe at T only picks out the grammatical subject is coincidental in the
sense that it is only the grammatical subject that appears in its search do-
main. We can discount this approach by looking again at Bantu. We will
see that in certain constructions, the aP is disallowed, so that the f-probe
is forced to be inherited by T. When this happens, a very di¤erent pattern
emerges, exactly like the pattern in the familiar Indo-European lan-
guages: the f-probe must pick out the grammatical subject. Since we
22 Chapter 1

know that the Bantu languages allow the object or the locative to enter
into the search domain of the f-probe at a, it cannot be the case that
when the f-probe occurs at T, the object and the locative are blocked
from moving to Spec,vP to be visible to the f-probe.
A more promising idea is to view f-probes as incapable of identifying a
goal by themselves. A goal must somehow be ‘‘activated’’ to be visible to
a f-probe, and the mechanism that typically activates it is Case (Chom-
sky 2001). Let us assume the traditional view (somewhat di¤erent from
the view in Chomsky 2001) that T assigns nominative Case; in minimalist
parlance, T values the Case on its target, which is the grammatical subject
(e.g., Pesetsky and Torrego 2001). The f-feature at T then picks out the
target of this Case assignment—the nominative subject—as its goal. This
is why the f-probe at T always picks out the grammatical subject.11
What about the Bantu case? The f-probe is inherited not by T but by
the higher head a, which is not a Case assigner. In chapter 4, I show
how this f-probe on a together with the topic feature, which is also inher-
ited by a, accounts for the several possible goals the probe can seek—
subject, object, or locative. As I also show, this analysis can explain
Baker’s (2003) polysynthesis parameter without stipulating that Case
does not play any role in Bantu.
The reason, then, that the f-probe is inherited by a lower head is that
it must seek a way to find its goal, being unable to do so by itself. The
situation is di¤erent for focus, for example. Focus is usually marked in
discourse-configurational languages—in Japanese, for example, it is mor-
phologically marked, as we will see in chapter 3 and especially in chapter
5. There is no need to activate the goal, and therefore the probe is able to
pick out its goal without depending on some activation mechanism. What
about topic? We will see in chapter 3 that topic is fundamentally di¤erent
from focus and also from the f-probe in that it does not seek a goal in
the sense of a probe-goal relation. It is similar to focus, though, in that
the probe responsible for topicalization does not require activation of the
goal.
Although I will not take up Case in this monograph, it does play an
important role particularly in tandem with the f-probe. Case’s role in
activation is prominently discussed in recent minimalist literature (e.g.,
Chomsky 2000, 2001). I assume that it has other roles to play as well,
one of which is to make a nominal phrase visible for y-marking, an as-
sumption from GB (Chomsky 1981). This is perhaps its main role. After
all, while we find inherent Case, which is Case that comes with a par-
Why Agree? 23

ticular y-role, we never find inherent agreement. The so-called default

agreement in antiagreement is a lack of agreement, not some inherent
agreement. On this view, Case is primarily an entity in the domain of lex-
ical relations, although it is typically assigned by a functional head such
as T, whereas agreement is an entity in the domain of functional relations.
Where the two come together is in those situations where a f-probe, in
establishing a functional relation, takes advantage of the ability of Case
to make a nominal visible for y-marking.
The above discussion makes clear when Agree takes place for the f-
probe and focus (regarding topic, see chapter 3). Whereas both the f-
probe and focus begin at C, the f-probe does not enter into an Agree
relation until it is inherited by a lower head that has an activation mech-
anism, most commonly Case. At this point, Agree takes place and a goal
for the f-probe is identified. In contrast, the focus feature requires no
activation, so it enters into an Agree relation when it is at C. Notice that
this raises an issue for complementizer agreement in West Flemish. How
can the f-probe at C be valued? I return to this problem in chapter 2.
The strong version of the Uniformity Principle that guides this work
predicts that a language like Japanese, which is considered to lack f-
feature agreement, should manifest it in some fashion. Let us look next
at this prediction.

1.4.3 Evidence for Person Agreement in Japanese

The strong interpretation of the Uniformity Principle assumes that all
languages share the same essential components. This means that all lan-
guages share the features we are looking at, f-features and topic/focus,
which are both merged initially on a phase head. Recall that Strong Uni-
formity does not state that these two features must always be present.
Rather, every language has both, and we should be able to find them in
some given construction in every language.
Although Japanese does not show the typical subject-verb agreement
found in many languages, it does exhibit person agreement involving ele-
ments that occur higher than T, in what Inoue (2006) calls D(iscourse)-
modals. These modals, which arguably occur in the C domain, express
some sort of attitude on the part of the speaker toward the utterance
and also typically the hearer. Many traditional Japanese grammarians
have examined issues of modality in the language and have found that
these modalities often impose a limitation on the kind of subject that is
allowed—person agreement, in other words. Examples (30)–(34) are
24 Chapter 1

from Ueda (2006:168–169, 174), who based them on examples in Nitta

1991; see also Tenny 2006 for relevant discussion.
(30) Exhortative [first person, *second person, *third person]
{Watasi/*Anata/*Yamada-sensei}-ga Taroo-ni tegami-o
I/*you/*Prof. Yamada-nom Taro-dat letter-acc
okuri-MASYOO.
send-let’s
‘Let’s (have) me/*you/*Prof. Yamada send Taro a letter.’
(31) oku [first person, *second person, *third person]
{Watasi/*Anata/*Yamada-sensei}-wa Taroo-ni tegami-o okutte
I/*you/*Prof. Yamada-top Taro-dat letter-acc send
OKU.
oku
‘I/*You/*Prof. Yamada will send a letter to Taro.’
(32) Prohibition [*first person, second person, *third person]
{*Watasi/Anata/*Yamada-sensei}-wa Taroo-ni tegami-o
*I/you/*Prof. Yamada-top Taro-dat letter-acc
okuru-NA.
send-don’t
‘Don’t *I/you/*Prof. Yamada send Taro a letter.’
(33) Negative supposition [first person, *second person, third person]
{Boku/*Kimi/Kare}-wa iku-MAI.
I/*you/he-top go-probably.not
‘I/*You/He probably won’t go.’
(34) Assertion (see also Inoue 2006) [first person, *second person, third
person]
{Watasi/*Anata/Yamada-sensei}-wa Taroo-ni tegami-o
I/*you/Prof. Yamada-top Taro-dat letter-acc
okut-TA.
send-past.assert
‘Asserted: I/*You/Prof. Yamada sent a letter to Taro.’
Exhortative (see (30)) and oku (see (31)) only allow the first person; pro-
hibition (see (32)) only allows the second person; negative supposition
(see (33)) and assertion (see (34)) only allow first and third persons. There
may be other such modalities. For example, Ueda (2006:174) lists ques-
tions as only allowing second and third person, but that may simply be a
matter of the meaning—normally, a speaker does not ask about an action
Why Agree? 25

or event whose subject is the speaker himself or herself, though in the

right context, it should be possible.
Another example of f-feature agreement in a discourse-configurational
language is provided by the so-called force markers in Korean (examples
from Pak 2006:295–296).
(35) a. Imperative
Cemsim-ul mek-e-la!
lunch-acc eat-imp
‘Eat lunch!’
b. Exhortative
Icey kongpwuha-ca.
now study-exh
‘Now, let’s study.’
c. Promissive
Nayil nay-ka cemsim-ul sa-ma.
tomorrow I-nom lunch-acc buy-prm
‘I will buy lunch tomorrow.’
Although earlier studies of these force markers treat them as distinct con-
struction types, Pak notes that the markers share several syntactic traits:
in embedding, none allows an overt subject; they do not allow mood par-
ticles; they allow a special negative marker, -mal; there is no tense mark-
ing; and they can be conjoined by -ko, which requires conjuncts to be
the same type of clause. The following examples demonstrate the force
markers’ ability to occur with -mal.
(36) a. Imperative
Mek-ci mal-a-la.
eat-nom neg-a-imp
‘Do not eat.’
b. Exhortative
Mek-ci mal-ca.
eat-nom neg-exh
‘Let’s not eat.’
c. Promissive
Mek-ci mal-u-ma.
eat-nom neg-u-prm
‘I promise not to eat.’
Pak concludes that these particles all represent the same type of construc-
tion and, specifically, that they all involve agreement at C.
26 Chapter 1

There are two further points to be made about these particles. First,
as Pak notes, the restrictions found in person with these markers are
imposed on the subject, and they appear best specified as Gspeaker,
Gaddressee. This makes sense because at the C level, which is the inter-
face with the universe of discourse, the two participants are the speaker
and the addressee. Pak proposes the following characterizations:
(37) a. Imperative: speaker, þaddressee
b. Exhortative: þspeaker, þaddressee
c. Promissive: þspeaker, addressee
The imperative marker -la indicates that the subject is the addressee, the
exhortative -ca indicates that the subject is the speaker and the addressee
together (expressed using the first person inclusive of the addressee), and
the promissive -ma indicates that the subject is the speaker. The same
approach applies by and large to the person restrictions observed earlier
for Japanese.
The second point has to do with how valuation of the f-probe pro-
ceeds. All of the Japanese and Korean examples of person agreement at
C involve some element, typically associated with modality, such as the
exhortative -masyoo and negative supposition -mai in Japanese and the
force markers in Korean. It is the presence of one of these elements that
imposes the person agreement/restriction. Crucially, it is not the subject
that values the f-probe at C, as is commonly the case in agreement lan-
guages. Later I will speculate on why the subject does not value the f-
probe in Japanese, but right now let us work through how valuation takes
place. First, it is important to note that the kind of person restriction
found in Japanese does not always occur; it emerges only when one of
the relevant modal elements appears in the construction. If there is no
such modal, there is no person restriction; hence, there is no f-feature
agreement. This is still consistent with the Strong Uniformity interpreta-
tion of the Uniformity Principle: it simply shows that in Japanese, f-
feature agreement does occur, but not in every clause.
When the f-probe does occur, I suggest that it is inherited by the
modal head, as shown in (38), and the modal head, which contains the
interpretable person feature, values it at that point. This suggestion is
based on the intuition that it is the modal that imposes the person
agreement.
Why Agree? 27

(38)

I am supposing here that the f-feature, by attaching directly to its goal,

is able to undergo valuation because it does not have to seek the goal
through some activation mechanism. Now the f-feature has a value; let
us say it has been valued as first person by the goal Mod. The valued f-
feature is now almost like a clitic in that it has an agreement value, and it
needs to be able to impose this restriction. But given that it is a f-probe,
it cannot assert its value by itself; instead, it has to depend on some other
mechanism. Case is the natural candidate, so the f-probe is inherited by
T, where the person restriction on the subject is imposed through the
nominative Case assigned by T. This is shown in (39).
(39)

One final question regarding person agreement in Japanese and Korean

is, why does it work the way it does? Unlike the typical f-feature agree-
ment, where a f-probe seeks its goal inside TP, in Japanese and Korean a
modal is the goal that values the f-probe. I speculate that this is because
nominals in Japanese and Korean are not associated with the appropriate
28 Chapter 1

interpretable person features to value a f-probe. The only other possibil-

ity is Case, and the nominal phrase that is picked out is the ‘‘subject,’’
similarly to what happens in languages like English. Why does the Japa-
nese and Korean nominal lack f-features? One possible answer could
come from Chierchia’s (1998) nominal parameter. Chierchia argues that
nominals are parameterized for whether they can refer to individuals
(e.g., English) or just to kinds (e.g., Japanese). We might speculate that
nominals that can only refer to kinds do not have the appropriate content
for carrying interpretable f-features. We will see in chapter 2 that Chi-
nese, the language that Chierchia uses to demonstrate the ‘‘kind’’ type
of nominal system, turns out to have a robust person agreement similar
to the ones found in Indo-European. But this will not negate the idea
that the ‘‘kind’’ nominal cannot carry f-feature agreement. As we will
see, the actual goal of the f-probe in Chinese is an empty agreement
head, similar to those studied by Alexiadou and Anagnostopoulou (1998)
in Greek. Even in Chinese, the fully specified nominal does not appear ca-
pable of valuing the f-probe. I will leave this issue open.

1.5 Summary

In this chapter, I posed two questions about agreement: why does agree-
ment occur, and why do some languages have it while others do not? I
suggested that agreement establishes what I call a functional relation be-
tween a functional head and an XP. Unlike lexical relations, which are
strictly local, any relation between a functional head and an XP must
be established by some rule that can relate two points, often two distant
points, in the structure. Agreement is tapped for this purpose in lan-
guages that have f-feature agreement. The relations that are established
between functional heads and XPs are critical: they substantially enhance
the expressive power of human language by making it possible to express
such notions as topic-comment, subject of a clause, focus, and content
questions.
To explain why agreement occurs in some languages but not others,
I first proposed that all languages have a uniform set of features that
includes f-features and topic/focus and that all languages should display
overt evidence of both. I called this proposal Strong Uniformity, reflect-
ing the idea that it is a stronger version of Chomsky’s (2001) Uniformity
Principle. In some cases, such as person agreement in Japanese and Ko-
rean, f-feature agreement does not occur in every sentence; instead, it
arises in constructions involving a variety of modals. Moreover, I argued
Why Agree? 29

that topic/focus in discourse-configurational languages such as Japa-

nese has a function equivalent to f-feature agreement in agreement lan-
guages in triggering A-movement. So, the di¤erence between agreement
languages and discourse-configurational languages boils down to what
triggers movement at T: the f-probe or topic/focus. I suggested that
topic/focus and f-feature agreement are both merged on a phase head
(C being the primary phase head that I dealt with) and that the f-probe,
if it occurs, is inherited by T or some related head.
The di¤erence between agreement and discourse-configurational lan-
guages boils down to whether or not the topic/focus feature is also
inherited by a lower head such as T. If it is inherited, we have a
discourse-configurational language, but if not, we have an agreement
language. Of course, agreement also occurs in discourse-configurational
languages, and, as we will see, how it functions di¤ers from one
discourse-configurational language to the next. This issue of variability
among discourse-configurational languages will be a focus of the discus-
sion in chapters 2–4.
2 Why Move?

2.1 Introduction

Chapter 1 dealt with the question, why does human language have agree-
ment? My answer was that agreement, or its discourse-configurational
counterpart, topic/focus, establishes functional relations—that is, rela-
tions between a functional head and an XP. Functional relations occur
at a distance because the functional head must ‘‘reach down’’ into the ar-
gument structure layer of language. Agreement and topic/focus, which
connect two points in the syntactic structure, are designed specifically to
create these links that occur at a distance. The ultimate purpose of agree-
ment (and topic/focus) is to imbue language with expressive power that it
could not otherwise have.
In this chapter, we turn to the second big question—why does human
language have movement? I will limit the discussion to movement opera-
tions that are (1) closely tied to agreement of some sort and (2) forced
(i.e., obligatory). I will exclude movements characterized as optional; for
this type of movement, see Fox 2000 and Miyagawa 2005a, 2006, among
others.

2.2 Movement Is a Record of Functional Relations

In chapter 1, we saw that the presence of an agreeing element, which

includes f-features and topic/focus, leads to movement. Specifically, the
goal of the agreement moves to the specifier of the head that has the un-
interpretable agreeing feature—what I have called the f-probe. For f-
feature agreement, the uninterpretable nature of the probe means that
this f-probe is a feature that has not been fully specified. For example,
an uninterpretable feature for number may simply be anumber, and
a must ultimately be valued as singular or plural in languages with
32 Chapter 2

singular/plural number agreement, such as English. In recent theory (e.g.,

Chomsky 2001, 2005, 2007, 2008), this valuation takes place when the
uninterpretable feature finds its goal (the controller of the agreement),
which contains a fully specified, hence interpretable, version of the
feature—for example, singular. This is the operation Agree (Chomsky
2001). We saw in chapter 1 that the Agree relation for the f-probe is
established after the f-probe is inherited by T. Under Agree, which links
the probe with the goal, the goal values the uninterpretable feature of the
probe, so that anumber takes the value singular. We might come up with
a more refined formalism, but the point is clear. For topic/focus, I will lay
out a similar system of valuation in chapter 3.
According to Chomsky (2001, 2005, 2007, 2008), valuation occurs
under Agree, so that as soon as the probe enters into an Agree relation
with its goal, valuation ensues. Under a di¤erent view (Bobaljik 2006,
Pesetsky and Torrego 2006, Sigur¶sson 2004, 2006), Agree is kept sepa-
rate from the process of valuation; Agree is a purely syntactic process by
which two positions in the syntactic structure are identified as being in
a relation. A proposal that extends separation of Agree and valuation is
to view valuation as taking place in the morphological component, a part
of PF (for the proposal that morphology is part of PF, see Halle and
Marantz 1993). This is essentially the position taken by Bobaljik (2006)
and Sigur¶sson (2004, 2006), who suggest that agreement is a matter of
morphology.
I will stay with the idea that Agree and valuation are part of the same
operation, where Agree establishes the probe-goal relation, and the goal
then values the probe. This is the simplest assumption to make, and it
enables us to address the central concern at hand—namely, the nature
of movement. This way of looking at Agree parallels, for example, y-
marking, the only di¤erence being that y-marking is always strictly local,
so there is no need for an operation like Agree to identify the target of the
y-marking operation. It is important to point out that, whereas valuation
takes place in narrow syntax as part of Agree, the actual phonological
features for the agreement values presumably are assigned in the mor-
phological component, which is assumed to be part of PF (Halle and
Marantz 1993). There is at least one possible exception to the idea that
valuation takes place in narrow syntax. Later in this chapter, when we
return briefly to the West Flemish data on complementizer agreement, I
will speculate that the complementizer portion of the agreement receives
its valuation not in narrow syntax but in PF. But this is an isolated case;
the more typical instances of valuation take place in narrow syntax as
part of Agree. We will see that, despite this di¤erence in valuation,
Why Move? 33

the complementizer agreement data provide evidence that f-probes begin

at C.
Recall the way that agreement is invoked by the probe-goal pair. The
f-probe is an uninterpretable feature that must be valued by the interpret-
able feature on the goal. Recall, too, that the purpose of agreement is in-
dependent of the content of agreement—it is to establish a functional
relation. Now let us consider what happens to the f-probe in the deriva-
tion. Before it reaches semantic interpretation, it must be erased (Chom-
sky 2001, 2005, 2007, 2008). The reason is simple. It makes no sense for
semantics to interpret a f-probe on a head, even if it is valued by the
goal. To give one example, it makes no sense for semantics to interpret
þfeminine on T. Deleting the f-probe scrubs the string clean of material
superfluous to semantic interpretation. But now we have a problem. The
probe-goal relation is established to create a functional relation between a
functional head and an XP, but if the f-probe is erased before the string
is transferred to semantic interpretation, there will be no record of the
functional relation that has been created. Semantic and information-
structure interpretation will never ‘‘know’’ that a functional relation was
ever established between a functional head and an XP. Functional rela-
tions are established across a distance, to bridge the functional layer and
the argument structure layer, and if the f-probe is erased, this link will be
lost. This is where movement comes in. By moving the goal to the probe,
human language has figured out a way to keep a record of functional
relations for semantic and information-structure interpretation, in the
absence of f-probes. This, I believe, is the true reason for movement, al-
though the way it is actually implemented by the probe-goal mechanism
does not readily reveal this reason.
(1) Why move?
Movement triggered by agreement takes place in order to keep a
record of functional relations for semantic and information-structure
interpretation. Given the architecture of human language, movement
is the only way to preserve functional relations beyond the interface
to interpretation of semantics and information structure.
In other words, language ensures that functional relations, which sig-
nificantly enhance expressive power, become part of the semantic and
information-structure interpretation of the sentence by forcing move-
ment. Of course, within narrow syntax, operations of agreement and
movement take place as formal computation, and they are not motivated
by considerations of meaning, expressiveness, and related matters. But
the e¤ect they have is to make the output of narrow syntax e¤ectively
34 Chapter 2

usable as an entity imbued with enormous expressiveness. In the next sec-

tion, I provide the technical details of what it means for the goal to move
to the probe, but the intuition is clear: movement takes place to retain
a record of the functional relations created initially by the operation of
Agree. The diagram in (2) illustrates what I have just described (see also
Chomsky 2007).
(2) Narrow syntax: The goal is moved to the probe
ƒƒ ƒ
ƒ ƒƒ ƒ
ƒ ƒƒ
ƒƒ
Probe given Probe erased, but the record of
ƒƒ
phonological ƒƒ the functional relation is retained
ƒ
ƒ ƒƒ
form ƒƒ
ƒ ƒƒ thanks to the movement
ƒƒ
ƒ
ƒƒ
ƒƒ ƒ
PF Semantic interpretation
The probe can survive in PF, where it is given phonological form (in most
cases). It does not receive semantic interpretation because it has been
erased, but by this time it has done its work of establishing a functional
relation and forcing movement to retain a record of that relation. In this
way, we can account for the intuition developed in earlier studies (e.g.,
Koopman 2000, 2003 and earlier works) that agreement always involves
a Spec-head relation. In discussing movement, I use the term specifier in
the traditional sense, although in the current system (bare phrase struc-
ture; see below) there is no ‘‘specifier’’ per se. I use the term strictly for
expository purposes, just as I use traces (t) and bar levels (X, X 0 , XP),
other carryovers from earlier practice.

2.3 How Movement Retains a Record of Functional Relations

To see how movement retains a record of functional relations, I turn to

the concepts in the bare phrase structure approach to structure building
(Chomsky 1995:241–249). In earlier work, a phrase such as the book is
characterized as having the structure in (3).
(3)
Why Move? 35

In the bare phrase structure approach, bar levels are dispensed with (see
Muysken 1982), and only the minimal and maximal nodes (D and DP
above) are visible at the interface (Chomsky 1995:242). Furthermore, the
label for the maximal projection is defined to be identical to its head. The
phrase the book would therefore have the informal characterization in (4),
informal because an item such as the is technically a bundle of features.
(4)

Merge builds structures by pairing two items, fa, bg, one of which proj-
ects to give the new item its label, either a or b.
Suppose that a f-probe a, an uninterpretable feature, occurs on a head,
X, which has the goal b. The diagram in (5) illustrates the structure prior
to the movement of the goal b.
(5)

The head X projects as the label for the constituent that contains X and
ZP. The label is the same as the head. Therefore, when the goal b moves
to the ‘‘specifier’’ of X, it merges with the label of the head that contains
the probe.
(6)

This movement has the e¤ect of transforming the nonlocal relation estab-
lished by Agree into a strictly local relation created by Merge, as stated
in (7).
(7) Probe-goal union (PGU)
A goal moves in order to unite with a probe.
36 Chapter 2

With this local structure in place, even when the f-probe is erased prior to
semantic interpretation, the functional relation established between X and
b is unmistakably retained for use in interpretation.
In e¤ect, the goal moves to the specifier of the head that contains the
probe. This explains the intuition that agreement requires a Spec-head re-
lation (e.g., Koopman 2000, 2003). The idea isn’t that agreement must be
represented in a Spec-head relation; rather, the Spec-head relation makes
it possible to attain PGU, which in turn retains for semantic interpreta-
tion (and other relevant interpretations such as information structure)
the functional relation that was established by Agree, which itself does
not require a Spec-head relation in order to apply. An obvious problem
with this view is the expletive construction, which appears to license
agreement without any movement. I will turn to this construction in the
next section and show that an analysis in the literature on expletives
makes this construction consistent with the PGU proposal.
Two other questions arise concerning movement as I have described it.
First, if, under the bare phrase structure approach, both the head and the
label are the same, are there cases where the goal raises directly to the
head instead of to the label to implement PGU? The answer is yes; these
are the cases of pro-drop in which a pronominal head containing rich
agreement raises to T and values the probe at T, making it unnecessary
to raise an XP to the specifier of TP (see Alexiadou and Anagnostopou-
lou 1998). I will comment on this in more detail below.
Second, at what point in the derivation does PGU take place? Clearly,
it must take place before transfer to semantic interpretation.
(8) PGU must be established by the point of transfer.
I have been assuming the existence of phases, a phase being a complete
unit of argument structure (vP) or expression (CP) (Chomsky 2001). In
Chomsky 2001, it is not the entire phase that is transferred (or ‘‘spelled
out’’); rather, it is only the interior (or ‘‘complement’’) of the phase. This
gives rise to the Phase Impenetrability Condition. For example, once a CP
is built, what is transferred is its complement, TP. Because the edge of
each phase remains intact for the next phase, elements are able to
pass from one phase to the next, as in the case of long-distance wh-
movement.1
What ultimately is pronounced in the specifier position of the head that
contains the probe may be the goal that fulfills the PGU requirement, as
in the case of the external subject in Spec,TP, or it may be a copy of the
goal. A copy in Spec,TP, for example, is fully capable of establishing
Why Move? 37

PGU so long as it is a fully specified copy. In chapter 4, we will look

at when a moved element leaves such a copy, but for the time being I
will assume that, whereas Ā-movement arguably leaves a full copy,
A-movement need not do so, although it can. Thus, the tail of an
Ā-movement is always able to fulfill the PGU requirement. In the rest
of this chapter, I will draw on studies in the literature to demonstrate
PGU and the fact that it must take place at the point of transfer.
Finally, the basic tenet of this monograph—that topic/focus plays
a role in discourse-configurational languages that is computationally
equivalent to the role of f-features—suggests that topic/focus in these
languages should undergo the kind of valuation found with f-probes
in agreement-based languages. This is di¤erent from the standard
view of topic/focus; for example, for focus it is typically assumed that
there is a focus head that carries the interpretable focus feature (e.g.,
Brody 1990; in chapter 3 I discuss this matter extensively). In chapters
3 and 4, I will develop a probe-goal system for focus in the discourse-
configurational language Japanese. I will suggest that topic works some-
what di¤erently, although it is in line with the overall assumptions of my
theory.

2.4 Expletives and Related Matters

I now turn to expletives, which, on the surface, appear to be an exception

to the PGU requirement.
Expletives have played a major role in the development of a number of
important aspects of GB and particularly in the Minimalist Program.
Chomsky (1981) based the Extended Projection Principle (EPP) on the
appearance of the expletive in examples like (9).
(9) There entered a boy into the room.
In more recent frameworks, the appearance of the expletive in such exam-
ples made it possible for T to enter into so-called long-distance agreement
with the postverbal nominal phrase (here, a boy), which led Chomsky
(2000, 2001) to decouple agreement from movement. In this construc-
tion, then, the agreement does not require movement; instead, it may ex-
ist as long-distance agreement, where the goal is not in the specifier of the
head that contains the probe. Under this view, expletives would be an ex-
ception to PGU. Below, I will present an alternative account that is con-
sistent with the view that agreement requires PGU to obtain. Later in
the chapter, I will take up other well-known cases cited as evidence for
38 Chapter 2

long-distance agreement in languages such as Hindi-Urdu and Itelmen,

and show that they too have an analysis compatible with PGU (although
problems remain).
A number of linguists have made proposals about the nature of exple-
tives. Bošković (1997, 2002) and Martin (1999) argue that an expletive
must have Case and that this makes it possible to predict the distribution
of expletives. Behind their studies is the desire to derive the EPP from
Case considerations, something they try to do by assuming that Case can
only be checked in the specifier position of the head that is responsible for
valuing Case (e.g., T) (Boeckx 2000, Epstein and Seely 1999; see also
Koopman 2000, 2003; Koopman and Sportiche 1991).
Another approach looks at expletives as the realization of the D feature
(Chomsky 1995:364; see also Lasnik 1997). Given that D is the locus of
f-feature agreement, this would mean that an expletive carries the fea-
ture(s) that function as the goal of the probe under Agree. The ‘‘EPP’’ ef-
fect would follow naturally from this relationship, given that the goal
must raise to the probe to value it. What would be involved is not Case,
as the linguists mentioned above suggest, but f-feature agreement associ-
ated with D—a state of a¤airs that would make the appearance of an ex-
pletive equivalent to the appearance of a DP.
What precisely is the source of expletives? A number of linguists have
argued that an expletive begins in the vicinity of the associate noun
phrase (e.g., Chomsky 1995). For Moro (1997), an expletive is a predi-
cate, and for Sabel (2000) it occupies D in the structure that contains the
associate noun phrase (see also Choe 2006, Kayne 2006 for relevant dis-
cussion). The following examples, with some additional detail, are from
Sabel 2000:414:2
(10) a. [TP There [T 0 is [DP t a man] in the garden]].
b. [TP A man [T 0 is t in the garden]].
In (10a), the expletive there starts out in the DP [DP there a man], and it
contains the formal features of the DP (see Lasnik 1997; see also Chom-
sky 1986 and Safir 1982, 1985, 1987 for earlier versions that share some
similarities with the proposals we are discussing). The probe enters into
Agree with the f-features of the DP, and movement raises the carrier of
the f-features—the expletive—to the probe, thereby establishing a PGU
at transfer. What is of interest here is that an expletive always starts out
with the associate DP, and it undergoes movement, just like a DP. From
this standpoint, the expletive construction is not an exception to the need
of the goal to move to the probe. Hence, the Moro/Sabel approach
allows us to unify all agreement-induced movement, including movement
Why Move? 39

of expletives, as the need of the goal to move to the probe to attain

PGU.3
Why does there split from the rest of its phrase and move? One possible
answer is that PGU requires movement; this movement targets the entity
that contains the interpretable feature; and in the expletive construction,
it is the expletive that has interpretable features. The expletive is morpho-
logically independent of the rest of phrase, so it is able to move without
having to pied-pipe the rest. In a normal construction where a full DP
moves, the element that carries the interpretable feature (N or D) cannot
split from the rest, so the entire DP must pied-pipe to the probe.
Does the expletive raise from the associate that is in its original posi-
tion? It appears that the associate itself first undergoes movement to a
position in the vicinity of v, something reminiscent of the family of pro-
posals that view the internal argument as having to undergo short move-
ment as in (11) (see Johnson 1991, Koizumi 1995, Lasnik 1999b, Runner
1995, Tanaka 1999).
(11) Short movement of the associate DP

Evidence that this movement occurs comes from the ordering of the DP
and the passive participle (Burzio 1986:154–158).
(12) a. There’ve been some men arrested.
b. *There’ve been arrested some men.
The DP, some men, must occur before the past participle. Lasnik (1995)
argues that this demonstrates movement of the DP to a higher position.
The internal argument in an active sentence occurs after the verb in En-
glish. To keep the picture consistent, let us assume, following the studies
cited earlier, that the internal argument moves to a higher position as in
(12a) in all cases, and that the di¤erence between the passive and the
active comes from the height of the verb: the passive participle stays low
in the structure, while the active verb moves up.
There is evidence for this. As mentioned by Caponigro and Schütze
(2003), Blight (1999), using Bowers’s (1993) observations about degree-
of-perfection adverbs, notes that such adverbs may occur before a passive
participle but not after an active participle.
40 Chapter 2

(13) a. The house was poorly built.

b. *They (have) poorly built the house.
c. They (have) built the house poorly.
The idea here is that, whereas the passive participle stays low, possibly in
its original position, the active participle in (13c) moves up higher across
the adverb and also the raised object. Let us assume that in the typical
expletive construction involving an unaccusative predicate, the associate
DP raises, and the verb then raises above this associate DP, giving the
surface form there V DP. Under this assumption, the expletive would be
launched in examples such as (13a) from a position adjoined to v.
(14)

An advantage of the ‘‘D’’ analysis of the expletive is that, as Sabel

(2000) argues, it can account for the definiteness e¤ect observed in the ex-
pletive construction (Milsark 1974).
(15) There appear(s) *the boy/*every boy/a boy/two boys in the picture.
The associate remains within the nuclear scope, forcing it to have an in-
definite reading (Diesing 1992).
The analysis also provides a straightforward account of an example
that Bošković (2002:191) presents in favor of his Case approach to
expletives.
(16) *There seems there to be someone in the garden.
Bošković correctly notes that if the only requirement that must be met is
that of the EPP on T, externally merging the expletive in the lower (and
the matrix) clause should be possible. He argues that the lower Spec,TP is
not a Case position, and because Case is settled for the nominal phrase
Why Move? 41

someone, nothing should happen in the lower Spec,TP aside from the nor-
mal external Merge. On the ‘‘D’’ approach to the expletive, there is no
way to derive (16) because there is a one-to-one relation between an ex-
pletive and its associate—yet (16) has two expletives for the one associate
someone.
There is, then, good evidence for the ‘‘D’’ account of the expletive. On
this account, the expletive carries the f-features that enter into Agree with
the f-probe, making the expletive the target of movement. The ‘‘EPP’’ ef-
fect is not one of Case; rather, it arises from the need of the goal to move
to the probe.4

2.5 The Resumptive Strategy

Danish and Swedish exhibit the that-t e¤ect, as shown by the Danish ex-
ample (17a). Interestingly, the two languages have a repair strategy that
is consistent with the analysis of the expletive just presented. (Examples
(17a–b) are from Jacobsen and Jensen 1982.)
(17) a. *Vennen [(som) han pastod [at havde lant] bogen]]
friend-def c he claimed c had borrowed book-def
var forsvundet.
was disappeared
‘The friend that he claimed had borrowed the book had
disappeared.’
b. Vennen [(som) han pastod [at der havde lant]
friend-def c he claimed c there had borrowed
bogen]] var forsvundet.
book-def was disappeared
‘The friend that he claimed had borrowed the book had
disappeared.’
In (17b), the that-t e¤ect is mitigated by the presence of the expletive der
in Spec,TP. Given my approach to probe-goal relations, the that-t e¤ect
in (17a) suggests that the subject wh-operator is unable to move to the
embedded Spec,TP when at ‘that’ occupies C, a point that must be stipu-
lated. As a result, PGU does not take place at the point where the TP is
transferred, which leads to ungrammaticality. Otherwise, a copy of the
goal would be left in Spec,TP, and PGU would hold with this copy
when the TP is transferred. Based on this, the proposed analysis provides
an account of (17b): the expletive acts as a proxy for the goal, just as we
saw for the expletive in English, and attains PGU with the probe on T by
42 Chapter 2

moving to Spec,TP. The subject wh-operator is able to move directly from

a lower position to Spec,CP—a phenomenon well documented in Italian
(Rizzi 1982). A similar phenomenon occurs in Swedish, except that what
acts as the proxy is a resumptive pronoun (Engdahl 1985:8).
(18) a. *Villet ord visste ingen [hur staves]?
which word knew no.one c is.spelled
‘Which word did no one know how (it) is spelled?’
b. Villet ordi visste ingen [hur deti staves]?
which word knew no.one c 3sg is.spelled
‘Which wordi did no one know how iti is spelled?’
Like Danish, Swedish exhibits the that-t e¤ect, as shown in (18a). In
(18b), the that-t e¤ect is mitigated by the occurrence of the resumptive
pronoun det in the lower Spec,TP. If the analysis proposed here is on the
right track, it suggests a direction for analyzing the that-t e¤ect in En-
glish.5 We would need to derive the stipulation that when that occurs,
the subject wh-phrase somehow cannot move to Spec,TP.6

2.6 Pro-Drop

As we have seen, the goal moves to establish a local relation with the
probe. So far, all examples of this movement but one have been of cate-
gory XP—for instance, the external argument moving to Spec,TP. The
one exception is movement of the expletive. Under the bare phrase struc-
ture approach to structure building, we predict that a goal ought to be
able to value a probe by moving directly to the head that contains the
probe. We find exactly this configuration in the so-called pro-drop lan-
guages of Romance and other language families. Because the goal moves
to the head that contains the probe, the goal itself must be a head, and
the movement responsible for attaining PGU is head movement. Other-
wise, all the details of PGU remain the same as in cases of movement to
the specifier that we have looked at so far. The pro-drop facts are well
known. I will just summarize them briefly to show that my proposed sys-
tem predicts the pro-drop phenomenon as reported in the literature.

2.6.1 Romance
In many languages of the world, the subject position may be left empty if
the referent is clearly understood from the discourse. This is the pro-drop
phenomenon, common in Romance languages, but also widespread in
many other language families.
Why Move? 43

(19) Spanish
baila bien.
dance.3sg well
‘She dances well.’
(20) Italian
verrà.
come.3sg.fut
‘He will come.’
It has long been noted that pro-drop languages, at least of the Romance
type, have rich inflection (Jespersen 1924, Perlmutter 1971, Rizzi 1978,
Taraldsen 1978). In their study of the pro-drop phenomenon in Romance
and related languages, Alexiadou and Anagnostopoulou (1998) make the
important observation that pro-drop is actually licensed by two factors,
rich agreement and a new factor they bring to light.
(21) Two necessary factors for pro-drop (of the Romance type)
a. Rich agreement
b. V-to-T movement, where the agreement shows up on T
Alexiadou and Anagnostopoulou demonstrate the rich agreement re-
quirement with Greek, a typical pro-drop language.
(22) English Greek
I love we love agapo agapame
you love you love agapas agapate
he loves we love agapa agapane
Unlike English, Greek has unique agreement forms for number and per-
son for each of the six possibilities shown. In addition, as Alexiadou and
Anagnostopoulou show, this rich agreement, which appears as inflection
on the verb, must occur on T, which is accomplished by the bare verb
raising to the richly inflected T. According to Alexiadou and Anagnosto-
poulou, the rich inflection is a pronominal with a D feature, and it is ca-
pable of checking the EPP feature on T, which they assume is a D feature
on T, following Chomsky (1995). The pronominal agreement on T takes
care of the EPP/D requirement, with the result that nothing needs to
move to Spec,TP. In fact, on Alexiadou and Anagnostopoulou’s analysis,
Spec,TP is never filled in Greek or any of the other pro-drop languages,
since the EPP feature on T is independently checked by the rich agree-
ment. I will return to this issue of V-to-T movement later.
Under the approach I am pursuing, the analysis of pro-drop would be
essentially the same as Alexiadou and Anagnostopoulou’s. A head that
44 Chapter 2

contains ‘‘rich’’ interpretable f-features—the goal—occurs in the gram-

matical subject position, and the f-probe at T enters into an Agree rela-
tion with it. Subsequently, the goal moves to the f-probe by head raising,
thereby establishing a PGU. The requirement that V move to T presum-
ably exists to morphologically support the agreement head at T.
Alexiadou and Anagnostopoulou’s system predicts that there is nothing
in Spec,TP in pro-drop languages—an account I would concur with be-
cause the goal in the form of the agreement head raises to T.7 There are,
however, instances where a subject occurs preverbally. Alexiadou and
Anagnostopoulou note that preverbal subjects have a special interpreta-
tion, something akin to a ‘‘topic,’’ which suggests that they occupy a
higher, Ā-position. Accordingly, Alexiadou and Anagnostopoulou ana-
lyze the preverbal subject in this way (example from 1998:506).
(23) SVO: S is not in Spec,TP in Greek/Romance
Indefinite preverbal subject: strong (partitive/specific) reading
a. Ena pedhi diavase to ‘‘Paramithi horis Onoma.’’
a child read the ‘‘Fairy Tale without a Title’’
‘A certain child/One of the children read ‘‘Fairy Tale without a
Title.’’ ’
b. Diavase ena pedhi to ‘‘Paramithi horis Onoma.’’
‘One of the children read ‘‘Fairy Tale without a Title.’’ ’
The preverbal indefinite ‘a child’ in (23a) has a partitive/specific read-
ing, whereas the postverbal subject in (23b) may receive a nonspecific
interpretation.
Finally, another typical property of pro-drop languages is what is
called ‘‘subject inversion.’’
(24) Subject inversion in Romance
a. Ha mangiato Giovanni. (Italian)
b. Ha comida Juan. (Spanish)
have.3sg eat.part.past John
‘John has eaten.’
The ‘‘inverted subject’’ never occupied Spec,TP because it never functions
as the goal of the probe; hence, it can occur in the ‘‘inverted’’ position to
begin with or be moved there from some lower position (Kenstowicz
1989, Rizzi 1982).

2.6.2 Bani-Hassan Arabic

In discussing f-feature agreement, I have abstracted away from the dif-
ferent types of agreement and their role in movement. I will continue
Why Move? 45

to do this to keep the exposition focused on movement and related mat-

ters. However, it is important to note for the record that there is a rich
literature on person, number, and gender agreements, with an apparent
consensus building that person appears always to function as ‘‘true’’
agreement whereas number and gender sometimes do not manifest all
the expected properties of genuine agreement (e.g., Baker 2008, Boeckx
2007, Ouhalla 2005, Sigur¶sson 2007). One example of this is found in
Bani-Hassan, an Arabic dialect spoken by ‘‘a Bedouin clan of the Jorda-
nian desert’’ (Kenstowicz 1989:264). According to Kenstowicz (1989:272),
this language has the following inflections:
(25) a. þtense, þperson finite
b. tense, þperson subjunctive
c. þtense, person participle
Quoting Kenstowicz (1989:272–273):
Finite verbs exhibit independent selection for the categories of tense/aspect. They
may thus appear in main clauses as well as in innu ‘that’ clauses complement to
verbs such as gaal ‘say’, iÀtigad ‘think’, etc. Subjunctive denotes verbs that exhibit
no independent selection for tense/aspect but nevertheless show full obligatory
agreement with the subject. . . . The participle . . . only shows distinctions in gender
(masculine versus feminine) and number (singular and plural).

Kenstowicz goes on to show that the participial inflection does not allow
pro-drop of the subject, unlike the forms with þperson; see (26), from
Kenstowicz 1989:265–266. This suggests that a head that can function as
the goal of a probe must have person agreement. (In some languages,
that is not su‰cient; Holmberg (2005) notes that in Finnish, first- and
second-person agreement license pro-drop but third-person agreement
does not.)
(26) a. Finite form
Al-binit gaalat innu ištarat al-libaas.
the-girl said that bought the-dress
b. Participle
Fariid gaal *innu/inn-ha mištarya al-libaas.
Fariid said that/that-she bought the-dress
Kenstowicz also notes that the finite form has the typical pro-drop prop-
erty of subject inversion that we observed in Romance and Greek, but the
participle does not allow it. This indicates that in the participial construc-
tion, the fully specified DP subject functions as the goal that must be
raised to Spec,TP to attain PGU.
46 Chapter 2

(27) a. Finite form (uninverted)

Fariid gaal innu al-binit ištarat al-libaas.
Fariid said that the-girl bought the-dress
b. Finite form (inverted)
Fariid gaal innu ištarat al-binit al-libaas.
Fariid said that bought the-girl the-dress
(28) a. Participle (uninverted)
Fariid gaal innu al-binit mištarya al-libaas.
Fariid said that the-girl bought the-dress
b. Participle (inverted)
*Fariid gaal innu mištarya al-binit al-libaas.
Fariid said that bought the-girl the-dress

2.7 Chinese as an Agreement Language

In this section, I will suggest that Chinese is an agreement language

that patterns in many ways like the pro-drop languages we looked at
in the previous section.8 This is di¤erent from the usual view of Chinese
as a topic-prominent language. This topic prominence is illustrated in
(29).
(29) Zhe-ben shu Zhangsan mai-le.
this-cl book Zhangsan buy-asp
‘This book, Zhangsan bought.’
It is generally believed that the topic in (29) occupies a position higher
than Spec,TP. From the present perspective this means that the topic fea-
ture remains at C, in turn suggesting that only the f-probe is inherited
by T, giving rise to an agreement language. I propose that Chinese shares
relevant features along these lines with agreement languages such as
Romance and Bani-Hassan Arabic—in particular, the idea that person
agreement at T licenses pro-drop.
Word order in Chinese is typically SVO, and adverbial modifiers occur
between the subject and the object.
(30) Zhangsan zuotian zai xuexiao kanjian-le Lisi.
Zhangsan yesterday at school see-asp Lisi
‘Zhangsan saw Lisi at school yesterday.’
One exception to SVO order is found in the ba construction, where the
object occurs preverbally, accompanied by ba.
Why Move? 47

(31) a. Ta pian-le Lisi.

he cheat-asp Lisi
‘He cheated Lisi.’
b. Ta ba Lisi pian-le.
he ba Lisi cheat-asp
‘He cheated Lisi.’
In the ba sentence in (31b), focus is placed on the ba phrase; and in cer-
tain cases, the presence of the ba phrase changes the aspectual interpreta-
tion (see, e.g., Li 1990, Li and Thompson 1981, Tenny 1994).
If we are to argue for an analysis that involves person agreement at T
in Chinese, like the analyses proposed for Romance and other pro-drop
languages, we first must establish that there is, in fact, a T projection in
this language. Because there is no T projection that can be identified,
many linguists have suggested that Chinese does not have T (see refer-
ences in Sybesma 2007). In contrast, Sybesma (2007) and Tang (1998)
provide arguments that it does have T. The examples and discussion are
from Sybesma 2007:581 (I omit diacritics for tones, following the practice
of other linguists).
The pair of examples in (32) appear to indicate that there is no T in
Chinese.
(32) a. Zhangsan zhu zai zher.
Zhangsan live at here
‘Zhangsan lives here.’
b. Zhangsan 1989 nian zhu zai zher.
Zhangsan 1989 year live at here
‘Zhangsan lived here in 1989.’
Example (32a) has a present tense interpretation, but when ‘1989 year’ is
added as in (32b), the interpretation is that of past tense. On the face of it,
this appears to indicate that Chinese does not have tense and that the in-
terpretation ‘past’ comes about with the inclusion of a temporal phrase
like ‘1989’ (similarly for other temporal interpretations). However, Syb-
esma notes, following Matthewson (2002), that this is the wrong way to
view the pair in (32). Specifically, he points out that (32a) has a present
tense interpretation. Where would this interpretation come from if Chi-
nese has no tense? It is truly a present tense interpretation because, for
example, if the subject refers to a person who is dead, the sentence is non-
sensical. (See Sybesma 2007 for other arguments.)
Now, if the T projection indeed occurs in Chinese, where is it located?
Because there is no overt manifestation of T as far as we can see, we have
48 Chapter 2

to locate it through indirect means. There are two considerations here.

First, the aspectual marker le occurs in mid-sentence, and it is often the
case that an aspectual marker occurs in the proximity of tense, as in Jap-
anese. Second, if Chinese is indeed a pro-drop language like those dis-
cussed in section 2.6, verb raising should raise the verb to T (Alexiadou
and Anagnostopoulou 1998). The verb in fact occurs in mid-sentence,
next to le. Let us, then, suppose that T occurs in Chinese in a position
similar to the one it occupies in English and Romance. If we look again
at the examples that contain an overt subject, and in particular at where
this subject occurs relative to T, we see that while it appears to occur next
to T in some cases, it can also occur away from T, as when an adjunct
(30) or the ba phrase in a ba construction (31b) intervenes. This is
expected under the pro-drop analysis because the overt subject is not the
goal of the probe at T; instead, the goal is the agreement head, so the
overt phrase corresponding to the subject is free to occur away from T.9
The next step is to establish that pro may occur in subject position but
nowhere else—something that turns out to be a challenge in Chinese be-
cause, unlike Romance, Chinese is a massively pro-drop language, allow-
ing gaps not only in subject position but also in, for example, object
position. Fortunately, C.-T. J. Huang (1984) has established that the gap
in subject position is pro, whereas the gap in object position is a variable
created by Ā-movement of an empty topic.
(33) Zhangsan shuo [e bu renshi Lisi].
Zhangsan say e not know Lisi
‘Zhangsan said that [he] did not know Lisi.’
(34) Zhangsan shuo [Lisi bu renshi e].
Zhangsan say Lisi not know e
‘Zhangsan said that Lisi did not know [him].’
Example (33) contains a subject gap in the subordinate clause, and (34)
an object gap. Huang observes that the subject gap in (33) can refer either
to the matrix subject Zhangsan or to someone else in the discourse. In
contrast, the object gap in (34) can only refer to someone else in the dis-
course, not to the matrix subject. Huang accounts for this asymmetry be-
tween subject and object gaps by proposing that the subject gap is a pro
that can refer either to the matrix subject or to someone else in the dis-
course. The object gap in (34) is instead a variable bound by a null dis-
course topic that begins in the subordinate object position and undergoes
Ā-movement to the topic position in the matrix clause.10 This means
Why Move? 49

that, in Chinese, the topic feature stays at C, and if the line we are pursu-
ing in this section is correct, the f-probe in the form of a person feature
is inherited by T. Further evidence for the subject/object asymmetry
shows up in topicalization, which involves a potential Crossover violation
(C.-T. J. Huang 1984:558).
(35) a. Zhangsani , tai shuo [ei mei kanjian Lisi].
Zhangsan he say e no see Lisi
‘Zhangsani , hei said that [hei ] didn’t see Lisi.’
b. *Zhangsani , tai shuo [Lisi mei kanjian ei ].
Zhangsan he say Lisi no see e
‘Zhangsani , hei said that Lisi didn’t see [himi ].’
The topicalized Zhangsan in (35a) is coreferential with the subordinate
subject, whereas the one in (35b) is coreferential with the subordinate ob-
ject. Huang notes that (35a) is grammatical, presumably because the sub-
ordinate subject is pro—hence, no movement has occurred to produce
topicalization, and a Crossover violation is thereby avoided. By contrast,
(35b) is ungrammatical because the subordinate object gap has a variable
created by Ā-movement of the topic—movement that triggers a Strong
Crossover violation.
As we have seen, Sybesma’s (2007) work (also Tang 1998) establishes
the possibility that Chinese has a tense projection, and C.-T. J. Huang’s
(1984) work on subject and object gaps shows that the subject gap is pro
but the object gap is created by movement. To argue that Chinese is an
agreement language, we also have to show that Chinese does in fact evi-
dence f-feature agreement associated with the subject. We can do this by
looking at what is called the ‘‘blocking e¤ect’’ for the reflexive anaphor
ziji (Y.-H. Huang 1984, Tang 1985, 1989). As (36) shows, ziji can func-
tion as a long-distance anaphor across a more local potential antecedent.
(All examples in this discussion of the blocking e¤ect are taken from Pan
2000.)
(36) Zhangsani zhidao Lisij dui zijii=j mei xinxin.
Zhangsan know Lisi to self not confidence
‘Zhangsan knows that Lisi has no confidence in him/himself.’
Note that in this example, both matrix and subordinate subjects are
third person. However, if the potential antecedent in the next higher
clause does not match the lower potential antecedent in person, the long-
distance construal of ziji is blocked, leaving only the local subject as its
antecedent.
50 Chapter 2

(37) a. Woi juede nij dui ziji i=j mei xinxin.

I think you to self not confidence
‘I think you have no confidence in yourself/*me.’
b. Nii juede woj dui ziji i=j mei xinxin ma?
you think I to self not confidence q
‘Do you think I have no confidence in myself/*you?’
c. Zhangsani juede wo/nij dui ziji i=j mei xinxin.
Zhangsan think I/you to self not confidence
‘Zhangsan thinks I/you have no confidence in myself/yourself/
*him.’
In these examples, the matrix and subordinate subjects do not match in
person ((37a): first-second; (37b): second-first; (37c): third-first/second).
This mismatch blocks ziji from long-distance construal. This blocking ef-
fect has been given as evidence that a subject-oriented anaphor undergoes
movement at LF (e.g., Battistella 1989, Cole, Hermon, and Sung 1990,
Huang and Tang 1991). One implementation consistent with the idea
that there is person agreement at T in Chinese is that ziji raises to its local
T, where it takes on the person value of the T (the person feature already
valued by the subject phrase). Ziji can, then, move to T in the higher
clause, and if the person feature on that T matches the person feature
already on ziji, long-distance construal is possible. This, or some other
approach, requires the existence of a person feature in Chinese.11
A point worth making here, though, is that Japanese does not exhibit
the blocking e¤ect.
(38) Taroo/Watakusi/Anata-wa [Taroo/watakusi/anata-ga zibun-no
Taro/I/you-top Taro/I/you-nom self-gen
syasin-o totta to] itta.
picture-acc take c said
‘Taro/I/You said that Taro/I/you took self ’s picture.’
Setting aside a certain pragmatic awkwardness with some possible inter-
pretations, it is possible in principle for the anaphor to refer to the sub-
ordinate or matrix subject in any combination. This indicates that the
anaphor in Japanese is sensitive to subjects but not to person agreement.
Recall that in the pro-drop languages we looked at in section 2.6, a
verb-initial order with postverbal ‘‘subject’’ is possible. Although such a
sequence is apparently not as common in Chinese, we do find examples
like (39) and (40), taken from Huang 1982.
Why Move? 51

(39) Yu xia-guo le.

rain fall-asp asp
‘It has rained.’
(40) Xia-guo le yu le.
fall-asp asp rain asp
‘It has rained.’
As a reviewer notes, the question remains why Chinese does not easily
allow the verb-subject order found in Romance.
Finally, the pro-drop analysis may account for a mystery about quanti-
fier scope. In Chinese, two quantifiers in a simple clause are limited to
surface scope; inverse scope is not possible. This was noted originally for
Japanese by Kuroda (1971), and Huang (1982:113) discovered that the
same fact holds in Chinese.
(41) Youyige xuesheng bu mai suoyonde shu.
one student not buy all book
‘There was a student who did not buy all the books (only some).’
(42) Meige xuesheng dou mai-le yiben shu.
every student all buy-asp one book
‘For every student x, there is one book y such that x bought y.’
A possible explanation is that the subject, which is not the goal, occurs in
a position higher than TP, most likely a topic position. Recall Alexiadou
and Anagnostopoulou’s (1998) Greek example (43), which has only sur-
face scope.
(43) Ena pedhi diavase to ‘‘Paramithi horis Onoma.’’
a child read the ‘‘Fairy Tale without a Title’’
‘A certain child/One of the children read ‘‘Fairy Tale without a
Title.’’ ’
The reason for the lack of ambiguity is that ‘a child’ here is not the true
subject, and it occurs in a higher position (Alexiadou and Anagnostopou-
lou claim that it is an Ā-position, but we will see in chapter 3 that at least
the comparable position in Japanese is best analyzed as an A-position).
If what I have proposed in this section is on the right track and Chinese
has person agreement, why doesn’t it ever appear as verbal inflection, as
in other agreement languages? There is no simple answer to this; rather,
a combination of factors leads to this unusual situation. First, it appears
that Chinese noun phrases are unable to value a probe; in other words,
the noun phrases do not have interpretable agreement features. Why
52 Chapter 2

would this be the case? We might speculate as follows, on the basis of a

proposal by Chierchia (1998). Chierchia argues for two types of nominal
expressions: those, like English and Romance nominals, that can refer to
individuals; and those, like Chinese nominals (and he specifically dis-
cusses Chinese), that refer to kinds. It could be that the English-type nom-
inal expressions are DPs and that D is typically the locus of f-feature
agreement (although it isn’t clear why); and that the Chinese-type expres-
sions are NPs, which would be unable to host f-features, so they don’t.
On the other hand, given the pro-drop nature of Chinese, and following
the analysis of Romance, we can speculate that the ‘‘small pro’’ in Chi-
nese does have an interpretable feature—specifically, a person feature—
and it is this feature that enters into an Agree relation with the f-probe.
The pro then moves to the probe for PGU, and its phonetically empty na-
ture leads to a lack of verbal agreement inflection.
But then, how can the language learner ever figure out that Chinese has
person agreement? Here, speculating again, and reflecting on the spirit
of the Uniformity Principle (Chomsky 2001), interpreted here as Strong
Uniformity, we might say that in the absence of any overt indication of
agreement, and in the absence of any indication that the language is of
the topic/focus type, where topic/focus is inherited by T, the language
learner simply turns to the default universal setting for agreement. This
f-feature agreement must be inherited by T. Given that there is no indi-
cation that the nominal can provide valuation, the language learner
assumes that an agreement head, which happens to be unpronounced,
provides the valuation. Finally, in the absence of overt manifestation of
f-features, the learner assumes the most basic form of agreement—
namely, person agreement. Other explanations exist, but whatever one
might come up with, it must account for the sheer lack of overt manifes-
tation of f-features in the inflectional system combined with a clear man-
ifestation of person agreement in the construal of the anaphor ziji.

2.8 Irish

VSO languages call into question the idea of PGU. Irish is one such lan-
guage. (All examples are from McCloskey 2001.)
(44) Do fuair sé nuachtán Meiriceánach óna dheartháir
past got he newspaper American from.his brother
an lá cheana.
the-other-day
‘He got an American paper from his brother the other day.’
Why Move? 53

McCloskey (1996) originally argued that the subject is in VP and that the
EPP is inactive in Irish. For the sentence in (45), he postulated the struc-
ture in (46).
(45) D 0 ól sı́ depcj uisce.
past drink she drink water
‘She drank a drink of water.’
(46)

In (46), there is no way for PGU to obtain because the subject remains
in VP. However, McCloskey has since argued that the subject raises to a
‘‘subject’’ position next to T (McCloskey 2001). From the perspective
taken here, this means that T Case-marks the subject, and the f-probe at
T is valued by the subject, which then moves to a position in the projec-
tion of T for the purpose of establishing a PGU. McCloskey (2001:169)
characterizes this position as a ‘‘nominative’’ position and a ‘‘subject’’ po-
sition. The process McCloskey proposes is similar to the raising of the
agreement morpheme to T in pro-drop languages, except that what has
moved is a DP. As one piece of evidence, McCloskey gives the following
periphrastic progressive aspect sentences:
(47) a. Tá ag neartú ar a ghlór.
is strengthen prog on his voice
‘His voice is strengthening.’
b. Tá a ghlór ag neartú.
is his voice strengthen prog
‘His voice is strengthening.’
In (47a), the complement is a PP (‘on his voice’), but in (47b), it is a bare
DP (‘his voice’). As shown, the PP in (47a) stays low, presumably in VP,
while the bare DP in (47b) raises above the progressive verb, which sug-
gests that it occupies a position in the vicinity of T. The f-probe on T is
appropriately valued, and PGU has taken place as required.
54 Chapter 2

2.9 A Word about Long-Distance Agreement

I need to say something about the so-called long-distance agreement

(LDA) reported in a variety of languages (e.g., Hindi-Urdu: Mahajan
1989, Bhatt 2005; Itelmen: Bobaljik and Wurmbrand 2005; Tsez: Polin-
sky and Potsdam 2001). Typical LDA has the form exemplified by the
Hindi-Urdu sentence (48a) from Mahajan 1989:234:
(48) a. Ram-ne [rotii khaa-nii] chaah-ii.
˙
Ram-erg bread.f eat-inf.f want-pvf.fsg
‘Ram wanted to eat bread.’
b. Ram-ne [rotii khaa-naa] chaah-aa.
˙
Ram-erg bread.f eat-inf.m want-pvf.msg
‘Ram wanted to eat bread.’
In (48a), the matrix verbal inflection agrees with the object ‘bread’ in the
infinitival, and this agreement also shows up on the infinitival. This LDA
is optional, and when it does not occur, as in (48b), the matrix inflection
takes on the default masculine singular, and the infinitival does not inflect
for agreement. Mahajan (1989) analyzes the LDA in Hindi-Urdu in a
way that is consistent with the PGU—that is, despite the appearance of
LDA, PGU for f-features is needed here. In his account, the object in the
infinitival clause first undergoes movement within the infinitival, where it
enters into agreement with the infinitival verbal complex. It then moves to
the matrix clause, to the position Mahajan identifies as Spec,AgrP. By
moving in this way, the embedded object, which functions as the goal of
the f-probe, moves to the probe and establishes a PGU.
Bobaljik and Wurmbrand (2005) propose a similar account for LDA in
Itelmen. They show, among other things, that when the embedded object
enters into LDA, it must take matrix scope, whereas when it does not, it
is limited to being interpreted inside the infinitival (Bobaljik and Wurm-
brand 2005:849).
These views of LDA are promising for PGU, but challenges remain.
Bhatt (2005) proposes an alternative account of the Hindi-Urdu facts
that does not necessarily involve movement of the subordinate object.
And in Tsez, the embedded object that apparently functions as the goal
of the f-probe in the matrix clause may be preceded by other elements
in the infinitive, as shown in (49) (Polinsky and Potsdam 2001:584).
(49) enir [užā magalu bāc’rułi] b-iyxo.
mother boy bread.III.abs ate.IV III-know
‘The mother knows (that) the boy ate the bread.’
Why Move? 55

The embedded object ‘bread’ agrees with the matrix verbal complex.
Note that this object is preceded by the subordinate subject ‘boy’, which
makes a simple movement analysis of the embedded object to the matrix
clause implausible. Polinsky and Potsdam do note that the embedded ob-
ject that enters into LDA has a ‘‘topic’’ interpretation, and they in fact
argue that the object undergoes movement to the subordinate topic posi-
tion in the CP domain. This argument is still not su‰cient to overcome
the di‰culty this example poses, however, given that the goal is still
in the subordinate structure; and besides, if Polinsky and Potsdam are
right, the movement only occurs at LF. I will leave these problems for fu-
ture study.12

2.10 A Brief Look Back at West Flemish

As the last item to be discussed in this chapter, let us go back to an issue

from chapter 1. Recall the West Flemish examples that show complemen-
tizer agreement as well as subject-verb agreement (from Carstens 2003,
based on Haegeman 1992).
(50) a. Kpeinzen dan-k (ik) morgen goan.
I.think that-I (I) tomorrow go
‘I think that I’ll go tomorrow.’
b. Kpeinzen da-j (gie) morgen goat.
I.think that-you (you) tomorrow go
‘I think that you’ll go tomorrow.’
c. Kvinden dan die boeken te diere zyn.
I.find that.pl the books too expensive are
‘I find those books too expensive.’
The problem is that we assume that a f-probe does not undergo valua-
tion until it is inherited by T or a related head where some mechanism is
available for the f-probe to identify its goal. This assumption predicts
that a f-probe at C should not undergo valuation. However, the comple-
mentizer agreement makes it seem as though not only does a copy of the
f-probe remain on C, but in fact this copy can seek a goal on its own
without the help of an activation mechanism such as Case. But the prob-
lem is in fact more complex. As Van Koppen (2006) and others point out,
the complementizer agreement and the agreement on the verbal inflection
do not always match. A particularly striking example is (51), from Bavar-
ian (Van Koppen 2006:3).
56 Chapter 2

(51) a. . . . dass-ds du und d’Maria an Hauptpreis gwunna

that-2pl [you.sg and the Maria]2pl the first.prize won
hab-ds.
have-2pl
b. . . . dass-sd du und d’Maria an Hauptpreis gwunna
that-2sg [you.sg and the Maria]2pl the first.prize won
hab-ds.
have-2pl
‘. . . that you and Maria have won the first prize.’
In (51a), the complementizer and the verbal inflection match, both show-
ing second person plural, which agrees with the subject. The interesting
fact is (51b), where the complementizer agrees with the first conjunct by
inflecting for singular, while the verbal agreement is plural. Van Koppen
proposes that the f-probe starts at C and is inherited by T, but that a
copy of the f-probe remains at C. The two f-probes then undergo valua-
tion independently, although each picks out the subject as its goal. Al-
though the verbal agreement always agrees with the full goal phrase, the
complementizer is able to agree with a feature that is a subcomponent of
the subject phrase.
Note that on this kind of analysis, we cannot assume the phase-based
approach to agreement for complementizer agreement. Although the sub-
ject and the f-probe on T represent a typical agreement configuration,
complementizer agreement does not. From Van Koppen’s data involving
first-conjunct agreement, it appears that in complementizer agreement,
the probe-goal relation is established strictly through string adjacency, of
the type familiar in phrasal phonology (e.g., Archangeli and Pulleyblank
1987, Odden 1994, Steriade 1987). The copy of the f-probe at C is not
valued in narrow syntax; instead, the valuation takes place at PF, after
the phases have been put back together, and the complementizer and the
subject are string adjacent. Chomsky (2007) suggests that complementizer
agreement is concordance, not agreement, which possibly points to the
same conclusion: that it is a PF phenomenon. Semantics—or, for that
matter, information structure—does not interpret this agreement because
the agreement relation only holds at PF. This, I believe, is correct: the rel-
evant functional relation is between the subject and T. Although comple-
mentizer agreement appears to be fundamentally di¤erent in this way, it
is important to note that, under my analysis, which is based on Van Kop-
pen’s work, the idea that the f-probe begins at C still stands. Comple-
mentizer agreement is a residue of this f-probe at C.
Why Move? 57

As the final note, if the proposed analysis is on the right track, we

would expect not only the subject to occur adjacent to the complemen-
tizer, but also the expletive, since we are assuming that the expletive is a
proxy for the goal. Consider the following West Flemish example, pointed
out to me by Liliane Haegeman:
(52) . . . dan/*da der morgen meer studenten goan kommen.
that.pl/*that.sg there tomorrow more students go come
‘. . . that there will be coming more students tomorrow.’
The expletive der occurs adjacent to the complementizer, as we would ex-
pect. Also presumably so that it can be adjacent to the complementizer,
the expletive occurs to the left of the time adverbial morgen ‘tomorrow’.
The following example, also pointed out to me by Liliane Haegeman,
demonstrates the same point:
(53) . . . dan/*da der dienen boek nie vee studenten kennen.
that.pl/*that.sg there that book not many students know
‘. . . that not many students know that book.’
The object has been moved to the left of the subject, possibly for reasons
of topicalization, and the expletive der occurs to the left of this object and
adjacent to the complementizer.13
This discussion also pertains to a problem I have ignored so far: the
problem of auxiliary inversion in English wh-questions.
(54) What does John want?
The auxiliary, does, is at C, but its goal, John, is in Spec,TP. Does and
John are not adjacent according to our account based on the bare phrase
structure approach. However, they are string adjacent. So, whatever the
reason for auxiliary inversion—possibly the same as the one given above
for the Bavarian and West Flemish phenomena: that there is a copy of
the f-feature at C that must be valued—its placement at C need not be
viewed as violating the PGU requirement. It is just that in this case, the
PGU requirement would hold at PF only.

2.11 Summary

In this chapter, I o¤ered a reason why movement exists in natural lan-

guage. Movement plays a critical role in enhancing the expressive power
of language, although the way movement works is intricate enough that
its function is not immediately obvious. Movement of the type we have
58 Chapter 2

looked at—A-movement triggered by a grammatical feature that starts

at C and is inherited by T—is a record of functional relations. A func-
tional relation is established by Agree, but if nothing else happens, the
record of this critical relation is erased before semantic interpretation
and interpretation for information structure because the probe must be
deleted. Natural language has figured out a way to tap the expressive
power of functional relations by installing an intricate system of agree-
ment, which includes f-features and topic/focus, and instituting the pro-
cess of valuation. Movement forces the goal to move to the probe to
retain a record of the functional relations. I called this probe-goal union.
A look at constructions in a number of languages, including mitigation
of the that-t e¤ect, pro-drop, and VSO word order, revealed that the goal
that values the probe may be an XP or a head, either being possible be-
cause of the notions inherent to the bare phrase structure approach on
which I built my arguments about movement. One language discussed
here, Chinese, poses an interesting question. Because it has no agreement
morphology, it has generally been considered to be a topic-prominent
language. But it gives evidence for f-feature agreement in the form of
person features even though it lacks person agreement morphology.
How can the language learner figure out that this language has person
agreement? As one possibility, I speculated that, given the approach that
every language has f-feature agreement, the language learner simply
assumes the basic universal—that is, initial-state—agreement, which
would be just person agreement.
I briefly discussed three additional issues. The first was surface VSO
word order in Irish, which seemingly contradicts the PGU requirement
because the subject could be viewed as staying in vP. However, Mc-
Closkey’s work shows that the subject in fact moves to a position close
to T, and we can assume that this movement satisfies the PGU require-
ment. The second issue was long-distance agreement, which appears to
blatantly violate the adjacency requirement for agreement. I noted some
analyses that propose that the embedded object moves to the matrix
clause, which would make this phenomenon compatible with PGU, but
there are problems that remain for future study. The third issue was com-
plementizer agreement. In Bavarian and West Flemish, the agreement on
the complementizer is not always identical to the agreement on the verbal
inflection. What appears to allow the f-probe ‘‘copy’’ to be valued at C is
string adjacency, something that I speculated might also apply to auxil-
iary inversion in English wh-questions.
3 Unifying A-Movements

3.1 Introduction

In Government-Binding (GB) Theory, A-movement was defined as move-

ment to a ‘‘potential’’ y-position (Chomsky 1981). The A in A-movement
comes from the idea that a phrase that undergoes this movement moves
to a potential argument (A-)position. In virtually all cases, this position
is what we today would call Spec,TP. In GB and all its predecessors, it
was postulated that Spec,TP is where the external argument of a predicate
is assigned. In the sentence John ate pizza, John merges directly into
Spec,TP, where it receives the external y-role; the source of this argument
y-role is the VP (Marantz 1984). Just in those constructions that do not
assign an external y-role, such as the passive, raising, and unaccusative
constructions, Spec,TP becomes available for A-movement. Although no
external y-role is assigned in these cases, Spec,TP is a ‘‘potential’’ position
for the external y-role, and in GB this was su‰cient to count it as an
argument position for the sake of movement. However, this way of char-
acterizing A-movement became obsolete with the advent of the predicate-
internal subject hypothesis (Kuroda 1988, Sportiche 1988; see also Fukui
1986, Kitagawa 1986). This hypothesis postulates that the external y-role,
as well as the internal y-role, is assigned inside the verbal projection, ei-
ther VP or vP depending on the version of the hypothesis, so that the ex-
ternal y-role is never assigned to Spec,TP. This hypothesis, which has
gained wide currency, is one of the major features that separate GB and
the current Minimalist Program. Obviously, if no y-role is ever assigned
to Spec,TP, this position is never an ‘‘A’’ position as far as y-roles are
concerned, so the idea of characterizing it as a ‘‘potential’’ argument
position for movement is no longer an option. How can we portray this
position in a way that is consistent with current theory?
60 Chapter 3

One possibility is that the distinction between A-movement and Ā-

movement is unnecessary. However, there is good empirical evidence
to suggest that these two types of movement do behave di¤erently. A-
movement does not trigger a Weak Crossover (WCO) violation, and it
can create a new binder (e.g., Mahajan 1990).
(1) Whoi ti seems to hisi mother ti to be smart?
(2) Johni seems to himselfi to be ti smart.
In (1), the wh-phrase who undergoes A-movement from the subordinate
subject position to the matrix Spec,TP, crossing the pronoun his. Never-
theless, the sentence is grammatical. A WCO violation is invoked if the
sentence contains a variable, and if there is a pronoun coreferential with
the variable that the variable fails to c-command. Not being a form of op-
erator movement, A-movement does not create a variable, so in (1) the
trace and the pronoun in the subordinate subject position are not subject
to WCO. In (2), John undergoes A-movement to Spec,TP and is able to
bind himself from this new position. Presumably, such binding only takes
place from A-positions.
By contrast, as (3) and (4) show, Ā-movement is incapable of suppress-
ing a WCO violation; it also cannot create a new binder.
(3) ?*Whoi does hisi mother love ti ?
(4) *To whomi did Mary introduce each otheri ’s friends ti ?
How can we define A-movement? If we limit ourselves to English, it is
possible to characterize it as movement triggered by the agreement fea-
ture on T. In (1) and (2), the phrase that moves into Spec,TP is the so-
called goal of the agreement at T, and it is this agreement that marks the
goal phrase for movement. The same holds for the typical ‘‘EPP’’ move-
ment of an external argument from Spec,vP to Spec,TP—again, by agree-
ment on T.
(5) A-movement (tentative)
An agreement feature on T targets the goal of the agreement for
movement.
However, attempting to define A-movement within the predicate-
internal subject hypothesis became complex when Mahajan (1990) iden-
tified the same properties discussed above for a certain subtype of
scrambling. A typical example of this kind of scrambling is movement of
the object to the head of the sentence above the subject. The following
Japanese examples are modeled after Mahajan’s work, and similar exam-
Unifying A-Movements 61

ples are discussed by Hoji (1985), Saito (1992), Tada (1993), and Yoshi-
mura (1989, 1992). As shown, ‘‘A’’ scrambling can suppress a WCO
violation.
(6) a. *[Kinoo proi proj atta hitoi ]-ga dare-oj hihansita no?
yesterday met person-nom who-acc criticized q
Lit. ‘The person who met (him) yesterday criticized whom?’
b. Dare-oj [kinoo proi proj atta hitoi ]-ga tj hihansita no?
who-acc yesterday met person-nom criticized q
Lit. ‘Who, the person who met (him) yesterday criticized?’
This type of scrambling can also create a new binder (Mahajan 1990,
Saito 1992).
(7) a. *Otagaii -no sensei-ga [Taroo-to Hanako]i -o
each.other-gen teacher-nom Taro-and Hanako-acc
suisensita.
recommended
‘Each other’s teachers recommended Taro and Hanaka.’
b. Taroo-to Hanako-oi otagai-no sensei-ga ti
Taro-and Hanako-acc each.other-gen teacher-nom
suisensita.
recommended
‘Taro and Hanako, each other’s teachers recommended.’
Unlike this kind of local scrambling, long-distance scrambling has solely
Ā properties, so that it is unable to suppress a WCO violation and cannot
create a new binder (Mahajan 1990, Saito 1992, Tada 1993, Yoshimura
1989, 1992).
(8) *Dare-oj [kinoo proi proj atta hitoi ]-ga
who-acc yesterday met person-nom
[Taroo-ga tj sitteiru to] itta no?
Taro-nom know c said q
Lit. ‘Who, the person who met (him) yesterday said that Taro
knows (him)?’
(9) ?*Taroo-to Hanako-oi otagai-no sensei-ga
Taro-and Hanako-acc each.other-gen teacher-nom
[koutyou-ga ti sikaru to] omotta.
principal-nom scold c thought
Lit. ‘Taro and Hanako, each other’s teachers thought that the
principal will scold.’
62 Chapter 3

There is no f-feature agreement in the relevant scrambling constructions

that would enable us to characterize the clearly ‘‘A’’ nature of these
movements in the same way that we portrayed English A-movements.
Another possible mechanism that Mahajan (1990) suggests is Case, which
works for English, where Spec,TP could be viewed as the position to
which nominative Case is assigned, though not necessarily. This, too,
will not work for A-scrambling, however.
I suggest that, as already noted in chapters 1 and 2, the cases where
scrambling can be characterized as A-movement involve a grammatical
feature that is computationally equivalent to f-feature agreement. Fol-
lowing the proposal of É. Kiss (1995, 1997, 2003), I assume that Japanese
is a discourse-configurational language and therefore that information-
structural notions such as topic and focus are grammaticalized and play
the same role in narrow syntax in Japanese that f-feature agreement
plays in agreement languages. In chapter 1, we saw that topic/focus and
f-features begin at C and that in Japanese-type languages, topic/focus
lowers to T. In this way, focus occupies the same position—T—as f-
feature agreement and attracts a topic/focus phrase to Spec,TP. A slight
change in the tentative definition of A-movement that I gave in (5) covers
the scrambling cases.
(10) A-movement (revised)
A grammatical feature on T targets the goal of the agreement for
movement.
In chapter 4, I will show that (10), although correct, is simply a descrip-
tion of one type of movement that I have been calling A-movement. As I
will show, (10) can be derived from considerations of chain formation
within a phase-based architecture of narrow syntax. But first, I need to
establish that topic/focus indeed triggers A-movement, which is the task
of this chapter. I begin with a discussion of focus.

3.2 Focus Movement as A-Movement to Spec,TP

Japanese has numerous expressions that are associated with focus. We

saw in chapter 1 that the indeterminate pronoun provides one piece of ev-
idence for focus movement to Spec,TP. In this section, I take up another
type of focus phrase: the XP-mo ‘XP-also’ expression studied by Hase-
gawa (1991, 1994, 2005) and Kuroda (1965, 1971).
Unifying A-Movements 63

3.2.1 The Japanese ‘Also’ Phrase

Hasegawa (2005) (see also Miyagawa 2007) has proposed that, just like
the indeterminate pronoun, the mo ‘also’ expression undergoes A-
movement, in most cases to Spec,TP.1 Here, I will give further evidence
that the movement is indeed A-movement.2 This will lay the groundwork
for showing that focus in Japanese is a grammatical feature that is com-
parable to f-feature agreement in agreement languages such as those of
Indo-European, as I argued in chapter 1, thereby further supporting the
analysis of focus as a feature relevant to syntactic movement (e.g., Brody
1990, Horvath 1981, 1986, 1995, É. Kiss 1995). Once I establish that this
focus movement takes place to Spec,TP, I will expand the analysis and
show that there is a second position, above TP and below CP, where
focus (or topic) may also end up. This projection, which I call aP, recalls
the analysis of Finnish by Holmberg and Nikanne (2002), of Hungarian
by É. Kiss (1995), and of Romance by Uriagereka (1995). These authors
propose that this position is a topic position (see also Saito 2006); and as
we will see, in Japanese, it can host a topic, but it can also host a focused
element under certain circumstances. Because of this flexibility, I propose
calling this projection simply aP as opposed to TopicP or FocusP. In
extending the ‘‘EPP’’ approach to scrambling (e.g., Kitahara 2002, Miya-
gawa 2001, 2003), Saito (2006) proposes this higher position for Japanese,
arguing that it is a topic position. In section 3.3, I will look in detail at
Saito’s proposal and incorporate its core idea into my analysis.
A mo expression carries focus stress, indicating that it is associated with
focus.
(11) a. Taroo-wa HON-o katta.
Taro-top book-acc bought
‘Taro bought a book.’
b. TAROO-mo hon-o katta.
Taro-also book-acc bought
‘Taro also bought a book.’
In (11a), which does not contain a mo expression, the sentence has neutral
intonation, with the object DP ‘book’ receiving default prominence. In
(11b), the stress falls not on the object but on the ‘also’ expression
Taroo-mo. Hasegawa (1991, 1994) notes that when it occurs with senten-
tial negation, the mo phrase is interpreted outside the scope of negation.
(12) a. John-mo ko-nakat-ta.
John-also come-neg-past
‘John (in addition to someone else) did not come.’
64 Chapter 3

b. John-ga hon-mo kaw-anakat-ta.

John-nom book-also buy-neg-past
‘A book is one of the things that John did not buy.’
Example (12a) only has the interpretation that there is at least one person
who did not come besides John. It does not mean that someone came, but
John didn’t come as well, which would be the interpretation if the mo
phrase were inside the scope of negation. Likewise, (12b) only means
that John did not buy at least two things, something and a book; it does
not mean that John bought something but not also a book.
Early on, Hasegawa (1991) described this phenomenon as the positive
polarity property of mo. According to Hasegawa, at LF the mo phrase
must be outside the scope of negation to stay true to its positive polarity
property. But this is not always true. As shown in (13), a mo phrase can
indeed occur inside the scope of negation if the negation is in the higher
clause.
(13) Taroo-ga [Hanako-ga suteeki-mo tabeta to] omotte-i-nai.
Taro-nom Hanako-nom steak-also ate c think-neg
O-susi-dake-da.
hon-sushi-only-cop
‘Taro doesn’t think that Hanako also ate steak. Just sushi.’
This example has the interpretation that Hanako ate sushi but not also
steak. This shows that mo is not necessarily a positive polarity item.
More recently, Hasegawa (2005) has revised her analysis of mo, arguing
that it involves a form of agreement (see also Hasegawa 1994). She sug-
gests that T in Japanese is associated with the EPP and that movement of
the mo phrase is EPP movement to Spec,TP.3 The mo phrase must raise
not at LF but in overt syntax. The mo phrase raises above negation,
which is between vP and TP (see Laka 1990, Pollock 1989). Tree dia-
grams (14) and (15) show subject and object mo phrases. In both cases,
the focus feature begins at C and is inherited by T. I represent just the
TP level here.
Unifying A-Movements 65

(14)

(15)

I follow Klima (1964) in assuming that for an element to occur in the

scope of negation, that element must be c-commanded by negation.
We see, then, that the focus on the mo phrase is targeted by the focus
feature on T. I will give evidence that the mo phrase in fact moves to
Spec,TP. One immediate observation that gives credence to this claim is
that this movement does not allow reconstruction. If it did, we would
expect the mo phrase to be interpretable inside the scope of sentential ne-
gation, something we have already seen is impossible. Lack of reconstruc-
tion is a common property of A-movement (see, e.g., Chomsky 1993, Fox
1999, Lasnik 1999a, Miyagawa and Arikawa 2007), and although there is
66 Chapter 3

debate about whether all instances of A-movement disallow reconstruc-

tion, the fact that the movement of mo does not allow reconstruction is a
safe indication that it has undergone A-movement. It is an accepted fact
that movement to Spec,TP is A-movement, which provides another indi-
cation that mo moves to Spec,TP, and this movement is triggered by the
grammatical feature of focus on T.4
The picture that emerges is the following. In languages that have f-
feature agreement, it is typically the case that the goal of the f-feature
agreement is targeted by the agreement on T (or the EPP requirement on
T in the traditional approach) and raised to Spec,TP. If there is an exter-
nal argument, it is this external argument that functions as the goal of f-
feature agreement, and this goal is raised to Spec,TP. We thus have the
following observation, repeated from (10).
(16) A-movement
A grammatical feature on T targets the goal of the agreement for
movement.
As our discussion of focus in Japanese showed, focus functions in this
language exactly the same way that f-feature agreement does in agree-
ment languages. This, then, is a case where focus functions as a grammat-
ical feature, an idea found in numerous works on focus (e.g., Brody 1990,
Horvath 1981, 1986, 1995, É. Kiss 1995). Although some linguists argue
that the focus feature is in the CP domain (see, e.g., Culicover and Roche-
mont 1983), I assume that, just like f-features in agreement languages,
this focus is inherited by T in the cases we observed in Japanese. In a re-
lated point, Horvath (1995) argues that in Hungarian, the focus feature
appears on T.
If the movement of mo raises it to Spec,TP, it is A-movement, not Ā-
movement. I will give two arguments that the mo phrase undergoes overt
A-movement. First, as already noted, A-movement scrambling may over-
come a WCO violation (Mahajan 1990, Saito 1992, Tada 1993, Yoshi-
mura 1989). Consider (17). Example (17a) is a typical WCO violation,
and (17b) is a typical case of WCO suppression by A-scrambling.
(17) a. ?*[Sakihodo ei ej yonda hitoi ]-ga futatu-izyou-no
[ just.now read person]-nom two-more.than-gen
meiwaku meeruj -o kesita.
spam mail-acc deleted
‘The person who read them just now deleted more than two
pieces of spam mail.’
Unifying A-Movements 67

b. Futatu-izyou-no meiwaku meeruj -o [sakihodo ei ej yonda

two-more.than-gen spam mail-acc [ just.now read
hitoi ]-ga tj kesita.
person]-nom deleted
Lit. ‘More than two pieces of spam mail, the person who read
them just now deleted.’
Now note that the WCO violation is suppressed even if the fronted phrase
contains mo.
(18) Futatu-izyou-no meiwaku meeruj -mo [sakihodo ei ej yonda
two-more.than-gen spam mail-also [ just.now read
hitoi ]-ga tj kesita.
person]-nom deleted
Lit. ‘More than two pieces of spam mail also, the person who read
them just now deleted.’
This example demonstrates that the mo phrase has undergone A-
movement. We can also see that a mo phrase, when locally scrambled,
may function as a new binder for the reciprocal ‘each other’.
(19) a. Taroo-to Hanako-oi otagai-no sensei-ga ti
Taro-and Hanako-acc each.other-gen teacher-nom
suisensita.
recommended
‘Taro and Hanako, each other’s teachers recommended.’
b. Taroo-to Hanako-moi otagai-no sensei-ga ti
Taro-and Hanako-also each.other-gen teacher-nom
suisensita.
recommended
Lit. ‘Taro and Hanako also, each other’s teachers
recommended.’
We thus have converging evidence that local movement of the mo phrase
is A-movement. I will assume that this movement raises mo to Spec,TP
(Miyagawa 2005b, 2007; see also Hasegawa 2005), although I will show
that Spec,TP is not the only landing site for A-movement.
A consequence of this analysis is that it predicts that any XP to the left
of the mo phrase must have moved there by scrambling. This is shown in
(20), where, according to the proposed analysis, the subject has under-
gone scrambling to the left of the mo object phrase.
68 Chapter 3

(20) John-ga [TP hon-mo [vP tSubj [VP tObj kaw-anaka-]] ta]
John-nom book-also buy-neg-past
!

!
scrambling
A-scrambling
‘A book is one of the things John did not buy.’
Saito (1985) argues that subjects do not scramble. However, Ko (2007)
argues that the subject can, in fact, scramble in Japanese and Korean. If
the analysis outlined here is correct, it supports Ko’s proposal. Miyagawa
and Arikawa (2007) provide further evidence for subject scrambling in
Japanese based on the distribution of floating numeral quantifiers.
In fact, we can use an example similar to the ones in Miyagawa and
Arikawa 2007 as evidence for the analysis given above. In the standard
analysis of floating numeral quantifiers, a subject separated from its float-
ing quantifier by the object is typically unacceptable (Haig 1980, Kuroda
1980; see Miyagawa and Arikawa 2007 for further discussion).
(21) *Gakusei-ga uisukii-o futa-ri nonda.
student-nom whiskey-acc two-cl drank
‘Two students drank whiskey.’
The subject floating numeral quantifier is inside the VP, where it cannot
be construed with the subject. Now note the following example:
(22) ?Gakusei-ga uisukii-mo futa-ri nonda.
student-nom whiskey-also two-cl drank
‘Two students also drank whiskey.’
Although not perfect, this sentence with object mo shows marked im-
provement over (20), at least for those speakers who accept this kind of
‘‘nonstandard’’ case (see Miyagawa and Arikawa 2007 for discussion of
the nonstandard cases, which are typically judged to be essentially gram-
matical as reported in the literature). This is an indication that the object
has moved to its left and that the subject, too, has moved to the left of the
object. Miyagawa and Arikawa present an analysis in which the object in
these nonstandard cases of floating numeral quantifiers occupies Spec,TP,
and the subject to its left has undergone Ā-movement. As Miyagawa and
Arikawa note, even the ‘‘ungrammatical’’ (21) can be made to sound bet-
ter with a pause between the object and the subject numeral quantifier,
indicating that ‘‘double’’ scrambling is possible even if the subject is not
marked by mo, which is what we expect.
Unifying A-Movements 69

I will depart from Miyagawa and Arikawa 2007 in one respect. Note
that, if Miyagawa and Arikawa are correct that the position to which
the subject scrambles in these cases is an Ā-position, we would expect a
WCO violation, and we also would not expect the subject to be able to
function as a binder (see Hoji and Ishii 2004, Miyamoto and Sugimura
2005). As it turns out, neither of these predictions holds.
(23) Darei -ga [mukasi proi proj hihansita hitoj ]-mo tj
who-nom long.time.ago criticized person-also
sukininatta no?
came.to.like q
Lit. ‘Who, the person who (he) criticized a long time ago also came
to like?’
(24) Hanakoi -ga zibun-zisini -mo ti hihansita.
Hanako-nom self-also criticized
‘Hanako also criticized herself.’
As we can see, the scrambled subject suppresses a WCO violation, and it
can create a new binder, both pointing to the fact that this subject is in an
A-position. Miyamoto and Sugimura (2005), in criticizing an earlier ver-
sion of Miyagawa and Arikawa 2007, argue that the scrambled subject is
not in an Ā-position but rather in the major subject position (a major
subject being an ‘‘additional’’ subject marked by the nominative that
appears higher in the structure than the normal subject; see Kuno 1973).
Contrary to Miyamoto and Sugimura’s proposal, however, the A-
position above Spec,TP is not the major subject position, as the following
examples show:
(25) a. *[ei ei sukizya-nai hitoj ]-mo darej -ni atta no?
like-not person-also who-dat met q
Lit. ‘The person who doesn’t like (him) met who?’
b. Darej -ni [ei ei sukizya-nai hitoj ]-mo tj atta no?
who-dat like-not person-also met q
Lit. ‘Who, the person who doesn’t like (him) met?’
Here, what occurs in the higher A-position is a dative phrase. Because
there is no such a thing as a ‘‘major dative,’’ the higher A-position must
be something other than the major subject position.5

3.2.2 aP
The evidence clearly points to the existence of an A-position above TP
and below CP. I will adopt a version of a proposal by Saito (2006), who
70 Chapter 3

suggests such a projection. He calls it the ‘‘Theme’’ projection, where

theme refers to what I am calling topic. I will use the more common term
topic, with the understanding that as I am using it, the term does not nec-
essarily refer to a discourse topic; rather, the topic can simply be the
‘‘topic’’ of the sentence—part of a categorical expression as opposed to a
thetic expression in Kuroda’s (1972–1973) terms. This way of using topic
is consonant with Saito’s usage of theme. In the examples above, the
phrase at the head of the sentence, and to the left of the mo phrase, is
most naturally understood as the topic of the sentence. However, I will
revise Saito’s proposal and suggest that there is simply a projection, aP,
that at times serves to host a topic but at other times may host a focused
element. In chapter 4, we will see that a may even host f-feature agree-
ment in some languages. This aP occurs as needed, and if it does occur,
the head a hosts the syntactic feature of topic/focus that it inherits from
C, making Spec,aP an A-position. If the clause contains just one focus or
topic, aP need not occur, and if it does not, the topic/focus feature is
inherited by T. There may also be more than one aP, again, if needed.
The structure of Japanese when there are two relevant elements—topic
and focus, for example—is as shown in (26).
(26)

The idea that there is a projection between TP and CP is certainly not

novel. For example, in Kikuyu, a Bantu language, this is the position
where focused elements occur (Horvath 1995:41); and it is the position
that hosts a topic in Finnish (Holmberg and Nikanne 2002) and in Ro-
mance (Uriagereka 1995).
Unifying A-Movements 71

For the examples with the mo phrase in Spec,TP and the topic in the
higher position, I suggest the structure in (27).
(27)

The identification of a with topic is, in this instance, an adoption of

Saito’s (2006) proposal. However, as we will see, a may host a focused
phrase. I have assumed that focus is a syntactic feature, as proposed by
Horvath (1981, 1986) and further developed by Brody (1990). I also take
from Horvath’s work the idea that the focus feature may be assigned to
di¤erent heads, an idea that departs from Brody 1990 and much other re-
search that assumes a special focus head. I will propose that the focus fea-
ture—or, more precisely, a probe that turns into the focus feature, as I
will show later—always merges initially on C, and it can stay there, or it
can be inherited by a lower head, T (or a if a occurs). What gets passed
down to T (or a) forms a parameter that I wish to exploit to unify all
A-movements—the ‘‘EPP’’ movement and A-scrambling. One question
is, what makes T and a a target for A-movement in a way that C is not?
I attempt to answer this question in chapter 4 when I discuss A- and Ā-
movements. In what follows, I will give further evidence for the structure
in (27).

3.2.3 Nominative Object

Many stative transitive predicates in Japanese allow the object to be
marked with nominative instead of accusative case.
72 Chapter 3

(28) Taroo-ga eigo-o/-ga hanas-e-ru.

Taro-nom English-acc/-nom speak-can-pres
‘Taro can speak English.’
It has been observed that the choice of case marking on the object leads
to a di¤erence in scope (Sano 1985, Tada 1992; see also Koizumi 1995,
2008).
(29) a. Kiyomi-wa migime-dake-o tumur-e-ru.
Kiyomi-top right.eye-only-acc close-can-pres
Lit. ‘Kiyomi can just close his right eye.’
[can > only] ‘Kiyomi can wink his right eye.’
b. Kiyomi-wa migime-dake-ga tumur-e-ru.
Kiyomi-top right.eye-only-nom close-can-pres
[only > can] ‘It is only the right eye that Kiyomi can close.’
In (29a), the accusative object is interpreted below the verb ‘can close’, so
this sentence means that Kiyomi has the ability to just close his right eye,
thus the translation ‘wink’. In contrast, the nominative object in (29b) has
the opposite scope, where the object is interpreted higher than the verb,
which gives the meaning that Kiyomi can only close his right eye. Tada
(1993) and Koizumi (1995) relate the wide scope reading of the nomina-
tive object to Case: the nominative Case on the object must raise to a po-
sition high in the structure, possibly to the region of T, where it can be
licensed. But this would be an instance of forced movement strictly due
to Case, something that should not occur, as we saw in chapter 1. A point
not noted before is that the nominative object, as opposed to the accusa-
tive, tends to be focused, and this focus is further enhanced in these exam-
ples by the occurrence of -dake ‘only’. Given what we have already seen,
the nominative object with focus and the nominative subject, presumably
as topic, both undergo A-movement (Miyagawa 2001), the trigger being
the grammatical features of focus and topic, respectively. This is shown in
the tree diagram in (30).
Unifying A-Movements 73

(30)

This analysis makes a clear prediction that distinguishes it from earlier

Case-based analyses: namely, that if the context allows for defocusing of
the nominative object, it should be possible to interpret it in its original
object position. Indeed, Nomura (2005) presents examples of precisely
this nature (see also Koizumi 2008). Thus, we can assume that the wide
scope interpretation of the nominative is focus-related and not due to
Case.
When the nominative object does raise, it moves into an A-position—
Spec,TP, I assume—because it does not reconstruct. What about the sub-
ject? It, too, occupies an A-position, as shown in (31) by the fact that it
can bind an anaphor.
(31) Tarooi -ga zibun-zisini -no sensei-ga hihan-deki-nakat-ta.
Taro-nom self-gen teacher-nom criticize-can-neg-past
‘Taro was unable to criticize his own teacher.’
The relevant portion of the structure for (31) is as follows.
(32) [CP [aP Taroo-gai [TP zibun-zisini -no sensei-ga
Taro-nom self-gen teacher-nom
[vP ti . . . tj . . . ] Tfocus ] atopic ] C]
Later I will give an explanation of how focus and topic emerge.
74 Chapter 3

3.3 Topic Movement

3.3.1 Scrambling as Topicalization

I now turn to the other discourse-configurational feature—topic. As men-
tioned earlier, topic as I am using it refers to the element, usually at the
left edge, that represents what the sentence is about. It can be the dis-
course topic, but it need not be. Japanese has a discourse topic marker
wa, which typically occurs on a phrase at the left edge of the sentence.
(33) Taroo-wa piza-o tabeta.
Taro-top pizza-acc ate
‘As for Taro, he ate pizza.’
The wa phrase is outside TP, in the CP region (Kuno 1973, Kuroda 1965;
see Kishimoto 2006 for evidence that the wa phrase is in the CP domain).
Wa marks the discourse topic, but the construction I am concerned with
involves scrambling of a phrase to the left edge of a sentence, as in (34).
(34) Piza-oi Taroo-ga ti tabeta.
pizza-acc Taro-nom ate
‘Pizza, Taro ate.’
I will argue that one instance of this local scrambling is due to the topic
feature on the head whose specifier is occupied by ‘pizza’.
To get started, let us review the evidence from earlier studies for the
proposal that A-scrambling is triggered by an independent EPP feature on
T. Since we now assume that there is no independent EPP feature, but that
instead, movement is triggered in conjunction with some grammatical fea-
ture, we need to search for what we can take from the earlier ‘‘EPP’’ analysis
and what parts of that analysis we can set aside. I will first present the argu-
ment for the earlier analysis, then point out some problems, and finally
propose a new analysis based on ‘‘topic’’ (and ‘‘focus’’) as a feature that
triggers this ‘‘EPP’’ movement, adopting an idea suggested by Saito (2006).
The test given in Miyagawa 2001 (see also Miyagawa 2003) to argue
for the ‘‘EPP’’ basis of A-scrambling involves the Japanese universal
quantifier zen’in ‘all’ and its interpretation relative to sentential negation.
As shown in (35), zen’in ‘all’ in object position may have the partial nega-
tion interpretation of ‘not all’. (The other reading, ‘all > not’, is probably
due to a collective reading of ‘all’.)
(35) Taroo-ga zen’in-o sikar-anakat-ta.
Taro-nom all-acc scold-neg-past
‘Taro didn’t scold all.’
not > all (all > not)
Unifying A-Movements 75

As noted by Kato (1988), when a universal expression is in subject

position, it is interpreted outside the scope of negation (with neutral
intonation).
(36) Zen’in-ga siken-o uke-nakat-ta.
all-nom test-acc take-neg-past
‘All did not take the test.’
*not > all, all > not
In contrast, as noted in Miyagawa 2001, if the object is scrambled,
the subject universal expression may be interpreted inside the scope of
negation.
(37) Siken-oi zen’in-ga ti uke-nakat-ta.
test-acc all-nom take-neg-past
‘All didn’t take the test.’
not > all, all > not
Thus, when ‘all’ is in subject position in the SOV order, as in (36), the
preferred reading is ‘all > not’. When the object is scrambled to the left
edge, as in (37), partial negation becomes possible.
How does negation take scope over ‘all’ to achieve the partial nega-
tion interpretation? In (36), where the subject ‘all’ is outside the scope of
negation, ‘all’ begins in Spec,vP and moves to a position outside the c-
command domain of negation. A reasonable assumption is that it moves
to Spec,TP as shown in (38). (The position of negation in (38) is roughly
as proposed by Laka (1990) and Pollock (1989).)
(38)
76 Chapter 3

In (38), where the subject ‘all’ occurs in the scrambled order OSV, it is
able to be interpreted inside the scope of negation. The simplest assump-
tion to make here is that this subject ‘all’ stays in situ in Spec,vP, a situa-
tion made possible by the movement of the object to Spec,TP.
(39)

These structures suggest that something—subject, object—raises to

Spec,TP, even though the raised XP is not focused, or need not be (Miya-
gawa 2001; see also Miyagawa 2003). If the subject moves to Spec,TP,
the object stays in situ, as in (38). On the other hand, if the object moves
to Spec,TP, this allows the subject to stay in situ, as shown in (39).6 Al-
though in (39) I only represent the movement from inside VP to Spec,TP,
if we are to follow the requirements imposed by the notion of phases, the
object must first move and adjoin to vP, a point I will return to at the end
of this chapter. Finally, the OSV order with the universal in subject posi-
tion has another interpretation in which the subject takes wide scope over
negation. This results from a derivation in which the subject first moves
to Spec,TP and the object then moves to a higher position. I will return
to both of these derivations later.

3.3.2 The Kumamoto Dialect of Japanese

There is dramatic evidence from a recent study of the Kumamoto dialect
of Japanese by Kato (2007) for the type of ‘‘EPP’’ analysis I gave above.
Unlike standard Japanese, this dialect, which is spoken on the southern
island of Kyushu, distinguishes between two types of nominative case
marking.
Unifying A-Movements 77

(40) Nominative marking

Standard Japanese: ga
Kumamoto: ga, no
(Yoshimura 1994)
According to Kato (2007), the two kinds of nominative marking occur in
di¤erent syntactic environments. Ga occurs as the nominative marker
outside vP, and no as the nominative marker inside vP. We can see this
in multiple-nominative constructions like (41), from Yoshimura 1994:20.
(41) Kumamoto-ga baniku-no umaka.
Kumamoto-nom horse.meat-no tasty
‘It is Kumamoto where horse meat tastes good.’
However, in a normal transitive construction, in the SOV order the sub-
ject must have ga.7
(42) a. Taroo-ga sakana-ba tabeta-bai.
Taro-nom fish-acc ate-final.particle
‘Taro ate fish.’
b. ?*Taroo-no sakana-ba tabeta-bai.
Taro-no fish-acc ate-final.particle
This is consistent if we assume that the subject in (42a) necessarily under-
goes ‘‘EPP’’ movement to Spec,TP, where it is outside vP, hence must
be marked by ga. Kato (2007:122) further notes the following crucial
example:
(43) Sakana-ba Taroo-no tabeta-bai.
fish-acc Taro-no ate-final.particle
Here, the object has scrambled to the head of the sentence. Crucially,
now it is possible for the subject to be marked with the ‘‘vP’’ nominative
no. This indicates that because of the scrambling of the object, the subject
may stay in situ in Spec,vP. This, again, shows that if something occurs in
Spec,TP, everything else, including the subject, may stay in situ—a typi-
cal EPP e¤ect. Finally, Kato notes that in a variant of (43), the subject
may be marked with ga as well.
(44) Sakana-ba Taroo-ga tabeta-bai.
fish-acc Taro-nom ate-final.particle
This is the derivation in which the subject first moves to Spec,TP, so that
it receives the nominative marking ga. The object then undergoes move-
ment to a position above this Spec,TP.
78 Chapter 3

Finally, recall that the focus phrase mo moves to Spec,TP.

(45) a. Taroo-mo hon-o katta.
Taro-also book-acc bought
‘Taro also bought a book.’
b. Taroo-ga hon-mo katta.
Taro-nom book-also bought
‘Taro bought a book, too.’
Earlier, I gave evidence based on Hasegawa’s (2005) analysis that the mo
phrase in both of these examples moves to Spec,TP overtly. This proposal
in turn predicts that, if the mo object phrase in (45b) scrambles to the left
of the subject, it moves to Spec,TP. This prediction is confirmed by the
Kumamoto dialect (thanks to Sachiko Kato for providing the examples).
(46) a. Inu-ga booru-mo kuwaetekita.
dog-nom ball-also chewed
‘The dog also chewed a ball.’
b. Booru-mo inu-no kuwaetekita.
ball-also dog-no chewed
‘The ball, the dog also chewed.’
As shown in (46b), scrambling the mo phrase makes it possible for the
subject ‘dog’ to be marked by no. It can also be marked by ga, but the
important point is that no is possible.

3.3.3 A Problem with the EPP Analysis

The EPP analysis of A-scrambling is couched in the earlier notion of
the EPP—that every T is associated with the EPP property (Alexiadou
and Anagnostopoulou 1998, Chomsky 1981, 1995). However, as we saw
in chapter 1, this is simply false; so-called EPP movement only emerges
when some relevant grammatical feature is present, typically f-feature
agreement (e.g., Chomsky 2000, 2001, 2005, Kuroda 1988, Miyagawa
2005b). If it is true that in languages such as Japanese, topic/focus consti-
tutes a grammatical feature equivalent to f-feature agreement, we would
expect ‘‘EPP’’ movement to occur only in the presence of topic/focus, not
every time the structure contains a T. In other words, there is no such
thing as ‘‘EPP’’ independent of relevant grammatical features (Miyagawa
2005b).
Indeed, a problem with applying the traditional EPP to scrambling is
noted in Miyagawa 2001. There, I point out that, contrary to Kato’s
(1988) observation, the universal quantifier in subject position even in
Unifying A-Movements 79

SOV order can take scope inside negation when put in an embedded con-
text. The following example, taken from Saito 2006, demonstrates the
same point:
(47) Zen’in-ga siken-o erab-ana-i to omou.
all-nom exam-acc choose-neg-pres c think
‘I think that all will not choose an exam (over a term paper).’
all > not, not > all
Saito concludes that the ‘‘EPP’’ e¤ects observed in Miyagawa 2001, 2003
are what he calls ‘‘interpretational’’ e¤ects that commonly appear in root
clauses. In his approach, this is a left-edge e¤ect of cartography as studied
by Rizzi (1997, 2004) and others. I have already shown that focus triggers
movement in Japanese to Spec,TP. I will extend this idea and adopt
Saito’s proposal that the EPP movement identified in Miyagawa 2001,
2003 and other works is a function of ‘‘topic,’’ thereby unifying move-
ment in Japanese under what É. Kiss (1995) has called discourse configu-
rationality. Thus, movement of the subject or the object, or even some
other element, in A-scrambling is a form of movement triggered by a
grammatical feature, which in Japanese happens to be either topic or fo-
cus, given that Japanese is a discourse-configurational language. If such a
feature occurs, as it usually does in the root clause, it results in the type of
EPP e¤ects we saw earlier. Thus, in SOV or OSV word order, the left-
most element (S or O) has undergone movement due to a grammatical
feature with an ‘‘interpretational’’ e¤ect, most likely ‘‘topic’’ in a broad
sense, although focus might play a role in certain cases. In the subordi-
nate clause, such an interpretational e¤ect typically does not arise because
topic is less apt to appear in such a context. The situation would be di¤er-
ent if focus were present, of course.
On this discourse-configurational view, the phrases on the left edge in
(36) and (37), repeated here, carry some information-structural meaning.
(48) Zen’in-ga siken-o uke-nakat-ta.
all-nom test-acc take-neg-past
‘All did not take the test.’
(49) Siken-oi zen’in-ga ti uke-nakat-ta.
test-acc all-nom take-neg-past
‘All didn’t take the test.’
not > all, all > not
The subject ‘all’ in (48) and the scrambled object ‘test’ in (49) are both
topics—they both represent what the sentence is about. It is possible
80 Chapter 3

with an identificational (‘‘narrow’’) focus intonation that ‘all’ in (48)

could alternatively be interpreted as focus, but I will keep to the neutral
intonation. The question is, where is this ‘‘topic’’ in (48) and (49)?
Let us look at Saito’s (2006) analysis of these sentences. Saito proposes
that there is a projection above TP, which he calls ThP (ThemeP). (This
is where I postulate instead the category aP.) In essence, his analysis
captures at this Theme level what I earlier suggested for the TP level
(Miyagawa 2001, 2003): either the subject or the object may occupy
Spec,ThP, allowing the other—object or subject—to stay inside the scope
of negation. The structure Saito suggests for (48), where ‘all’ in the SOV
order takes scope over sentential negation, is shown in (50).
(50)

That is, the subject first moves to Spec,TP, apparently because of the EPP
requirement on T (see Saito’s example (25)). It then moves to Spec,ThP
and gets interpreted as the topic of the sentence. An argument Saito gives
to explain why ‘all’ has to move to Spec,ThP to escape the scope of nega-
tion has to do with the fact that in English, a universal quantifier in
Spec,TP is known to be able to take scope inside negation.
(51) Everyone had not left the party. There were still people talking and
drinking.
As Saito suggests, topic is a left-edge e¤ect, and it does not, or need not,
occur in subordinate structures. This explains the fact observed in Miya-
gawa 2001 that in subordinate clauses, the universal in subject position
may take scope inside negation even in SOV order. I will adopt Saito’s
idea that topic does not, or need not, occur in subordinate structures pre-
cisely for the reason he gives, that it is a left-edge e¤ect. However, I will
depart from his analysis in not identifying topic with a particular projec-
tion. Rather, I will suggest that topic as I am using the term may occur
Unifying A-Movements 81

on one head or another—T or a—depending on other factors in the

structure.
Although the fact illustrated in (51) is all that Saito notes about En-
glish, there is, in fact, evidence that something comparable to ThP exists
even in English. On the basis of comparing English with Italian and
French, Belletti (1990) suggests that a subject that precedes a sentential
adverb, or something akin to it, is in a topicalized position (in Belletti’s
framework, this means an adjunction position, but that does not concern
us here).
(52) John probably/unfortunately has already talked to Mary.
In a related point, in Miyagawa 1993, I gave evidence that the univer-
sal quantifier in subject position in English cannot take scope inside nega-
tion if it is separated from the rest of the sentence (Everyone has been not
turning in their papers). A more appropriate example for the issue at hand
is this:
(53) Everyone probably/unfortunately/as far as I know has not done the
homework.
Here it is di‰cult, if not impossible, to interpret the universal inside the
scope of negation. This, then, is empirical evidence for Saito’s suggestion
that at least in English, the universal in Spec,TP can be in the scope of
sentential negation. But does the same hold in Japanese? I will return to
this question after completing the description of Saito’s proposal. For the
scrambled order in (49), in which the object has moved to sentence-initial
position across the subject, Saito proposes that the initial scrambling is
adjunction to TP (see also Saito 1985).
(54)
82 Chapter 3

Saito suggests two possible derivations besides (54). In one option, the
object moves to Spec,ThP, and the subject ‘all’ stays in Spec,TP and gets
interpreted inside the scope of negation. If, however, the object does not
move to Spec,ThP, it reconstructs at LF to its original object position.
This opens the way for the subject ‘all’ to move to Spec,ThP at LF, giving
the other possible reading where ‘all’ takes scope over negation.
Saito assumes that Spec,TP is within the scope of negation, basing his
assumption on the English example in (51) (Everyone had not left the
party . . .). There is a second way to view the lack of reconstruction in En-
glish and Japanese. In both languages, topic movement (and presumably
also focus movement) to Spec,TP or Spec,aP does not reconstruct. In
English, the topic may move only to aP, since Spec,TP is reserved for f-
feature agreement, but in Japanese, the topic may move either to Spec,TP
or to Spec,aP. I will assume this alternative view of ‘‘lack of reconstruc-
tion.’’ As one piece of evidence, recall that in the double-nominative con-
struction, the second nominative, which is the object, is normally focused
and occupies Spec,TP. As noted earlier, the nominative object is typically
interpreted high in the structure; thus, it does not reconstruct if moved
to Spec,TP. Sentence (55) provides another example of this (Miyagawa
2001).
(55) Taroo-ga zen’in-ga home-rare-nakat-ta.
Taro-nom all-nom praise-can-neg-past
‘Taro was not able to praise all.’
all > not, *not > all
The nominative object in Spec,TP takes scope outside of negation. To en-
sure that we are dealing with the relevant structure, we need to verify that
the nominative object is in Spec,TP and the nominative subject is in the
A-position above it. Example (56) gives evidence for this.
(56) Taroo-to Hanakoi -ga [otagaii -no zen’in-no sensei]-ga
Taro-and Hanako-nom each.other-gen all-gen teachers-nom
home-rare-nakat-ta.
praise-can-neg-past
‘Taro and Hanako could not praise every one of each other’s
teachers.’
It is possible to interpret the nominative object outside the scope of nega-
tion, which, after all, is the more natural interpretation, and at the same
time understand the reciprocal to refer to ‘Taro and Hanako’, indicating
that ‘Taro and Hanako’ is in an A-position—that is, in Spec,aP.
Unifying A-Movements 83

Saito suggests that in the OSV order, the nominative subject moves to
Spec,TP because of the EPP feature on T (Saito’s example (30)). How-
ever, with the ‘‘agreement’’ approach to the EPP that I am assuming, un-
less the subject is focused or is a topic, there is no reason for it to move to
Spec,TP; the EPP e¤ect only arises if some relevant grammatical feature
is present. I will assume that if a sentence contains just one topic, and no
focus, its structure has no aP (Saito’s ThP) and the topic feature occurs
directly on T, having been inherited from C (a point I take up below).
Tree (57) illustrates this configuration.
(57)

This structure contains just one grammatical feature, topic, and it raises
the object to Spec,TP. (Later I will argue that topic and focus are the
same feature, di¤erentiated essentially by the context in which they
occur.) The subject stays in situ in Spec,vP, where it can be interpreted
inside the scope of negation. In the other possible scope situation with the
same OSV word order, in which the subject takes scope outside negation,
the subject is raised to Spec,TP either as topic or as focus. The object may
be raised to aP, an instance of A-movement, or to a higher position, pos-
sibly Spec,CP—an Ā-position.
Finally, if a sentence contains two elements—some combination of
topic and/or focus—aP is projected. We expect that Spec,aP can host a
focused phrase as well as a topic. Sentence (58) illustrates this situation:
it contains a mo phrase followed by a scrambled phrase that can antecede
an anaphor. This scrambled phrase is not associated with focus in the de-
fault intonation; hence, we can assume that it is a topic.
(58) John-ni-moi Taroo-to Hanako-oj [otagaij -no tomodati]-ga
John-dat-also Taro-and Hanako-acc each.other-gen friends-nom
ti tj syookaisita.
introduced
84 Chapter 3

Lit. ‘To John also, Taro and Hanako, each other’s friends
introduced.’
(59) [aP¼focus John-ni-mo [TP¼topic Taroo-to Hanako-o
John-dat-also Taro-and Hanako-acc
[vP . . . ] Ttopic ] afocus ]
Next, I will discuss the feature relevant to topic and focus movement.8

3.4 Assigning Topic and Focus

In this section, I will develop an analysis of topic and focus consistent

with the idea that they are computationally equivalent to f-feature agree-
ment. When inherited by T (or in certain circumstances, a), either of these
features triggers A-movement. The idea that topic and focus play a syn-
tactic role in some languages is not new. In her work on discourse config-
urationality, É. Kiss identifies these two as being expressed in the syntax
of languages such as Hungarian (É. Kiss 1995, 1997, 2003). In the carto-
graphic approach to linguistic structure (e.g., Rizzi 1997, 2004), topic and
focus are part of the articulated CP system. The topic and focus that we
have been discussing start at C, and in the Japanese-type languages they
are inherited by T. This picture di¤ers from most treatments of topic and
focus in the literature. In É. Kiss’s and Rizzi’s work, topic and focus typ-
ically are each associated with a special head with the relevant property,
Topic or Focus (on the latter, see also Brody 1990). There are exceptions
to this treatment of topic and focus as being associated with topic and
focus heads. One is Maki, Kaiser, and Ochi’s (1999) proposal that the
topic feature begins at T and raises to C. Another is Horvath’s (1995)
argument that focus in Hungarian occurs as a grammatical feature on T,
and not on a special focus head.
A number of questions arise when we view topic and focus as syntactic
features. First, what precisely is the feature for each? Second, why is it
that one feature does not interfere with the other? If they did, we would
expect either that only one of them could show up within a given clause,
or that the two would be in a highly restrictive ordering relation. Neither
is the case: topic and focus can both occur in the same clause, and the
ordering of phrases associated with each appears to be free both before
and after movement (they do need to be in the local domain for A-
movement). An obvious answer is that topic and focus are two distinct
features, inherently already valued as such, and because of this, they do
Unifying A-Movements 85

not interfere with each other. That seems reasonable, and much of the lit-
erature on the issue assumes this. However, by simply separating topic
and focus, we are potentially missing some important insights. As a pre-
lude to showing this, I will briefly discuss Holmberg and Nikanne’s (2002)
proposal about the notion of topic, which I will adopt and extend.
Finnish has a topic position above TP and below CP. The following
examples are from Holmberg and Nikanne (H&N) 2002:78:
(60) a. Graham Greene on kirjoittanut tämän kirjan.
Graham Greene has written this book
‘Graham Greene has written this book.’
b. Tämän kirjan on kirjoittanut Graham Greene.
this book has written Graham Greene
The two examples mean essentially the same, although, as H&N note,
(60b) most naturally translates into a passive sentence in English (This
book was written by Graham Greene). H&N propose that in both exam-
ples, the phrase at the head of the sentence (‘Graham Greene’ in (60a),
‘this book’ in (60b)) occurs in the topic position, which they label as FP
for Finite P, a term taken from cartography. To avoid confusion with
F(ocus)P, I will use the label FinP instead.
(61) a. [CP [FinP Graham Greene [TP . . . this book . . . ]]]
b. [CP [FinP this book [TP . . . Graham Greene . . . ]]]
H&N assume that the topic moves to Spec,FinP through the typical pro-
cess of agreement, and for this, they postulate the feature focus as the
feature for topic, which they assign to the Fin head. This focus on the
Fin head is comparable to the probe. Moreover, they postulate that
the same focus is automatically assigned to every phrase in the string,
unless a phrase has the þfocus feature to begin with. The ‘‘probe’’ focus
on Fin picks out one of these focus phrases and raises it to Spec,FinP.
The focus features on the remaining phrases are deleted.
Is topicalization always required in Finnish? With very few exceptions,
the focus feature apparently occurs in every sentence, but even with this
feature, topicalization does not take place if the ‘‘topic expletive’’ sitä
occupies Spec,FinP.
(62) Sitä ovat nämä lapset jo oppineet imaan.
expl have these children already learned to.swim
‘These children have already learned to swim.’
86 Chapter 3

As H&N note, and as (62) shows, the expletive is not related to any item
in the sentence. It is ‘‘a pure expletive. . . . [I]t has no f-features and is thus
not directly involved in any Case or agreement checking. It is also not a
placeholder for the subject. Its function is just to check the EPP feature
[of focus]’’ (H&N 2002:90). H&N suggest that the expletive sitä is asso-
ciated not with any f-feature (or Case), but with the ‘‘topic’’ feature
focus, although sitä is obviously not a topic. Focus requires some-
thing in its specifier, which is normally the topic, but the expletive may
occur instead, preventing topicalization from taking place. From the pres-
ent perspective, the occurrence of this expletive that is associated with
topic is further confirmation that topic (and, by implication, focus) func-
tions as a grammatical feature that triggers movement in some languages.
Finnish also has f-feature agreement—so here is a language that com-
bines both agreement and discourse-configurational properties, a point I
return to in chapter 4.
H&N’s system leads to two general observations.
(63) Generalizing H&N’s approach to Finnish topic, we can say that
a. topic is default, whereas focus is marked; and
b. topic is not uniquely associated with any particular phrase in the
structure.
Topic is default and focus is marked, in that the focus feature, which
represents topic, is automatically assigned to all phrases. The one excep-
tion to this is a phrase that is focused because it already has the þfocus
feature; þfocus is therefore marked. Turning to the second point, there
are two ways in which the topic is not associated with any particular
phrase in the structure. First, the ‘‘probe’’ focus on Fin can pick any
phrase that has focus; it just needs to pick one. Essentially, any phrase
that isn’t marked þfocus will do. Second, the ‘‘probe’’ focus can be sat-
isfied by inserting the ‘‘topic expletive’’ sitä, and this expletive is clearly
not associated with any associate NP in the structure. It simply fills in
the gap to satisfy the feature focus.
Given these observations, let us make the following assumption:
(64) Topic/Focus
The default feature for topic/focus is focus (topic).
For the marked property of focus, I propose the following character-
ization:
Unifying A-Movements 87

(65) Focus as a marked feature

If focus at C enters into agreement with a focused phrase
(þfocus), focus is valued as þfocus.
That is, the default focus undergoes valuation as þfocus under agree-
ment with a focused phrase. The focused phrase brings the þfocus feature
with it from the lexicon, as does the mo phrase in Japanese; or the þfocus
feature may simply be assigned to it at numeration. Finally, suppose the
following characterization for the idea that focus at C (the probe) is not
associated with any particular phrase in the structure, which is close to
what H&N propose:
(66) Topic as free movement
The focus topic feature, once inherited by T or a, simply requires
its specifier to be filled.
The focus feature that does not enter into an agreement with a focused
phrase will be inherited by T or a without picking out a goal. The only
thing this focus requires is that its specifier be filled, as we saw from
the ‘‘topic expletive’’ construction in Finnish. In H&N’s terminology,
focus in Finnish has the ‘‘pure’’ EPP feature. I assume this for all
instances of topic of the type we are dealing with in discourse-
configurational languages. Below, I give some sample derivations to illus-
trate the default, ‘‘pure’’ EPP nature of topics and the marked nature of
focus.
The simplest cases are those that contain just one topic or just one
focus. A one-topic sentence has the derivation in (67).
(67) One-topic sentence
a.
88 Chapter 3

b.

As shown, the focus feature that does not enter into an agreement (with
þfocus) is inherited by T from C. This focus feature on T simply
requires that its specifier be filled, which has the e¤ect of marking what-
ever fills this position as focus (i.e., as topic). The one exception is the
merging of the expletive sitä into this position in Finnish; this merger
plugs up the topic position, so that the sentence is without a topic.
A one-focus sentence has the derivation in (68).
(68) One-focus sentence
a.
Unifying A-Movements 89

b.

As shown in (68a), if focus at C enters into agreement with a focused

phrase, this focus is valued as þfocus by the goal, thus becoming a
marked feature.
What if there are two phrases, one topic and one focus? The proposed
analysis predicts that either order should be possible. The reason is that
while focus involves a probe-goal relation, topic does not. The derivation
for a topic-focus order is illustrated in (69).
(69) Two phrases: Topic and focus
a.
90 Chapter 3

b.

One of the focus features at C agrees with þfocus via agreement, which
values the focus as þfocus, and when þfocus is inherited by T, the goal
comes up to its specifier. The other focus feature is inherited by a, and
the phrase that moves to its specifier is given the attribution of topic. Be-
cause there is no probing by the focus feature, the features do not inter-
fere with each other, and as a result, phrases that end up as topic and
focus can occur in either order, both before and after movement. This
also holds true for two-topic structures, which are commonly found in
Japanese in double-scrambling constructions.
(70) Hanako-nii tegami-oj Taroo-ga ti tj okutta.
Hanako-dat letter-acc Taro-nom sent
Lit. ‘Hanako, letter, Taro sent.’
Finally, let us look at a two-focus construction. Here, phrases are
strictly ordered.
(71) a. Taroo-mo piza-mo tabeta.
Taro-also pizza-also ate
‘Taro also ate pizza, too.’
b. *Piza-mo Taroo-mo tabeta.
pizza-also Taro-also ate
Unifying A-Movements 91

Example (71b) can only have the improbable meaning that a pizza (and
something else) ate Taro, too. We can predict this on the assumption
made earlier that focus is marked, and that this markedness comes from
agreement. The fact that there is agreement means that there is a probe-
goal relation, and we would expect two identical features on two di¤erent
heads (T and a) to be impossible given Locality. This can only mean that
there is just one focus feature that enters into multiple agreement with
the two focus phrases. As Richards (2001) notes, when there is one fea-
ture that attracts two elements, the closer element (here, the subject
Taroo-mo) is attracted first, then the lower one (here, piza-mo) is tucked
in underneath the first, giving rise in this case to the ordering Taroo-mo
– piza-mo. The other ordering violates Locality as defined by Richards.
In contrast, for topics, no such ordering restriction exists. Thus, the
scrambled phrases in the double-scrambling example (70) could occur in
either order. This is expected because the ‘‘topic’’ focus feature does not
probe.

3.5 Summary

In this chapter, I extended the agreement–topic/focus parameter pro-

posed in chapter 1 and showed how this parameter can characterize A-
movements across languages, both A-movement of the typical ‘‘EPP’’
type found in English and other Indo-European languages and the A-
scrambling found in many scrambling languages. The idea is that topic/
focus is a grammatical feature in discourse-configurational languages
that functions in a manner equivalent to f-feature agreement in agree-
ment languages. When a grammatical feature (topic/focus, f-feature) is
inherited by T, the goal of the grammatical feature raises to T by A-
movement. I introduced a projection, aP, above TP and below CP; the a
head may also host a grammatical feature, and I showed how, when there
are two grammatical features—topic and topic/focus, for example—one
occurs on a and the other on T, and both give rise to A-movements.
Finally, I extended Holmberg and Nikanne’s (2002) treatment of topic
in Finnish, suggesting that ‘‘topic’’ and ‘‘focus’’ arise from the same fea-
ture. The default is ‘‘topic,’’ as indicated by the feature focus. If this fea-
ture does not enter into agreement with a goal, it occurs as focus on T
(or a). Any given element may undergo free movement to the specifier of
T (or a) with focus, and that element is marked as the topic of the sen-
tence. If focus enters into agreement with a focused element—that is, an
element bearing the feature þfocus—it is valued as þfocus, and inherited
92 Chapter 3

by T or a. This valued þfocus feature attracts the focused goal to its spec-
ifier. The di¤erence between topic (default) and focus (marked) in turn
explains the distribution of these items. Topics occur freely in either
order, and a topic can also occur with one focus, again in free order
(topic-focus or focus-topic). It is in two-focus structures where we find
an ordering restriction: a strict superiority e¤ect arises when one þfocus
enters into multiple agreement with two focus elements, and the lower
focus element must tuck in under the higher one.
I left one question unanswered: why is the head a an ‘‘A’’ head like T
and unlike C? I turn to this question in the next chapter.
 aP, f-Features, and the A/Ā Distinction
4

4.1 Introduction

In chapter 3, I proposed aP as a projection that occurs between TP and

CP, following similar proposals in the literature (e.g., Holmberg and
Nikanne 2002, É. Kiss 1995, Uriagereka 1995).
(1)

In a language such as Japanese, we saw that aP occurs in the root clause,

but need not occur in subordinate clauses.
In this chapter, I explore cases in which a inherits not only topic/focus
from C but also the f-probe, as diagrammed in (2).
94 Chapter 4

(2)

What will a language look like in this situation? First of all, given that the
f-probe is incapable of seeking its goal by itself, and a is not a Case
assigner, we predict that something other than the f-probe must be inher-
ited on the same head a to allow the f-probe to enter into an Agree rela-
tion. In other words, the fact that the f-probe is inherited in tandem with
–focus is not coincidental, but necessary. The –focus feature is responsi-
ble for locating the goal for the f-feature in the absence of Case. Suppose
that –focus itself has not entered into an Agree relation. This means that
it requires an XP to raise to its specifier position, where the XP is given
the interpretation of topic. It is at this point that the interpretable f-
feature on the XP values the f-probe on a. On this view, we predict the
following:
(3) The goal of the f-probe at a is the topic of the sentence.
This is in essence the biconditional that Baker (2003) observes for a range
of languages including Kinande, a Bantu language that we will look at in
detail in this chapter.
(4) A verb X agrees with an NP Y if and only if Y is in a dislocated,
adjunct position. (Baker 2003:109)
Dislocated, adjunct position here means a topic position, something I will
return to. Baker explains this biconditional by stating that the languages
to which it applies lack Case. I will show that Baker is correct that, when
the biconditional does apply, it does so because Case is absent. This is so
because the f-probe occurs on a and not on T. As I will also show, how-
ever, Case e¤ects do show up in certain environments; one cannot say
that in these languages Case simply doesn’t matter. Where aP is dis-
allowed, the f-probe is inherited by T, and in this environment, we
aP, f-Features, and the A/Ā Distinction 95

observe the familiar subject-verb agreement, showing the e¤ect of nomi-

native Case.
I begin in section 4.2 by discussing Kinande as analyzed by Baker
(2003) and others. I first present Baker’s analysis and show that it is con-
sistent with a structure that contains the projection aP. I also introduce
new data based on consultation with a native speaker of Kinande that
further support my analysis. In section 4.3, I take up Kilega, another
Bantu language, basing the discussion on work by Carstens (2005) and
Kinyalolo (1991). I look in particular at wh-movement, a form of Ā-
movement. I then explore ways to distinguish A- and Ā-movements in
the kind of approach I am pursuing. I suggest that the phase architecture
provides a natural way to distinguish these two types of movement.

4.2 Kinande and aP

For this discussion of Kinande, I depend on the analysis and insights

found in Baker 2003 but also draw from Progovac 1993 and Schneider-
Zioga 2007.
Kinande has agreement—in fact, quite an elaborate agreement system
that allows agreement not only between the subject and the verbal a‰x
but also between the verbal a‰x and the object or the locative. These pos-
sibilities are illustrated in (5) (Baker 2003:113).
(5) a. Omukali mo-a-seny-ire olukwi (lw’-omo-mbasa). (SVO)
woman.1 aff-1.s/t-chop-ext wood.11 lk11-loc.18-axe.9
‘The woman chopped wood (with an axe).’
b. Olukwi si-lu-li-seny-a bakali (omo-mbasa). (OVS)
wood.11 neg-11.s-pres-chop-fv women.2 loc.18-axe.9
‘Women do not chop wood (with an axe).’
c. ?Omo-mulongo mw-a-hik-a omukali. (LocVS)
loc.18-village.3 18.s-t-arrive-fv woman
‘At the village arrived a woman.’
Example (5a) shows the SVO order, and (5b) illustrates the ‘‘object rever-
sal’’ order of OVS. In (5c), the locative phrase occurs in the agreeing po-
sition. As in English, this locative inversion occurs in Kinande only with
unaccusative and passive verbs, but unlike in English, the locative enters
into agreement with the verbal inflection, which suggests that the locative
phrase in Kinande is a DP, not a PP.
Along with this three-way option for agreeing (subject, object, locative),
Kinande exhibits a property that Baker characterizes as ‘‘dislocation.’’ As
96 Chapter 4

noted by Baker (2003) and Progovac (1993), the agreeing phrase must be
interpreted as definite (or specific). This is illustrated with the object re-
versal order in (6).
(6) Eritunda, n-a-ri-gul-a.
fruit.5 1sg.s-t-om5-buy-fv
‘The fruit, I bought it.’
In this reversal construction, the object is in a position to trigger agree-
ment on the verbal inflection, and it must be interpreted as definite
(i.e., as a topic). This is claimed to be true for all agreeing phrases,
a point I will evaluate later. Baker notes that ‘‘[t]rue polysynthetic
languages . . . always have agreement and always have dislocation’’
(2003:112). By dislocation, Baker means that he views the agreeing
phrase, such as the object in (6), as being somewhere above Spec,TP—he
assumes that it is in a higher Spec,TP—in a position comparable to the
dislocation position in languages such as Italian. In Romance, dislocation
is possible only if the phrase is definite or specific (Rizzi 1986). The agree-
ment, therefore, occurs with a phrase that is in a specifier higher than the
normal Spec,TP. Baker expresses this observation as a biconditional for
languages such as Kinande (and Mohawk, etc.) and contends that it is a
parameter for polysynthetic languages such as Kinande.
(7) A verb X agrees with an NP Y if and only if Y is in a dislocated,
adjunct position. (Baker 2003:109)
How is the agreeing phrase ‘‘dislocated’’? Baker’s analysis forces the
agreeing phrase to occur in a higher position—thus forces the agreement
to hold between this higher specifier and some head—by requiring pro to
occur in Spec,TP (Baker 2003:124).
(8) [TP NPi [TP proi ThAgri iþVerb . . . [VP ti . . . ]]]
The occurrence of pro in the lower Spec,TP meets the EPP requirement of
T (the assumption here being that there is an independent EPP feature,
something that I do not assume); thus, in Kinande the EPP feature is
located on T (Baker 2003:125). Moreover, this pro, by virtue of occurring
in the ‘‘normal’’ Spec,TP, is the actual agreeing phrase; but because it is
unpronounced, a fully specified NP that corresponds to the pro, NPi , may
occur in the higher specifier of TP. If a fully specified DP does not occur,
a pro-drop construction results, and in Bantu, only the agreeing phrase
(subject, object, locative) can be pro (Vicki Carstens, pers. comm.).
What forces the pro to occur in the lower Spec,TP? According to Baker
(2003), the reason is that Kinande has no structural Case. Because no
aP, f-Features, and the A/Ā Distinction 97

Case is assigned to the lower Spec,TP (or any other position), no fully
specified DP can occur in the lower Spec,TP—yet something must occur
in this position to satisfy the EPP requirement on T. Following earlier
work (Baker 1996), Baker (2003) suggests that pro fits the bill: not requir-
ing Case itself, it can occur in Spec,TP without Case and meet the EPP
requirement.
Baker (2003) suggests the following way of comparing Kinande and
Romance:
(9) Kinande: agreement, EPP, Case
Romance: agreement, EPP, Case
In Kinande, agreement occurs in tandem with the EPP. Because there is
no Case, any DP within the local domain of the agreement can become
the goal of agreement—the subject, the object, or the locative. Once
agreement is established, the DP is raised to Spec,TP to satisfy the EPP.
In Romance, Spec,TP is not filled, according to Alexiadou and Anagnos-
topoulou (1998), and agreement always goes with the subject. Baker
describes this state of a¤airs by claiming that T lacks the EPP require-
ment in Romance (however that lack is to be accounted for—perhaps
along the lines suggested in Alexiadou and Anagnostopoulou 1998). The
fact that agreement always goes with the subject in Romance is due to the
occurrence of Case—specifically, nominative Case.
The other relevant property of Kinande is that the agreeing phrase
must occur on the left edge, where it is interpreted as definite or specific.
One piece of evidence for this involves augment vowels (Baker 2003, Pro-
govac 1993, Schneider-Zioga 2007). Nouns in Kinande sometimes begin
with an augment vowel that matches the vowel of the class that the noun
prefix belongs to. This augment vowel may drop under the scope of nega-
tion and in some other contexts. The following sentences exemplify an
object with and without its augment vowel:
(10) a. Yohani si-a-nzire o-mu-kali.
John neg-1.s/t-like aug-cl1-woman
‘John does not like the woman.’
b. Yohani si-a-nzire mu-kali.
John neg-1.s/t-like cl1-woman
‘John does not like a(ny) woman.’
Whenever a noun lacks an augment vowel, it has an indefinite inter-
pretation, as in (10b). Whenever a noun has an augment vowel, it is inter-
preted as definite, as indicated by the translation in (10a) (later we will see
98 Chapter 4

exceptions to this when the subject carries the augment vowel). A point
relevant here is that an agreed-with phrase can never drop its augment
vowel (Baker 2003), indicating that the agreed-with phrase must always
be definite. By transitivity of reasoning, an indefinite phrase can only
occur in a lower position, most typically in its originally merged position
within vP/VP. In (11a), the subject has the augment vowel, whereas in
(11b) it does not.
(11) a. O-mu-kali mo-a-teta-gul-a ki-ndu.
aug-cl1-woman aff-1.s-neg/past-buy-fv cl7-thing
‘The woman didn’t buy anything.’
b. *Mu-kali mo-a-teta-gul-a eritunda.
cl1-woman aff-1.s/t-neg/past-buy-fv fruit.5
‘No woman bought a fruit.’
As (11b) shows, it is not possible for a phrase without the augment vowel,
which is nonspecific and indefinite, to occur in the agreement position.
What we have seen predicts that a completely nonspecific expression
such as ‘nobody’ ought not be able to occur in the agreed-with position;
Schneider-Zioga (2007:406) gives the following example as evidence that
this is correct:
(12) *Si-ha-li mundu eriyenda.
neg-there-be person canonical-agr.left
‘Nobody left.’
The augment-vowel phenomenon clearly shows that, in Kinande,
something that occurs on the left edge is marked both as being the topic
of the sentence and for agreement with the verbal inflection. Simply occu-
pying the normal Spec,TP does not force a phrase to be interpreted as
specific or definite. This topic interpretation arises from the phrase’s mov-
ing to a higher region of the structure.
If we look at what Baker, Progovac, and Schneider-Zioga have
observed about Kinande from the perspective of the Uniformity Principle
interpreted as Strong Uniformity, which states that languages are entirely
uniform, we are led to a picture of the language somewhat di¤erent from
Baker’s. Baker’s analysis rests crucially on the idea that Case, particularly
nominative Case, does not have any role to play in Kinande. But the
Uniformity Principle/Strong Uniformity prescribes that Case ought to be
present, and that it should have exactly the same role in this language as
elsewhere in making arguments visible for y-marking. Below, we will see
instances where nominative Case emerges in Kinande exactly the way it
emerges in languages where the function of Case is indisputable.
aP, f-Features, and the A/Ā Distinction 99

To begin, recall the primary insight in Baker’s analysis of polysynthetic

languages.
(13) A verb X agrees with an NP Y if and only if Y is in a dislocated,
adjunct position.
As we saw, this property of dislocation is marked by the ‘‘topic’’ nature
of the agreed-with phrase. What I propose is that this property of poly-
synthetic languages identified by Baker, together with the seeming inert-
ness of Case, points to the fact that the f-probe in Kinande is inherited
by a head higher than T, namely, a. a is not a Case assigner, so we would
not expect the typical ‘‘nominative Case’’ e¤ect with the f-probe. But the
f-probe by itself cannot find a goal; to make this possible, focus is also
inherited by a, as shown in (14).
(14)

At the point where the f-probe is inherited, it is not in any Agree relation.
It enters such a relation only when an XP is raised to Spec,aP to satisfy
the –focus feature. In this way, the polysynthetic biconditional that Baker
observed falls out of the interactions of the grammatical features to-
gether with a particular structure, aP, all consistent with the Uniformity
Principle/Strong Uniformity.
I noted earlier that the aP projection does not always occur. If there are
environments in Kinande where aP is disallowed, the f-probe will be
inherited by T instead of a. If that happens, we expect that Case will be-
come a factor and that the f-probe will only find the grammatical subject
as its goal. Although Baker 2003 has no data bearing on this prediction,
precisely the relevant situation turns up in wh-questions, as reported by
Schneider-Zioga (2007), and in other, embedded contexts, as work with
a native-speaker consultant, Pierre Mujomba, has revealed.
100 Chapter 4

As noted by Progovac (1993), the agreed-with phrase in Kinande must

be definite, reflecting its topic nature—a fact Baker (2003) draws upon
extensively in his analysis. Thus, a nonspecific, nondefinite expression is
inappropriate in this position.
(15) *Si-ha-li mundu eriyenda.
neg-there-be person canonical-agr.left
‘Nobody left.’
Schneider-Zioga (2007) points out that this requirement that the agreed-
with phrase must be definite is suspended if wh-movement takes place. (I
have changed the verb form to atahuka from Schneider-Zioga’s syanga-
huka at consultant Pierre Mujomba’s suggestion.)
(16) Ekihi kyo mukali atahuka?
what that woman not.cook
‘What didn’t any woman cook?’ (or ‘What did no woman cook?’)
Here the phrase ‘woman’ is interpreted as indefinite and nonspecific even
though it agrees with the verb. Schneider-Zioga uses this fact to argue
that the agreed-with phrase need not always move to a higher position, a
conclusion with which I concur. Moreover, this lower phrase is limited to
the subject, as I have confirmed with Pierre Mujomba.
(17) Iyondi yo u-kandi-gul-a esyongoko?
who that agr-will-buy the.chickens
‘Who will buy the chickens?’
(18) *Iyondi yo esyongoko si-kandi-gul-a?
who that the.chickens agr-will-buy
‘Who will buy the chickens?’
Example (18) is an instance of object reversal, so the agreement goes with
‘the chickens’. The fact that ‘the chickens’ is definite in (18) is not the
problem, as other data from the consultant will show. I suggest that in
certain environments, such as wh-questions, aP is not allowed to occur in
Kinande, so that the f-probe is inherited directly by T. Once this occurs,
the f-probe depends on the nominative Case of T and finds the grammat-
ical subject, and nothing else, for its goal. This, then, is a clear indication
that Case does matter in Kinande, contrary to Baker’s (2003) analysis.1
We find the same restriction in relative clauses and clefts, which argu-
ably also involve some sort of operator movement to C, apparently an en-
vironment where aP is not allowed. First, note the following declarative
sentence:
aP, f-Features, and the A/Ā Distinction 101

(19) Abakali ba-ka-gul-a esyongoko.

the.women buy chickens
‘The women buy chickens.’
The object reversal version of this sentence is (20), where the verb now
agrees with the object in the agreed-with position.
(20) Esyongoko si-ka-gul-a bakali.
the.chickens agr-pres-buy women
‘The WOMEN buy the chickens.’
Now, what happens to these sentences under relativization? It turns out
that they fare di¤erently. In (21) and (22), the subject ‘women’ from (19)
and (20) has been relativized.
(21) abakali ba-ka-gul-a esyongoko
the.women buy chickens
‘the women who buy the chickens’
(22) *abakali esyongoko si-ka-gul-a
the.women the.chickens buy
‘the women who buy the chickens’
This is the same pattern found with the wh-question (16) discussed by
Schneider-Zioga, but I have filled out the paradigm a bit more by show-
ing that object reversal is not possible, demonstrating clearly that aP does
not occur if operator movement takes place. As a result, the f-probe is
inherited by T, and because of T’s dependence on nominative Case, only
the grammatical subject can be the goal, and it is the subject that is raised
to Spec,TP. Note, too, the following relative clauses:
(23) a. omulongo ogo abakali ba-ka-gul-a-ko esyongoko
the.village where the.women 3pl-pres-buy-at/from the.chickens
‘the village where the women buy the chickens’
b. omulongo ogo bakali ba-ka-gul-a-ko esyongoko
the.village where women 3pl-pres-buy-at/from the.chickens
‘the village where (some) women buy the chickens’
(24) a. omulongo ogo esyongoko si-ka-gul-a-ko bakali
the.village where the.chickens agr-pres-buy-at women
‘the village where WOMEN buy the chickens’
b. omulongo ogo ngoko si-ka-gul-a-ko bakali
the.village where chickens agr-pres-buy-at women
‘the village where WOMEN buy chickens’
102 Chapter 4

Examples (23a–b) show again that the preverbal subject is fine whether it
is definite or indefinite. What is surprising is that the two object reversal
examples (24a–b) were judged acceptable by the consultant. Apparently
these examples are interpreted not as relative clauses with operator move-
ment, which would block aP and make the object reversal impossible, but
as complex NPs, which arguably do not, or need not, involve operator
movement. This demonstrates that just the consideration of informational
structure (see note 1) cannot explain the pattern of grammaticality.
The following examples illustrate the restriction with clefts:
(25) Esyongoko esi syo bakali ba-kandi-gul-a.
chickens these that women agr-will-buy
‘It’s these chickens that women will buy.’
(26) Abakali aba bo ba-kandi-gul-a esyongoko.
women these that agr-will-buy the.chickens
‘It’s these women who will buy the chickens.’
(27) *Abakali aba bo esyongoko si-kandi-gul-a.
women these that the.chickens agr-will-buy
‘It’s these women who will buy the chickens.’
The pattern of grammaticality in these operator-movement
constructions—wh-question, relative clause, and cleft—suggests that
nominative Case is a factor in Kinande.

4.3 Kilega Wh-Questions

Kilega is a Bantu language that resembles Kinande in many ways (Cars-

tens 2005, Kinyalolo 1991). For example, the agreed-with phrase may be
the subject, the object, or the locative. (All Kilega examples are taken
from Carstens 2005, which in turn drew some of them from Kinyalolo
1991.)
(28) a. Mutu t-á-ku-sol-ág-á maku wéneéné.
1person neg-1agr-prog-a-sleep-fv 6beer alone
‘A person does not usually drink beer alone.’
b. Maku ta-má-ku-sol-ág-á mutu wéneéné.
6beer neg-6agr-prog-drink-hab-fv 1person alone
‘No one usually drinks beer alone.’
c. Mu-zı́zo nyumbá mu-á-nyám-é bána wálúbı́.
18-10-that 10house 18sa-a-sleep-fv 2child one.day.period
‘There will sleep children in those houses tomorrow.’
aP, f-Features, and the A/Ā Distinction 103

One di¤erence between Kinande and Kilega is that in Kilega wh-

questions, the fronted wh-phrase is what the verb agrees with. Examples
(29)–(31) illustrate subject, object, and locative wh-phrase fronting in
Kilega. (I will return to why the subject agreement á in (29) is not possi-
ble. The gloss CA stands for complementizer agreement, which I will also
discuss below.)
(29) Názı́ ú-(*á)-ku-kı́t-ag-a búbo?
1who 1ca-(*1sa)-prog-do-hab-f-fv 14that
‘Who (usually) does that?’
(30) Bikı́ bı́-á-kás-il-é bábo bı́kulu mwámi mu-muwı́lo?
8what 8ca-a-give-perf-fv 2that 2woman 1chief 18-3village
‘What did those women give the chief in the village?’
(31) Kúnı́ ku-ta-bá-ku-yan-ág-á mukindi?
16where 16ca-neg-2sa-prog-play-hab-fv 18-7night
‘Where don’t they usually play at night?’
Two additional points are relevant to our discussion of wh-questions.
First, Kilega (and also Kinande) has the wh-in-situ option; (32) is the in-
situ counterpart of (30).
(32) Bábo bı́kulu b-á-kás-il-é mwámi bikı́ mu-muwı́lo?
2that 2woman 2sa-a-give-perf-fv 1chief 8what 18-3village
‘What did those women give the chief in the village?’
I will return to the in-situ option. Second, in an example like (30), where
the verb agrees with the fronted object wh-phrase, the verb cannot also
carry subject agreement.
(33) *Bikı́ bı́-b-á-kás-il-é bábo bı́kulu mwámi mu-muwı́lo?
8what 8ca-2sa-a-give-perf-fv 2that 2woman 1chief 18-3village
‘What did those women give the chief in the village?’
Therefore, as Carstens (2005) has argued, examples like (33) show that
there can be only one formal agreement relation, and if a wh-phrase is to
be moved, C must enter into agreement with the wh-phrase. Note, how-
ever, that it is not clear how we can accomplish this. In keeping with the
analysis of Kinande proposed here, in the non-wh examples in (28) (the
subject, the object, or the locative in the agreed-with position), the agreed-
with phrase occupies Spec,aP. But a is not a head that hosts the feature rel-
evant for wh-questions—namely, wh or Q, which we would expect to be
borne instead on the higher head, C. I will sketch a possible analysis of
the relevant portion of Kilega that distinguishes it from Kinande.
104 Chapter 4

Let us suppose that the f-probe is inherited by a in Kilega just as in

Kinande. The di¤erence is that unlike in Kinande, in Kilega aP may proj-
ect in wh-questions. As a result, the wh-phrase can raise as the goal of the
f-probe on a. In Kinande, –focus is inherited together with the f-probe,
which gives rise to the topic interpretation of the goal. But a topic inter-
pretation is inappropriate for a wh-phrase. How can the f-probe on a be
valued? Let us suppose that nothing else is inherited by a. Instead, there
is a –focus feature on C that enters into an Agree relation with the wh-
phrase, turning it into þfocus, and this Agree relation requires the wh-
phrase to raise ultimately to Spec,CP. However, on its way, the wh-phrase
first merges with a and values the f-probe (see (34)). Because the f-probe
and focus are di¤erent types of features, no Minimality violation ensues.
(34)

The first leg of the movement (to Spec,aP) values the f-probe, and the
second leg (to Spec,CP) places the wh-phrase properly in the domain of
Q. This, then, shows that both topic and focus are relevant for the aP
projection. The di¤erence is that if –focus is inherited by a, it cannot be
in a wh-question environment because there will be no way to raise the
wh-phrase all the way to C.2
Note one crucial point here, however. Along with the movement of the
wh-phrase, the a head must also raise to C. This is to ensure that at the
next phase, when the interior of that higher phase is transferred to PF
and semantic interpretation, the wh-phrase and its probes—focus and f-
probe—are in a ‘‘Spec-head’’ relation, thus fulfilling the probe-goal union
(PGU) requirement. The idea that a raises is consistent with the Kilega
aP, f-Features, and the A/Ā Distinction 105

data found in Carstens 2005. The agreement element has a di¤erent shape
depending on whether it is pronounced at C or lower (a in the present
analysis). I repeat example (29), which illustrates this.
(35) Názı́ ú-(*á)-ku-kı́t-ag-a búbo?
1who 1ca-(*1sa)-prog-do-hab-f-fv 14that
‘Who (usually) does that?’
In this wh-question, the wh-phrase agrees in class with the agreement ele-
ment in the verbal morphology. Crucially, this agreement element is ú,
not á. Ú is the agreement element for class 1 when the agreement element
(and the verbal complex that contains it) occurs at C, whereas á is the
agreement element for class 1 when the verbal complex occurs lower (a
in the present analysis) (Carstens 2003, Kinyalolo 1991). The latter case
is illustrated here (from Carstens 2005:265).
(36) Mutu t-á-ku-sol-ág-á maku wéneéne.
1person neg-1sa-prog-drink-hab-fv 6beer alone
‘A person does not usually drink beer alone.’
In this way, a is raised to C, after being picked up by the verbal complex,
so that the entire verbal complex is at C. This meets the PGU require-
ment with the wh-phrase in Spec,CP. This is why the verbal complex
always immediately follows the wh-phrase in Spec,CP.
Finally, if a non-wh phrase instead of the wh-phrase were raised to
Spec,aP, the result would be an in-situ construction of the type we saw
earlier in (32). In a wh-in-situ construction, the verbal complex does not
raise to C; hence, the agreement is of the ‘‘lower’’ type, not the type that
occurs at C.
(37) *Miwána ú-ku-kı́-ag-a bikı́?
1child 1ca-prog-do-hab-fv 2what
‘What does a/the child usually do?’
The agreement element should be the ‘‘lower’’ one, á, but instead ú
occurs, the agreement element for class 1 at C. Finally, in the wh-in-situ
construction, the f-probe at a finds its goal thanks to –focus, which is
also inherited by a.

4.4 Finnish

We saw in chapter 3 that Finnish has a topic projection above TP and

below CP, which I labeled aP (in contrast to F[in]P in Holmberg and
Nikanne 2002).3
106 Chapter 4

(38) a. Graham Greene on kirjoittanut tämän kirjan.

Graham Greene has written this book
‘Graham Greene has written this book.’
b. Tämän kirjan on kirjoittanut Graham Greene.
this book has written Graham Greene
As we saw in chapter 3, in both cases here the subject is ‘Graham
Greene’. However, the topics di¤er: the subject ‘Graham Greene’ occu-
pies topic position in (38a), but the object ‘this book’ occupies topic posi-
tion in (38b).
(39) a. [CP [aP Graham Greene [TP . . . this book . . . ]]]
b. [CP [aP this book [TP . . . Graham Greene . . . ]]]
Up to this point, Finnish looks very much like Kinande and Kilega in
having the aP projection that hosts a topic. Holmberg and Nikanne
(2002) go into some detail to show that the subject and the object in these
examples occupy the same position. For example, (40a–c) demonstrate
that a subject or an object in topic position can control a subject floating
quantifier, but a phrase in Spec,CP, which is higher than the topic, cannot
(Holmberg and Nikanne 2002:88).
(40) a. Ilmeisesti kriitikot ovat (kaikki) ylsäneet tätä kirjaa.
evidently the.critics have all praised this book
b. Ilmeisesti nämä kirjat on (kaikki) kirjoittanut Graham Greene.
evidently these books has all written Graham Greene
‘Evidently Graham Greene has written all these books.’
c. Nämä kirjat Graham Greene on (*kaikki) kirjoittanut (kaikki).
these books Graham Greene has all written all
The subject (40a) and object (40b) in topic position can control the float-
ing quantifier ‘all’ in the subject position immediately after the auxiliary
verb. What (40c) shows is that ‘these books’, which occurs in Spec,CP,
cannot control the subject floating quantifier; instead, it is limited to con-
trolling the postverbal floating quantifier.
Finnish di¤ers from Kinande and Kilega in agreement possibilities. In
the Bantu languages, agreement may hold with the subject, object, or
locative. This led us to postulate that the f-probe is inherited by a in this
language, not T, and that this f-probe at a finds its goal through topical-
ization, the latter implemented by the –focus feature that also lowers to a
from C. In Finnish, despite its similarity with Bantu in topicalization pos-
sibilities, agreement always goes with the subject. Given our assumptions
about the f-probe, this means that, unlike in the Bantu languages, in
aP, f-Features, and the A/Ā Distinction 107

Finnish the f-probe is inherited by T, not a, as shown in (41). The f-

probe finds its goal through the nominative Case of T, which limits the
goal to the nominative phrase. Presumably, the –focus feature is inherited
by a, just as in the Bantu languages.
(41)

The f-probe seeks its goal while at T, as in (41), but the T bearing this f-
probe ultimately shows up on a, presumably raising there with other rele-
vant material (Holmberg and Nikanne 2002:72). One argument for this
involves pro-drop (Holmberg 2005:539).
(42) a. (Minä) puhun englantia.
I speak.1sg English
b. (Sinä) puhut englantia.
you speak.2sg English
c. *(Hän) puhuu englantia.
he/she speak.3sg English
d. (Me) puhumme englantia.
we speak.1pl English
e. (Te) puhutte englantia.
you speak.2pl English
f. *(He) puhuvat englantia.
they speak.3pl English
As shown, the verb ‘speak’ with the appropriate agreement occurs next to
where the pro is postulated to occur. Moreover, first and second person
agreement, but not third, licenses pro in this language. Recall from
108 Chapter 4

chapter 3 that the expletive sitä is merged at Spec,aP to satisfy the –focus
feature, not the f-probe. The relevant example is (43), from Holmberg
and Nikanne 2002:72.
(43) Sitä ovat nämä lapset jo oppineet uimaan.
expl have these children already learned to.swim
‘These children have already learned to swim.’
Sitä is a ‘‘pure’’ expletive in that it is not related to f-features or Case
(Holmberg and Nikanne 2002:90). The topic position is not just a posi-
tion relevant to information structure; it imparts a subjectlike quality to
any element that appears in it, even an object. Returning to pro-drop, we
see that the expletive sitä and pro are incompatible (Holmberg 2005:543).
(44) *Sitä puhun englantia.
expl speak.1sg English
Holmberg argues that this indicates that the pro occurs in topic position,
what I am calling aP. Since pro occurs in topic position, the expletive sitä
cannot also occur there.
The similarities and di¤erences between the Bantu languages we looked
at earlier and Finnish are clear: both have the aP projection, which in
declarative sentences hosts a topic, but the f-probe occurs in di¤erent
positions. In the Bantu languages, the f-probe is inherited by a, whereas
in Finnish, it is inherited by T. This di¤erence leads to di¤erences in
agreement possibilities, including licensing of pro. Despite this di¤erence,
agreement (the head with the f-probe) occurs at a in both cases, the f-
probe having been inherited directly by a in the Bantu languages, and T
with the f-probe having raised to a in Finnish. This is all consistent with
the type of approach developed here so far. The one inconsistency has to
do with the fact that in Finnish, it is apparently possible for a nongoal as
well as the goal to occur in Spec,aP. Recall (38b), repeated here.
(45) Tämän kirjan on kirjoittanut Graham Greene.
this book has written Graham Greene
In this example, agreement goes with the subject, ‘Graham Greene’, and
it shows up on the auxiliary verb on ‘has’ at a. Surprisingly, the object
‘this book’ occurs in Spec,aP as the topic even though on agrees with the
subject; the subject occurs low in the structure, as we can see. This is an
exception to the idea that a functional relation established by Agree must
be replicated by movement of the goal to the probe before transfer to
interpretation.
aP, f-Features, and the A/Ā Distinction 109

What causes this failure of correlation between f-probe and PGU? To

repeat the problem, in Finnish the f-probe always enters into agreement
with the subject, indicating that the f-probe occurs on T at the point
where agreement is established. Yet the f-probe ultimately moves to a,
and the subject may or may not move with it. If it does not, it fails to ful-
fill the PGU requirement. Why is it able to ignore PGU? What appears to
be happening in Finnish is that functional relations such as topic and sub-
ject-of are established not by f-feature agreement, although it occurs, but
by the topic structure (–focus). So, for example, it is topic position that
allows control of the subject floating quantifier regardless of whether the
subject or the object occurs in the topic position. The f-feature agreement
appears to be a strictly surface matter. The combination of Case and f-
features requires that agreement be implemented so that the features that
need to be erased before being transferred can be erased. But because
the functional relations that are retained beyond the interface are those
established by the topic feature, the relations established by f-feature
agreement do not have any role to play beyond establishing agreement.
Hence, the goal of the agreement need not move to the probe to fulfill
PGU, and this relation is erased prior to transfer. It is possible that f-
feature agreement does have a role to play after transfer just in those
cases where PGU is met—when the subject moves into topic position. I
will leave this issue open.

4.5 Distinguishing A- and Ā-Movements without Reference to Case

We saw in the discussion of Kilega that the PGU requirement holds re-
gardless of whether the movement that takes place is A- or Ā-movement.
This is, in fact, what we wish to see: the condition(s) for movement ought
to be the same for all movements that are triggered by a probe-goal rela-
tion, and this relation should hold for both A- and Ā-movements. I
assume that all probes begin at a phase head such as C; if the probe is
inherited by T, movement stops at T, but a probe that is retained at C
triggers movement to the C domain. As far as probe-goal relations for
movement are concerned, this is essentially all that narrow syntax pro-
vides. But if the triggering mechanism (probe-goal) does not distinguish
A- and Ā-movements, how are they distinguished? We saw clear evidence
in chapter 3 that these two movements behave di¤erently. A-movement
can overcome WCO violations and can create a new binder, but Ā-
movement can do neither. It is also well known that while A-movement
110 Chapter 4

can overcome a Condition C violation, in most cases Ā-movement

cannot.
These three facts about the two types of movement—WCO suppres-
sion, creation of a new binder, and Condition C suppression—are typi-
cally viewed in the context of reconstruction: although A-movement
does not need to undergo reconstruction, Ā-movement typically does.
This is often characterized in terms of A-movement not being required
to leave a (full) copy in the original position and Ā-movement leaving
such a copy. Fox (1999) and Lasnik (1999a) describe it in terms of Ā-
traces being copies and A-traces being unstructured. (Unstructured simply
means that something is there but is not a fully specified copy.) Lasnik
(1999a, 2003) has argued that A-movement never leaves a (full) copy.
Nevertheless, I will follow other works (e.g., Fox 1999 and references
therein) that suggest that A-movement allows reconstruction—hence is
allowed to leave a full copy, although it is not required to do so—but
that Ā-movement always leaves a full copy.

4.5.1 A- and Ā-Movements

A typical case of reconstruction associated with Ā-movement is (46)
(based on an example from Fox 1999:172).
(46) [Which of hisi students] do you think [every professor]i talked to?
We can see that this construal is mediated by movement from the fact
that (47) (also based on an example from Fox 1999:172) is ungrammati-
cal with the intended interpretation.
(47) *[Which of hisi students] do you think talked to [every professor]i ?
In the grammatical example (46), the wh-phrase containing the pronoun
begins in a position lower than the universal expression every professor;
but in the ungrammatical example (47), the wh-phrase starts out higher
than the universal expression. Example (46) is an instance of reconstruc-
tion, and in the copy theory of movement (Chomsky 1993, 1995), this
means that a full copy of the wh-phrase occurs in the original position.
(48) [which of his students] do you think [every professor] talked to
[which of his students]
It is the lower copy that allows the pronoun inside it to be bound by the
universal expression. No such copy exists lower than the universal quan-
tifier in (47).
The occurrence of the copy in (46) allows the pronoun inside the wh-
phrase to be bound by the universal quantifier, but there are instances
aP, f-Features, and the A/Ā Distinction 111

where the copy leads to ungrammaticality. Relevant examples were noted

by Freidin (1986) and Van Riemsdijk and Williams (1981); (49a–b) come
from Freidin 1986:179.
(49) a. *[Which report that Johni was incompetent] did hei submit?
b. [Which report that Johni revised] did hei submit?
In (49a), the copy of the Ā-movement contains the R-expression John and
it is c-commanded by the pronoun with which it is coindexed, triggering a
Condition C violation. This is what we expect if Ā-movement obligatorily
leaves a copy. I will return to (49b), which is unexpectedly grammatical
even though it also involves Ā-movement.
Turning to A-movement, we see from (50) that, like Ā-movement, A-
movement allows reconstruction (example from Fox 1999:161).
(50) [Someone from hisi class] seems to [every professor]i to be a genius.
However, unlike Ā-movement, A-movement is apparently not forced to
leave a copy, as the following examples show ((51a) from Chomsky
1993:37); ((51b) from Lebeaux 1988:23).
(51) a. [The claim that Johni was asleep] seems to himi to be correct.
b. [Johni ’s mother] seems to himi to be wonderful.
These examples do not show a Condition C violation, which indicates
that A-movement does not leave a copy in the relevant sense. The gener-
alization to be drawn for the two types of movement is this (Takahashi
2006, Takahashi and Hulsey 2009):
(52) a. A-movement optionally leaves a copy.
b. Ā-movement obligatorily leaves a copy.

4.5.2 Takahashi 2006 and Takahashi and Hulsey 2009

To capture the generalizations in (52), Takahashi (2006) and Takahashi
and Hulsey (2009) (henceforth TTH) propose an interesting theory of
the two types of movement. I will briefly introduce their approach, then
show a problem with it, and finally propose an alternative that overcomes
the problem.
TTH’s central idea is that late Merge (Lebeaux 1988; see also Chom-
sky 1993) applies widely in grammar. What is late Merge? Recall Frei-
din’s (1986:179) pair in (49), repeated here.
(53) a. *[Which report that Johni was incompetent] did hei submit?
b. [Which report that Johni revised] did hei submit?
112 Chapter 4

Example (53a) is ungrammatical, as expected, because this is an instance

of Ā-movement, which obligatorily leaves a full copy that, in this case,
triggers a Condition C violation. But what about (53b)? Lebeaux (1988)
argues that the di¤erence between these two examples arises from the
fact that, whereas (53a) contains a full copy of the entire wh-phrase, in
(53b) the copy consists only of the head of the relative clause and the
wh-element (which report); the relative clause that John revised does not
occur in the lower copy. The notion that Lebeaux introduced to account
for examples like (53b) is late Merge, which allows some unit of expres-
sion to be merged into the structure countercyclically. Specifically, in
(53b), the relative clause that John revised is introduced by late Merge
into the wh-phrase, after the wh-phrase has moved to Spec,CP. Prior to
movement, the structure includes only which report, and because this
does not contain the R-expression, there is no Condition C violation.
Why is late Merge not possible in (53a)? Lebeaux argues that in (53a),
the expression that contains the R-expression (that John was incompetent)
is the argument of the head report, and as such, it must have been merged
with the head to begin with; as a result, the entire wh-phrase, including
the R-expression, occurs in the original position of the wh-phrase. But in
(53b), the relative clause is an adjunct, and adjuncts need not be merged
initially with their head. Instead, they can be merged late, avoiding a
Condition C violation.
Following the suggestion of Bhatt and Pancheva (2004, 2007), TTH
argue that late Merge is a more widespread phenomenon than Lebeaux
first proposed. They argue that the ‘‘bleeding’’ of Condition C demon-
strated for A-movement in examples like (51a–b) indicates that a type of
late Merge has occurred (I repeat (51a) here). In fact, TTH propose that
a structure bigger than what Lebeaux suggested may be introduced by
late Merge; they call this process wholesale late merger.
(54) [The claim that Johni was asleep] seems to himi to be correct.
Extending the analysis of late Merge in Fox 2002, TTH suggest that in
one derivation, which results in a grammatical sentence with the intended
interpretation, only the D, the, and its projection, DP, are merged at the
base. This the is the target of raising, and after raising takes place, the
matrix Spec,TP hosts just this the. To this DP in Spec,TP, the NP struc-
ture claim that John was asleep is merged by wholesale late merger. The
lower copy consists of just D/DP the, which makes it possible to avoid a
Condition C violation.
aP, f-Features, and the A/Ā Distinction 113

TTH extend this idea of wholesale late merger to well-known examples

of quantifier scope noted by May (1977).
(55) Some argument seems to be correct.
This sentence has one reading in which there exists an argument that
seems to be correct (some argument > seem) and another reading in
which it seems that some argument is correct (seem > some argument).
The second reading involves reconstruction. What TTH argue is that in
the first reading, which does not involve reconstruction, wholesale late
merger takes place in the following way. For the first reading associated
with (55), the lower clause has the representation [some] correct. This
some raises to Spec,TP, resulting in the structure (56a).4 To this, whole-
sale late merger applies and inserts the restrictor argument, as in (56b).
(56) a.

b.

This structure does not have the NP argument in the lower copy; hence,
the reconstruction e¤ect is absent. The lower copy does receive some in-
terpretation; TTH suggest that this ‘‘small’’ copy undergoes Fox’s (2002)
Trace Conversion for proper interpretation (see TTH 2006, 2009 for
details). For the reconstruction interpretation of (55), argument is merged
to begin with, as shown in (57); in semantic interpretation, it is this lower
full copy that gets interpreted.
114 Chapter 4

(57)

Given that both A- and Ā-movements allow late Merge, how do we

distinguish between the two in a way that captures the observation that
A-movement leaves a copy optionally but Ā-movement does so obligato-
rily? TTH make the point that a copy is obligatorily left in both types of
movement. For example, in (56a) some is merged as the subject of correct,
and it is left as the copy of the movement. The di¤erence arises from the
particular positions that allow late Merge. For this, TTH turn to consid-
erations of Case (Chomsky 1981), arguing that in a DP structure, both
the DP and the NP inside it require Case. In wh-movement, which is
Ā-movement, the landing site is a non-Case position so that if an NP is
introduced by wholesale late merger into the DP that has been wh-moved,
the NP will be without Case. On the other hand, the landing site of
A-movement is a Case position, so that an NP that is introduced into
the moved DP by wholesale late merger will receive Case. Hence, A-
movement always allows the possibility of wholesale late merger, and
this is why it appears that A-movement only optionally leaves a copy.

4.5.3 A Problem with TTH’s Proposal, and a Solution: Phase-Based

Characterization of Chains
TTH’s proposal can account for the core data regarding A- and Ā-
movements in English. A problem arises when we look beyond English,
however. As we saw earlier, in scrambling languages such as Hindi and
Japanese, A-movement scrambling can overcome WCO, and it can also
create a new binder, both indicating that a full copy is not present in the
original position. Yet the movement involved is clearly not driven by
Case since typical A-movement scrambling takes the object and puts it
to the left of the subject. The following examples are repeated from
chapter 3:
(58) a. *[Kinoo proi proj atta hitoi ]-ga dare-oj hihansita no?
yesterday met person-nom who-acc criticized q
Lit. ‘The person who met (him) yesterday criticized whom?’
aP, f-Features, and the A/Ā Distinction 115

b. Dare-oj [kinoo proi proj atta hitoi ]-ga tj hihansita no?

who-acc yesterday met person-nom criticized q
Lit. ‘Who, the person who met (him) yesterday criticized?’
Example (58a) is a straightforward WCO violation; ‘who-acc’ fails to
c-command the pro inside the subject phrase. In (58b), this object ‘who-
acc’ has scrambled by A-movement to a position that c-commands the
pro, allowing the sentence to overcome a WCO violation (e.g., Hoji
1985, Saito 1992). This movement of the object to the head of the sen-
tence clearly doesn’t happen for Case reasons, yet it has the same ‘‘no
copy’’ property as the A-movement in English that targets a Case posi-
tion. How can we capture the optional nature of the lower copy in A-
movement and its obligatoriness in Ā-movement without reference to
Case?
I will propose a solution based on the phase architecture of narrow syn-
tax. According to this view, structures are built from the bottom up in
chunks, called phases. The phases that have been identified are vP and
CP (and possibly DP). Once a phase is completed, its complement (or
‘‘interior’’) is transferred to PF and to semantic interpretation, leaving
the edge of the phase to be transferred with the complement of the next
higher phase (Chomsky 2001, 2005, 2007, 2008). Ultimately, the phases
have to be put back together in PF and at semantic interpretation to
make the sentence whole, and this is where I suggest that the A/Ā distinc-
tion with respect to copies come into play. Let us suppose that chains are
characterized as follows:
(59) Phase-based characterization of chains
A full copy of a moved item must be available for interpretation if
the movement crosses a transfer domain boundary.
The idea is simple, and the two possibilities are illustrated in (60).
(60) a.
116 Chapter 4

b.

The intuitive idea is that when it comes time to put the phases back to-
gether in semantic interpretation, a chain must be made whole, and if a
chain is dispersed between two phases (or transfer domains, more accu-
rately), a record of the chain in its entirety must be kept. In other words,
a full copy of the moved element must be present. If the movement occurs
within the same transfer domain, the chain as a whole is transferred in-
tact, so there is no need for a fully specified copy to occur at the point
where the movement originated, although there is nothing wrong with
leaving such a copy. With this in mind, let us look concretely at three
cases (NCN stands for no copy needed ).
(61) Movement of a DP to Spec,TP from vP
[CP [TP DP [vP NCN [VP . . . ]]]]
!

(62) Raising
[TP DP [seem [TP NCN . . . ]]]
!

(63) Wh-movement
[CP wh-phrase [TP . . . [vP wh-phrase [vP . . . [VP wh-phrase . . . ]]]]]
!

The movement illustrated in (61) could be the movement of the external

argument to Spec,TP, or it could be the scrambling of the object (or some
other VP-internal XP) to the TP domain. The movement illustrated in
(62) is raising. In these two cases, the movement takes place within the
same transfer domain so that a copy isn’t needed, though there’s always
the option of leaving a full copy. In the case of raising, I assume that the
matrix v is defective and does not form a phase (Chomsky 2001); the
same analysis applies to A-movement in passives. Wh-movement, illus-
trated in (63), does cross a transfer domain boundary—in fact, two (VP,
aP, f-Features, and the A/Ā Distinction 117

TP), because the wh-phrase here is an object wh-phrase—so a copy must

be available in each domain for re-creating the sentence as a whole at
semantic interpretation. Note that whether a copy is needed or not can
be evaluated at each phase level, avoiding look-ahead. To summarize:
Once a phase is built—say, CP—if it contains a chain formed by move-
ment, a copy is left to be interpreted if the chain crosses the transfer
domain boundary—the complement of the phase. Otherwise, a copy is
unnecessary.
In TTH’s system, late Merge is intimately connected to Case. Late
Merge (wholesale late merger) is always possible with instances of A-
movement because A-movements terminate at a Case position, and the
Case licenses the NP that is introduced into the moved DP by late Merge.
With instances of Ā-movement, late Merge of an NP is impossible be-
cause Ā-movements do not terminate at a Case position. Although this
picture is plausible if the data are limited to English, we saw that once
we bring in A-movement scrambling, Case considerations must be set
aside. But if that is true, the possibility arises that late Merge can also
be set aside for A-movement. Everything else being equal, that would be
desirable given that late Merge violates two principles, Cyclicity and the
Extension Condition. It may still be right, but if we can do without it,
the design of narrow syntax would be simpler. I will suggest this for A-
movement, leaving open the question of whether late Merge can also be
eliminated in Ā-movement.
Under the phase-based characterization of chains (PBCC), there is
no reason to assume the late Merge model for A-movement; instead, we
can go back to the simpler view that A-movement optionally leaves a
copy. With the PBCC, we need not make any reference to A-movement
versus Ā-movement. Instead, any movement that does not cross a trans-
fer domain boundary is free to not leave a copy (although it can),
whereas a movement that crosses a transfer domain boundary must
leave a copy so that the chain can be put back together in the semantic
component.
There is another point we need to observe about A-movement. Let us
suppose that a movement that occurs strictly within a transfer domain
leaves a copy. In TTH’s approach, this leads to reconstruction, which
means that the higher copy is not interpreted. But is this true? A simpler
view is that even in these cases of ‘‘reconstruction,’’ the higher copy,
which is the pronounced one, is capable of being interpreted as well. The
following example from Japanese demonstrates this point:5
118 Chapter 4

(64) [Dono zibuni -no e]-oj [ej sorej -o kaita] ekakii -ga tj
which self-gen picture-acc it-acc drew artist-nom
kiratteiru no?
hate q
‘Which of self ’s pictures does the artist who drew it hate?’
In this example, the object wh-phrase has scrambled locally to the left of
the subject. This wh-phrase binds the pronoun sore ‘it’ inside the relative
clause modifying the subject ‘artist’. The pronoun does not c-command
the trace of the moved wh-phrase, yet this binding construal is possible,
which indicates that we are dealing with an A-movement, and the wh-
phrase is interpreted at the moved position above the subject. At the
same time, the wh-phrase contains the anaphor zibun ‘self ’, which is
bound by the subject ‘artist’; this requires that the copy of the A-moved
wh-phrase must also be interpreted in its original position below the
subject. Clearly, in this example, the lower copy is available for interpre-
tation even though the movement took place wholly within a transfer do-
main. The simplest way to view this sort of movement is to say that the
higher copy is always available for interpretation, and the lower copy, if
left, can also be interpreted. As (65) shows, without this movement, the
sentence is ungrammatical because it violates WCO, as expected.
(65) *[ej Sorej -o kaita] ekakii -ga [dono zibuni -no e]-oj
it-acc drew artist-nom which self-gen picture-acc
kiratteiru no?
hate q
‘Which of self ’s pictures does the artist who drew it hate?’
Of the several native speakers I consulted, all found (64) fine, although
one hesitated until a proper context could be imagined, in which the artist
drew many pictures of himself. Judgments for (65) ranged from ungram-
matical to marginal (* to ??), which is typical of WCO violations. Finally,
the English version of (64) evidences exactly the same property: both
copies are available for interpretation simultaneously.6
(66) [No picture of himi ]j seems to every studenti who drew itj tj to be
nice.
Just as with the Japanese example (64), it takes a moment to imagine
an appropriate context for (66)—one in which each student drew many
pictures of himself—but once we can do so, the sentence appears to be
grammatical with the intended reading in which the A-moved quantifier
phrase no picture of him binds the pronoun it from the moved position,
aP, f-Features, and the A/Ā Distinction 119

and the pronoun him inside this quantifier is bound in the lower copy by
the quantifier every student.
Next, I will return to Finnish and show how the PBCC approach to A-
and Ā-movement correctly accounts for an unusual ‘‘mixed’’ position.

4.5.4 A ‘‘Mixed’’ Position in Finnish

The constructions we have observed so far are fairly conventional for
the most part, and there are surely other ways to account for them.
What the PBCC predicts, however, is the existence of some exotic move-
ments whose landing sites combine the properties of A- and Ā-positions.
Such a movement is found in Finnish (Holmberg and Nikanne 2002:87);
see É. Kiss 1995:233 for a similar example in Hungarian. In Finnish,
there is a topic position above TP. In chapter 3, I labeled this as aP, but
for the sake of the exposition here, I will use Top(ic)P. The following ex-
ample is judged grammatical by Holmberg and Nikanne:
(67) Ilmeisesti itseää äänesti vain Jussi.
apparently for.himself voted only Jussi
‘Apparently the only person who voted for himself was Jussi.’
The odd property of this example is that an anaphor occurs in a topic
position, which Holmberg and Nikanne clearly identify as an A-position,
not an Ā-position—yet it must also be an Ā-position because the anaphor
must reconstruct. As a result, Holmberg and Nikanne characterize this
topic position as a mixed A- and Ā-position. This is an example where
A-movement allows the higher copy to be interpreted even though the
lower copy must also be interpreted: the higher copy receives an interpre-
tation such that ‘for.himself ’ is given a topic reading in the moved posi-
tion. Its lower copy also must receive an interpretation, for the obvious
reason.
The fact that ‘for.himself ’ occupies an A-position is indicated by the
fact that a phrase in this position (whether subject or object) can control
a subject floating quantifier. Recall (40a–c), repeated here.
(68) a. Ilmeisesti kriitikot ovat (kaikki) ylsäneet tätä kirjaa.
evidently the.critics have all praised this book
b. Ilmeisesti nämä kirjat on (kaikki) kirjoittanut Graham Greene.
evidently these books has all written Graham Greene
‘Evidently Graham Greene has written all these books.’
c. Nämä kirjat Graham Greene on (*kaikki) kirjoittanut (kaikki).
these books Graham Greene has all written all
120 Chapter 4

Although the subject (68a) and the object (68b) in the topic position can
control the subject floating quantifier, which occurs between the auxiliary
and the verbal participle, a phrase in Spec,CP cannot do so (68c).
The PBCC makes the correct prediction without resorting to a
‘‘mixed’’ position.
(69) Object topicalization
[CP . . . [TopP XPObj [TP . . . [vP NCN . . . [VP . . . XP . . . ]]]]]
!

The first leg of the movement crosses the transfer domain boundary, VP.
As a result, a copy must be left in the original position, which accounts
for the reconstruction e¤ect. The second leg of the movement occurs
within the same transfer domain, from the edge of vP to TopP. Hence,
this second piece of the chain has the option of leaving or not leaving a
copy. Even if the lower copy is not interpreted, the higher copy, which is
in the same transfer domain as this lower copy, can be construed with the
full copy that has been left in the original, object position in the previous
transfer domain. The series of movements just described would constitute
an improper movement in the traditional approach since the first is Ā-
movement, and this is followed by A-movement. Under the present anal-
ysis, there is nothing wrong with this movement; this analysis makes
no reference to the A or Ā nature of movement, and Case is properly
assigned to the lower copy but not the higher copy given that the last
movement does not terminate in a Case position. This sequence of chains
that terminates at Spec,TopP has two copies, one in Spec,TopP, and the
other in the original position of the moved item. Both copies receive inter-
pretation: the copy in Spec,TopP, which associates the anaphor with a
topic interpretation, and the copy in the original object position, which
makes it possible for the anaphor to be bound by its antecedent. Finally,
in those cases where the object in Spec,TopP controls a subject floating
quantifier, no copy is left at the edge of vP, and no reconstruction takes
place.
As the final point in this chapter, recall from chapter 3 that the follow-
ing examples show that the object in Japanese may scramble to Spec,TP
and allow the subject to stay in Spec,vP (Miyagawa 2001):
(70) a. Zen’in-ga siken-o uke-nakat-ta.
all-nom test-acc take-neg-past
‘All did not take the test.’
all > not, */??not > all
aP, f-Features, and the A/Ā Distinction 121

b. Siken-oi zen’in-ga ti uke-nakat-ta.

test-acc all-nom take-neg-past
‘Test, all didn’t take.’
all > not, not > all
In the SOV order in (70a), the preferred reading is for the subject ‘all’ to
take wide scope over negation (Kato 1988); but in (70b), with the object
scrambled, the partial negation interpretation becomes much easier for
many speakers. I argued in Miyagawa 2001 that these examples indicate
that in (70a) the subject has moved to Spec,TP, outside the domain of ne-
gation, whereas in (70b) it is the object that has moved to Spec,TP, allow-
ing the subject to stay in Spec,vP and inside the scope of negation.
In response to this argument, Saito (2006) gives the following example:
(71) Zibun-zisin-oi zen’in-ga ti seme-nakat-ta.
self-self-acc all-nom blame-neg-past
‘Self, all didn’t blame.’
all > not, not > all
As Saito correctly notes, it is possible to get the partial negation interpre-
tation here, which indicates that the scrambled item is raised to Spec,TP
by A-movement; but given that the scrambled item is an anaphor, it must
be reconstructed, which in the traditional view means that the movement
is in fact Ā-movement. Saito concludes that the e¤ect that is identified in
Miyagawa 2001 as partial negation is possible even with Ā-movement,
hence that it has nothing to do with the EPP feature of T. Although this
conclusion is well founded in the traditional view that ties reconstruction
solely to Ā-movement, we have seen that, under the PBCC, A-movement
can also take part in reconstruction, with concomitant interpretation
of the higher copy. Under this view, in (71), the anaphor has moved to
Spec,TP, where it meets the requirement that I earlier called the EPP
and now would call topicalization. But at the same time, its copy in VP,
which is a transfer domain, is also interpreted and the anaphor is properly
bound by the subject ‘all’.
One question that arises concerns (70a), which indicates that in the
SOV order, the universal subject that has undergone A-movement from
Spec,vP to Spec,TP does not reconstruct, making it di‰cult to get a par-
tial negation interpretation.8 Under the PBCC approach, it ought to be
possible for the subject to leave a copy in Spec,vP; hence, a partial nega-
tion interpretation should emerge. One possible reason why it doesn’t is
that, as extensively discussed in chapter 3, this particular movement is
conditioned by either topic or focus, and, at least for focus, it has been
122 Chapter 4

shown that the focused element must be interpreted in its surface position
(É. Kiss 1998). I will assume that the lack of a copy in this instance is not
an exception to the PBCC, but instead results from a particular factor
surrounding the movement, such as topic/focus.

4.6 Summary

In the first part of this chapter, I explored the consequence for the current
approach when both focus (topic) and the f-probe are inherited by a.
Given that the f-probe does not lower to T, it does not interact with
nominative Case. But because a f-probe must depend on something else,
in this case focus, to identify its goal, the current approach predicts that
the goal of the f-probe can be any appropriate DP and not just the sub-
ject, and it must always be a topic that occurs in Spec,aP. This is what we
found for Kinande and Kilega, two languages from the Bantu family. We
also found that in certain instances—apparently those that have operator
movement—Kinande does not allow the aP projection. This forces the f-
probe to be inherited by T. This, in turn, led to a very di¤erent pattern of
agreement in which the goal of the agreement is limited to the grammati-
cal subject because the f-probe depends on nominative Case on T to seek
its goal. We also looked at wh-question formation in Kilega. Kilega has
the unusual property that, in wh-questions, agreement goes with the wh-
phrase at Spec,CP. I analyzed this as a case in which the focus probe
(focus turned into þfocus under agreement) agrees with the wh-phrase
and moves the wh-phrase up to Spec,CP, but on its way the wh-phrase
stops by Spec,aP to value the f-probe on a.
Another language that has both topic/focus and f-feature agreement is
Finnish. However, as we saw, Finnish works di¤erently from Kinande
and Kilega in one important respect. Unlike in the Bantu languages, in
Finnish the f-probe is always inherited by T, so that agreement always
goes with the subject. The topic feature (focus) is inherited by a just as
in the Bantu languages, but because the topic feature and the f-probe are
on di¤erent heads, Finnish allows a topic to occur in Spec,aP that is not
the goal of the f-probe, such as the object.
In the second half of the chapter, I outlined a theory of A- and Ā-
movements that I called the phase-based characterization of chains
(PBCC). According to this approach, we need not specify the landing
site as an A- or Ā-position to predict whether a particular movement is
A- or Ā-movement insofar as A-movement need not leave a fully speci-
fied copy whereas Ā-movement must. The PBCC addresses how the se-
aP, f-Features, and the A/Ā Distinction 123

mantic interface must combine all the transfers that were sent to it into
one whole sentence. The central issue is whether a movement has crossed
a transfer domain boundary. For the two adjacent phases to be combined
properly, a chain that crossed the transfer domain boundary must have a
copy in the lower transfer domain, so that the tail of the chain can be
properly linked with the head in the higher transfer domain. This gives
the typical reconstruction e¤ect. If, however, a movement takes place en-
tirely within a transfer domain, there is no need for a copy, although
there is always the option of leaving one. We saw that the PBCC can
solve a problem that arises with the A/Ā analysis of Takahashi (2006)
and Takahashi and Hulsey (2009), and it also provides an explanation
for the exotic ‘‘mixed’’ position in Finnish.
5 Wh-Questions and Focus

5.1 Wh-Questions

In a wh-question, a question operator at C binds a bound variable. The

bound variable itself is contained in the restriction term that defines the
range of the objects that the bound variable may stand for (people, ob-
jects, etc.). Linguists typically represent this operator–bound variable re-
lation as the Q feature on C binding the bound variable that is contained
in the wh-phrase. Let us assume this. The question I wish to ask is, how
does the Q feature come to bind the relevant element contained in the
wh-phrase? It is commonly believed that the Q feature enters into an
Agree relation with the relevant feature(s) of the wh-phrase (e.g., Baker
1970, Hagstrom 1998, Richards 2001), and this Agree relation makes the
operator–bound variable interpretation possible. It also triggers, in some
fashion that is not well understood, the movement of the wh-phrase to
Spec,CP for those languages that have wh-movement.1
How does the Q feature manage to enter into an Agree relation with
the wh-phrase? There is an assumption that it enters into Agree with an
appropriate feature on the wh-phrase, something like a wh-feature. How-
ever, this is inconsistent with the recent Minimalist Program approach to
Agree, including the approach taken here, because Agree presumes the
existence of an uninterpretable feature, and there is no reason to believe
that Q is an uninterpretable feature. The most sensible way to view it is
as a fully interpretable feature that endows the expression with the
speech-act force of interrogation.
We might take a di¤erent tack and account for the Q-variable relation
as one of unselective binding. One leading assumption that could serve
as the basis for this approach is to view the operator in a wh-question as
an existential quantifier (Karttunen 1977). Heim (1982) proposes that the
typical existential quantifiers (some X ) are themselves not quantifiers;
126 Chapter 5

rather, they are bound variables of quantification. The actual existential

quantifier occurs higher in the structure, and it unselectively binds the
bound variable. This accounts for the surprising behavior of existentials,
including the so-called donkey sentences. Karttunen, refining the ap-
proach to questions by Hamblin (1973), has argued that the quantifica-
tion essential to questions is existential quantification. If we combine this
view of wh-questions with the view of Tsai (1994) and Reinhart (1995)
that the wh-operator, which may or may not be an existential, unselec-
tively binds the wh-phrase to implement the operator-variable chain, we
essentially get a Heim-style approach to quantification being applied to
wh-questions. On this view, Q behaves like an uninterpretable feature
without being one because it must (unselectively) bind a bound variable.
Is this the correct way to view the Q feature? Certainly, the fact that it
appears to be able to bind more than one wh-phrase, as in the multiple-
wh construction (Baker 1970), makes the unselective binding approach
credible.
There is one problem with the unselective binding approach: it does not
explain why the wh-phrase must move to Spec,Q in some languages. In
operator-variable chains, it is never required that the variable must move
to the specifier of the head that hosts the operator. From the perspective
taken here, the movement requirement in wh-questions (at least for the
wh-movement languages) is not a matter of an operator-variable relation;
rather, it points to the occurrence of a probe that attracts its goal (wh-
phrase) by movement. By postulating such a probe for wh-questions, we
can keep the Q feature as a pristine interpretable feature. The Q feature is
not a probe, in other words. We can also account for the movement of
wh-phrases. What goal feature could the probe seek in a wh-question?
In the previous chapters, I gave ample evidence that in addition to f-
features, topic/focus is a probe available universally. I have assumed that
this holds at the TP level. For wh-questions, we are looking at agreement
at the CP level, but is there any reason to expand the range of possible
goals and their features and the corresponding probe? Ideally, we would
want to keep exactly the same possibilities: topic/focus features and f-
features. I will argue for this. Typically, wh-questions are associated with
focus, and, as I will show, this is no accident. Not only is focus semanti-
cally compatible with wh-questions; it also happens to be one of the two
common types of grammatical features available as a probe, the other
being f-features. We saw in chapter 4 how focus operates in wh-questions
in Kilega, and the analysis in this chapter will apply that assumption to
wh-questions in general.
Wh-Questions and Focus 127

In this chapter, I develop an analysis of wh-questions, paying particular

attention to the mechanism that makes the Q-wh construal possible. As I
will show, focus plays a central role in creating this linkage. Along with
wh-questions, I take up the related issue of intervention, which Kim
(2002, 2006) argues is induced by focus, a point also developed by Beck
(2006). I will provide additional evidence for this focus approach to inter-
vention based on my analysis of wh-questions.
To set the stage, let us begin by briefly looking at Watanabe’s (2002)
analysis of wh-movement in Old Japanese. This study, combined with
Hagstrom 1998 and Miyagawa 2001, hints at the basic elements that go
into wh-questions across all languages.

5.2 Old Japanese

Watanabe (2002), basing his analysis on descriptive work by Nomura

(1993), argues that the Old Japanese of the eighth century had overt wh-
movement. The data are drawn from Man’yoshu, an anthology of poems
compiled in the eighth century. Nomura observes that in Old Japanese,
the wh-phrase tends to occur on the left periphery of the clause—
crucially, to the left of the subject phrase. The following example is taken
from Watanabe 2002:182:
(1) Kado tate-te to-mo sashi-taru-wo izuku-yu-ka
gate close-conj door-also shut-past-acc where-through-ka
imo-ga iriki-te yume-ni mie-tsuru? (3117)
wife-nom enter-conj dream-loc appear-perf
‘From where did my wife come and appear in my dream, despite the
fact that I closed the gate and shut the door?’
In this example, the object wh-phrase izuku-yu-ka ‘where-through-ka’
occurs to the left of the subject ‘wife’. What is of particular interest is
that beginning in the ninth century, movement of ka-marked phrases
began to be lost. Apparently, what triggered this loss was separation of
ka from the wh-phrase. In Isobe’s (1990) study of wh-questions in the
Tale of Genji, a psychological novel written by Lady Murasaki Shikibu
in the tenth century, roughly one-third of the examples have ka on the
wh-phrase, whereas in another third, the wh-phrase occurs by itself and
ka appears toward the end of the sentence, just as it does in modern Jap-
anese. The remaining third of the examples are similar to those with
clause-final ka, except that the particle that occurs at the end is a di¤erent
focus particle, zo, as in (2) (Watanabe 2002:187).
128 Chapter 5

(2) Kono nisi-naru ie-ha nani-bito-no sumu-zo?

this west-be house-top what-person-nom live-q
‘What person lives in this house to the west?’
In Watanabe’s (2002) study, we observe Japanese changing from a wh-
movement language to a wh-in-situ one. It is clear from the details of this
change that the trigger is the status of the focus particle. During the
movement stage of the language, a focus particle such as ka attached to
the wh-phrase or to any focused phrase. With the change in the language
that led to ka (or zo) appearing at question (or focus) C, the wh-phrase
and other focus phrases stopped moving. This historical change in one
language embodies the essence of what constitutes wh-questions in gen-
eral. This is the topic I turn to in the next section.

5.3 Wh-Questions Contain a Focus Probe

Returning to wh-questions across languages, recall that one problem I

raised in section 5.1 is that the Q feature on the question C is not uninter-
pretable, so it cannot probe for a feature contained in a wh-phrase, which
would be its goal. Yet we know that some sort of Agree relation holds be-
tween the Q feature and a wh-phrase. We can see this, for example, from
the fact that multiple-wh constructions exhibit superiority e¤ects. What
enables this Agree relation to obtain in a wh-question? Old Japanese pro-
vides an obvious clue: focus. Limiting our discussion to ka, the crucial
factor is the location of the focus particle ka. If it attaches to a wh-phrase,
the wh-phrase moves, but if it occurs on the question C, the wh-phrase
does not move. Let us suppose the following characterization of wh-
questions:
(3) Wh-questions
A focus probe merges on the question C. It enters into Agree with
the focus feature of the closest wh-phrase.
Recall from chapter 3 that the ‘‘focus probe’’ is in fact focus, which, by
itself, is topic, but when it enters into Agree with a focus phrase, it is
valued as þfocus. For simplicity, though, I will refer to the probe as focus
probe. The idea in (3) is that the Q feature, like the f-probe, needs a
mechanism for seeking its goal. We saw that the f-probe typically
depends on Case, although we also saw in Kinande and Kilega that the
topic probe or focus probe can play this role. Viewed in this way, wh-
questions are similar to focus constructions, the one di¤erence being that
wh-questions have a Q feature on C.
Wh-Questions and Focus 129

(4)

The Agree relation establishes a functional relation between the C head

and the wh-phrase, in turn allowing the Q feature on C to establish an
operator-variable relation with the wh-phrase. This is similar to how the
topic feature ‘‘helps’’ the f-probe at a to attain valuation in Kinande. An
advantage to this way of viewing the relationship that the Q feature
establishes with the wh-phrase is that this makes it possible for the Q fea-
ture to enter into multiple agreements that lead to multiple-wh questions.
The focus probe may enter into agreement with however many wh-
phrases exist, and with each instance of agreement, the Q feature is
marked as being linked to the particular wh-phrase.
Three additional points are relevant. First, the relation that is estab-
lished between the Q feature and the wh-phrase via the focus probe is
retained beyond narrow syntax, as is necessary for semantic interpreta-
tion. Although the focus probe on C is erased in the transfer to semantic
interpretation, Q is not erased because it is an interpretable feature. In
this way, it is possible for Q to enter into Agree with multiple wh-phrases,
the relations made possible via multiple agreement by a single focus
probe. The focus probe is ultimately erased, but all of the relations that
Q has established with wh-phrases are retained for semantic interpretation
thanks to the interpretable nature of Q.
Second, why does focus play this role of establishing a functional rela-
tion between the question C and the wh-phrase? There are two reasons.
First, as we have seen in the discussions up to now, focus (and topic) is a
probe that is always potentially available along with the f-probe. Second,
focus and questions share the semantics of alternatives (Beck and Rull-
mann 1998, Hamblin 1973, Hagstrom 1998, Karttunen 1977, Kratzer
and Shimoyama 2002, Ramchand 1997, Rooth 1992, 1996). In fact, if
we excluded the Q feature of questions, questions and focus would be
very similar in their semantic representation, although not necessarily
identical (see Beck 2006).
Third, recall from the discussion of pro-drop that narrow syntax has
two ways of attaining a probe-goal union (PGU). One is to move an XP
130 Chapter 5

‘‘goal’’ to the label of the head that contains the f-probe (Spec-head rela-
tion), and the other is to move an agreement head ‘‘goal’’ to the f-probe
(head movement). The historical change in Japanese wh-questions repre-
sents a change from one (Spec-head) to the other (head movement). The
general pattern here is what Cheng (1991) calls ‘‘clause-typing’’ of ques-
tions. A wh-question is typed as such by either moving a wh-phrase to
Spec,CP or inserting a question particle in lieu of wh-phrase movement.
In Miyagawa 2001, I argued that this clause-typing of questions parallels
the phenomenon that Alexiadou and Anagnostopoulou (1998) have iden-
tified for the EPP at T: either move the goal to Spec,TP or move the rich
verbal morphology to T, a point originally due to Hagstrom (1998) (see
also Landau 2007).
The proposed analysis also explains an otherwise mysterious fact about
wh-questions and focus. In some languages (e.g., Old Japanese, Sinhala;
see Kishimoto 2005), the same verbal inflection agrees with a wh-phrase
or a focus phrase. In Old Japanese, the verb inflected for a variety of
forms; two main forms were conclusive and attributive. For the most part,
conclusive forms occurred in the matrix clause, and attributive forms in
the subordinate clause.
(5) a. Sakana-o tabu. (conclusive)
fish-acc eat
‘(I/She/They/etc.) eat fish.’
b. taburu sakana (attributive)
eat fish
‘the fish that (I/she/they/etc.) eat’
The attributive form occurs unexpectedly in the matrix clause when there
is a focus particle. I demonstrate this for the three focus particles ya,
zo, and ka. (Examples (6a–c) are from Sansom 1928, and (6d) is from
Kokinshuu.)
(6) a. Isi-wa kawa-ni otu. (conclusive)
rock-top river-in fall
‘Rocks fall into the river.’
b. Isi zo kawa-ni oturu. (attributive)
c. Isi ya kawa-ni oturu. (attributive)
d. Ikito si ikeru mono, izure ka uta-o (attributive)
all-the-living things which kakari poem-acc
yomazarikeru.
compose.neg.e(attrib)
‘Every living creature sings.’
Wh-Questions and Focus 131

As we can see, in the presence of zo, ya, or ka, the verb must be in the
attributive form, a form of agreement with the focus particle. This con-
struction is called kakarimusubi. Note that the same ‘‘focus agreement’’
occurs in wh-questions.
(7) Ikani motenai-tamahan-to suru ni-ka?
how treat.give-hon.v-c do loc-ka
‘How is he going to treat me?’
The verb here is inflected for attributive, not conclusive.
For a typical wh-movement language such as English, the analysis pro-
posed here would identify focus on wh-phrases. Because there is no focus
particle that can split from the wh-phrase, the wh-phrase itself must move
to the specifier of the focus probe head to value the focus probe. This is
essentially the analysis of Watanabe (1992) (although he proposes that
what moves is an empty wh-operator) and, of course, Hagstrom (1998)
(although in his view the question particle ka is an existential quantifier,
not focus). Also, Cable (2007) presents an extensive analysis of wh-
questions based on Tlingit, which has both a wh-phrase and a question
particle, similar to Japanese and Sinhala. His analysis independently
develops many ideas that are similar to those in this chapter, particularly
the role of focus in wh-questions and the need for the Q feature (which he
calls ‘‘interrogative’’) to depend on a probe for establishing a link with a
variable (in his case, a choice function).2
To summarize, the semantics of wh-questions contain a question (or
existential) operator and a variable. Because the Q feature on C, which
is the operator, is fully interpretable, it cannot probe for the relevant fea-
ture on the wh-phrase in order to enter into an Agree relation with it. A
focus probe (focus valued as þfocus) is merged on the same C as the
Q feature, and it functions as the probe to link the C head with the wh-
phrase, by entering into an Agree relation with the focus feature on the
wh-phrase.

5.4 Intervention Effects

In a variety of languages, certain items (called intervenors) cause ungram-

maticality if one of them occurs between a wh-phrase in situ and the C at
which the wh-phrase takes scope. Example (8b) shows a typical interven-
tion e¤ect in German.
(8) a. Was glaubt Hans, wen Karl gesehen hat?
what believes Hans whom Karl seen has
‘Who does Hans believe that Karl saw?’
132 Chapter 5

b. *Was glaubt niemand, wen Karl gesehen hat?

what believes nobody whom Karl seen has
‘Who does nobody believe that Karl saw?’
In this wh-question, the ‘‘content’’ wh-phrase is wen ‘whom’; it is this wh-
phrase that must take matrix scope semantically. However, in this partic-
ular case, what occurs in the matrix Spec,CP is was, which in its normal
usage means ‘what’ but here is a placeholder equivalent to an expletive
(see McDaniel 1989). The problem arises with the occurrence of the neg-
ative universal expression niemand ‘nobody’ as shown in (8b), which
somehow blocks the ‘‘real’’ wh-phrase wen ‘whom’ from taking scope at
the matrix C. If we replace niemand with an R-expression, as in (8a), the
example is grammatical (Beck 1996, Rizzi 1992), which suggests that
some lexical property of niemand not shared by R-expressions induces
intervention. The standard view of intervention is that it is caused by a
scope-bearing item. It is, therefore, an e¤ect induced by a quantificational
expression. This is the view found in Hoji 1985, which is the first system-
atic study of intervention. Other studies adopting this view include Beck
1995, 1996, Beck and Kim 1997, Chang 1997, Hagstrom 1998, Miyagawa
1998, Pesetsky 2000, Rizzi 1992, Takahashi 1990, and Tanaka 1997.
More recently, however, a di¤erent view has emerged that points to
focus as the source of intervention e¤ects (Beck 2006, Kim 2002, 2006).
Kim (2002), who first proposed this, observes that not all quantifiers in-
duce an intervention e¤ect (see Beck and Kim 1997) and that those that
do are associated with focus. Following Kim’s work, Beck (2006) makes a
similar observation: crosslinguistically, it is focus expressions such as the
‘only’ phrase that induce the strongest intervention e¤ects. In this section,
I will argue for this focus-based approach to intervention. I will give a
syntactic argument to support the view that only focused phrases in Jap-
anese function as intervenors. This provides further evidence that what I
have called focus in Japanese, ka, is in fact a bearer of focus.
The focused phrases in Japanese relevant to intervention are morpho-
logically transparent; they are existential expressions with the particle ka
(e.g., dare-ka ‘someone’) (Miyagawa 1998) and universal expressions with
the universal particle mo (e.g., dare-mo ‘everyone’) (Hagstrom 1998, Shi-
moyama 2001, 2006). This demonstrates that along with the semantic
function of indicating existence (ka) or universality (mo), these particles
show that the phrase they attach to is associated with focus. The idea
that focus induces intervention receives clear support when we compare,
for example, two universal expressions, one with mo and the other with-
out: only the expression with mo induces intervention. This shows that
Wh-Questions and Focus 133

intervention is not quantificational in nature; rather, it indicates focus.

Relevant examples will follow shortly.
Beck (2006) provides a formal semantic account of Kim’s original
‘‘focus’’ insight. I will adopt the core of their insight—that intervention
is a matter of focus. I will capture this within a focus-probe approach to
wh-questions. After reviewing the intervention data in Japanese, I will
show that the syntactic approach that Kim (2006) proposes follows
straightforwardly from the focus-probe approach.

5.4.1 Quantifier-Induced Barrier

Abstracting away from the details of their analyses, the linguists adopting
the standard view of intervention e¤ects see an intervenor as blocking
movement of the wh-expression, as the German example (8b) illustrates.
Under this analysis, in (8b), wen ‘whom’ must move to the matrix
Spec,CP for proper interpretation of the wh-question, but the intervenor,
niemand ‘nobody’, somehow blocks this movement. The following is an
example from Japanese (Hoji (1985) first noted this type of example):3
(9) *Daremo-ga nani-o katta no?
everyone-nom what-acc bought q
‘What did everyone buy?’
Here the intervenor is the quantifier daremo ‘everyone’, which is com-
posed of the indeterminate pronoun/wh-phrase dare ‘who’ and the par-
ticle mo, which creates universal expressions when combined with an
indeterminate pronoun. In both German and Japanese, the ostensible
wh-movement that is blocked by an intervenor would be covert move-
ment (see Huang 1982). Does an intervenor block overt movement? Ex-
ample (10) shows that the answer is no, given that overt scrambling of
the wh-phrase over the universal quantifier leads to a grammatical ques-
tion (Hoji 1985).
(10) Nani-oi daremo-ga ti katta no?
what-acc everyone-nom bought q
‘What did everyone buy?’
The same covert/overt di¤erence emerges in the following French
examples from Chang 1997:
(11) a. *?Jean ne mange pas quoi?
Jean neg eat neg what
b. Que ne mange-t-il pas?
what neg eat-he neg
‘What doesn’t Jean eat?’
134 Chapter 5

In (11a), the wh-phrase presumably stays in situ, a type of construction

allowed in the root clause in French. The intervenor ne ( pas) blocks co-
vert raising of quoi to Spec,CP. In (11b), the overt movement of que to
Spec,CP is not blocked by the negative intervenor.
What are the items in a language that induce an intervention e¤ect?
Hoji (1985) explicitly identifies quantifiers as blocking covert wh-
movement. He notes that if a quantifier c-commands a wh-phrase, and
the quantifier itself is c-commanded by the C associated with the question
force, the sentence is ungrammatical because the wh-phrase must raise
covertly over the quantifier.
(12) *[ . . . [C[þQ] [TP . . . quantifier . . . wh-phrase . . . ]]]
Hoji reduces this ungrammaticality to the language-specific property of
quantifier scope rigidity (Kuroda 1971)—namely, the property that only
surface scope is permitted at LF for scope-bearing items including wh-
phrases. In (12), the quantifier undergoes Quantifier Raising (QR) to TP,
but the wh-phrase must raise to Spec,CP, thereby reversing the surface
c-command relation in violation of scope rigidity.
Beck (1996) formalizes the intervention e¤ect as follows (see also Beck
and Kim 1997):
(13) Quantifier-induced barrier (QUIB)
The first node that dominates a quantifier, its restriction, and its
nuclear scope is a quantifier-induced barrier.
(14)

In (14), a is a quantifier, and the node that immediately dominates it is a

QUIB that blocks movement of the XP from the position of b. Beck fur-
ther states that the type of chain susceptible to a QUIB is created at LF,
by covert movement. This presumably is because, in this type of approach,
scope-taking quantifiers induce the QUIB e¤ect, and scope-taking is typi-
cally viewed as an LF phenomenon (e.g., QR). By specifying that only
covert movement is blocked by a QUIB, this approach takes account of
the fact that a QUIB does not block overt movement of a wh-phrase. I
should note, however, that in the end, we are not certain why this overt/
covert distinction exists.
Wh-Questions and Focus 135

Some of the analyses mentioned above address the overt/covert issue

by assuming a neutral stance on whether the wh-movement is overt or co-
vert. Rizzi (1992) analyzes the German examples in (8) as involving the
movement of a piece of the ‘‘real’’ wh-phrase, possibly the wh-operator
portion that is not pronounced, as in Watanabe 1992. Because the wh-
operator portion is just a piece of the argument wh-phrase wen ‘whom’,
this movement does not count as movement of an argument; hence, the
movement is subject to the Empty Category Principle, a version of which
is Rizzi’s (1990) Relativized Minimality. Because there is a quantifier in
(8b), that quantifier is closer to the wh-phrase than C, which causes a
Minimality violation. Taking a similar tack, Pesetsky (2000) assumes
that what moves in the wh-in-situ construction is the wh-feature; what
remains is the semantic restriction of the wh-phrase. He proposes the
following constraint to account for the intervention e¤ect (Pesetsky
2000:67):4
(15) Intervention e¤ect (universal characterization)
A semantic restriction on a quantifier (including wh) may not be
separated from that quantifier by a scope-bearing element.
Both Rizzi’s and Pesetsky’s accounts depend on the intervenor being a
quantifier. Hagstrom’s (1998) analysis is also developed along these lines,
except that in his case what moves in wh-in-situ constructions is the ques-
tion particle. As noted earlier, Cheng (1991) observes that languages
often have a way to overtly mark a wh-question, either by moving the
wh-phrase to the head of the sentence or, if the wh-phrase does not move,
by marking the construction with a question particle. Hagstrom (1998)
argues that these two movements are equivalent; in both cases, something
moves to C, and it is the question particle that is subject to intervention
e¤ects (see his work for details)—again, assuming that the intervenor is a
quantifier.

5.4.2 An Alternative View of Intervention

In discussing her own work with Beck (Beck and Kim 1997), Kim (2002)
argues that the earlier work has a serious flaw—namely, not all quan-
tifiers induce an intervention e¤ect (Beck and Kim 1997:369–372).
Although negative polarity items (NPIs) and a handful of other scope-
bearing elements are intervenors, the following quantificational expres-
sions do not induce intervention in Korean: taepupun-ui N ‘most N’,
hangsang ‘always’, and chachu ‘often’.5 These happen not to be associated
with focus, and this leads Kim to the focus-based analysis of intervention.
136 Chapter 5

In the remainder of this chapter, I argue that the focus approach to in-
tervention is correct and that it gives evidence that there is focus in wh-
questions that is identical to the focus in focused expressions such as mo
and ka phrases.

5.4.2.1 Focus Barriers To substantiate the focus approach to interven-

tion, let us look at some data from Japanese. The intervenors in Japanese
are morphologically transparent: they include either the existential par-
ticle ka (Miyagawa 1998) or the universal particle mo (Hagstrom 1998).
The following is, I believe, an exhaustive, or close to exhaustive, set of
intervenors in Japanese:
(16) With universal mo
Universal quantifier (Hoji 1985)
?*Daremo-ga nani-o katta no?
everyone-nom what-acc bought q
‘Everyone bought what?’
(17) With ka
a. NPI sika-nai ‘only’ (Takahashi 1990)6
*Taroo-sika nani-o kawa-nakat-ta no?
Taro-only what-acc buy-neg-past q
‘What did only Taro buy?’
b. Existential quantifier (Hoji 1985)
??Dareka-ga nani-o katta no?
someone-nom what-acc bought q
‘Someone bought what?’
The e¤ect of any of these intervenors can be overcome by overtly moving
the wh-phrase over it. I demonstrate this for the NPI sika-nai ‘only’ in
(18).
(18) NPI sika-nai ‘only’
Nani-oi Taroo-sika ti kawa-nakat-ta no?
what-acc Taro-only buy-neg-past q
‘What did only Taro buy?’

5.4.2.2 Mo and Ka Phrases Are Focus We have already seen that mo

phrases are associated with focus. As noted in chapter 1, the mo expres-
sion carries focus stress. Note the minimal pair in (19).
(19) a. Taroo-wa HON-o katta.
Taro-top book-acc bought
‘Taro bought a book.’
Wh-Questions and Focus 137

b. TAROO-mo hon-o katta.

Taro-also book-acc bought
‘Taro also bought a book.’
In (19a), which has no special focused element, the object receives the de-
fault prominence. In (19b), the mo subject phrase attracts this promi-
nence, showing that it carries focus. Moreover, as Hasegawa (1991, 1994)
points out, when occurring with sentential negation, the mo phrase is
interpreted outside the scope of negation.
(20) a. John-mo ko-nakat-ta.
John-also come-neg-past
‘John (in addition to someone else) did not come.’
b. John-ga hon-mo kaw-anakat-ta.
John-nom book-also buy-neg-past
‘A book is one of things that John did not buy.’
Example (20a) has only the interpretation that there is at least one person
who did not come besides John. It does not mean that someone came, but
John didn’t come as well, which would be the interpretation if the mo
phrase were inside the negative scope.7 Likewise, (20b) means only that
John did not buy something besides a book; it does not mean that John
bought something but not also a book. I have argued that this phenome-
non whereby the mo phrase takes scope outside the sentential negation
of its clause indicates that movement has occurred (see Hasegawa 2005).
Movement, in turn, shows that there is a probe; and in the case of mo, the
probe agrees with the focus feature on the mo phrase.

5.4.2.3 Intervenors Are Focused Phrases If we apply the ‘‘focus’’ test to

the intervenors exemplified in (16) and (17), we see that each one has the
property Hasegawa (1991, 1994) identified—namely, it is interpreted out-
side the scope of negation (Hasegawa 1991 notes this for ka as well as
mo).
(21) With universal mo
a. Universal quantifier (Hoji 1985)
Taroo-ga daremo-o mi-na-akat-ta.
Taro-nom everyone-acc see-neg-past
‘Taro did not see everyone.’
*not > everyone, everyone > not
b. ‘Almost every’ (Miyagawa 1998)
Taroo-ga hotondo daremo-o mi-nakat-ta.
Taro-nom almost everyone-acc see-neg-past
138 Chapter 5

‘Taro did not see almost everyone.’

*not > almost every, almost every > not
c. ‘Also’
Taroo-ga hon-mo kaw-anakat-ta.
Taro-nom book-too buy-neg-past
‘Taro did not also buy a book.’
*not > also, also > not
(22) With ka
a. NPI sika-nai ‘only’ (Takahashi 1990)
Not applicable because this is an NPI.
b. Existential quantifier (Hoji 1985)
Taroo-ga dareka-o mi-nakat-ta.
Taro-nom someone-acc see-neg-past
‘Taro did not see someone.’
*not > someone, someone > not
c. Disjunction ka (Hoji 1985)
Taroo-ga [John-ka Mary]-o mi-nakat-ta.
Taro-nom John-or Mary-acc see-neg-past
‘Taro did not see John or Mary.’
or > not, *not > or
As shown, all mo and ka phrases are focused and enter into Agree with
the focus probe on T, making it mandatory for them to move to Spec,TP
and be interpreted outside sentential negation. This explains why, in
(22c), disjunction ‘or’ in Japanese does not have the ‘universal’ reading
associated with or in English (Taro didn’t see John or Mary). Although
the object disjunctive phrase in English may stay in object position, where
it is in the scope of negation that leads to the universal interpretation of
or, in Japanese, the focus probe that enters into Agree with focus of the
disjunction phrase with ka ‘or’ forces this phrase to raise to Spec,TP and
outside the scope of negation, thus depriving it of the universal reading.8

5.4.2.4 Nonfocused Quantifiers Are Not Intervenors We have seen that

universal expressions formed with mo and existential expressions formed
with ka are focused expressions that raise to Spec,TP. Following Kim
(2002, 2006) and Beck (2006), I assume that it is this focus property, and
not the quantificational nature of these expressions, that makes them
intervenors. One clear piece of evidence for this is found with expressions
that are very close in meaning to the intervenors. For example, the word
minna ‘all/everyone’ has a meaning similar to that of the intervenor
Wh-Questions and Focus 139

daremo ‘everyone’, yet it does not induce an intervention e¤ect (Hoji

1985).
(23) Minna-ga nani-o katta no?
all/everyone-nom what-acc bought q
‘What did all/everyone buy?’
Minna is not formed with one of the ‘‘focus’’ particles (e.g., mo), so it is
not lexically marked for focus.9 One possible counterargument is that
minna in (23) has the group reading common to the word all, hence it
is not behaving like a quantifier, so (23) really doesn’t tell us that it is
focus, not quantification, that is relevant for intervention. We can re-
spond to this with (24), in which minna clearly has distributional—hence
quantificational—meaning.
(24) Minna-ga sorezore nani-o katta no?
everyone-nom each what-acc bought q
‘What did everyone each buy?’
This question is most naturally interpreted as a pair-list question, which
signifies that ‘what’ distributes over the members of the set designated by
minna sorezore ‘everyone each’. This indicates that minna functions as a
quantifier. Even so, minna does not induce an intervention e¤ect.10
Kim (2002) (see also Beck and Kim 1997) points out that the Korean
expressions taepupun-ui N ‘most N’, chachu ‘often’, and hangsang
‘always’ are not intervenors despite their quantificational nature. In Jap-
anese, too, these do not induce an intervention e¤ect.
(25) a. Hotondo-no hito-ga nani-o katta no?
most-gen people-nom what-acc bought q
‘What did most people buy?’
b. Hanako-ga yoku nani-o asa taberu no?
Hanako-nom often what-acc morning eat q
‘What does Hanako often eat in the morning?’
c. Taroo-ga itu-mo nani-o asa taberu no?
Taro-nom always what-acc morning eat q
‘What does Taro always eat in the morning?’
Neither hotondo-no N ‘most N’ nor yoku ‘often’ is a focused expression,
in that neither contains a particle equivalent to mo or ka. What is surpris-
ing is itu-mo ‘always’, which is composed of the indeterminate pronoun
itu ‘when’ and the universal particle mo. This has the same morphology
as intervenors such as dare-mo ‘everyone’, yet it apparently does not in-
140 Chapter 5

duce an intervention e¤ect. I do not know why itu-mo is able to be free of

focus, but the facts are clear.
There are two di¤erences, very much related, between itu-mo ‘always’
and an intervenor like dare-mo ‘everyone’. First, unlike dare-mo, itu-mo
does not attract stress in sentences with neutral prosody. This means that
focus does not arise every time mo appears.
(26) Taroo-ga itu-mo PIZA-o taberu.
Taro-nom always pizza-acc eat
‘Taro always eats pizza.’
This suggests that itu-mo ‘always’ is not inherently focused. This in turn
predicts that it is not the goal of a focus probe. Example (27) shows that
this prediction is borne out.
(27) Taroo-ga itu-mo piza-o tabe-nai.
Taro-nom always pizza-acc eat-neg
‘Taro does not always eat pizza.’
not > always, always > not
As shown, itu-mo can be interpreted inside the scope of local sentential
negation (‘not > always’), something that is di‰cult, if not impossible,
with dare-mo ‘everyone’. This shows that itu-mo may stay in situ in a po-
sition under Spec,TP. For the other reading, ‘always > not’, one could
imagine that itu-mo undergoes QR, which it is free to do because it does
not enter into an Agree relation with a focus probe.

5.4.3 Intervention as a Relativized Minimality Violation

In this section, I analyze intervention e¤ects as being induced by focus.
Drawing on data from Korean and Hindi (see Lahiri 1998 for Hindi),
Kim (2002) argues that the true source of intervention e¤ects is focus.
Later (Kim 2006), she provides a syntactic approach to intervention by
incorporating into her analysis the notions of interpretable and uninter-
pretable features.
(28) *[CP C[iQ, iF] [ . . . Foc[iF] . . . [ . . . [wh[uQ, uF] . . . ]]]]
!

The question C contains the interpretable Q and interpretable focus fea-

tures, and they must be linked together to their uninterpretable counter-
parts contained in a wh-phrase. The problem with (28) is that it violates
Relativized Minimality: the focus feature on C finds as its closest counter-
Wh-Questions and Focus 141

part the interpretable focus feature on the focus expression Foc instead of
on the wh-phrase.11
We can improve on Kim’s Relativized Minimality approach by simpli-
fying the features that are involved, and by taking into account what we
found with the focused ka and mo phrases: that they enter into an Agree
relation with a focus probe. Let us suppose that the focus feature on the
focused phrase—the intervenor—and the focus feature on the wh-phrase
are identical, contrary to Kim’s analysis in (28). This makes intuitive the
fact that they interact in a way that induces a Relativized Minimality vi-
olation under certain structural conditions. Also, each focus feature in the
structure requires a focus probe to enter into an Agree relation. I propose
(29) as the initial structure that leads to intervention. Following Hag-
strom (1998), I will assume that the question particle ka is merged next
to the wh-phrase. In giving the structure, I will arbitrarily use the head-
initial representation.
(29) CQ, focus probe, focus probe [ . . . T . . . XPfocus . . . wh-phrase-kafocus ]
There are two focus probes on C: one for the focus phrase, the other for
the wh-phrase. They enter into Agree with the respective phrases that
carry focus.
(30) CQ, focus probe, focus probe [ . . . T . . . XPfocus . . . wh-phrase-kafocus . . . ]
!

Presumably, the first Agree relation is with the focus XP, and the second
is with the focus feature on the wh-phrase. Conceivably, the second Agree
relation may be considered to be impossible owing to a defective interven-
tion e¤ect, but let us keep to the simplest assumption, namely, that once
the first Agree relation is established, the focus feature on the XP is neu-
tralized for the purposes of Agree, and the second Agree relation can then
establish itself across the focus XP. At this point, one of the focus probes
is inherited by T.
(31) CQ, focus probe [TP . . . Tfocus probe . . . XPfocus . . . wh-phrase-kafocus . . . ]
!

At this point, the goal must move to the probe for PGU. The focus XP
encounters no problem; it moves to Spec,TP. But a problem does arise
for the wh-phrase. The focus particle ka, merged by the wh-phrase, must
142 Chapter 5

move to CQ, focus probe , but the closest focus probe is the focus probe on T.
This focus probe on T blocks ka from moving all the way to C, thereby
prohibiting the functional relation between the focus probe on C and the
focus feature on the wh-phrase that is needed to attain PGU.12 In con-
trast, if the wh-phrase with ka scrambles above the TP, the intervention
is avoided. I presume that this phrase lands in aP.
(32) CQ, focus probe [aP wh-phrase-kafocus [TP . . . Tfocus probe . . . XPfocus . . . ]]

!
From this position, ka is able to move to C and implement a PGU with
the appropriate focus probe at C.

5.5 Summary

In this chapter, I argued that wh-questions contain a focus probe (focus

that is valued as þfocus) at C that agrees with the focus feature on the
wh-phrase. This makes it possible for the Q feature at C to enter into a
relation with the wh-feature on the wh-phrase. The focus probe is neces-
sary for Q to acquire this relation because Q, being an interpretable fea-
ture, cannot probe for a goal on its own. The focus probe can enter into
an agreement relation with one or more wh-phrases. In each case, Q,
which is not a probe, benefits by having a true probe (focus) create the
needed relation. Even after the focus probe is erased, Q retains its relation
to the wh-phrase and this makes the appropriate semantic interpretation
possible. Q is retained beyond narrow syntax because of its interpretable
nature. The movement of the wh-phrase is due to the focus portion of
the construction, which requires PGU to take place. Likewise, in the wh-
in-situ language Japanese, movement takes place, but what moves is
the minimal focus element, ka. We saw that historically, in terms of wh-
questions, Japanese changed from a wh-phrase movement language to the
current ‘‘particle’’ movement (wh-in-situ) language. Finally, in an exten-
sive look at intervention e¤ects, I showed that the recent focus-based
approach to intervention provides another argument for the occurrence
of the focus probe in wh-questions.
6 Concluding Remarks

I set out to try to provide plausible answers to the questions ‘‘Why is

there agreement in human language?’’ and ‘‘Why is there movement?’’ If
any of what I argued for is on the right track, the real answer, setting
aside the intricacies of how agreement and movement work in narrow
syntax, is that their interaction imbues language with the enormous
expressiveness that makes it what it is. Without them, we humans would
be abysmally handicapped in expressing our thoughts. Topic-comment
(or theme-rheme), subject of a clause, focus, content questions, and other
notions would be absent from our cognitive life as far as language is
concerned.
For this system to work properly, all the pieces must fit together per-
fectly; one piece missing, and the entire system falls flat. The sparseness I
tried to attain—and I’m sure there are a number of components that can
be simplified further with deeper understanding—led to some open and
critical questions that I must leave for future study. But my hope is that
where I was able to provide answers, those answers proved to be deeper
and more interesting because I attempted to capture them with as few
assumptions as possible.
Notes

Preface

1. Relational Grammar (RG) also has a linearization component, although it is

nowhere as extensively developed or formalized as the linearization component
in Kayne’s work. Blake (1990) describes the RG linearization as follows, basing
his characterization in part on Perlmutter’s comments in the afterword to Perl-
mutter and Postal 1983. The need for linearization arises because ‘‘[r]elational net-
works are unordered representations’’ (Blake 1990:20), all relations being strictly
hierarchical, and linearization occurs as the last step in the derivation of a
sentence.

Chapter 1

1. Fassi Fehri (1993:34–38) provides a brief overview of asymmetries found in a

number of languages. Thanks to Omer Preminger for pointing me to this work.
2. In these dialects, tonic pronouns and full NPs in preverbal position are accom-
panied by a subject clitic that also agrees with the subject. The occurrence of
this clitic has no bearing on my argument that movement is correlated with full
agreement. See Brandi and Cordin 1989 for discussion of subject clitics in these
dialects.
3. As mentioned, a reasonable way to view (7a) (There appears to be a man in the
garden) is that nothing appears in the embedded Spec,TP, and nothing is ‘‘raised,’’
but instead, the expletive is inserted directly into Spec,TP of the root clause. This
would mean that in the following example, where the subject Mary clearly has
raised from the embedded clause, it did not go through Spec,TP:
(i) Maryi seems to him to like herselfi (*heri ).
Presumably, Mary raises directly from the lower Spec,vP to the matrix Spec,TP.
Thanks to Noam Chomsky for bringing examples like these to my attention.
4. As Noam Chomsky (pers. comm.) has noted, there are questions about the va-
lidity of structure preservation, as in the case of the structure that arises from ob-
ject shift or from stacking of wh-phrases found in some languages (e.g., Richards
146 Notes

2001). If these turn out to be genuine exceptions to structure preservation, we will

need to qualify the idea that Merge, external and internal, is a unified operation. I
will leave this issue open.
5. There is another, equally important function of agreement. As I explain later
in this chapter, probes (‘‘uninterpretable features’’) are initially merged only on
phase heads such as C and v (Chomsky 2005, 2007, 2008). In essence, therefore,
agreement, and specifically probes, define phases, which are the domains of com-
putation in narrow syntax. A particularly significant point is that agreement
appears to keep phases as small as possible (Chomsky 2007), possibly for compu-
tational e‰ciency and simplicity. I will not discuss this function of agreement in
this monograph, although it is important to note that it complements what I will
discuss, and does not in any way contradict it.
6. A proposal consonant with mine is Chung’s (1998) notion of associate relation.
Associate relations are very close to—possibly the same as—the functional rela-
tions that I propose. The associate relation is basically a relation that holds be-
tween a functional head and its specifier or a projection of the functional head
(Chung 1998:179). Chung describes this relation of a nominal to a functional
head as critical to agreement, a proposal I concur with. Like the functional rela-
tion, the associate relation establishes a relation between a nominal (mostly) and a
functional head, and it ‘‘contributes significantly to syntactic licensing’’ (Chung
1998:5). I take this notion of syntactic licensing to be equivalent to (possibly iden-
tical with) the notion of functional relation, whereby a nominal is licensed within
the larger syntactic-expression clause in which it occurs. Although Chung does
not give a clear motivation for the associate relation, if the proposal in this mono-
graph is in any way on the right track, the evidence I present for the existence of
agreement will also be evidence for the associate relation.
7. A reviewer notes that agreement sometimes shows up in nominal clauses, and
it is not always certain whether this agreement involves the D head. I will assume
that such agreement always involves the D head or some other functional relation,
but I will not attempt to defend this assumption here.
8. A number of linguists have pointed out to me that the indeterminate pronoun
is associated with two distinct prominence patterns, the ‘‘normal’’ pattern where
the first mora in dare ‘who’ receives prominence (DAre) and a second pattern
where no prominence is associated with this word. The first pattern leads to the
ungrammatical judgment of (20) noted by Kishimoto (2001), but the second pat-
tern improves the sentence. If this is indeed the case, then we can surmise that the
first pattern associates focus with the indeterminate pronoun, forcing it to raise to
Spec,TP, but the second pattern does not, allowing the indeterminate pronoun to
stay in situ within the scope of mo. This would be a further confirmation that
focus forces movement. Throughout this monograph, I will use the first pattern.
9. In other work, Chomsky (2007), adopting a suggestion by Marc Richards,
argues that the need for inheritance arises from the probe-goal system within the
phase architecture of narrow syntax. The idea is that an uninterpretable feature
(probe) must be erased before being transferred to semantic interpretation. In the
probe-goal system, the goal values the probe, and it is assumed that, once valued,
Notes 147

the uninterpretable feature looks exactly like the interpretable feature of the goal.
To ensure that the system can identify which features were formerly uninter-
pretable features, the probe and the goal must be in the same transfer (spell-out)
domain. Because the complement of the phase head (e.g., TP, which is the com-
plement of the phase head C) undergoes transfer, inheritance is needed to place
the probe as well as the goal in the same transfer domain (e.g., TP). Since in later
chapters I will generalize the probe-goal system for agreement not only at TP, but
also at CP, I am unable to fully adopt the approach just described; instead, I will
continue to make the earlier assumption that inheritance occurs to make A-chains
available to language.
10. When the object becomes the agreed-with phrase, it moves from within VP to
Spec,vP. I will give arguments showing why the object, being part of agreement,
must move to Spec,vP. This will be one of the central themes of this monograph.
On a phase-based approach to movement, this is not the only movement that
occurs, however. The object presumably first moves from within VP to Spec,vP,
the edge of the first phase. What drives this movement? It is clearly not triggered
by agreement. Chomsky (2007) proposes that all heads may be associated with
what he calls an edge feature that attracts an element to it, and that this edge fea-
ture is independent of any other feature such as agreement. I will not attempt to
justify this proposal, but will simply assume it for these ‘‘intermediate’’ move-
ments that set the stage for the kinds of movement I wish to deal with.
11. There are some potential problems with the idea that f-probes require Case.
For example, in Greek (Iatridou 1993), subject-verb agreement takes place in sub-
junctive as well as finite clauses, the former apparently without tense. There are a
number of possible solutions for these problems, possibly along the lines of those
proposed for Chinese in chapter 2 and Bantu in chapter 4. I will leave this open.
See George and Kornfilt 1981 for similar issues related to Case and agreement.

Chapter 2

1. In some ways, this view recalls the model of movement found in the earliest
minimalist literature (e.g., Chomsky 1993), where movement takes place to
‘‘check’’ a feature. As part of this checking process, the entire XP that contains
the relevant feature is pied-piped, because phonology does not allow just the fea-
ture to raise. That is also true in the present proposal.
2. In indicating that movement has occurred in the examples, I will use the GB
practice of marking the original position of the moved item with t (trace). Not
only is this still the most widely used practice; it also stays neutral about the na-
ture of the entity in this original position—a full copy or something reduced. In
chapter 4, I will propose a way to predict the nature of the entity in the original
position.
Also, I will use bar levels such as V, V 0 , and VP, strictly for expository
purposes.
3. As a reviewer notes, whereas *the a book is ungrammatical, Sabel’s approach
has to allow [there a book]. On Sabel’s account, the and there are both D, so a way
148 Notes

must be found to block one from occurring with a (*the a) but not the other (there
a). Moro’s account is superior in this regard since it identifies the expletive as a
predicate and not D.
4. I have focused on the there construction and have ignored the other expletive,
it. I will simply mention that there is an analysis of expletive it that is similar to
the analysis of there that I have described. The following is a typical expletive con-
struction with it (Akmajian and Heny 1975:280).
(i) It is obvious that the world is round.
Stroik (1996) suggests that it in this construction begins in the Spec,CP of the
embedded clause and raises to the matrix Spec,TP. One piece of evidence for this
is the asymmetry shown in (ii) and (iii).
(ii) a. I just knew that Mary would fire John today.
b. I just knew it that Mary would fire John today.
(iii) a. I just knew where Mary would fire John today.
b. *I just knew it where Mary would fire John today.
The reason for the ungrammaticality of (iiib) is that it and where both occur in
Spec,CP, violating the Doubly Filled Comp Filter; doubly filled Comp is not a
problem in (iib) because that is a head. Although many details need to be filled
in, this analysis of expletive it parallels the analysis given in the text for there in
the sense that it is the goal with the interpretable feature and that it values the
probe and moves to the probe under PGU.
5. The idea that a proxy for the goal meets the PGU requirement may also be
applicable to the well-known que/qui facts in French (Rizzi 1990). The following
examples are taken from Taraldsen 2002:29:
(i) Quel livre crois-tu que/*qui les filles vont acheter?
which book think-you that the girls will buy
‘Which book do you think that the girls will buy?’
(ii) Quelles filles crois-tu *que/qui vont acheter ce livre-là?
which girls think-you that will buy that book-there
‘Which girls do you think will buy that book?’
Rizzi suggests that the qui form of the complementizer agrees with the subject wh-
phrase and that this agreement licenses the movement of the subject wh-phrase
into the lower Spec,CP.
Taraldsen (2002) has argued that qui is que plus the expletive i.
(iii) quelles filles . . . [CP que [IP i vontþI . . . ]]
The subject wh-phrase starts out lower in the structure, as in Italian (see Rizzi
1990, Taraldsen 2002). On this account, i is a proxy for the goal and, as such, is
able to attain PGU with the embedded T’s probe. This analysis brings the que/qui
alternation into line with the resumptive/expletive construction in Danish and
Swedish, overcoming a potential that-t violation. See Taraldsen 2002 for details.
6. The stipulation that when there is a complementizer, the subject wh-phrase is
not allowed to move into the Spec,TP below the complementizer also can account
Notes 149

for typical cases of antiagreement (thanks to a reviewer for suggesting that I look
at antiagreement in Berber). Berber has subject-verb agreement, with the agreeing
inflection on the verb showing up as both pre-stem and post-stem (Ouhalla 1993,
2005). The following is a simple example. (Unless otherwise noted, the data are
from Ouhalla 1993, 2005.)
(i) Lsa-nt tifruxin ijllabn.
wore-3f.pl girls jellabas
‘The girls wore jellabas.’
As Ouhalla notes, when the subject is extracted in relative clause and cleft con-
structions, the agreement inflection cannot show up, resulting in what is com-
monly called antiagreement.
(ii) TAFRUXT ay sqad-n/*t-sqad tabratt.
girl c send-part/3f-send letter
‘It was the girl who sent the letter.’
(iii) SHEK ay iuggur-n/*t-ggurt-t.
you.m.sg c leave-part/2m-leave-2m.sg
‘YOU are the one who left.’
Examples (iv)–(vi), taken from Elouazizi 2005:122–123, show the contrast be-
tween subject and object extractions, where subject extraction leads to antiagree-
ment (AA) but object extraction does not.
(iv) Wh-questions
a. Uw (g) y-w§i-n/(*y-w§a) lktab (AA)
who X part-give.perf-part/(*3m.sg-give.perf) book
i Mena?
to Mena
‘Who gave the book to Mena?’
b. Min y-w§a/(*y-w§i-n) Jamal (no AA)
what 3m.sg-give.perf/(*part-give.perf-part) Jamal
i Mena?
to Mena
‘What did Jamal give to Mena?’
(v) Relative clauses
a. Y-ssqad w-ar yaz yabrat i Mena. (no AA)
3m.sg-send.perf cs-man.m.sg letter to Mena
‘The man sent the letter to Mena.’
b. Zri-x ar yaz i (g) (AA)
see.perf-1sg man.m.sg rm X
y-ssqad-n/(*y-ssqad) yabrat i Mena.
part-send.perf-part/(*3m.sg-send.perf) letter to Mena
‘I saw the man who sent the letter to Mena.’
c. yabrat i (g) y-ssqad/(*y-ssqad-n) (no AA)
letter rm X 3m.sg-send.perf/(*part-send.perf-part)
w-ar yaz i Mena y-x§§ d.
cs-man.m.sg to Mena 3f.sg-arrive.perf cl.dir
‘The letter which the man sent to Mena has arrived.’
150 Notes

(vi) Cleft constructions

a. Y-ssqad w-ar yaz yabrat. (no AA)
3m.sg-send.perf cs-man.m.sg letter
‘The man sent the/a letter.’
b. (¶) ar yaz i (g) y-ssqad-n/(*y-ssqad) (AA)
cop man.m.sg cm X part-send.perf-part/(*3m.sg-send.perf)
yabrat.
letter
‘It is the man who sent the letter.’
c. (T) yabrat i (g) y-ssqad w-ar yaz. (no AA)
cop letter cm X 3m.sg-send.perf cs-man.m.sg
‘It is a letter that the man sent.’
Why is agreement suppressed when the subject, but not the object, wh-phrase is
extracted to Spec,CP? I hypothesize that movement is triggered by the need to
attain PGU. In the subject extraction cases, the subject wh-phrase has been
extracted across a complementizer, and this complementizer, by stipulation, didn’t
allow the wh-phrase to move to Spec,TP below the complementizer. As a result,
PGU is not attained at the point where this TP is transferred.
7. A reviewer notes that at least for languages such as Italian, the idea that
Spec,TP is never filled has been disputed.
8. I thank Jim Huang and Dylan Tsai for assistance with this section. All mis-
takes are my responsibility.
9. For additional evidence that Chinese has tense, see Tsai, to appear. Tsai argues
that, for tense, Chinese employs morphosyntactic measures instead of the mor-
phological representation typically found in languages where tense is clearly
marked. He calls this ‘‘tense anchoring.’’ From this perspective, Chinese is a syn-
tactic tense language, and the di¤erence between English and Chinese lies in the
way they realize the underlying event argument for tense operator binding—
English by morphology and Chinese through syntax.
10. See Xu 1986 for a di¤erent approach to these examples, and for counter-
examples.
11. Pan (2000) notes another possible issue regarding the blocking e¤ect—
namely, not all person conflicts lead to a blocking e¤ect. If the local antecedent
is third person, then a nonthird person in the higher clause can function as the
antecedent.
(i) a. Woi zhidao Lisij bu xihuan ziji?i=j .
I know Lisi not like self
‘I knew that Lisi did not like me/himself.’
b. Nii xiang mei xiang guo Lisij conglai jiu mei xihuan guo ziji?i=j ?
you think not think asp Lisi never conj not like asp self
‘Have you ever thought about the idea that Lisi never liked you/himself?’
c. Woi yizhi yiwei Zhangsanj xihuan zijii=j , keshi wo cuo le.
I so.far think Zhangsan like self but I wrong prt
‘I always thought that Zhangsan liked me/himself, but I was wrong.’
Notes 151

In these examples, the lower subject is third person, and it does not block a first-
or second-person subject in the higher clause from functioning as the antecedent.
This is not at all inconsistent with what is known about person agreement. In
Finnish, for example, which is a pro-drop language, first- and second-person
agreements, but not third-person, sanction pro-drop (Holmberg 2005). Observing
these types of asymmetries between first/second and third person, Alexiadou
(2003:25) comments that ‘‘in some languages 3rd person is actually ‘non-person’.’’
There are two problems with Pan’s observation, however. First, if it were cor-
rect, we would expect subject pro-drop in Chinese to behave like subject pro-drop
in Finnish in being licensed only by first and second person, third person being a
‘‘nonperson.’’ But the data on pro-drop in the literature do not show this distinc-
tion, allowing pro-drop with all persons. Second, a number of native Chinese
speakers consulted about the data strongly favored the local antecedent, suggest-
ing that for these speakers, there is a blocking e¤ect even with a third-person
nominal.
On the other hand, the possibility that first and second person function as the
genuine agreement in Chinese is interesting given that there is no morphological
manifestation of this agreement and that the language learner has to figure it out
essentially from the default setting of Universal Grammar. Is this default setting
not just that person agreement is the primary agreement, but in fact that within
this category, first- and second-person agreement, which might reflect the dis-
course nature of the origin of agreement, hold primacy? I leave this issue open.
12. Another well-known case of LDA is found in Icelandic. I will note the core
data here, and one recent proposal. The following data (from Boeckx 2008) pro-
vide the basic facts (see, e.g., Sigur¶sson 1991, 1996, Taraldsen 1995, 1996).
(i) Henni voru gefnar bækurnar.
she.dat were.pl given.pl books.nom.pl
‘She was given the books.’
As shown, when the subject has the quirky dative case, the object receives nomi-
native Case and the verb agrees with this object in number. This is LDA. If the
subject is nominative, the agreement goes with the subject and includes person as
well as number.
(ii) Vi¶ kusum *hún/hana.
we.nom elected.1pl she.nom/her.acc
‘We elected her.’
With the quirky-case example, the agreement between the verb and the object is
limited to number; inserting person agreement leads to ungrammaticality.
(iii) *Henni leiddumst vi¶.
her.dat bored.1pl us.nom
‘She was bored with us.’
Focusing on the fact that person agreement is impossible in LDA, Boeckx (2008)
observes that agreements that occur inside vP tend not to have person agreement,
and he suggests that (i) is in fact not a case of LDA but a case of agreement inside
vP. This is possible, although there are well-known cases of person agreement that
152 Notes

involve elements inside vP (see Baker 2008 for example). I leave the issue of Ice-
landic open.
13. Regarding the idea that complementizer agreement occurs at PF by string
adjacency, Liliane Haegeman has pointed to me that, although it is true that the
subject must typically be adjacent to the complementizer in West Flemish, there
are two exceptions. One is that the object clitic may occur on the complementizer.
(i) . . . dan t Valère en Marie a weten.
that.pl it.cl Valère and Marie already know
‘. . . that Valère and Marie already know.’
The other is tet, a pronounlike particle that may intervene between the comple-
mentizer and the subject in finite clauses.
(ii) . . . dan tet Valère en Marie da weten.
that.pl tet Valère and Marie that know
‘. . . that Valère and Marie know.’
We need not consider the object clitic as a counterexample to the need for string
adjacency, given that it cliticizes to C. On the other hand, tet is apparently not a
clitic; I will leave its intervention between subject and complementizer as an issue
to be addressed in the future. See Guéron and Haegeman 2007, Haegeman 2008,
and Haegeman and Van de Velde 2008 for discussion about tet.

Chapter 3

1. The relevance of focus for scrambling has been suggested in Abe 2003, Bailyn
2003, Ishihara 2000, Jung 2002, Miyagawa 1997, 2006, 2007, Otsuka 2005, and
Yang 2004, among other works. Kitahara (1994) first suggested that scrambling
involves what today we would call minimal Attract, which assumes some sort of
feature.
2. Following Aoyagi (1998, 2006; see Sells 1995 for the original idea), I assume
that the particle mo attaches to the DP (or some other XP) as something like a
clitic, thereby not changing the original category’s identity as DP, PP, and so on.
3. To be precise, Hasegawa (2005) adopts the idea in Miyagawa 2005b that the
focus feature begins at C but is inherited by T, where it attracts the mo phrase.
4. In chapter 4, I will introduce an analysis of A- and Ā-movement that allows A-
movements to optionally leave a full copy. On this account, there is a further rea-
son why the A-moved focused element does not reconstruct: the A-moved element
carries the information-structural element of focus, so that this A-moved chain
involving focus is deprived of the option of leaving a trace (see É. Kiss 1998 for
relevant discussion about narrow focus and lack of reconstruction).
5. This analysis also responds to a point made about an earlier version of Miya-
gawa and Arikawa 2007 by Hoji and Ishii (2004)—namely, that the higher posi-
tion is not an Ā-position. Hoji and Ishii are correct, but with the revision I have
introduced in this monograph, their point no longer needs to be considered as a
criticism of the overall approach in Miyagawa and Arikawa 2007.
Notes 153

6. Kuroda (1988) was the first to propose that, for example, the object in Japa-
nese can move to Spec,TP. In his approach, this movement is purely optional,
but in the approach taken in Miyagawa 2001, it is one way of obligatorily fulfill-
ing the EPP requirement, which can be achieved by moving the object or subject
or some other category into Spec,TP. See Kitahara 2002 for an analysis of scram-
bling that also utilizes the EPP feature on T.
7. Yoshimura (1994) presents some SOV sentences in which the subject has the
other nominative marking, no. Kato (2007) notes that this no independently
appears in honorific environments, and Yoshimura’s examples involve such hon-
orification. Otherwise, the subject in SOV must have ga.
8. There is acquisition evidence in Japanese for the ‘‘theme’’ view of scrambling.
In an early study on acquisition of scrambling in Japanese, Hayashibe (1975)
noted that there appears to be a period, sometimes up to five years of age, where
children tend to interpret scrambled sentences like (ib) as if they were non-
scrambled sentences like (ia).
(i) a. SOV
Kamesan-ga ahirusan-o osimasita.
turtle-nom duck-acc pushed
‘A turtle pushed a duck.’
b. OSV
Ahirusan-o kamesan-ga osimasita.
duck-acc turtle-nom pushed
Hayashibe concludes from this that scrambling is acquired late in language devel-
opment. However, Otsu (1994) challenges this assumption by questioning Haya-
shibe’s experimental design. Otsu shows that children even before the age of
three have no problem with scrambling when they are presented with a discourse
context that makes the scrambled sentence sound natural. The following is an ex-
ample used in Otsu’s experiment:
(ii) Kooen-ni ahirusan-ga imasita. Sono ahirusan-o kamesan-ga osimasita.
park-in duck-nom was the duck-acc turtle-nom pushed
‘There was a duck in the park. A turtle pushed the duck.’
As is clear, the first sentence registers ‘duck’ in the discourse, which makes it pos-
sible for it to be the theme (also discourse topic) in the second sentence, and in
turn making the scrambling of ‘duck’ natural. This suggests that scrambling must
be motivated. It also shows that scrambling emerges early in language develop-
ment, something not at all surprising if it is akin to grammatical agreement, which
is apparently acquired quite early (e.g., Hoekstra and Hyams 1998, Wexler 1998).
In chapter 4, I discuss the function of the feature ‘‘topic’’ in detail.

Chapter 4

1. A reviewer points out that, although the noted data are consistent with the idea
that Case is active in Kinande, other issues might conspire to give rise to the data
without requiring us to adopt the Case analysis. For example, it is well known
154 Notes

that the object reversal construction has a complex set of information-structural

properties: although the agreed-with object is a topic, the subject that remains
postverbal must be contrastive. These facts are still poorly understood; possibly,
with a deeper understanding of them, we will be able to explain the lack of object
reversal (and locative reversal) in certain constructions without resorting to Case.
2. A reviewer notes that for (33), it is possible to have an agreement marker for
subject along with the agreement marker for wh-phrase if the subject is an empty
pro. According to the reviewer, this is noted in work by Kasangati Kinyalolo.
Such a ‘‘double-agreement’’ structure would be possible under the current analysis
precisely in the environment of pro-drop in the following way. First, two f-probes
are merged at C. One f-probe lowers to a and is valued by the wh-phrase on its
way to C, as already noted. The other f-probe lowers to T. This f-probe is valued
by the agreement head in Spec,vP, just as we saw for pro-drop in Greek and other
languages in chapter 2. This head does not cause a Minimality violation for the
movement of the wh-phrase precisely because it is a head, and not an XP. This
recalls Anagnostopoulou’s (2003) analysis of the Greek double-object construc-
tion, in which the theme can be passivized over the dative only if the dative is a
clitic, thus a head.
3. I am grateful to Anders Holmberg for assistance with Finnish. All mistakes are
my own responsibility.
4. See Sportiche 1999, 2006 for ideas that are similar in the overall rationale al-
though not in the implementation.
5. Thanks to Kyle Johnson for the suggestion that led to this example.
6. Thanks to David Pesetsky for modeling this English example on the original
Japanese sentence (64). Kyle Johnson presented examples in his fall 2008 lectures
at MIT that had similar properties.
7. I thank Noam Chomsky for suggesting (75) and Kyle Johnson for suggesting
an example similar to (76).
8. There are speakers who can get the partial negation interpretation for (77a).
However, as noted in Miyagawa and Arikawa 2007, this interpretation appears
to be possible with a nondefault prosody; with the default prosody, it is di‰cult,
if not impossible. See Ishihara 2007 for relevant discussion about prosody in this
construction.

Chapter 5

1. For a promising approach to wh-movement languages and wh-in-situ lan-

guages based on prosodic considerations, see Richards, to appear.
2. A reviewer raises the question of what triggers the operator movement in rela-
tive clauses. The reviewer then suggests as one possibility that it may be topic, as
opposed to focus in wh-questions, that triggers this movement. This is plausible.
Kuno (1973) notes a number of parallels in Japanese between topicalization and
relativization. Pesetsky and Torrego (2006) suggest that certain forms of relativ-
ization involve topicalization, although the cases they discuss are those that do
Notes 155

not involve wh-movement (the book I read ). Finally, there is a possible correlation
between topicalization and WCO. As Chomsky (1982) notes, wh-movement with-
in a relative clause does not trigger a WCO violation (the boy who his mother
loves). Independently, Lasnik and Stowell (1991) point out that topicalization
does not trigger a WCO violation. I will not pursue relativization and topicali-
zation further here. These observations provide the interesting possibility that
wh-movement is triggered by the focus feature, which remains focus (topic) in
relative clauses but is valued as þfocus (focus) in wh-questions.
3. Some native speakers might find this example only marginally awkward. For
these speakers, a clearer judgment obtains if we add ‘almost’.
(i) *Hotondo daremo-ga nani-o katta no?
almost everyone-nom what-acc bought q
‘What did almost everyone buy?’
4. Rizzi’s (1992) and Pesetsky’s (2000) idea that a phonologically null element
moves for wh-in-situ recalls Watanabe’s (1992) analysis of wh-in-situ.
5. In a similar vein, Tomioka (2007) notes that the set of intervenors in Japanese
is not a natural class. He develops a pragmatic approach to intervention in which
the intervenors are what he calls ‘‘antitopic’’ items that, in the intervention envi-
ronment, are inappropriately forced into a topic position.
6. The etymology of ka in the NPI sika is not known (Konoshima 1983). I will
simply assume that this ka is the same ka as those in the existential expressions.
7. See Lee 2004 for an extensive discussion of similar constructions in Korean.
8. What we just observed may also explain what has been termed scope rigidity in
Japanese—the notion that quantifier scope is limited to surface scope and inverse
scope is impossible (Hoji 1985, Kuroda 1971). The quantifiers most commonly
used from early on to show this are typically existential and universal quantifiers
marked by ka and mo, respectively. Some linguists have informally observed that
inverse scope is easier to obtain with, for example, numerals.
9. As a reviewer notes, if minna ‘all’ is not lexically marked for focus, it should
be able to stay inside the scope of negation, a fact confirmed in the following
example:
(i) Minna-ga piza-o tabe-nakat-ta.
all-nom pizza-acc eat-neg-past
‘All didn’t eat pizza.’
not > all, all > not
10. As a reviewer notes, the other possibility is that the universal quantifier
‘everyone/all each’ raises to a position high in the structure, higher than the ques-
tion C, so that intervention never takes place. This is the notion of ‘‘quantifying
in’’ (see Beck 1996). This is consistent with the idea that minna is not focused; be-
cause it doesn’t enter into agreement with a probe, it is free to move above the CP
by QR. I simply note this alternative; I will not pursue it here.
11. I should note that the original Relativized Minimality approach to interven-
tion is found in Hagstrom 1998:63.
156 Notes

12. In the German intervention case, where no question-particle movement takes

place, we could imagine a couple of scenarios. One is the Rizzi-type analysis, in
which a piece of the wh-phrase that contains the focus feature tries to move to C
but is blocked by the intervening focus phrase niemand ‘nobody’. Alternatively,
under the expletive account given earlier, the ‘‘expletive’’ wh-phrase was merges
right above the wh-phrase. Was takes on the focus feature of the wh-phrase, there-
by acting as the goal of the focus feature at C. It then tries to move to C to attain
PGU, but its movement is blocked by the intervening focus phrase niemand.
References

Abe, Jun. 2003. Economy of scrambling. Unpublished manuscript, Tohoku Gai-

kuin University.
Akmajian, Adrian, and Frank Heny. 1975. An introduction to the principles of
transformational syntax. Cambridge, Mass.: MIT Press.
Alexiadou, Artemis. 2003. On nominative case features and split agreement. In
Ellen Brandner and Heike Zinsmeister, eds., New perspectives on Case theory,
23–52. Stanford, Calif.: CSLI Publications.
Alexiadou, Artemis, and Elena Anagnostopoulou. 1998. Parametrizing word
order, V-movement, and EPP-checking. Natural Language & Linguistic Theory
16:491–539.
Anagnostopoulou, Elena. 2003. The syntax of ditransitives: Evidence from clitics.
Berlin: Mouton de Gruyter.
Anderson, Stephen. 1992. A-morphous morphology. Cambridge: Cambridge Uni-
versity Press.
Aoyagi, Hiroshi. 1998. Particles as adjunct clitics. In Pius Tamanji and Kiyomi
Kusumoto, eds., NELS 28, 17–31. Amherst: University of Massachusetts, Gradu-
ate Linguistic Student Association.
Aoyagi, Hiroshi. 2006. Nihongo no joshi to kinoohanchu [Particles and functional
categories in Japanese]. Tokyo: Hituzi Syobo.
Archangeli, Diana, and Douglas Pulleyblank. 1987. Maximal and minimal rules:
The e¤ects of tier scansion. In Joyce McDonough and Bernadette Plunkett, eds.,
NELS 17, 16–35. Amherst: University of Massachusetts, Graduate Linguistic
Student Association.
Bahloul, Maher, and Wayne Harbert. 1993. Agreement asymmetries in Arabic. In
Jonathan Mead, ed., WCCFL 11, 15–31. Stanford, Calif.: CSLI Publications.
Bailyn, John F. 2003. Does Russian scrambling exist? In Simin Karimi, ed., Word
order and scrambling, 156–176. Oxford: Blackwell.
Baker, C. L. 1970. Notes on the description of English questions: The role of an
abstract question morpheme. Foundations of Language 6:197–219.
Baker, Mark. 1996. The polysynthesis parameter. Oxford: Oxford University
Press.
158 References

Baker, Mark. 2003. Agreement, dislocation, and partial configurationality. In

Andrew Carnie, Heidi Harley, and MaryAnn Willie, eds., Formal approaches to
function in grammar, 107–132. Amsterdam: John Benjamins.
Baker, Mark. 2008. The syntax of agreement and concord. Cambridge: Cambridge
University Press.
Battistella, Edwin. 1989. Chinese reflexivization: A movement to INFL approach.
Linguistics 27:987–1012.
Beck, Sigrid. 1995. Negative islands and reconstruction. In Uli Lutz and Jürgen
Pafel, eds., Extraction and extraposition in German, 121–143. Amsterdam: John
Benjamins.
Beck, Sigrid. 1996. Quantified structures as barriers for LF movement. Natural
Language Semantics 4:1–56.
Beck, Sigrid. 2006. Intervention e¤ects follow from focus interpretation. Natural
Language Semantics 14:1–56.
Beck, Sigrid, and Shin-Sook Kim. 1997. On wh- and operator scope in Korean.
Journal of East Asian Linguistics 6:339–384.
Beck, Sigrid, and Hotze Rullmann. 1998. Presupposition projection and the inter-
pretation of which questions. In Devon Strolovitch and Aaron Lawson, eds.,
SALT 8, 215–232. Ithaca, N.Y.: Cornell University, CLC Publications.
Belletti, Adriana. 1990. Generalized verb movement. Turin: Rosenberg and Sellier.
Benmamoun, Elabbas. 1992. Functional and inflectional morphology: Problems
of projection, representation, and derivation. Doctoral dissertation, University of
Southern California, Los Angeles.
Bhatt, Rajesh. 2005. Long distance agreement in Hindi-Urdu. Natural Language
& Linguistic Theory 23:757–807.
Bhatt, Rajesh, and Roumyana Pancheva. 2004. Late merger of degree clauses.
Linguistic Inquiry 35:1–45.
Bhatt, Rajesh, and Roumyana Pancheva. 2007. Degree quantifiers, positions of
merger e¤ects with their restrictors, and conservativity. In Chris Barker and Pau-
line Jacobson, eds., Direct compositionality, 306–335. Oxford: Oxford University
Press.
Blake, Barry. 1990. Relational Grammar. London: Croom Helm.
Blight, Ralph C. 1999. Verb positions and the auxiliary be. Unpublished manu-
script, University of Texas, Austin.
Bobaljik, Jonathan David. 2006. Where’s phi? Agreement as a post-syntactic op-
eration [working paper version]. In Marjo van Koppen, Pepijn Hendriks, Frank
Landsbergen, Mika Poss, and Jenneke van der Wal, eds., Leiden working papers
in linguistics 3(2), 1–23. Leiden: Leiden University Centre for Linguistics.
Bobaljik, Jonathan David, and Susi Wurmbrand. 2005. The domain of agree-
ment. Natural Language & Linguistic Theory 23:809–865.
Boeckx, Cedric. 2000. Quirky agreement. Studia Linguistica 54:354–380.
Boeckx, Cedric. 2003. Islands and chains: Resumption as stranding. Amsterdam:
John Benjamins.
References 159

Boeckx, Cedric. 2007. Case and agreement: The syntax of A-dependencies.

Unpublished manuscript, Harvard University.
Boeckx, Cedric. 2008. Aspects of the syntax of agreement. London: Routledge.
Bošković, Željko. 1997. The syntax of nonfinite complementation. Cambridge,
Mass.: MIT Press.
Bošković, Željko. 2002. A-movement and the EPP. Syntax 5:167–218.
Bowers, John. 1993. The syntax of predication. Linguistic Inquiry 24:591–656.
Brandi, Luciana, and Patricia Cordin. 1989. Two Italian dialects and the null sub-
ject parameter. In Osvaldo Jaeggli and Kenneth J. Safir, eds., The null subject pa-
rameter, 111–142. Dordrecht: Kluwer.
Bresnan, Joan. 2001. Lexical-Functional Syntax. Oxford: Blackwell.
Brody, Michael. 1990. Remarks on the order of elements in the Hungarian focus
field. In István Kenesei, ed., Approaches to Hungarian 3, 95–121. Szeged: JATE.
Burzio, Luigi. 1986. Italian syntax: A Government-Binding approach. Dordrecht:
Reidel.
Cable, Seth. 2007. The Grammar of Q: Q-Particles and the Nature of Wh-Fronting,
as Revealed by the Wh-Questions of Tlingit. Doctoral dissertation, MIT.
Caponigro, Ivano, and Carson T. Schütze. 2003. Parameterizing passive participle
movement. Linguistic Inquiry 34:293–308.
Carstens, Vicki. 2003. Rethinking complementizer agreement: Agree with a case-
checked goal. Linguistic Inquiry 34:393–412.
Carstens, Vicki. 2005. Agree and EPP in Bantu. Natural Language & Linguistic
Theory 23:219–279.
Chang, Lisa. 1997. Wh-in-situ phenomena in French. Master’s thesis, University
of British Columbia.
Cheng, Lisa L.-S. 1991. On the typology of wh-questions. Doctoral dissertation,
MIT.
Chierchia, Gennaro. 1998. Reference to kinds across languages. Natural Lan-
guage Semantics 6:339–405.
Choe, Hyon Sook. 2006. The syntax of existential there and the structure of nom-
inal expressions. Unpublished manuscript, MIT/Yeungnam University.
Chomsky, Noam. 1955/1975. Logical structure of linguistic theory. New York:
Plenum.
Chomsky, Noam. 1965. Aspects of the theory of syntax. Cambridge, Mass.: MIT
Press.
Chomsky, Noam. 1981. Lectures on government and binding. Dordrecht: Foris.
Chomsky, Noam. 1986. Reflections on language. New York: Pantheon.
Chomsky, Noam. 1993. A minimalist program for linguistic theory. In Kenneth
Hale and Samuel Jay Keyser, eds., The view from Building 20, 1–52. Cambridge,
Mass.: MIT Press.
Chomsky, Noam. 1995. The Minimalist Program. Cambridge, Mass.: MIT Press.
160 References

Chomsky, Noam. 2000. Minimalist inquiries. In Roger Martin, David Michaels,

and Juan Uriagereka, eds., Step by step: Essays on minimalism in honor of Howard
Lasnik, 89–155. Cambridge, Mass.: MIT Press.
Chomsky, Noam. 2001. Derivation by phase. In Michael Kenstowicz, ed., Ken
Hale: A life in language, 1–52. Cambridge, Mass.: MIT Press.
Chomsky, Noam. 2005. Three factors in language design. Linguistic Inquiry 36:1–
22.
Chomsky, Noam. 2007. Approaching UG from below. In Uli Sauerland and
Hans-Martin Gärtner, eds., Interfaces þ recursion ¼ language?: Chomsky’s min-
imalism and the view from syntax-semantics, 1–29. Berlin: Mouton de Gruyter.
Chomsky, Noam. 2008. On phases. In Robert Freidin, Carlos Otero, and Maria
Luisa Zubizarreta, eds., Foundational issues in linguistic theory, 133–166. Cam-
bridge, Mass.: MIT Press.
Chung, Sandra. 1998. The design of agreement: Evidence from Chamorro. Chi-
cago: University of Chicago Press.
Cinque, Guglielmo. 1999. Adverbs and functional heads: A cross-linguistic perspec-
tive. Oxford: Oxford University Press.
Cole, Peter, Gabriella Hermon, and Li-May Sung. 1990. Principles and parame-
ters of long-distance reflexives. Linguistic Inquiry 21:1–22.
Corbett, Greville G. 2006. Agreement. Cambridge: Cambridge University Press.
Culicover, Peter, and Michael Rochemont. 1983. Stress and focus in English. Lan-
guage 59:123–165.
Diesing, Molly. 1992. Indefinites. Cambridge, Mass.: MIT Press.
Elouazizi, Noureddine. 2005. Anti-agreement e¤ects as (anti)-connectivity. In
John Alderete, Chung-hye Han, and Alexei Kochetov, eds., WCCFL 24, 120–
128. Somerville, Mass.: Cascadilla Proceedings Project.
Emonds, Joseph. 1976. A transformational approach to English syntax. New York:
Academic Press.
Engdahl, Elisabet. 1985. Parasitic gaps, resumptive pronouns, and subject extrac-
tions. Linguistics 23:3–44.
Epstein, Samuel David, and T. Daniel Seely. 1999. SPEC-ifying the GF ‘subject’:
Eliminating A-chains and the EPP within a derivational model. Unpublished
manuscript, University of Michigan and Eastern Michigan University.
Fassi Fehri, Abdelkader. 1993. Issues in the structure of Arabic clauses and word
order. Dordrecht: Kluwer.
Fox, Danny. 1999. Reconstruction, binding theory, and the interpretation of
chains. Linguistic Inquiry 30:157–196.
Fox, Danny. 2000. Economy and semantic interpretation. Cambridge, Mass.: MIT
Press.
Fox, Danny. 2002. Antecedent-contained deletion and the copy theory of move-
ment. Linguistic Inquiry 33:63–96.
Freidin, Robert. 1986. Fundamental issues in the theory of binding. In Barbara
Lust, ed., Studies in the acquisition of anaphora, 151–188. Dordrecht: Reidel.
References 161

Fukui, Naoki. 1986. A theory of category projection and its applications. Doc-
toral dissertation, MIT.
Gazdar, Gerald, Ewan Klein, Geo¤rey Pullum, and Ivan Sag. 1985. Generalized
Phrase Structure Grammar. Cambridge, Mass.: Harvard University Press.
George, Leland, and Jaklin Kornfilt. 1981. Finiteness and boundedness in Turk-
ish. In Frank Heny, ed., Binding and filtering, 105–127. London: Croom Helm,
and Cambridge, Mass.: MIT Press.
Givón, Talmy. 1976. Topic, pronoun and grammatical agreement. In Charles Li,
ed., Subject and topic, 149–188. New York: Academic Press.
Grewendorf, Günther. 2005. The discourse configurationality of scrambling. In
Joachim Sabel and Mamoru Saito, eds., The free word order phenomenon: Its syn-
tactic sources and diversity, 75–136. Berlin: Mouton de Gruyter.
Guéron, Jacqueline, and Liliane Haegeman. 2007. Subject positions, point of
view, and the neuter pronoun tet in West Flemish. Unpublished manuscript, Paris
III and STL Lille III.
Haegeman, Liliane. 1992. Some speculations on argument shift, clitics and cross-
ing in West Flemish. Unpublished manuscript, University of Geneva.
Haegeman, Liliane. 2008. Pleonastic tet in West Flemish and the cartography of
subject positions. In Sjef Barbiers, Margreet van der Ham, Olaf Koeneman, and
Marika Lekakou, eds., Microvariations in syntactic doubling, 277–290. Amster-
dam: Elsevier.
Haegeman, Liliane, and Danièle Van de Velde. 2008. Pleonastic tet in the Laps-
cheure dialect. Catalan Journal of Lingsuistics 7:157–199.
Hagstrom, Paul. 1998. Decomposing questions. Doctoral dissertation, MIT.
Haig, John H. 1980. Some observations on quantifier floating in Japanese. Lin-
guistics 18:1065–1083.
Halle, Morris, and Alec Marantz. 1993. Distributed Morphology and the pieces
of inflection. In Kenneth Hale and Samuel Jay Keyser, eds., The view from Build-
ing 20, 111–176. Cambridge, Mass.: MIT Press.
Hamblin, Charles Leonard. 1973. Questions in Montague English. Foundations of
Language 10:41–53.
Hasegawa, Nobuko. 1991. A‰rmative polarity items and negation in Japanese.
In Carol Georgopoulos and Roberta Ishihara, eds., Interdisciplinary approaches
to language: Essays in honor of S.-Y. Kuroda, 271–285. Dordrecht: Kluwer.
Hasegawa, Nobuko. 1994. Economy of derivation and Ā-movement in Japanese.
In Masaru Nakamura, ed., Current topics in English and Japanese, 1–25. Tokyo:
Hituzi Syobo.
Hasegawa, Nobuko. 2005. The EPP materialized first, Agree later: Wh-questions,
subjects and mo ‘also’-phrases. Scientific Approaches to Language 4:33–88.
Hayashibe, Hideo. 1975. Word order and particles: A developmental study in
Japanese. Descriptive and Applied Linguistics 8:1–18.
Heim, Irene. 1982. The semantics of definite and indefinite noun phrases. Doc-
toral dissertation, University of Massachusetts, Amherst.
162 References

Hoekstra, Jarich, and László Marácz. 1989. On the position of inflection in West
Germanic. Working Papers in Scandinavian Syntax 44:75–88.
Hoekstra, Teun, and Nina Hyams. 1998. Aspects of root infinitives. Lingua
106:81–112.
Hoji, Hajime. 1985. Logical Form constraints and configurational structures in
Japanese. Doctoral dissertation, University of Washington, Seattle.
Hoji, Hajime, and Yasuo Ishii. 2004. What gets mapped to the tripartite structure
of quantification in Japanese. In Vineeta Chand, Ann Kelleher, Angelo J. Rodrı́-
guex, and Benjamin Schmeiser, eds., WCCFL 23, 346–359. Somerville, Mass.:
Cascadilla Press.
Holmberg, Anders. 2005. Is there a little pro? Evidence from Finnish. Linguistic
Inquiry 36:533–564.
Holmberg, Anders, and Urpo Nikanne. 2002. Expletives, subjects, and topics in
Finnish. In Peter Svenonius, ed., Subjects, expletives, and the EPP, 71–106. Ox-
ford: Oxford University Press.
Horvath, Julia. 1981. Aspects of Hungarian syntax and the theory of grammar.
Doctoral dissertation, UCLA.
Horvath, Julia. 1986. FOCUS in the theory of grammar and the syntax of Hungar-
ian. Dordrecht: Foris.
Horvath, Julia. 1995. Structural focus, structural Case, and the notion of feature-
assignment. In Katalin É. Kiss, ed., Discourse configurational languages, 28–64.
Oxford: Oxford University Press.
Huang, C.-T. James. 1982. Logical relations in Chinese and the theory of gram-
mar. Doctoral dissertation, MIT.
Huang, C.-T. James. 1984. On the distribution and reference of empty pronouns.
Linguistic Inquiry 15:531–574.
Huang, C.-T. James, and C.-C. Jane Tang. 1991. The local nature of the long-
distance reflexive in Chinese. In Jan Koster and Eric Reuland, eds., Long-distance
anaphora, 263–282. Cambridge: Cambridge University Press.
Huang, Yun-Hua. 1984. Reflexives in Chinese. Studies in English Literature and
Linguistics 10:163–188.
Iatridou, Sabine. 1993. On nominative case assignment and a few related things.
In Colin Phillips, ed., Papers on Case and agreement II, 175–196. MIT Working
Papers in Linguistics 19. Cambridge, Mass.: MIT, MIT Working Papers in
Linguistics.
Inoue, Kazuko. 2006. Nihongo-no jookenbun-to shubun-no modaritii [Condi-
tional construction and modality in the matrix clause in Japanese]. Scientific
Approaches to Language 5:9–28.
Ishihara, Shinichiro. 2000. Stress, focus, and scrambling in Japanese. In Elena
Guerzoni and Ora Matushansky, eds., A few from Building E39: Papers in syntax,
semantics, and their interface, 142–175. MIT Working Papers in Linguistics 39.
Cambridge, Mass.: MIT, MIT Working Papers in Linguistics.
References 163

Ishihara, Shinichiro. 2007. Major phrase, focus intonation, and multiple spell-out.
The Linguistic Review 24:137–167.
Isobe, Yoshihiro. 1990. Chuuko-wabun-no yoosetsumei gimonhyoogen [Question
expressions in Middle Japanese]. Nihon-Bungaku Kenkyuu 26:165–176.
Jacobsen, Bent, and Per Anker Jensen. 1982. Some remarks on Danish weakly
stressed der. Unpublished manuscript, Aarhus Business School.
Jaeggli, Osvaldo. 1982. Topics in Romance syntax. Dordrecht: Foris.
Jespersen, Otto. 1924. The philosophy of grammar. London: George Allen and
Unwin.
Johnson, Kyle. 1991. Object positions. Natural Language & Linguistic Theory
9:577–636.
Jung, Yeun-Jin. 2002. Scrambling, edge e¤ects, and A/Ā-distinction. The Linguis-
tics Association of Korea Journal 10(4):41–64.
Karttunen, Lauri. 1977. Syntax and semantics of questions. Linguistics and Phi-
losophy 1:3–44.
Kato, Sachiko. 2007. Scrambling and the EPP in Japanese: From the viewpoint of
the Kumamoto dialect in Japanese. In Formal Approaches to Japanese Linguistics:
Proceedings of FAJL 4, 113–124. MIT Working Papers in Linguistics 55. Cam-
bridge, Mass.: MIT, MIT Working Papers in Linguistics.
Kato, Yasuhiko. 1988. Negation and the discourse-dependent property of relative
scope in Japanese. Sophia Linguistica 23:31–37. Tokyo: Sophia University.
Kayne, Richard. 1984. Connectedness and binary branching. Dordrecht: Foris.
Kayne, Richard. 1994. The antisymmetry of syntax. Cambridge, Mass.: MIT
Press.
Kayne, Richard. 2006. Expletives, datives, and the tension between morphology
and syntax. Unpublished manuscript, New York University.
Keenan, Edward. 1974. The functional principle: Generalizing the notion ‘subject
of ’. In Michael W. La Galy, Robert A. Fox, and Anthony Bruck, eds., Papers
from the 10th Regional Meeting of the Chicago Linguistic Society, 298–309. Chi-
cago: University of Chicago, Chicago Linguistic Society.
Kenstowicz, Michael. 1989. The null subject parameter in modern Arabic dialects.
In Osvaldo Jaeggli and Kenneth J. Safir, eds., The null subject parameter, 263–
275. Dordrecht: Kluwer.
Kim, Shin-Sook. 2002. Intervention e¤ects are focus e¤ects. In Noriko Akatsuka
and Susan Strauss, eds., Japanese/Korean linguistics 10, 615–628. Stanford, Calif.:
CSLI Publications.
Kim, Shin-Sook. 2006. Questions, focus, and intervention e¤ects. In Susumu
Kuno et al., eds., Harvard studies in Korean linguistics 11, 520–533. Cambridge,
Mass.: Harvard University, Department of Linguistics.
Kinyalolo, Kasangati K. W. 1991. Syntactic dependencies and the Spec-head
agreement hypothesis in Kilega. Doctoral dissertation, UCLA.
164 References

Kishimoto, Hideki. 2001. Binding of indeterminate pronouns and clause structure

in Japanese. Linguistic Inquiry 32:597–633.
Kishimoto, Hideki. 2005. Wh-in-situ and movement in Sinhala questions. Natural
Language & Linguistic Theory 23:1–51.
Kishimoto, Hideki. 2006. Japanese as a topic-movement language. Scientific
Approaches to Language 5:85–105.
É. Kiss, Katalin. 1995. Introduction. In Katalin É. Kiss, ed., Discourse configura-
tional languages, 3–27. Oxford: Oxford University Press.
É. Kiss, Katalin. 1997. Discourse-configurationality in the languages of Europe.
In Anna Siewierska, ed., Constituent order in the languages of Europe, 681–727.
Berlin: Mouton de Gruyter.
É. Kiss, Katalin. 1998. Identificational focus versus informational focus. Lan-
guage 74:245–273.
É. Kiss, Katalin. 2003. Argument scrambling, operator movement, and topic
movement in Hungarian. In Simin Karimi, ed., Word order and scrambling, 22–
43. Oxford: Blackwell.
Kitagawa, Yoshihisa. 1986. Subject in Japanese and English. Doctoral disserta-
tion, University of Massachusetts, Amherst.
Kitahara, Hisatsugu. 1994. Restricting ambiguous rule-application: A unified
analysis of movement. In Masatoshi Koizumi and Hiroyuki Ura, eds., Formal
Approaches to Japanese Linguistics: Proceedings of FAJL 1, 179–209. MIT Work-
ing Papers in Linguistics 24. Cambridge, Mass.: MIT, MIT Working Papers in
Linguistics.
Kitahara, Hisatsugu. 2002. Scrambling, Case, and interpretability. In Samuel
David Epstein and T. Daniel Seely, eds., Derivation and explanation in the Mini-
malist Program, 167–183. Oxford: Blackwell.
Klima, Edward. 1964. Negation in English. In Jerry A. Fodor and Jerrold Katz,
eds., The structure of language, 246–323. Englewood Cli¤s, N.J.: Prentice-Hall.
Ko, Heejeong. 2007. Asymmetries in scrambling and cyclic linearization. Linguis-
tic Inquiry 38:49–83.
Koizumi, Masatoshi. 1995. Phrase structure in minimalist syntax. Doctoral disser-
tation, MIT.
Koizumi, Masatoshi. 2008. Nominative object. In Shigeru Miyagawa and
Mamoru Saito, eds., The Oxford handbook of Japanese linguistics, 141–164. Ox-
ford: Oxford University Press.
Kokinshuu. 1975. Tokyo: Shintensha.
Konoshima, M. 1983. Jodooshi, joshi gaisetsu [Auxiliary verbs and case marking:
An outline]. Tokyo: Ofusha.
Koopman, Hilda. 2000. The syntax of specifiers and heads: Collected essays of
Hilda J. Koopman. New York: Routledge.
Koopman, Hilda. 2003. Inside the ‘‘noun’’ in Maasai. In Anoop Mahajan, ed.,
Syntax at sunset 3: Head movement and syntactic theory, 77–116. Los Angeles:
UCLA, Department of Linguistics.
References 165

Koopman, Hilda. 2005. Agreement: In defense of ‘‘Spec head.’’ In Cedric Boeckx,

ed., Agreement systems, 159–199. Amsterdam: John Benjamins.
Koopman, Hilda, and Dominique Sportiche. 1991. The position of subjects. Lin-
gua 85:211–258.
Koppen, Marjo van. 2006. Complementizer agreement and the relation between
C 0 and T 0 . Unpublished manuscript, Utrecht University.
Kornfilt, Jaklin. 2000. Some syntactic and morphological properties of relative
clauses in Turkish. In Artemis Alexiadou, Chris Wilder, and Paul Law, eds., The
syntax of relative clauses, 121–159. Amsterdam: John Benjamins.
Kornfilt, Jaklin. 2004. Unmasking covert complementizer agreement. Paper
presented at the annual meeting of the Linguistic Society of America, Boston, 10
January.
Kratzer, Angelika, and Junko Shimoyama. 2002. Indeterminate pronouns: The
view from Japanese. In Yukio Otsu, ed., Proceedings of the 3rd Tokyo Conference
on Psycholinguistics, 1–25. Tokyo: Keio University.
Kuno, Susumu. 1973. The structure of the Japanese language. Cambridge, Mass.:
MIT Press.
Kuroda, S.-Y. 1965. Generative grammatical studies in the Japanese language.
Doctoral dissertation, MIT.
Kuroda, S.-Y. 1971. Remarks on the notion of subject with reference to words
like also, even, or only, illustrating certain manners in which formal systems are
employed as auxiliary devices in linguistic descriptions: Part 2. Annual Bulletin 4,
Logopedics and Phoniatrics Research Institute, University of Tokyo, 127–152.
[Reprinted in Papers in Japanese Linguistics 11:157–202.]
Kuroda, S.-Y. 1972–1973. The categorical and the thetic judgment. Foundations
of Language 9:153–185.
Kuroda, S.-Y. 1980. Bun kouzou no hikaku [The comparison of sentence struc-
tures]. In Tetsuya Kunihiro, ed., Niti-eigo hikaku kouza 2: Bunpou [Lectures on
Japanese-English comparative studies 2: Grammar], 23–61. Tokyo: Taisyukan.
Kuroda, S.-Y. 1988. Whether we agree or not: A comparative syntax of English
and Japanese. Lingvisticae Investigationes 12:1–47.
Lahiri, Utpal. 1998. Focus and negative polarity in Hindi. Natural Language Se-
mantics 6:57–123.
Laka, Itziar. 1990. Negation in syntax: On the nature of functional categories and
projections. Doctoral dissertation, MIT.
Landau, Idan. 2007. EPP extensions. Linguistic Inquiry 38:485–523.
Lasnik, Howard. 1995. Case and expletives revisited: On Greed and other human
failings. Linguistic Inquiry 26:615–633.
Lasnik, Howard. 1997. A gap in an ellipsis paradigm: Some theoretical implica-
tions. Linguistic Analysis 27:166–185.
Lasnik, Howard. 1999a. Chains of arguments. In Samuel David Epstein and Nor-
bert Hornstein, eds., Working minimalism, 189–215. Cambridge, Mass.: MIT
Press.
166 References

Lasnik, Howard. 1999b. Minimalist analysis. Oxford: Blackwell.

Lasnik, Howard. 2003. Miminalist investigations in linguistic theory. New York:
Routledge.
Lasnik, Howard, and Tim Stowell. 1991. Weakest Crossover. Linguistic Inquiry
22:687–720.
Lebeaux, David. 1988. Language acquisition and the form of the grammar. Doc-
toral dissertation, University of Massachusetts, Amherst.
Lee, Youngjoo. 2004. The syntax and semantics of focus particles. Doctoral dis-
sertation, MIT.
Levin, Magnus. 2001. Agreement with collective nouns in English. Stockholm:
Almqvist & Wiksell.
Li, Charles N., and Sandra A. Thompson. 1981. Mandarin Chinese: A functional
reference grammar. Los Angeles: University of California Press.
Li, Yen-hui Audrey. 1990. Order and constituency in Mandarin Chinese. Dor-
drecht: Kluwer.
Mahajan, Anoop. 1989. Agreement and agreement phrases. In Itziar Laka and
Anoop Mahajan, eds., Functional heads and clause structure, 217–252. MIT
Working Papers in Linguistics 10. Cambridge, Mass.: MIT, MIT Working Papers
in Linguistics.
Mahajan, Anoop. 1990. The A/A 0 distinction and movement theory. Doctoral
dissertation, MIT.
Maki, Hideki, Lizanne Kaiser, and Masao Ochi. 1999. Embedded topicalization
in English and Japanese. Lingua 114:809–847.
Marantz, Alec. 1984. On the nature of grammatical relations. Cambridge, Mass.:
MIT Press.
Martin, Roger. 1999. Case, the EPP, and minimalism. In Samuel David Epstein
and Norbert Hornstein, eds., Working minimalism, 1–25. Cambridge, Mass.:
MIT Press.
Marty, A. 1918. Gesammelte Schriften II. Band 1. Halle: Max Niemeyer.
Marty, A. 1965. Psyche und Sprachstruktur. Bern: Verlag A. Francke.
Matthewson, Lisa. 2002. An underspecified Tense in St’á’imcets. Paper presented
at the Western Conference on Linguistics, University of British Columbia, 1–3
November.
May, Robert. 1977. The grammar of quantification. Doctoral dissertation, MIT.
McCloskey, James. 1996. On the scope of verb-movement in Irish. Natural Lan-
guage & Linguistic Theory 14:47–104.
McCloskey, James. 2001. The distribution of subject properties in Irish. In
William Davies and Stanley Dubinsky, eds., Objects and other subjects: Grammat-
ical functions, functional categories, and configurationality, 157–192. Dordrecht:
Kluwer.
McDaniel, Dana. 1989. Partial and multiple wh-movement. Natural Language &
Linguistic Theory 7:565–604.
References 167

Milsark, Gary. 1974. Existential sentences in English. Doctoral dissertation, MIT.

Miyagawa, Shigeru. 1993. LF Case-checking and Minimal Link Condition. In
Colin Phillips, ed., Case and agreement II, 213–254. MIT Working Papers in Lin-
gustics 19. Cambridge, Mass.: MIT, MIT Working Papers in Linguistics.
Miyagawa, Shigeru. 1997. Against optional scrambling. Linguistic Inquiry 28:1–
26.
Miyagawa, Shigeru. 1998. On wh-scope. Unpublished manuscript, MIT.
Miyagawa, Shigeru. 2001. The EPP, scrambling, and wh-in-situ. In Michael Ken-
stowicz, ed., Ken Hale: A life in language. Cambridge, Mass.: MIT Press.
Miyagawa, Shigeru. 2003. A-movement scrambling and options without option-
ality. In Simin Karimi, ed., Word order and scrambling, 177–200. Oxford:
Blackwell.
Miyagawa, Shigeru. 2005a. EPP and semantically vacuous scrambling. In Joa-
chim Sabel and Mamoru Saito, eds., The free word order phenomenon, 181–220.
Berlin: Mouton de Gruyter.
Miyagawa, Shigeru. 2005b. On the EPP. In Martha McGinnis and Norvin
Richards, eds., Perspectives on phases, 201–236. MIT Working Papers in Linguis-
tics 49. Cambridge, Mass.: MIT, MIT Working Papers in Linguistics.
Miyagawa, Shigeru. 2006. On the ‘‘undoing’’ property of scrambling: A response
to Bošković. Linguistic Inquiry 37:607–624.
Miyagawa, Shigeru. 2007. Unifying agreement and agreementless languages. In
Meltem Kelepir and Balkiz Öztürk, eds., Proceedings of the Workshop on Altaic
Formal Linguistics 2, 47–66. MIT Working Papers in Linguistics 54. Cambridge,
Mass.: MIT, MIT Working Papers in Linguistics.
Miyagawa, Shigeru, and Koji Arikawa. 2007. Locality in syntax and floating nu-
meral quantifiers. Linguistic Inquiry 38:645–670.
Miyamoto, Yoichi, and Mina Sugimura. 2005. A subject/object asymmetry and
its implication for clausal architecture in Japanese. Nanzan Linguistics 2:33–46.
Moro, Andrea. 1997. The raising of predicates. Cambridge: Cambridge University
Press.
Muysken, Pieter. 1982. Parametrizing the notion head. Journal of Linguistic Re-
search 2:57–75.
Nishigauchi, Taisuke. 1990. Quantification in the theory of grammar. Dordrecht:
Kluwer.
Nitta, Yoshio. 1991. Nihongo-no modaritii-to ninshoo [Modality and person in
Japanese]. Tokyo: Hituzi Syobo.
Nomura, Masashi. 2005. Nominative Case and AGREE(ment). Doctoral disser-
tation, University of Connecticut, Storrs.
Nomura, Takashi. 1993. Joudaigo no no to ga nitsuite [No and ga in Old Japa-
nese]. Kokugo-Kokubun 62(2):1–17, 62(3):30–49.
Odden, David. 1994. Adjacency parameters in phonology. Language 70:289–
330.
168 References

Otsu, Yukio. 1994. Early acquisition of scrambling in Japanese. In Teun Hoekstra

and Bonnie D. Schwartz, eds., Language acquisition studies in generative gram-
mar, 253–264. Amsterdam: John Benjamins.
Otsuka, Yuko. 2005. Scrambling and information focus: VSO-VOS alternation in
Tongan. In Joachim Sabel and Mamoru Saito, eds., The free word order phenom-
enon: Its syntactic sources and diversity, 243–280. Berlin: Mouton de Gruyter.
Ouhalla, Jamal. 1993. Subject-extraction, negation, and the anti-agreement e¤ect.
Natural Language & Linguistic Theory 11:477–518.
Ouhalla, Jamal. 2005. Agreement features, agreement, and anti-agreement. Natu-
ral Language & Linguistic Theory 23:655–686.
Pak, Miok. 2006. Jussive clauses and agreement of sentence final particles in Ko-
rean. In Tim Vance and Kimberly Jones, ed., Japanese/Korean linguistics 14, 295–
306. Stanford, Calif.: CSLI Publications.
Pan, Haihua. 2000. Why the blocking e¤ect? In Peter Cole, C.-T. James Huang,
and Gabriella Hermon, eds., Long-distance reflexives, 279–316. San Diego, Calif.:
Academic Press.
Perlmutter, David. 1971. Deep and surface structure constraints in syntax. New
York: Holt, Rinehart and Winston.
Perlmutter, David, and Paul Postal. 1983. The Relational Succession Law. In
David Perlmutter, ed., Studies in Relational Grammar 1, 30–80. Chicago: Univer-
sity of Chicago Press.
Pesetsky, David. 2000. Phrasal movement and its kin. Cambridge, Mass.: MIT
Press.
Pesetsky, David, and Esther Torrego. 2001. T-to-C movement: Causes and conse-
quences. In Michael Kenstowicz, ed., Ken Hale: A life in language, 355–426.
Cambridge, Mass.: MIT Press.
Pesetsky, David, and Esther Torrego. 2006. Probes, goals, and syntactic catego-
ries. In Yukio Otsu, ed., Proceedings of the 7th annual Tokyo Conference on Psy-
cholinguistics, 25–60. Tokyo: Hituzi Syobo.
Pesetsky, David, and Esther Torrego. 2007. The syntax of valuation and the inter-
pretability of features. In Simin Karimi, Vida Samiian, and Wendy Wilkins, eds.,
Phrasal and clausal architecture, 262–294. Amsterdam: John Benjamins.
Polinsky, Maria, and Eric Potsdam. 2001. Long-distance agreement and topic in
Tsez. Natural Language & Linguistic Theory 19:583–646.
Pollard, Carl, and Ivan A. Sag. 1994. Head-Driven Phrase Structure Grammar.
Chicago: University of Chicago Press, and Stanford, Calif.: CSLI Publications.
Pollock, Jean-Yves. 1989. Verb movement, Universal Grammar, and the structure
of IP. Linguistic Inquiry 20:365–424.
Progovac, Ljiljana. 1993. Negative polarity: Entailment and binding. Linguistics
and Philosophy 16:149–180.
Ramchand, Gillian. 1997. Aspect and predication: The semantics of argument
structure. Oxford: Oxford University Press.
References 169

Reinhart, Tanya. 1995. Interface strategies. Utrecht: OTS Working Papers.

Richards, Norvin. 2001. Movement in language: Interactions and architectures.
Oxford: Oxford University Press.
Richards, Norvin. To appear. Uttering trees. Cambridge, Mass.: MIT Press.
Riemsdijk, Henk van, and Edwin Williams. 1981. NP structure. The Linguistic
Review 1:171–217.
Rizzi, Luigi. 1978. Violations of the Wh-Island Constraint in Italian and the
Subjacency Condition. In Colette Dubisson, David Lightfoot, and Yves-Charles
Morin, eds., Montreal working papers in linguistics 11, 155–190. McGill Univer-
sity, Montreal.
Rizzi, Luigi. 1982. Issues in Italian syntax. Dordrecht: Foris.
Rizzi, Luigi. 1986. Null objects in Italian and the theory of pro. Linguistic Inquiry
17:501–557.
Rizzi, Luigi. 1990. Relativized Minimality. Cambridge, Mass.: MIT Press.
Rizzi, Luigi. 1992. Argument/Adjunct (a)symmetries. In Kimberley Broderick,
ed., NELS 22, 365–381. Amherst, Mass.: University of Massachusetts, Graduate
Linguistic Student Association.
Rizzi, Luigi. 1997. The fine structure of the left periphery. In Liliane Haegeman,
ed., Elements of grammar: Handbook in generative syntax, 281–337. Dordrecht:
Kluwer.
Rizzi, Luigi, ed. 2004. The cartography of syntactic structures. Vol. 2, The struc-
ture of CP and IP. Oxford: Oxford University Press.
Rooth, Mats. 1992. A theory of focus interpretation. Natural Language Semantics
1:75–116.
Rooth, Mats. 1996. Focus. In Shalom Lappin, ed., Handbook of contemporary se-
mantic theory, 271–297. Oxford: Blackwell.
Runner, Je¤rey T. 1995. Noun phrase licensing and interpretation. Doctoral dis-
sertation, University of Massachusetts, Amherst.
Sabel, Joachim. 2000. Expletives as features. In Roger Billerey and Brook Dan-
ielle Lillehaugen, eds., Proceedings of WCCFL 19, 411–424. Somerville, Mass.:
Cascadilla Press.
Safir, Kenneth J. 1982. Syntactic chains and the definiteness e¤ect. Doctoral dis-
sertation, MIT.
Safir, Kenneth J. 1985. Syntactic chains. Cambridge: Cambridge University Press.
Safir, Kenneth J. 1987. What explains the definiteness e¤ect? In Eric J. Reuland
and Alice G. B. ter Meulen, eds., The representation of (in)definiteness, 71–97.
Cambridge, Mass.: MIT Press.
Saito, Mamoru. 1985. Some asymmetries in Japanese and their theoretical impli-
cations. Doctoral dissertation, MIT.
Saito, Mamoru. 1989. Scrambling as semantically vacuous A 0 -movement. In
Mark Baltin and Anthony Kroch, eds., Alternative conceptions of phrase structure,
182–200. Chicago: University of Chicago Press.
170 References

Saito, Mamoru. 1992. Long-distance scrambling in Japanese. Journal of East

Asian Linguistics 1:69–118.
Saito, Mamoru. 2006. Optional A-scrambling. In Yukinori Takubo, Tomohide
Kinuhata, Szymon Grzelak, and Kayo Nagai, eds., Japanese/Korean linguistics
16, 44–63. Stanford, Calif.: CSLI Publications.
Sano, Masaki. 1985. LF movement in Japanese. Descriptive and Applied Linguis-
tics 18:245–259.
Sansom, G. B. 1928. A historical grammar of Japanese. Oxford: Clarendon Press.
Schneider-Zioga, Patricia. 2007. Anti-agreement, anti-locality and minimality:
The syntax of dislocated subjects. Natural Language & Linguistic Theory 25:403–
446.
Sells, Peter. 1995. Korean and Japanese morphology from a lexical perspective.
Linguistic Inquiry 26:277–325.
Shieber, Stuart M. 1986. An introduction to unification-based approaches to gram-
mar. Stanford, Calif.: CSLI Publications.
Shimoyama, Junko. 2001. Wh-constructions in Japanese. Doctoral dissertation,
University of Massachusetts, Amherst.
Shimoyama, Junko. 2006. Indeterminate phrase quantification in Japanese. Natu-
ral Language Semantics 14:139–173.
Sigur¶sson, Halldór Ármann. 1991. Icelandic case-marked PRO and the licensing
of lexical arguments. Natural Language & Linguistic Theory 9:327–363.
Sigur¶sson, Halldór Ármann. 1996. Icelandic finite verb agreement. Working
Papers in Scandinavian Syntax 57:1–46.
Sigur¶sson, Halldór Ármann. 2003. Meaningful silence, meaningless sounds.
Unpublished manuscript, Lund University.
Sigur¶sson, Halldór Ármann. 2004. Agree and agreement: Evidence from Ger-
manic. In Werner Abraham, ed., Focus on Germanic typology, 61–103. Berlin:
Akademie Verlag.
Sigur¶sson, Halldór Ármann. 2006. Remarks on features. Unpublished manu-
script, Lund University.
Sigur¶sson, Halldór Ármann. 2007. On the EPP e¤ects. Unpublished manuscript,
Lund University.
Simpson, Andrew, and Zoe Wu. 2001. Agreement, shells, and focus. Language
78:287–313.
Sportiche, Dominique. 1988. A theory of floating quantifiers and its corollaries
for constituent structure. Linguistic Inquiry 19:425–449.
Sportiche, Dominique. 1999. Reconstruction, constituency, and morphology.
Paper presented at the 1999 Generative Linguistics in the Old World (GLOW)
Colloquium, Berlin.
Sportiche, Dominique. 2006. Reconstruction, binding, and scope. In Martin Ever-
aert and Henk van Riemsdijk, eds., The Blackwell companion to syntax: Volume
IV, 35–93. Oxford: Blackwell.
References 171

Steele, Susan. 1978. Word order variation: A typological study. In Joseph H.

Greenberg, Charles A. Ferguson, and Edith A. Moravcsik, eds., Universals of
human language 4: Syntax, 585–623. Stanford, Calif.: Stanford University Press.
Steriade, Donca. 1987. Locality conditions and feature geometry. In Joyce
McDonough and Bernadette Plunkett, eds., NELS 17, 595–618. Amherst, Mass.:
University of Massachusetts, Graduate Linguistic Student Association.
Stroik, Thomas. 1996. Extraposition and expletive-movement: A minimalist ac-
count. Lingua 99:237–251.
Sybesma, Rint. 2007. Whether we tense-agree overtly or not. Linguistic Inquiry
38:580–587.
Tada, Hiroaki. 1992. Nominative objects in Japanese. Journal of Japanese Lin-
guistics 14:91–108.
Tada, Hiroaki. 1993. A/A 0 partition in derivation. Doctoral dissertation, MIT.
Takahashi, Daiko. 1990. Negative polarity, phrase structure, and the ECP. En-
glish Linguistics 7:129–146.
Takahashi, Shoichi. 2006. Decompositionality and identity. Doctoral dissertation,
MIT.
Takahashi, Shoichi, and Sarah Hulsey. 2009. Wholesale late merger: Beyond the
A/Ā distinction. Linguistic Inquiry 40.3.
Tanaka, Hidekazu. 1997. Invisible movement in sika-nai and the Linear Crossing
Constraint. Journal of East Asian Linguistics 6:143–178.
Tanaka, Hidekazu. 1999. LF wh-islands and the Minimal Scope Principle. Natu-
ral Language & Linguistic Theory 17:371–402.
Tang, C.-C. Jane. 1985. A study of reflexives in Chinese. Master’s thesis, National
Taiwan Normal University.
Tang, C.-C. Jane. 1989. Chinese reflexives. Natural Language & Linguistic Theory
7:93–121.
Tang, Sze-Wing. 1998. Parametrization of features in syntax. Doctoral disserta-
tion, University of California, Irvine.
Taraldsen, Knut Tarald. 1978. On the NIC, vacuous application, and the That-t
Filter. Unpublished manuscript, MIT.
Taraldsen, Knut Tarald. 1995. On agreement and nominative objects in Icelandic.
In Hubert Haider, Susan Olsen, and Sten Vikner, eds., Studies in comparative Ger-
manic syntax, 307–327. Dordrecht: Kluwer.
Taraldsen, Knut Tarald. 1996. Reflexives, pronouns, and subject/V agreement in
Icelandic and Faroese. In James Black and Virginia Motapanyane, eds., Micro-
parametric syntax and dialect variation, 189–212. Amsterdam: John Benjamins.
Taraldsen, Knut Tarald. 2002. The que/qui alternation and the distribution of
expletives. In Peter Svenonius, ed., Subjects, expletives, and the EPP, 29–42. Ox-
ford: Oxford University Press.
Tenny, Carol. 1994. Aspectual roles and the syntax-semantics interface. Dordrecht:
Kluwer.
172 References

Tenny, Carol. 2006. Evidentiality, experiencers, and the syntax of sentence in Jap-
anese. Journal of East Asian Linguistics 15:245–288.
Tomioka, Satoshi. 2007. Pragmatics of LF intervention e¤ects: Japanese and Ko-
rean interrogatives. Journal of Pragmatics 39:1570–1590.
Tsai, Wei-Tien Dylan. 1994. On nominal islands and LF extraction in Chinese.
Natural Language & Linguistic Theory 12:121–175.
Tsai, Wei-Tien Dylan. To appear. Tense anchoring in Chinese. Lingua.
Ueda, Yukiko. 2006. Ninshoo-seigen-to tougo-kouzou [Person restriction and
syntactic structure]. Scientific Approaches to Language 5:161–180.
Uriagereka, Juan. 1995. Aspects of the syntax of clitic placement in Western Ro-
mance. Linguistic Inquiry 26:79–123.
Watanabe, Akira. 1992. Subjacency and S-structure movement of wh-in-situ.
Journal of East Asian Linguistics 1:255–291.
Watanabe, Akira. 2000. Feature copying and binding. Syntax 3:159–181.
Watanabe, Akira. 2002. Loss of overt wh-movement in Old Japanese. In David
W. Lightfoot, ed., Syntactic e¤ects of morphological change, 179–195. Oxford:
Oxford University Press.
Wexler, Kenneth. 1998. Very early parameter setting and the Unique Checking
Constraint. Lingua 106:23–79.
Xu, Liejiong. 1986. Free empty category. Linguistic Inquiry 17:75–93.
Yang, Dong-Whee. 2004. Scrambling, interpretive complex, and cyclic spell-out.
Unpublished manuscript. Seoul National University.
Yoshimura, Noriko. 1989. Parasitic pronouns. Paper presented at the Southern
California Conference on Japanese/Korean Linguistics, University of California,
Los Angeles.
Yoshimura, Noriko. 1992. Scrambling and anaphora in Japanese. Doctoral dis-
sertation, University of Southern California, Los Angeles.
Yoshimura, Noriko. 1994. ‘Ga’ no mondai [Issues of ‘ga’]. In Henyoo-suru gengo
bunka kenkyuu, 13–28. University of Shizuoka, Japan.
Zwart, C. Jan-Wouter. 1993. Dutch syntax: A minimalist approach. Doctoral dis-
sertation, University of Groningen.
Zwart, C. Jan-Wouter. 1997. A minimalist approach to the syntax of Dutch. Dor-
drecht: Kluwer.
Author Index

Alexiadou, Artemis, 2, 28, 36, 43–44, Choe, Hyon Sook, 38

48, 51, 78, 97, 130 Chomsky, Noam, ix–x, 1–2, 3, 4, 5, 7,
Anagnostopoulou, Elena, 2, 28, 36, 9, 10, 11, 16–17, 19–20, 22, 28, 32,
43–44, 48, 51, 78, 97, 130 33, 34–36, 37, 38, 43, 52, 56, 59, 65,
Anderson, Stephen, 7 78, 110–111, 114, 115–116
Archangeli, Diana, 56 Cinque, Guglielmo, 11
Arikawa, Koji, 65, 68–69 Cole, Peter, 50
Cordin, Patricia, 3
Bahloul, Maher, 4
Baker, C. L., 125, 126 Diesing, Molly, 40
Baker, Mark, 19, 20–22, 45, 94–100
Battistella, Edwin, 50 Emonds, Joseph, 5
Beck, Sigrid, 127, 129, 132–134, 135, Engdahl, Elisabet, 42
138, 139 Epstein, Samuel David, 3, 38
Belletti, Adriana, 81
Benmamoun, Elabbas, 4 Fassi Fehri, Abdelkader, 4
Bhatt, Rajesh, 54, 112 Fox, Danny, x, 31, 65, 110–111, 112,
Blight, Robert C., 39 113
Bobaljik, Jonathan David, 32, 54 Freidin, Robert, 111
Boeckx, Cedric, 3, 16, 38, 45 Fukui, Naoki, 1, 59
Boškovi, Željko, 2–3, 5, 38, 40–41
Bowers, John, 39 Gazdar, Gerald, 7
Brandi, Luciana, 3 Givón, Talmy, 18
Bresnan, Joan, 8 Grewendorf, Günther, 12
Brody, Michael, 13, 37, 63, 66, 71, 84
Burzio, Luigi, 39 Haegeman, Liliane, 16, 55, 57
Hagstrom, Paul, 125, 127, 129, 130,
Cable, Seth, 131 131, 132, 135, 136, 141
Caponigro, Ivano, 39 Haig, John H., 68
Carstens, Vicki, 16, 21, 55, 95, 96, Halle, Morris, 32
102–103, 105 Hamblin, Charles Leonard, 126, 129
Chang, Lisa, 132, 133–134 Harbert, Wayne, 4
Cheng, Lisa L.-S., 130, 135 Hasegawa, Nobuko, 15, 62, 63–64, 67,
Chiercia, Gennaro, 28, 52 78, 137
174 Author Index

Heim, Irene, 125–126 Lahiri, Utpal, 140

Hoekstra, Jarich, 16 Laka, Itziar, 64, 75
Hoji, Hajime, 61, 69, 115, 132, 133, Landau, Idan, 3, 130
134, 136, 137, 138, 139 Lasnik, Howard, 38, 39, 65, 110
Holmberg, Anders, 17, 45, 63, 70, 85– Lebeaux, David, 111, 112
87, 93, 105–108, 119 Levin, Magnus, 6
Horvath, Julia, 13, 63, 66, 70, 71, 84 Li, Charles N., 47
Huang, C.-T. James, 48–49, 50–51, 133 Li, Yen-hui Audrey, 47
Huang, Yun-Hua, 49
Hulsey, Sarah, 111–114, 117, 123 Mahajan, Anoop, 15, 54, 60–62, 66
Maki, Hideki, 84
Inoue, Kazuko, 23, 24 Marácz, László, 16
Ishii, Yasuo, 69 Marantz, Alec, 32, 59
Isobe, Yoshihiro, 127 Martin, Roger, 2–3, 38
Marty, A., 13
Jacobsen, Bent, 41 Matthewson, Lisa, 47
Jensen, Per Anker, 41 May, Robert, 113
Jespersen, Otto, 43 McCloskey, James, 52–53
Johnson, Kyle, 39 McDaniel, Dana, 132
Milsark, Gary, 40
Kaiser, Lizanne, 84 Miyagawa, Shigeru, x, 3, 15, 19, 31,
Karttunen, Lauri, 125–126, 129 63, 65, 67, 68–69, 72, 74–76, 78–82,
Kato, Sachiko, 76–78 120–121, 127, 130, 132, 136, 137
Kato, Yasuhiko, 75, 78, 121 Miyamoto, Yoichi, 69
Kayne, Richard, x, 9, 38 Moro, Andrea, 38–39
Keenan, Edward, 7 Mujomba, Pierre, 99–100
Kenstowicz, Michael, 44, 45 Muysken, Pieter, 35
Kim, Shin-Sook, 127, 132–133, 134,
135, 138–139, 140–141 Nikanne, Urpo, 17, 63, 70, 85–87, 91,
Kinyalolo, Kasangati K. W., 95, 102, 93, 105–108, 119
105 Nishigauchi, Taisuke, 13
Kishimoto, Hideki, 14–15, 74, 130 Nitta, Yoshio, 24
É. Kiss, Katalin, xi, 1, 12, 13, 15, 19, Nomura, Masashi, 73
20, 62, 63, 66, 79, 84, 93, 119, 122 Nomura, Takashi, 127
Kitagawa, Yoshihisa, 1, 59
Kitahara, Hisatsugu, 15, 63 Ochi, Masao, 84
Klein, Ewan, 7 Odden, David, 56
Klima, Edward, 65 Ouhalla, Jamal, 45
Ko, Heejeong, 68
Koizumi, Masatoshi, 39, 72, 73 Pak, Miok, 25–26
Koopman, Hilda, 3, 9, 34, 36, 38 Pan, Haihua, 49
Koppen, Marjo van, 55–56 Pancheva, Roumyana, 112
Kornfilt, Jaklin, 16 Perlmutter, David, 43
Kratzer, Angelika, 129 Pesetsky, David, 3, 7, 9, 17, 22, 32,
Kuno, Susumu, 8, 13, 69, 74 132, 135
Kuroda, S.-Y., 1, 3, 10–11, 13, 51, 59, Polinsky, Maria, 54–55
62, 68, 70, 74, 78, 134 Pollard, Carl, 7
Author Index 175

Pollock, Jean-Yves, 64, 75 Thompson, Sandra A., 47

Potsdam, Eric, 54–55 Torrego, Esther, 3, 7, 9, 17, 22, 32
Progovac, Ljiljana, 95, 96, 97, 98, 100 Tsai, Wei-Tien Dylan, 126
Pulleyblank, Douglas, 56
Pullum, Geo¤rey, 7 Ueda, Yukiko, 24
Uriagereka, Juan, 20, 63, 70, 93
Ramchand, Gillian, 129
Reinhart, Tanya, 126 Watanabe, Akira, 16, 127–128, 131,
Richards, Norvin, 91, 125 135
Riemsdijk, Henk van, 111 Williams, Edwin, 111
Rizzi, Luigi, 11, 15, 42, 43, 44, 79, 84, Wu, Zoe, 18
96, 132, 135 Wurmbrand, Susi, 54
Rochemont, Michael, 15, 18, 66
Rooth, Mats, 129 Yoshimura, Noriko, 61, 66, 77
Rullmann, Hotze, 129
Runner, Je¤rey T., 39 Zwart, C. Jan-Wouter, 16

Sabel, Joachim, 38–39, 40

Safir, Kenneth J., 38
Sag, Ivan A., 7
Saito, Mamoru, 11, 15, 61, 63, 66, 68,
69–71, 74, 79–83, 115, 121
Sano, Masaki, 72
Sansom, G. B., 130
Schneider-Zioga, Patricia, 95, 97, 98,
99, 100, 101
Schütze, Carson T., 39
Seely, T. Daniel, 3, 38
Shieber, Stuart M., 7
Shimoyama, Junko, 129, 132
Sigurðsson, Halldór Ármann, 9, 11,
32, 45
Simpson, Andrew, 18
Sportiche, Dominique, 1, 3, 9, 38, 59
Steele, Susan, 7
Steriade, Donca, 56
Sugimura, Mina, 69
Sybesma, Rint, 47, 49

Tada, Hiroaki, 15, 61, 66, 72

Takahashi, Daiko, 132, 136, 138
Takahashi, Shoichi, 111–114, 117, 123
Tanaka, Hidekazu, 39, 132
Tang, C.-C. Jane, 49, 50
Tang, Sze-Wing, 47, 49
Taraldsen, Knut Tarald, 43
Tenny, Carol, 24, 47
Subject Index

Greek letters are alphabetized according to the English spelling of their names; for
example, f-probe is alphabetized as Phi probe would be, after Phases but before
Polysynthesis parameter.

A-/Ā-movement distinction, 37, 60, Agreement–topic/focus parameter, 1,

109–122 15–19, 71
binding and, 60, 66–67, 69, 109–110 historical evidence, 18
copy theory of movement and, 110– and inheritance of features, 19, 71
111 aP, 17, 20–22, 63, 69–71
inheritance of features and, 19–20 in Finnish (aP/TopP), 70, 105–107,
late Merge and, 111–114, 117 119–120
‘‘mixed’’ positions, 119–120 in Japanese, 80–83
phases and, 115–122 in Kikuyu, 70
reconstruction and, 65–66, 82, 110, in Kilega, 103
111, 113, 119–121 in Kinande, 95–102
Activation, 22 in Romance, 22, 70
by Case, 22–23, 27 Alternatives, semantics of, 129
by topic/focus, 94, 128–129 A-movement, 59–60, 62, 66
and valuation of features, 22 not Case-driven, 2–5, 11, 62, 72–73,
Agree, 9, 23, 31–32 114–115, 117
uninterpretable features, 7, 31–32 Ā-movement, 15, 19–20, 48, 49, 61.
valuation and, 32 See also Wh-movement
Agreement, purpose of. See Purpose of Antiagreement, 23
agreement A-positions, 59, 60
Agreement, syntactic or morphologi- in Japanese, 69, 73, 82
cal, 32 Spec,aP, 69–70, 82
Agreement asymmetry, 3–4 Spec,TP, 59, 73
Agreement head, T vs. C, 16 Arabic, agreement asymmetry in, 4
Agreement languages, 10, 11 Arabic, Bani-Hassan, pro-drop in, 44–
Agreementless languages, xi, 1, 11 46
Japanese, 11 Arbitrariness, of grammatical gender, 6
and the nominal parameter, 27–28 Argument/adjunct distinction, 112
Strong Uniformity and, 23–26 Argument structure, 10
178 Subject Index

Bantu. See also Kikuyu; Kilega; Crossover, 49

Kinande Strong, 49
agreement patterns, 21 Weak, 60, 66–67, 69, 118
aP in, 17, 20 Cyclicity, 117
Bare phrase structure, 34–35, 36, 42
Bavarian complementizer agreement, Danish, 41
55–56 Definiteness e¤ect, 40
Binding. See also Crossover Definiteness/specificity in Kinande,
Condition C, 111 95–96, 97–98
of long-distance anaphor, 49 Discourse configurationality, 11, 12–
new positions created by A- 23, 29, 62, 79
movement, 60, 67, 69 D(iscourse)-modals in Japanese, 23–25

C, 16, 18 Edge feature, 10

Cartography, 79, 84, 85 Empty Category Principle, ix, 135
Case English, topic position in, 81
activation by, 22–23, 27 EPP movement, ix, 14, 60
agreement and, 22–23 Expletives, 1, 37–38, 57, 85–86
in expletives, 38, 40–41 carriers of f-features, 38
inherent, 22–23 carriers of topic/focus feature, 86
lexical relations and, 22–23 in Danish, 41
not the trigger for A-movement, 2–5, definiteness e¤ect with, 40
11, 62, 72–73, 114–115, 117 and the Extended Projection
putative lack of, in Kinande, 22, 96– Principle, 1, 37
97, 98, 99–100 and probe-goal union, 37–42
Categories. See Functional heads; and resumption, 41–42
Lexical heads Expression structure, 10
Chains. See Phase-based Expressive power of human language,
characterization of chains xi, 31, 33–34, 143
Chinese Extended Projection Principle (EPP),
evidence for T in, 47–48 1–5, 37
nominals, 51–52 attempts to derive, 2–3, 5, 38
pro-drop in, 46–52 Extension Condition, 117
Clefts in Kinande, 100–101
Complementizer agreement, 55–57 Feature, focus. See Topic/focus
Bavarian, 55–56 feature
with first conjunct of conjoined Feature, Q. See Q feature
subject, 56 Feature, topic. See Topic/focus
Kilega, 103 feature
as a PF phenomenon, 56 Features
West Flemish, 16–17, 23, 55–57 EPP, 1–2, 74, 96
Completeness, 10 f-features, 11, 45
Concordance, 56 (un)interpretable, 7, 31–32
Condition C. See Binding universality of (see Strong Uniformity)
Copy theory of movement, 37, 110 valuation of, 31–32
Covariance, agreement as, 7 Finnish
Covert movement, 133–135 aP/TopP in, 70, 105–107, 119–120
Subject Index 179

compared to Bantu, 106–108 Greek, pro-drop in, 43

floating quantifiers, 106, 119–120 Greek, quantifier scope in, 51
‘‘mixed’’ positions, 119–120
pro-drop, 45, 107 Head movement, 42
‘‘topic expletive,’’ 85–86, 108 Hindi-Urdu long-distance agreement,
topic position in, 70, 85, 106 54
Fiorentino dialect of Italian, 3 Historical change, 18, 128
Flemish, West, complementizer Hungarian
agreement, 16–17, 23, 55–57 discourse-configurational, 84
Floating quantifiers. See Quantifiers, focus position in, 22, 66
floating topic position in, 63
Focus
in Chinese, 47 Improper movement, 120
in Japanese, 13–15, 62–67, 78, 127– Inheritance, 19–20
128, 132–133 of f-probe, 19, 20–22, 26–27, 49, 99,
marked, relative to topic, 86–87 100, 101
movement, 12, 13, 15, 62–73 (see also of topic/focus feature, 17, 18–19, 64–
Wh-movement) 65, 71, 84, 89–90
movement, as A-movement, 13–15, Intervention e¤ects in wh-questions,
62–67 127, 131–142
particles, in Old Japanese and defective, 141
Japanese, 127–128, 136–137, 141 focus approach, 132–133, 135–142
scope of negation and, 63–64, 81 in German, 131–132
Focus feature. See Topic/focus feature intervenors, characterization of, 131,
Focus probe. See Topic/focus feature 132, 134, 136, 138–140
Focus projection, 15–16, 63, 71, 84. in Japanese, 132, 133, 136–140
See also aP overt vs. covert movement, 133–135
Force markers in Korean, 25–26 quantificational approach, 132, 133–
French, intervention e¤ects in, 133– 135
134 Irish, VSO order in, 52–53
Functional explanation for agreement, Italian, 3
6 Itelmen long-distance agreement, 38,
Functional heads, 7–8 54
Functional relations, 8–9, 33–37, 109,
129 Japanese. See also Old Japanese
agreementless, 11
GB. See Government-Binding Theory aP in, 80–83
Gender, 6, 45 A-positions in, 69, 73, 82
German, intervention e¤ects in, 131– discourse-configurational, 12–13, 62,
132 79
German, scrambling in, 12 D(iscourse)-modals, 23–25
Goal, 7 floating numeral quantifiers, 68
Government-Binding Theory, ix, 59 focus constructions, 13–15, 62–67, 78
Grammatical relations. See Functional intervention e¤ects in, 132, 133, 136–
relations; Lexical relations 140
Greek preverbal subjects, interpreta- Kumamoto dialect, 76–78
tion of, 44 lacks f-features in nominals, 27–28
180 Subject Index

Japanese (cont.) Modality, 25–28

multiple-nominative constructions, 77 Morphology, 32
nominals, 27–28 Move a, ix
nominative objects, 71–73, 82 Movement
scope rigidity, 134 copy theory of (see Copy theory of
scrambling, 60–61, 66–69, 74–91, movement)
120–121 and Merge, relationship of, 2, 5
topic construction with wa, 74 purpose of (see Purpose of
movement)
Kikuyu, 70 Multiple agreement, 91, 129
Kilega, 102–105
Kinande Nominal parameter, 28
agreement, 95–98 Nominals
aP in, 95–102 Chinese, 51–52
augment vowels, 97–98 Japanese and Korean, 27–28
clefts in, 100–101 Nominative objects in Japanese, 71–
compared to Kilega, 102–104 73, 82
definiteness/specificity, 95–96 Number agreement, not always ‘‘true’’
putative lack of Case, 22, 96–97, 98, agreement, 45
99–100
relativization in, 101 Object agreement, 95, 100, 101–102,
and Strong Uniformity, 98–99 103
wh-questions in, 100 Old Japanese
Korean conclusive and attributive verb forms,
force markers, 25–26 130–131
intervention e¤ects in, 135 wh-movement and wh-in-situ in, 127–
nominals, 27–28 128
Operator movement, 20, 100–102
Language change. See Historical
change Parameter, agreement–topic/focus. See
Late Merge, 111–114, 117 Agreement–topic/focus parameter
Left-edge phenomena, 79, 80, 97 Parameter, nominal. See Nominal
Lexical heads, 7–8 parameter
Lexical relations, 8–9 Parameter, polysynthesis. See
LF movement, 50, 134 Polysynthesis parameter
Linearization, x Passive, 59, 116
Locality, 9, 10, 32, 35, 91 Person agreement, marker of ‘‘true’’
Locative agreement, 21, 95, 103 agreement, 45
Long-distance agreement, 1, 17, 37, PGU. See Probe-goal union
54–55 Phase-based characterization of
chains, 115–122
Merge, 2, 5 Phase Impenetrability Condition, 36
Merge, countercyclic. See Late Merge Phases, 10, 36, 56, 115–122
Merge, late. See Late Merge f-feature agreement, 11
Minimalist Program, ix–x, 5, 10, 59, f-features, person vs. number and
125 gender, 45
Minimality, 104, 135, 140–142 f-probe, 17, 20–22, 33, 55
Subject Index 181

Polysynthesis parameter, 22, 96, 99 Romance. See also French; Italian

Polysynthetic languages, 96, 99 aP in, 22, 70
Predicate-internal subject hypothesis, compared to Kinande, 97
59 pro-drop in, 42–44
Principle C. See Binding, Condition C
Principles-and-parameters approach. Scope
See Government-Binding Theory of quantifiers, 51
Probe-goal approach to agreement, 7, rigidity in Japanese, 134
9 scrambling and, 74–76
Probe-goal union, 35–37 Scope-bearing elements, intervention
counterexamples, 52–57, 108–109 by, 132, 134, 138–140
and expletives, 37–42 Scrambling, 60–61
and historical change, 129–130 A-scrambling, 15, 61, 114–115
and phases, 104–105 Ā-scrambling, 15, 61
and pro-drop, 42–45, 52 binding and, 61, 67, 118
in wh-questions, 141–142 double, 90
Pro-drop, 2, 42–52, 129 EPP analysis of A-scrambling, 64,
Bani-Hassan Arabic, 44–46 74–78, 83
Chinese, 46–52 German, 12
Finnish, 45, 107 Japanese, 60–61, 66–69, 74–91, 120–
Greek, 43 121
as head movement, 42–43 and reconstruction, 15, 65–66, 121
Romance, 42–44 and scope, 74–76
Purpose of agreement, x–xi, 5–10 of subjects, 67–69
Purpose of movement, ix, xi, 33–34 as topic/focus movement, 12, 15, 62
Short (vP-internal) movement, 39–40
Q feature, 125–126, 128, 129, 131 Silence Principle, 11
Quantifier-induced barrier, 133–134 Sinhala agreement, 130
Quantifier Raising, 134 Spec-head agreement, 9, 34, 36
Quantifiers, 125–126, 132, 134, 138– String adjacency, 56, 57
140. See also Scope Strong Crossover. See Crossover
floating, 68, 106, 119–120 Strong Uniformity, 11–12, 23, 26,
Questions. See Wh-questions 52
Kinande, 98–99
Raising, 59, 116 Subjacency, ix
Reconstruction Subjects, 1, 44, 45–46, 53
A-/Ā-movement distinction and, 15, Subjects, scrambling of, in Japanese,
65–66, 82, 110, 111, 113, 119–121 67–69
in A-movement, 111, 117–118 Superiority e¤ects, 92, 128
copy theory of movement and, 110 Swedish, that-t e¤ect in, 41
impossibility of, diagnostic of A- Swedish resumptive pronouns, 42
movement, 65–66, 82
Redundancy of agreement, 5–6 Tense, 47, 48
Relativization in Kinande, 101 That-t e¤ect, 41
Relativized Minimality, 135 Theme. See Topic
Rich agreement, prerequisite for pro- y-roles, 22–23, 32, 59
drop, 43 Tlingit, wh-questions in, 1
182 Subject Index

Topic, 13, 74 focus and, 125–142

of discourse, 13, 70, 74 German, 131–132
in Finnish, 63, 70, 85–86, 105–109 Kilega, 102–105
Japanese wa, 74 Kinande, 100
long-distance agreement and, 55 multiple, 126, 128
movement, 48, 55, 74–84, 87 Old Japanese, 127–128
of sentence, 13, 70, 74 Tlingit, 131
unmarked, relative to focus, 86–87 unselective-binding approach to,
Topicalization, 49, 57 125–126
‘‘Topic expletive’’ in Finnish, 85–86 wh-in-situ, 103, 105, 128, 135
Topic/focus feature, 11, 13, 71, 84–91 Word order, 39–40, 46–47, 50–51, 52–
agreement, 13–15 53, 57
default value, 86
trigger for movement, 13, 62, 66
uninterpretable, 17
Topic projection, 63, 80–81, 84. See
also aP
Topic-prominent languages, 12, 46
Trace Conversion, 113
Transfer, 36, 115
Transfer domains, 115–118
Trentino, dialect of Italian, 3
Tsez long-distance agreement, 54–55
Tucking in, 91

Unification-based frameworks, 7
Uniformity Principle, 11. See also
Strong Uniformity

Valuation, 31–32. See also Agree

VP-internal subject hypothesis. See
Predicate-internal subject hypothesis
VSO word order in Irish, 52–53
V-to-T movement, 36, 43–44, 48

Weak Crossover. See Crossover, Weak

Wh-feature. See Q feature
Wh-movement, 116, 125
English, 131
as focus movement, 126
Kilega, 103–105
Kinande, 100–102
not triggered by Q, 126
Old Japanese, 127–128
Wh-questions, 125, 128. See also
Intervention e¤ects in wh-questions
clause-typing of, 130
Linguistic Inquiry Monographs
Samuel Jay Keyser, general editor

1. Word Formation in Generative Grammar, Mark Arono¤

2. X̄ Syntax: A Study of Phrase Structure, Ray Jackendo¤
3. Recent Transformational Studies in European Languages, S. Jay Key-
ser, editor
4. Studies in Abstract Phonology, Edmund Gussmann
5. An Encyclopedia of AUX: A Study of Cross-Linguistic Equivalence,
Susan Steele
6. Some Concepts and Consequences of the Theory of Government and
Binding, Noam Chomsky
7. The Syntax of Words, Elisabeth O. Selkirk
8. Syllable Structure and Stress in Spanish: A Nonlinear Analysis, James
W. Harris
9. CV Phonology: A Generative Theory of the Syllable, George N. Clem-
ents and Samuel Jay Keyser
10. On the Nature of Grammatical Relations, Alec P. Marantz
11. A Grammar of Anaphora, Joseph Aoun
12. Logical Form: Its Structure and Derivation, Robert May
13. Barriers, Noam Chomsky
14. On the Definition of Word, Anna-Maria Di Sciullo and Edwin
Williams
15. Japanese Tone Structure, Janet Pierrehumbert and Mary E. Beckman
16. Relativized Minimality, Luigi Rizzi
17. Types of Ā-Dependencies, Guglielmo Cinque
18. Argument Structure, Jane Grimshaw
19. Locality: A Theory and Some of Its Empirical Consequences, Maria
Rita Manzini
20. Indefinites, Molly Diesing
21. Syntax of Scope, Joseph Aoun and Yen-hui Audrey Li
22. Morphology by Itself: Stems and Inflectional Classes, Mark Arono¤
23. Thematic Structure in Syntax, Edwin Williams
24. Indices and Identity, Robert Fiengo and Robert May
25. The Antisymmetry of Syntax, Richard S. Kayne
26. Unaccusativity: At the Syntax–Lexical Semantics Interface, Beth
Levin and Malka Rappaport Hovav
27. Lexico-Logical Form: A Radically Minimalist Theory, Michael Brody
28. The Architecture of the Language Faculty, Ray Jackendo¤
29. Local Economy, Chris Collins
30. Surface Structure and Interpretation, Mark Steedman
31. Elementary Operations and Optimal Derivations, Hisatsugu Kitahara
32. The Syntax of Nonfinite Complementation: An Economy Approach,
Željko Bošković
33. Prosody, Focus, and Word Order, Maria Luisa Zubizarreta
34. The Dependencies of Objects, Esther Torrego
35. Economy and Semantic Interpretation, Danny Fox
36. What Counts: Focus and Quantification, Elena Herburger
37. Phrasal Movement and Its Kin, David Pesetsky
38. Dynamic Antisymmetry, Andrea Moro
39. Prolegomenon to a Theory of Argument Structure, Ken Hale and
Samuel Jay Keyser
40. Essays on the Representational and Derivational Nature of Grammar:
The Diversity of Wh-Constructions, Joseph Aoun and Yen-hui Audrey Li
41. Japanese Morphophonemics: Markedness and Word Structure, Junko
Ito and Armin Mester
42. Restriction and Saturation, Sandra Chung and William A. Ladusaw
43. Linearization of Chains and Sideward Movement, Jairo Nunes
44. The Syntax of (In)dependence, Ken Safir
45. Interface Strategies: Optimal and Costly Computations, Tanya
Reinhart
46. Asymmetry in Morphology, Anna Maria Di Sciullo
47. Relators and Linkers: The Syntax of Predication, Predicate Inversion,
and Copulas, Marcel den Dikken
48. On the Syntactic Composition of Manner and Motion, Maria Luisa
Zubizarreta and Eunjeong Oh
49. Introducing Arguments, Liina Pylkkänen
50. Where Does Binding Theory Apply?, David Lebeaux
51. Locality in Minimalist Syntax, Thomas S. Stroik
52. Distributed Reduplication, John Frampton
53. The Locative Syntax of Experiencers, Idan Landau
54. Why Agree? Why Move?: Unifying Agreement-Based and Discourse-
Configurational Languages, Shigeru Miyagawa