University of Southern California Dissertations and Theses

Syntax-prosody interactions in the clausal domain: head movement and coalescence

(USC Thesis Other)

Syntax-prosody interactions in the clausal domain: head movement and coalescence

PDF

Download a page range

Download transcript

Copy asset link

Request this asset

Transcript (if available)

Content

SYNTAX-PROSODY INTERACTIONS IN THE CLAUSAL DOMAIN:
HEAD MOVEMENT AND COALESCENCE

Brian Hsu

Submitted to the USC Graduate School
in partial fulfillment of the requirements for the degree of

Doctor of Philosophy in Linguistics
at the
University of Southern California

August, 2016

Approved by:

____________________________________
Andrew Simpson, chair

____________________________________
Karen Jesney, member

____________________________________
Roumyana Pancheva, member

____________________________________
Rachel Walker, member

____________________________________
Krishna S. Nayak, member

ii

© Copyright by Brian Hsu

2016

iii
Acknowledgements

I owe a great debt of gratitude to the many people who have helped to make this work
possible. First and foremost, I must thank my advisor and dissertation chair, Andrew
Simpson. For years, he has provided invaluable advice, encouragement, and patience,
guiding me through false starts and helping me to focus on the questions and data that
matter the most. The other members of my committee have been no less instrumental in
the writing of this work. I thank Karen Jesney, Roumyana Pancheva, and Rachel Walker
for their extensive comments and criticisms, which were unfailingly insightful and have
greatly improved the dissertation. I would also like to thank Krishna Nayak for being a
formidable external member who ensured that everything went as smoothly as I could
have hoped.
I am grateful to the many people in the Department of Linguistics at USC who have
made the past few years so rewarding. Among past and present faculty, I would like to
thank Hagit Borer, Louis Goldstein, Elena Guerzoni, Hajime Hoji, Khalil Iskarous, Elsi
Kaiser, Audrey Li, Toben Mintz, and Maria Luisa Zubizarreta. I also thank my fellow
cohort, Emily Fedele, Assaf Israel, Grace Livingstone, Mythili Menon, and Saurov Syed,
for their support and friendship. I am indebted as well to my fellow students in syntax
and phonology with whom I have had many valuable discussions: Priyanka Biswas, Reed
Blaylock, Arunima Choudhury, Samantha Gordon, Peter Ara Guekguezian, Jessica
Harmon, Charlie O'Hara, Alfredo García Pardo, and Caitlin Smith. I would also like to
thank Janet Anderson, Andrés Benítez Pozo, Ana Besserman, Monica Do, Christina
Hagedorn, Xiao He, Lucy Kim, Maury Lander-Portnoy, Cynthia Lee, Katy McKinney-

iv
Bock, Sarah Ouwayda, Iris Ouyang, Barbara Tomaszewicz, Aaron Walker, and Yi-Hsien
Walker for making the department a great place to call home.
This dissertation has benefitted greatly from discussions with numerous people
outside of USC. For their time and input, I thank Theresa Biberauer, Emily Elfner, Bruce
Hayes, Sharon Inkelas, Sameer ud Dowla Khan, Norvin Richards, Ian Roberts, Elisabeth
Selkirk, and Sam Wolfe. I would also like to thank audiences at ETI3, AMP 2014, NELS
45, NELS 46, and Rethinking Verb Second for their feedback on portions of this
research.
This work would not have happened without the brilliant teachers who led me to
pursue graduate studies in linguistics. I would like to thank Brook Danielle Lillehaugen,
who introduced me to the field while I was a high school student taking Linguistics 1 at
UCLA in the summer of 2004. As an undergraduate at Haverford College, I was
privileged to have been taught by Jason Kandybowicz, Shizhe Huang, Vera Lee-
Schoenfeld, K. David Harrison, and Ted Fernald, who made the Tri-College Linguistics
Program such a wonderful place.
Last but certainly not least, I am deeply grateful to my parents, Shiu-Mei Liu and Ta-
Chuan Hsu, who have been endlessly supportive throughout my academic career.

v
Abstract of the dissertation

SYNTAX-PROSODY INTERACTIONS IN THE CLAUSAL DOMAIN:
HEAD MOVEMENT AND COALESCENCE

by

Brian Hsu

B. A., Haverford College
M.A., University of Southern California
Ph.D., University of Southern California

Chair: Professor Andrew Simpson

This dissertation addresses two core questions in linguistic theory. [1] What are the
sources of variation in word order, both within and across languages? [2] What
information is shared between the syntactic and phonological modules of natural
language grammar? It makes several proposals regarding possible mappings of syntactic
structures to surface word order, and the role that phonological information plays in
syntactic derivations. It argues that phonological considerations can influence word order
by determining the optimal pronunciation of movement copies, and that some syntactic
operations themselves are driven by a requirement to produce phonologically licit
structures.
I first focus on the role of the Phonetic Form (PF) grammar in the linearization of
movement copies and in the creation of prosodic structure. I present an analysis of a
puzzling pattern in Bangla (a.k.a. Bengali) that involves variation in the placement of the
subordinating complementizer je. I show that previous analyses, which rely purely on
grammatical or discourse properties, do not adequately account for the data, and present
new data on Bangla prosody to argue that variation in complementizer placement is

vi
driven by a grounded phonological constraint against placement of je in the initial
position of an Intonational Phrase. I argue that the PF component can be formalized as an
Optimality-Theoretic grammar in which phonological and syntactic well-formedness
conditions compete to determine the optimal pronunciation of movement copies.
The remainder of the dissertation turns to the accessibility of phonological
information to syntactic operations. I propose the existence of a syntactic operation,
Coalescence, which bundles structurally adjacent heads into a single, featurally complex
head. This operation derives both variation in how category features are realized on
heads, and certain types of head-adjunction and cliticization. I argue that the application
of Coalescence is properly motivated and constrained by a requirement to eliminate heads
that would be deficient in their phonological realization, suggesting that a restricted
amount of information about phonological exponence is visible to syntactic operations.
The approach is illustrated in analyses of cross-linguistic variation in the realization of
positions within the clausal left periphery and the inflectional domain, with a focus on
verb second effects.

vii
Table of Contents
Chapter 1: Introduction ....................................................................................................1
1.1 Overview ................................................................................................................................ 1
1.2 Theoretical background ....................................................................................................... 4
1.2.1 The architecture of the grammar ..................................................................................... 4
1.2.3 The Minimalist Program and head movement ................................................................ 4
1.2.3 The Copy Theory of Movement ................................................................................... 12
1.2.4 Prosodic Phonology and syntax-prosody mapping ....................................................... 16
1.2.5 The role of phonological defectivity in syntax ............................................................. 19
1.3 Empirical background ....................................................................................................... 22
1.3.1 The clausal left periphery and the Cartographic Program ............................................ 22
1.3.2 Head movement in the left periphery ............................................................................ 26
1.3.3 Verb second effects ....................................................................................................... 32
1.4 Organization of the dissertation ........................................................................................ 35
Chapter 2: Bangla complementizer order and copy pronunciation ...........................38
2.1 Introduction: The Bangla je puzzle .................................................................................. 38
2.2 Preliminaries on Bangla clause structure ........................................................................ 40
2.3 Topicalization and focus in embedded clauses ................................................................ 44
2.3.1 Postverbal embedded clauses ........................................................................................ 44
2.3.2 Preverbal embedded clauses ......................................................................................... 48
2.4 The syntax of two jes .......................................................................................................... 52
2.4.1 Copy pronunciation and je placement ........................................................................... 53
2.4.2 Subjects and non-initial je ............................................................................................. 57
2.5 Conclusion ........................................................................................................................... 61
Chapter 3: Prosodic constraints on Bangla complementizer order ............................62
3.1 Introduction ........................................................................................................................ 62
3.2 Prosodic properties of je and embedded clauses ............................................................. 62
3.2.1 Subjects and data collection .......................................................................................... 63
3.2.2 The intonational prosody of Bangla .............................................................................. 64
3.2.3 Postverbal embedded clauses ........................................................................................ 73
3.2.4 Preverbal embedded clauses ......................................................................................... 76
3.2.5 Syntax-prosody mapping of embedded clauses ............................................................ 80

viii
3.3 The prosodic representation of je ...................................................................................... 82
3.3.1 je in postverbal embedded clauses ................................................................................ 83
3.3.2 je in preverbal embedded clauses .................................................................................. 86
3.4 Syntax-Prosody interaction in phrasing and copy pronunciation ................................. 88
3.4.1 Constraints on prosodic organization and STRONGSTART ............................................ 88
3.4.2 HIGHESTCOPY and lower copy pronunciation ............................................................. 92
3.4.3 The architecture of the PF grammar ............................................................................. 99
3.5 Alternative accounts of je placement .............................................................................. 101
3.5.1 The verbal clitic analysis ............................................................................................ 101
3.5.2 Emphatic topicalization .............................................................................................. 103
3.6 Conclusion ......................................................................................................................... 110
Chapter 4: Verb second and its deviations: Bundling in the left periphery .............112
4.1 Introduction ...................................................................................................................... 112
4.1.1 Cartography and the problem of restricted instantiation ............................................. 112
4.1.2 V2 as a movement restriction: The Bottleneck Effect ................................................ 114
4.2 The problem of relaxed V2 .............................................................................................. 116
4.2.1 [XP Subject V] V3 ...................................................................................................... 117
4.2.2 [Topic Focus/Wh V] V3 ............................................................................................. 122
4.2.3 V>3 and multiple bottlenecks ..................................................................................... 126
4.2.4 The Stacked Head Hypothesis .................................................................................... 130
4.3 V2 as variation in head bundling .................................................................................... 135
4.3.1 The Feature Scattering Hypothesis ............................................................................. 136
4.3.2 Bundled heads and relaxed V2 ................................................................................... 137
4.3.3 Variation in active features within bundled heads ...................................................... 141
4.4 Additional ingredients of V2 ........................................................................................... 144
4.4.1 Single versus multiple specifiers ................................................................................ 144
4.4.2 The targets of verb movement .................................................................................... 146
4.5 Conclusion ......................................................................................................................... 148
Chapter 5: Coalescence .................................................................................................150
5.1 Introduction ...................................................................................................................... 150
5.2 The place of bundling in the grammar ........................................................................... 150
5.2.1 Feature Scattering: Bundling before syntax? .............................................................. 150
5.2.2 Morphological Merger: Bundling after syntax? ......................................................... 152

ix
5.2.3 Bundling before and after syntax? .............................................................................. 155
5.3 Defining Coalescence ........................................................................................................ 156
5.3.1 Dominance and Recession .......................................................................................... 157
5.3.2 Labels of bundled heads .............................................................................................. 162
5.3.3 Coalescence and head movement ............................................................................... 164
5.3.4 Coalescence by Internal or External Merge ................................................................ 170
5.4 Coalescence case studies .................................................................................................. 174
5.4.1 Relaxed V2 .................................................................................................................. 174
5.4.2 Kashmiri V2/V3 .......................................................................................................... 177
5.4.3 English negative contraction ....................................................................................... 180
5.4.4 The Catalan perfect ..................................................................................................... 184
5.4.5 Wolof C/T˚ bundling ................................................................................................... 188
5.5 Complex heads as minimal projections .......................................................................... 197
5.5.1 Coalescence and phrase structure uniformity ............................................................. 198
5.5.2 Coalescence and the identification of copy chains ..................................................... 202
5.5.3 Coalescence and affix ordering ................................................................................... 204
5.6 Coalescence and the architecture of the grammar ....................................................... 208
5.7 Conclusion ........................................................................................................................ 210
Chapter 6: Conclusion ...................................................................................................211
6.1 Overview of results ........................................................................................................... 211
6.2 Theoretical implications ................................................................................................... 212
6.2.1 Constraint interaction in copy linearization ................................................................ 212
6.2.2 The effects of phonological defectivity on word order ............................................... 213
6.2.3 Variation in the instantiation of functional projections .............................................. 214
6.3 Directions for future research ......................................................................................... 215
References .......................................................................................................................218

1
Chapter 1: Introduction

1.1 Overview

The goal of this dissertation is to undertake several studies of syntax-phonology
interactions, focusing on word order and prosody at the left edge of the clause, or CP. It is
primarily concerned with the mechanisms by which syntactic and prosodic factors
determine word order, and makes several proposals regarding the relation between syntax
and the Phonetic Form (PF) interface. The main proposals of the dissertation are itemized
below:

(1) Summary of main proposals
i. Patterns in which individual complementizers alternate between initial and
clause-medial placement result from variation in the pronunciation of
movement copies. Clause-medial complementizer placement is due to lower
copy spell-out, the pronunciation of a non-highest copy of the
complementizer's movement chain (Franks 1998; Pesetsky 1998; Bošković
2001; Bobaljik 2002). Such cases provide support for analyses of the left edge
of the clause as an articulated series of functional projections (Rizzi 1997;
Benincà and Poletto 2004, among many others).

ii. The pronunciation of a lower copy of a movement chain can be compelled by
language-particular restrictions on prosodic organization. Building on
previous works, (Samek-Lodovici 2005; Vogel 2006; Anttila et al. 2010;

2
Elfner 2012; Clemens 2014; Bennett et al. 2016), the PF component is
formalized as an Optimality-Theoretic grammar in which syntactic and
phonological well-formedness conditions compete to determine the optimal
pronunciation of copy chains. The restriction of prosodically-motivated word
order mismatches to those allowed by manipulating copy pronunciation makes
clear predictions about possible syntax-prosody mismatches.

iii. There is a syntactic bundling operation, Coalescence, that applies between
structurally adjacent heads to create a single terminal node that inherits their
features. Coalescence allows for a unified analysis of bundling effects
previously attributed to the lexicon (Feature Scattering: Giorgi and Pianesi
1996) and postsyntactic phonology (Morphological Merger: Marantz 1988;
Halle and Marantz 1993; Matushansky 2006). Furthermore, it
straightforwardly accounts for cross-linguistic and within-language variation
in the realization of functional heads, and permits a resolution of longstanding
problems posed by head movement in Minimalist syntax.

iv. The application of Coalescence is motivated by a requirement to eliminate
heads that would be deficient for interpretation at the PF interface by being
affixal or null in their phonological representation. This indicates that some
information about the phonological exponence of syntactic objects is
accessible to syntactic operations, contra strong claims of Phonology-Free
Syntax (Zwicky and Pullum 1986; Pullum and Zwicky 1989) and Late

3
Insertion of phonological features (Halle and Marantz 1993).

Chapters 2 and 3 of the dissertation address proposals (i) and (ii) in a study of the
syntax and prosody of embedded clauses and the complementizer je in Bangla (a.k.a.
Bengali). I show that possible non-initial placements of the je are best analyzed as the
pronunciation of a lower copy of the complementizer's movement chain within the
clausal left periphery. New data on Bangla prosody is presented to argue that lower copy
pronunciation takes place uniquely to avoid the placement of je at the left edge of an
intonational phrase. This motivates an analysis in which the PF grammar has access to
movement copies, and the surface pronunciation of a movement chain depends on the
interaction of well-formedness constraints that are both phonological and syntactic in
nature.
Proposals (iii) and (iv) are developed in Chapters 4 and 5, which examine a variety of
"bundling" processes in which multiple syntactic category features are realized on single
heads. In Chapter 4, I propose that cross-linguistic differences in the instantiation of left-
peripheral functional projections results from variation in whether certain category
features are realized on separate heads or bundled into single, featurally complex heads.
This is shown to allow for a successful account of the typology of "relaxed" verb-second
effects, which have been problematic to previous approaches to verb-second. Chapter 5
proposes that this type of bundling, along with adjunction structures previously attributed
to postsyntactic operations, can be analyzed as the result of a single syntactic operation,
Coalescence, if a limited amount of information about the phonological exponence is
visible to syntactic operations.

4
1.2 Theoretical background

This section provides a brief overview of existing theories of syntax, phonology, and
their interface assumed in the present work. It summarizes their basic claims and some
remaining theoretical questions that they raise, and previews the main proposals of the
dissertation.

1.2.1 The architecture of the grammar

Generative linguistic theory in the Principles and Parameters and contemporary
Minimalist traditions assumes the Y-model architecture of grammar (Chomsky and
Lasnik 1977; Chomsky 1995), also referred to as the Inverted Y-Model or Inverted T-
Model. Although many variants and refinements of the basic architecture have been
proposed, its basic structure is given in (2).

(2) Lexicon

Computational system (Syntactic operations)

LF PF

The key claims of the model are that the computational system of human language, which
includes core syntactic operations, manipulates sets of items selected from the Lexicon to
create an abstract, hierarchical, syntactic structure. At a point in the derivation called
Spell-Out or Transfer, this structure is transferred to two interpretive interfaces, the
Articulatory-Perceptual system (a.k.a. Phonetic Form; PF), which assigns a phonological

5
structure to the utterance, and the Conceptual-Intentional system (a.k.a Logical Form;
LF), which assigns a semantic representation.
I will additionally assume the principle of Single Output Syntax (Bobaljik 1995,
2000; Brody 1995; Groat and O’Neil 1996; Pesetsky 1998), which proposes that PF and
LF interpret an identical representation produced by the syntax, transfered to both
interfaces at Spell-Out. In other words, there are no syntactic operations that apply to
uniquely affect the structure interpreted by only LF or only PF. However, both modules
are presumed to allow a restricted set of language-specific interpretational mechanisms
that can create an indirect correspondence between their input syntactic representations
and their interpreted outputs (pronunciation at PF, semantic interpretation at LF). In other
words, this architecture of the grammar permits certain 'mismatches' in sound-meaning
correspondenc, which will be reviewed in more detail in section 1.2.3.
Following previous works, I assume that after syntactic structures undergo Spell-Out,
the PF grammar performs the following operations: [1] Lexical Insertion - the insertion of
the segmental and autosegmental exponents of syntactic terminals, [2] Linearization - the
organization of phonological segments into a linear order, and [3] Prosodic Structure
Formation - the creation of a hierarchical prosodic constituent structure. Identifying how
these processes interact has been a central question within a broad range of theories and
frameworks, including Lexical Phonology (Mohanan and Mohanan 1984; Kiparsky 1985;
Mohanan 1986; Hargus and Kaisse 1993 and works therein), Prosodic Phonology
(Selkirk 1981; Nespor and Vogel 1986; Zec and Inkelas 1990; Elfner 2012; Clemens
2014), Distributed Morphology (Halle and Marantz 1993; Embick and Noyer 2001;
Embick 2010; Kramer 2010), Stratal Optimality Theory (Kiparsky 2000, 2008;

6
Bermúdez-Otero 1999, 2012), Optimal Interleaving (Wolf 2008, 2011, 2015), and certain
works within Minimalist syntax (Pesetsky 1998; Zubizarreta 1998; Bošković 2001;
Nunes 2005; Bošković and Nunes 2007; Kandybowicz 2009, 2015), among many others.
In Chapters 2 and 3, I argue for the organization of the PF grammar shown in (3). The
unidirectional arrows indicate the sequence of events within the PF branch, while the bi-
directional arrows indicate patterns of interaction between components of the
phonological grammar. The key claim of the present study is that the pronunciation of
movement copies can be determined by constraints on prosodic organization.

(3) Syntactic operations
(Output: Hierarchical constituent structure)

PF Grammar

Copy Pronunciation Lexical Insertion

(Output: Linearized representation containing prosodic, segmental structure)

Phonetic Interpretation

While early works on linearization proposed a universal linearization mechanism,
such as the Linear Correspondence Axiom (Kayne 1994), insensitive to external factors
such as phonological restrictions, a substantial body of work has shown that the mapping
from structure to linear order can be influenced by well-formedness conditions defined on
PF representations. For instance, López (2009), Elfner (2010), Clemens (2014), and
Linearization,
Prosodic Structure
Formation

7
Bennett et al. (2016) argue that word order depends on the interaction of syntactic and
phonological conditions. In other words, linearization and the creation of prosodic
structure take place simultaneously, rather than sequentially. I argue in the following
chapters that restrictions on complementizer order in Bangla provide evidence for this
type of parallel interaction. This crucially contrasts with theories in which linearization
takes place independently of prosodic structure formation, and serialist theories of the PF
branch like that of Embick and Noyer (2001) in which the building of prosodic domains
happens after linearization and lexical insertion.
A separate question addressed in the dissertation involves the nature of the objects
manipulated by the syntactic derivation. A traditional interpretation of the Y-model holds
that PF interpretability considerations cannot drive syntactic operations (Zwicky and
Pullum 1986; Pullum and Zwicky 1989). Similarly, a key claim of works in Distributed
Morphology (Halle and Marantz 1993; Embick and Noyer 2001) is that syntax
manipulates structures with formal features but no phonological content, and that the
segmental exponents of syntactic features are inserted only after syntactic structures are
transferred to PF, a principle known as Late Insertion. A key prediction of such
approaches is that phonological properties should not be able to condition the application
of syntactic operations.
This prediction has previously been challenged in several works. For instance,
Zubizarreta (1998) proposes that movement in syntax may be triggered in order to satisfy
conditions on the prosodic realization of focus, while Richards (2010, 2016) argues that
wh-movement takes place in order to minimize the number of "active" prosodic
boundaries that intervene between wh-words and [+wh] complementizers. A crucial

8
implication of these analyses is that some information about PF structure must be visible
to the syntactic component. In Chapter 5, I argue that information about prosodic
deficiency, being phonetically null or dependent on linear adjacency with the exponents
of other features, triggers the syntactic head-bundling operation Coalescence, providing
further support for the view that some phonological features are accessible to syntax,
contra the strongest claims of Late Insertion.

1.2.2 The Minimalist Program and head movement

In foundational works of the Minimalist Program, Chomsky (1993, 1995, 2000) proposes
significant theoretical departures from the theory of Government and Binding, seeking to
reduce the permitted types of syntactic operations to a minimum. The Minimalist
Program takes a derivational approach to the building of syntactic structures; Each
derivation begins with a selection of lexical items, the Numeration, which are then
combined into a hierarchical structure. Chomsky proposes that the fundamental structure-
building operation is Merge, which under minimal additional stipulations constructs a
single syntactic object {X, Y} from two objects X and Y.

(4) {X,Y}
3
X Y

A major goal of Minimalist research is a theory of how the application of Merge is
constrained, and how the outputs of Merge are interpreted. For instance, Merge is
constrained by properties of selection; Individual lexical items can be restricted in what

9
they can combine with (Chomsky 1965; Grimshaw 1979; Pesetsky 1982), and some
items within a derivation must be Merged before others. Here, I assume that the
application of syntactic operations depends on the formal features of lexical items.
Concretely, I will adopt the proposal that all heads contain an interpretable feature and an
uninterpretable feature that selects for a property of its complement, and that Merge
allows uninterpretable features to be checked (Julien 2002; Adger 2003; Di Sciullo and
Isac 2008). X is Merged with Y in order to check an uninterpretable [uY] feature on X.

(5) {X,Y}
3
X Y
[X, uY] [Y]

There are additional constraints on the properties of the objects created by Merge.
Specifically, when two items are combined by Merge, the newly formed object appears to
inherit properties of only one of its daughter constituents. As an informal illustration, the
concatenation of the verb teach with the noun lessons, teach lessons, has the distribution
of a verb, not a noun.

(6) V
3
V N
teach lessons

We can say that a property of V projects to the higher node formed by Merge. I will refer
to features that can project as category features, and to projecting features as the labels of
newly formed objects. The item X that projects its features is the head (notationally

10
represented as X˚), and all nodes that inherit its label are phrasal projections or phrases
of X (represented as XP).

(7) XP
3
X˚ Y˚

A large body of research has shown that many cases of cross-linguistic word order
variation, as well as within-language differences in word order across clause types, can
be accounted for if heads are displaced within syntactic structures, an operation known as
head movement (Koopman 1983; Travis 1984; Baker 1988; Pollock 1989). To illustrate
with an English example in (8), declarative clauses and yes/no questions differ in the
relative placement of subjects and auxiliaries, such that subjects precede auxiliaries in
declaratives (8a) while auxiliaries precede subjects in yes/no questions (8b). Such
patterns can be accounted for by positing a base position of auxiliary heads below
subjects, and the movement of auxiliaries in yes/no questions to a higher head position
above the subject.

(8) a. [you should [hold the door for the president]
b. [should [you t [hold the door for the president?]

Head movement is also proposed to feed the creation of morphologically complex
words (see esp. Baker 1988; Julien 2002), and higher head-movement paths are often
correlated with a greater number of afffixes. Wolof, for instance, shows evidence for the
movement of finite verbs in declarative clauses from within VP to a high position within

11
the clause (Martinović 2015). When this happens, morphemes that indicate aspect, clause
type, and subject agreement are realized as suffixes on the verb.

(9) Xale yi lekk-oon-na-ñu ceeb bi
child DEF.PL eat-PERF-CLAUSETYP-3PL.SUB cake DEF.SG
'The children ate the rice (a long time ago).' (Martinović 2015)

Traditionally, the concatenation of morphemes fed by head movement is accounted for
by proposing that head movement results in an adjunction structure in which both the
moved head and attracting head are dominated by a head-level projection, and that
successive steps of head movement produce additional branching structure within the
complex head (Travis 1984; Baker 1988; Matushansky 2006). A Wolof form like lekk-
oon-na-ñu is potentially derived by movement of V˚ through the heads of AspectP,
FinitenessP, and a subject agreement phrase.

(10) AgrSubjP
3
AgrSubj˚ ...
3
Finiteness˚ AgrSubj˚
3 -ñu
Aspect˚ Finiteness˚
3 -na
V˚ Aspect˚
lekk -oon

Head adjunction will be discussed in greater detail in Chapter 5. To preview the analysis,
I present a number of conceptual problems raised by adjunction structures within
Minimalism. I will argue for an alternative theory of head bundling in which head
movement always produces a syntactic configuration with a single terminal node that

12
contains features of both the attracting and attracted heads. In other words, the head
corresponding to lekk-oon-na-ñu has the syntactic representation in (11).

(11)

AgrSubjP
wp
V/Asp/Fin/Subj˚

...
[V]
[Asp]

[Fin]

[Subj]

1.2.3 The Copy Theory of Movement

A key proposal in Chomsky (1995) is the reconception of movement as the result of
Merge extending the tree with an item already present in the derivation (a.k.a. Internal
Merge: Chomsky 2001), creating a copy of the existing object. Movement chains
therefore consist of a series of copies (indentified by subscript indices in this paper). The
primary conceptual benefit of such an approach is the maintenance of the claim that
syntactic operations apply only to lexical items present in the Numeration and to their
combinations (Chomsky 1995's Inclusiveness Condition); There is no need to stipulate
the existence of objects created during the derivation, such as traces (Chomsky 1973)
employed within Government and Binding.

(12) ...
3
X
i
...
6
... X
i
...

13
In the cases considered by Chomsky and many subsequent Minimalist works, there is an
assumed preference for all but the highest copy within the chain to be left unpronounced
when the structure is interpreted at the PF interface. This produces the effect of
displacement, as shown in (13). Nunes (2004) proposes that the preference to pronounce
only the highest copy of a chain is motivated by interpretational pressures. Because the
highest copy has checked the most features during the derivation, it should be the most
informative with respect to its semantic interpretation.

(13) [which son]
i
does Lucille prefer [which son
i
]?

Numerous subsequent works have argued that under some circumstances, however,
languages can pronounce lower copies or multiple copies of a chain (Franks 1998;
Bošković 2001; Bobaljik 2002; Bošković 2002; An 2007; Bošković and Nunes 2007;
Kandybowicz 2007; Stjepanović 2007; a.o.). Collectively, they propose that lower copy
and multiple copy pronunciation are motivated by well-formedness conditions defined on
PF, in other words, phonological constraints. Furthermore, because copy pronunciation
must take place at the PF interface, it should not be surprising that PF restrictions can
come into play. Consider a common attested pattern in languages with multiple wh-
movement, such as Serbo-Croatian, Russian, Bulgarian, and Romanian (Bošković 2001,
2002; Bošković and Nunes 2007). As illustrated in the Romanian examples below, all
wh-words in these languages typically front to a clause-initial position, and cannot be left
in their first-merged positions (14). However, if a clause contains two phonologically
identical wh-words that would be adjacent if fronted, one of them must be "left behind" in
a lower position (15).

14
(14) a. Cine ce precede?
who what precedes

b. *Cine precede ce?
who precedes what
'Who precedes what?'
(15) a. Ce precede ce?
what precedes what

b. *Ce ce precede
what what precedes
'what precedes what?'

(Bošković and Nunes 2007)

Bošković and Nunes propose that all wh-words are copied in the edge of the clause
during the syntactic derivation, and that lower copies are unpronounced by default: [cine
k

ce
i
precede cine
k
ce
i
]. However, if the pronunciation of the highest copy would violate a
constraint against adjacent identical wh-words, a lower copy is pronounced: [ce
k
ce
i

precede ce
i
]. Other PF conditions proposed to trigger lower or multiple copy
pronunciation include restrictions on sentential stress placement (Stjepanović 2007),
constraints on the marking of prosodic edges (Kandybowicz 2009; An 2007), and
requirements for functional morphemes to be affixed
(Franks 1998; Bošković 2001;
Bobaljik 2002).

Bobaljik (2002) observes that if both high and low copies can be available for
pronunciation by PF, and different copies can be privileged for interpretation at LF, the
following four-way typology is predicted (table adapted from Bobaljik 2002, ex. 3).

15
(16) Privileged copies at LF and PF

a. high PF, high LF
...
3
copy ...
| 3
PF,LF copy ...

b. high PF, low LF
...
3
copy ...
| 3
PF copy ...

|
LF
c. low PF, high LF
...
3
copy ...
| 3
LF copy ...

|
PF
d. low PF, low LF
...
3
copy ...
3
copy ...

|
PF,LF

Crucially, there are two types of "mismatch" cases, shown in (16b) and (16c). If PF
interprets a low copy while LF interprets a high one, the item will be pronounced in a
lower position than would be expected based on its semantic interpretation. On the other
hand, if PF interprets a high copy while LF interprets a lower one, its phonological
exponents will be pronounced in a higher position than expected based on its semantic
interpretation.
Although this interpretation of the Copy Theory permits such mismatches between
syntactic structure and word order, it maintains an important prediction that surface word
orders must still correspond with some syntactic structure. This is because an object
cannot be pronounced in some position at PF if there is no corresponding copy produced
by the syntactic derivation. For instance, if the phonologically motivated "postposing" of
an object is due to lower copy pronunciation, one should be able to find evidence based
on semantic interpretation or overt featural composition that it has a copy in the
corresponding position. This prediction will become crucial in the analysis of Bangla.

16
1.2.4 Prosodic Phonology and syntax-prosody mapping

We now turn to the question of what representation is created when syntactic structures
undergo Spell-Out at PF. Specifically, what are the units of representation that are
produced by the PF grammar? This dissertation adopts the main tenets of Prosodic
Phonology (Selkirk 1981, 2011; Nespor and Vogel 1986; among many others). The
principle claim of the theory is that utterances are organized in a hierarchically organized
set of prosodic constituents, whose organization corresponds closely (but not always
exactly) to syntactic constituency. The set of prosodic constituents is known as the
Prosodic Hierarchy, shown in (17). Prosodic constituents form the domains for certain
phonological restrictions at the segmental level, as well as domains for the phonetic
interpretation of pitch and intonation.

(17) Intonational Phrase
|
Phonological Phrase
|
Prosodic Word
|
Foot
|
Syllable

The Prosodic Structure Hypothesis additionally assumes that the phonological
grammar contains a procedure that determines how syntactic structure is mapped onto a
prosodic representation. This dissertation adopts the Match Theory (Selkirk 2009, 2011)
of syntax-prosody mapping assumed in most current research. Within Match theory, the
creation of a prosodic constituent structure is modeled in an Optimality-Theoretic

17
grammar, where the input is a syntactic constituent structure and the output is a prosodic
representation. Match Theory proposes that the creation of prosodic representations is
driven by a universal set of constraints that call for a direct correspondence between
syntactic and prosodic constituents. This includes constraints of the following types,
based on the schema of Elfner (2012).

(18) MATCH(α, π)
For any constituent α in the syntactic representation that dominates one or more
terminal nodes, assign a violation mark if there is no corresponding prosodic
constituent π that exhaustively dominates the phonological exponents of α.

(19) MATCH(π, α)
For any prosodic constituent π, assign a violation mark if there is no syntactic
constituent α such that π exhaustively dominates the phonological exponents of α.

A dominant claim within the literature is that the universal inventory of MATCH
constraints only calls for three types of correspondence between syntactic and prosodic
constituent types. This includes the mapping of clauses to intonational phrases, lexical
phrases to phonological phrases, and heads to prosodic words (Selkirk 2009; Itô and
Mester 2013; Sabbagh 2014; Clemens 2014). For any given syntactic constituent
structure, the corresponding output prosodic representation is determined by interaction
between MATCH constraints and conflicting well-formedness constraints on prosodic
structure. Languages can permit mismatches between syntactic and prosodic constituent
structure, in violation of MATCH constraints, in order to prevent the violation of other
markedness constraints on prosodic organization.
To illustrate a syntax-prosody mismatch within Match theory, Selkirk (2011)
discusses the case of Xitsonga (Bantu). Syllables in Xitsonga carry either high (H) or low
(L) tones, and surface tone structure depends on the distribution of lexical H tones and

18
phonological phrase (PPh) boundaries. If a word contains a lexical H tone, the tone
spreads to all syllables to the right, but does not cross a PPh boundary. Based on this
diagnostic for phonological phrasing, the prosodic structure assigned to VPs that consist
of verbs and NP direct objects depends on whether the NP contains internal branching
structure. Unmodified nouns are grouped with the verb in a single PPh, as indicated by
the fact that high tones spread from the verb to the noun (20). However, verb high tones
do not spread to object nouns that are followed by modifiers, suggesting that nouns with
modifiers are grouped into a separate PPh (21).
1

(20) vá-súsá [n-gúlú:ve]
'they are removing a pig'

Syntactic constituency: V [N]
NP

Prosodic constituency: (
PPh
vá-súsá n-gúlú:ve)

(21) vá-súsá [n-guluve y!á vo:n!á]
'they are removing their pig'

Syntactic constituency: V [N Mod]
NP

Prosodic constituency: vá-súsá (
PPh
n-gulu:ve t!á vo:n!á) (Kisseberth 1994)

The fact that unmodified nouns are not mapped to distinct phonological phrases violates
MATCH PHRASE, a constraint that requires XPs in syntax to correspond to a PPh in
prosodic representation. However, this non-correspondence allows for the satisfaction of
a common crosslinguistic preference for prosodic constituents to dominate at least two
instances of the prosodic category one step down in the Prosodic Hierarchy. This is
implemented within the grammar in the form of a constraint BINARYMINIMUM (Zec and

1
A highly similar pattern is discussed for Kinyambo by Bickmore (1990). Additional deviations from
expected prosodic phrasing due to binarity requirements are discussed by Zec and Inkelas (1990) and
Elfner (2012).

19
Inkelas 1990; Selkirk 2011). The patterns are phonological phrasing in Xitsonga are
generated by the constraint ranking BINMIN >> MATCH PHRASE.

(22) a. Prosodic phrasing of unmodified postverbal objects
[
Clause
[
VP
verb [
NP
noun] ] ] BINMIN
MATCH
PHRASE
a. {
IntP
(
PPh
verb (
PPh
noun) ) } *!
☞ b. {
IntP
(
PPh
verb noun) } *

b. Prosodic phrasing of modified postverbal objects
[
Clause
[
VP
verb [
NP
noun adj] ] ] BINMIN
MATCH
PHRASE
☞ a. {
IntP
(
PPh
verb (
PPh
noun adj ) ) }
b. {
IntP
(
PPh
verb noun adj ) } *!

In the analysis of Bangla complementizer order in Chapter 3, I propose that the same
grammar that produces prosodic representations has access to movement copies created
during the syntactic derivation. The decision of which copies to pronounce depends on
the interaction of prosodic markedness constraints with syntactic preferences on
linearization, implemented as violable constraints like HIGHESTCOPY. This accounts for
the potential influence of prosodic constraints on the mapping from syntactic structure to
surface word order.

1.2.5 The role of phonological defectivity in syntax

While syntactic operations are generally characterized as being driven by purely formal
syntactic features, this thesis examines the role of various types of phonological
"defectivity" in triggering the application of syntactic processes, and how this possibility
is to be modeled.

20
First, some syntactic operations appear to be triggered in order to provide
phonological support for affix-like morphemes. A particularly well-known case is the
analysis of do-support in English (Chomsky 1965; Lasnik 1981). In declarative
affirmative clauses, tense is realized as an affix on the main verb. In negative clauses,
however, tense morphology is realized on an auxiliary do, which precedes negation (24).
Do-support also applies in questions, in which do and tense suffixes are realized in a
position preceding the subject (25).

(23) Lindsay cares about the wetlands

(24) a. Lindsay does not care about the wetlands
b. *Lindsay cares not about the wetlands

(25) a. Does Lindsay care about the wetlands?
b. *Cares Lindsay about the wetlands?

Standardly, do-support receives an explanation like the following. In English, tense
inflection is always realized as a suffix, either on main verbs or on auxiliaries like do or
have. Head movement of main verbs in English is restricted: They cannot precede
negation, and cannot move to positions preceding subjects (24b, 25b). However, tense
morphology must be realized above negation and above subjects in questions. The only
way for this requirement to be met is for a dummy auxiliary do to be inserted in order to
carry the affixal tense morphology. Lasnik (1981) proposes that do-support occurs in
order to satisfy a requirement for tense, which is obligatorily affixal, to be realized as the
phonological dependent of another head. This is formalized as the STRAY AFFIX FILTER.

(26) STRAY AFFIX FILTER: A morphologically realized affix must be a syntactic
dependent of a morphologically realized category, at surface structure. (Lasnik
1981)

21

The notion that the STRAY AFFIX FILTER is a well-formedness condition on surface
structure is supported by the fact that languages can use either Merge or movement to
prevent violations of the condition. This can be illustrated by comparing the English
examples above with their equivalents in French. The crucial observation about the
following examples is that the present tense morphology, orthographic -sent, is realized
in the same positions as it is in English. Tense is realized above negation (we can
consider pas to be the instantiation of NegP, while treating ne as a clitic that attaches to
the verb), and above subject pronouns in questions. Unlike in English, however, tense is
uniformly realized on the main verb, rather than an auxiliary.

(27) a. Ils construisent des maisons
they build INDEF.PL houses
'They build houses'

b. Ils ne construisent pas des maisons
they NE build not INDEF.PL houses
'They don't build houses'

c. Construisent-ils des maisons
build-they INDEF.PL houses
'Do they build houses?'

Because French permits verb movement to relatively high positions, the Stray Affix Filter
with respect to affixal tense is satisfied by head movement. On the other hand, English
requires the auxiliary do to be Merged in order to support the affixal tense morpheme.
If we are correct that these processes affecting word order take place in order to avoid
an unlicensed affix, two general analyses are possible. The first possibility is that some
information about a morpheme's phonological realization is present in the syntax, and

22
able to trigger syntactic processes. Alternatively, if phonological information is present
only after syntactic structures are transfered to PF, this implies that these movement
operations occur after syntax. Indeed, a variety of postsyntactic operations affecting word
order have been proposed in many works of Distributed Morphology (Halle and Marantz
1993; Embick and Noyer 2001). Similarly, several works have proposed that head
movement applies uniquely on the PF branch, rather than in syntax (Chomsky 2001;
Boeckx and Stjepanović 2001; Platzack 2013).
On the other hand, head movement has been shown to have semantic effects,
particularly related to scope and the licensing of negative polarity items (Lechner 2006;
Matushansky 2006; Roberts 2010; Hartman 2011), indicating that not all head movement
can take place at PF. If operations like head movement are indeed triggered by
requirements on phonological well-formedness, then some information about the
phonological form of lexical items must be available to the syntactic module. These
issues will be discussed in further detail in chapter 5.

1.3 Empirical background

1.3.1 The clausal left periphery and the Cartographic Program

This dissertation's main empirical focus is on the structure of the complementizer
domain. Following Bresnan (1972), research during the Government and Binding period
largely assumed the existence of a single projection above the inflectional domain of the
clause, the Complementizer Phrase (CP). The head of this projection C˚ was argued to

23
host overt complementizers (e.g. English that, for), with the specifier of CP being the
target of wh- and focus-related movement.

(28) CP
3
3
C˚ InflP
6

Following Rizzi (1997), a variety of evidence has emerged to suggest that the
traditional CP contains additional internal structure. This so-called Cartographic
Approach proposes that rather than a single projection, the left edge of the clause
includes a series of distinct functional projections. These functional heads, collectively
referred to as the extended left periphery, perform the various functions of
complementizer-like elements, generally related to clause typing and the encoding of
information structure. While the large body of research within the approach has given rise
to many proposed structures, we will consider for illustrative purposes the 'core' structure
of the analysis of Benincà and Poletto (2004).

(29) ForceP
3
3
Force˚ TopicP
3
3
Topic˚ FocusP
3
3
Focus˚ FinP
3
3
Fin˚ InflP
6

24
This expanded inventory of functional projections has proven instrumental in the analysis
of several empirical generalizations. First, it allows for an account of restrictions on the
relative orderings of different types of discourse-marked phrases in the left periphery. As
shown in the following examples from Benincà and Poletto (2004; 54-55), while Italian
allows both topics and foci to be fronted to the left edge of the clause, topics must
precede focus. Such patterns can be taken to result from the placement of TopicP above
FocusP.

(30) A Gianni, domani, QUESTO gli dovremmo dire
to Gianni, tomorrow this 3.DAT should-1PL tell.
'Tomorrow we should tell THIS to Gianni.'

(31) *A GIANNI, un libro di poesie, lo regalerete
TO GIANNI, a book of poems, 3.ACC give-2SG

The expanded functional domain also accounts for ordering restrictions between
individual complementizers and other left-peripheral phrases. For example, Rizzi (1997)
discusses several differences in placement between che, which introduces finite
embedded clauses, and di, which introduces non-finite embedded clauses. Consider their
possible distribution relative to fronted topics; che must precede topics, while di follows
them.

(32) Credo (che) il tuo libro, (*che) loro lo apprezzerebbero molto
I.believe that your book that they it would appreciate a.lot
'I believe that your book, they would appreciate it a lot.' (Rizzi 1997)

(33) Credo (*di) il tuo libro, (di) apprezzarlo molto
I.believe of your book of to.appreciate.it a.lot
'I believe to appreciate your book a lot.' (Rizzi 1997)

25
While these restrictions do not receive an obvious account if one assumes only a single
C˚ head, they are straightforwardly explained if che occupies a relatively high position
(Force), while di is in a lower head below TopicP. More generally, the existence of the
low Fin(iteness) complementizer position has allowed for an analysis of modal particles
whose distribution does not naturally fit in the InflP domain (Roussou 2000; Roberts
2004; Aboh 2006; Morin 2006).
Further evidence for the expanded functional structure is found in languages that
appear to simultaneously realize multiple left-peripheral heads. The following example
from Aboh's (2006) analysis of Gungbe (Gbe language; Ghana and Nigeria) shows the
required orderings of the declarative subordinator (Force), topic marker (Topic), focus
marker (Focus), and deontic modal particle (Fin).

(34) Ùn ɖɔ̀ ɖɔ ̀ Àsíbá yà làn wɛ ́ é ní xɔ ̀ ná mì
1SG say that Asiba TOP meat FOC 3SG DEON buy for 1SG
'I said that, as for Asiba, she should buy me some MEAT.' (Aboh 2006)

Beyond the empirical investigations into the details of clause structure, the
Cartographic Program makes several strong claims about the permitted space of cross-
linguistic variation. First, it proposes that the inventory of functional categories is
universal, and that their corresponding functional projections are rigidly ordered across
languages (Cinque 1999). Second, it proposes a one-to-one mapping of features to heads
of syntactic structure, such that each feature is realized on a distinct syntactic head within
the hierarchy of functional projections (Cinque and Rizzi 2010). Crucially, the "one
feature, one head" requirement holds on items as they are first Merged, since head

26
movement permits the creation of featurally complex heads. In chapter 5, I argue that the
syntax can also create complex heads in the absence of movement.

1.3.2 Head movement in the left periphery

While such cases have not been as thoroughly investigated as the head movement of
verbs through the InflP domain, there is evidence to suggest that complementizers also
undergo head movement within the extended left periphery (Watanabe 1993; Poletto
2000; Grewendorf and Poletto 2011). The first argument in favor of this is based on the
observation that some complementizers reflect semantic properties of multiple left-
peripheral features. For instance, Rizzi (1997) observes that English that has multiple
semantic specifications: that appears only in declarative clauses and only in finite
clauses. If clause type is a property specified in Force˚ and finiteness is specified in Fin˚,
that displays semantic properties of two left-peripheral heads.

(35) Buster wondered [*that/if he should join the army]
(36) George would like [*that/for Oscar to leave his house]

However, it appears that that is pronounced only in Force˚, rather than Fin˚, as it
obligatorily precedes frame-setting adverbials, as well as focused items that trigger
subject-verb inversion, which presumably occur in FocusP.

(37) a. We know [that for all intents and purposes, George-Oscar is the best man for
the job]
b. The president believes [that never in his life would he make a mistake]

27
This pattern can be accounted for if that-like complementizers are first Merged in Fin˚
before moving to Force˚ (Rizzi 1997; Shlonsky and Rizzi 2007), as shown below.

(38) ForceP
3
3
Force˚ TopicP
[Force: decl] 3
[Fin: finite] 3
that Topic˚ FocusP
3
3
Focus˚ FinP
3
3
Fin˚ InflP
[Force: decl] 5
[Fin: finite]
that

More compelling empirical evidence for head movement of complementizers is
presented in Grewendorf and Poletto's (2011) study of complementizers in Cimbrian, a
Germanic language spoken in Northern Italy. Specifically, they show that certain
complementizers with relatively high surface positions in Force˚ preclude verb movement
to Fin˚, suggesting that these complementizers originate in the lower position before
moving to Force˚.
A host of phenomena in Cimbrian suggest that the presence of a certain class of
complementizers (including az 'if/that,' bal 'when/if', and the relative complementizer bo)
appears to block verb movement into a low left-peripheral position. Here, I will briefly
present their discussion of negation and subject clitics. In main clauses with negation, the
finite verb obligatorily precedes the negative marker net 'not,' and cannot follow

28
negation. In embedded clauses with an az-type complementizer, however, the opposite
pattern holds; finite verbs must follow net. Under the assumption that net occupies a
fixed position like NegP, this suggests that verbs generally move to a position above
NegP, and that this movement is blocked if an az complementizer is present.

(39) a. dar is net khent
he is not come
'He has not come'

b. *dar net is khent
he not is come

(40) a. dar has geböllt [azz-e net vortgea]
he has wanted that-I not away-go

b. *dar hat geböllt [azz-e vortgea net]
he has wanted that-I away-go not

Similar conclusions are supported by considering the distribution of subject clitics. In
main clauses where the first position item is a non-subject, no overt subject can be
realized, and instead a subject clitic pronoun follows the verb.

(41) a. Gestarn hatt-ar gisekk in has
yesterday has-he seen a hare
'Yesterday, he has seen a hare.'

b. *Gestarn ar hatt gisekk in has
yesterday he has seen a hare

In non-subject-initial embedded clauses with an az-type complementizer, however,
subject clitics are not permitted.

(42) *I boas [az gestarn hatt-ar-z-en gett]
'I know that yesterday has-he-it-him given

29
There is substantial evidence that in verb-second languages like Cimbrian, verbs move to
Finiteness˚ in declarative main clauses (the analysis of V2 will be returned to in greater
detail in chapters 4 and 5). Additionally, there is evidence that subject agreement takes
place through a low complementizer position, either in Finiteness˚ or a similar projection
(Poletto 2000; Aboh 2006; Branigan 2011). The data appear to indicate then, that the
presence of an az-type complementizer prevents verb-movement to Fin. On the other
hand, az-type complementizers are obligatorily clause-initial, and precede both topics and
foci, suggesting that their surface position is in ForceP.

(43) a. Dar libar [bo dar-r IN GIANNI has get]
the book that part-he the Gianni has given
'the book that he has given TO GIANNI.'

Grewendorf and Poletto argue that the apparent surface position of az-type
complementizers in Force˚ and their effects on verb movement to Fin˚ can be accounted
for as follows: In clauses without an az-type complementizer, the verb raises past NegP
to Fin˚, where subject agreement takes place. In clauses with an az-type complementizer,
the complementizer is first Merged in Fin˚. This prevents verb movement to Fin˚,
resulting in its low position relative to negation and the impossibility of subject
agreement. Later in the derivation, the complementizer is copied and pronounced in
Force˚.

(44) [
ForceP
... [
TopicP
... [
FocusP
... [
FinP
V˚ [
InflP
... [
NegP
... [... V˚ ...]]]]]]

(45) [
ForceP
az [
TopicP
... [
FocusP
... [
FinP
az [
InflP
... [
NegP
... [... V˚ ...]]]]]]

30
As additional evidence for head movement in the left periphery, we can consider
blocking effects on movement. This is illustrated in the Béarnais dialect of Gascon,
which is characterized by the use of exclamative particles that appear in both main and
embedded clauses (Bouzet 1932; Campos 1992; Pusch 2002; Morin 2006). The three
particles (declarative que, interrogative e, and exclamative be) typically follow preverbal
subjects and fronted topics. However, they can appear clause-initially if no phrase
precedes the main verb. All Gascon examples are from Morin (2006).

(46) a. Maria que parla gascon
Maria DECL speaks Gascon
'Maria speaks Gascon'

b. E parla gascon Maria
Q speaks Gascon Maria
'Does Maria speak Gascon?'

c. Be m'agrada la toa pelha
BE me-pleases the your dress
'How I like your dress!'

However, the distribution of exclamative particles in embedded clauses depends on the
presence of complementizers or overt topics. Consider the distribution of e with respect
to the complementizer se 'if', which appears in embedded interrogative clauses. Unlike in
main clauses, it is not possible for an embedded clause to contain only the particle e,
regardless of whether or not there is a preverbal constituent. The complementizer se
appears in all interrogative embedded clauses.

(47) Que m' demandi [se drom lo gat]
DECL me-ask if sleeps the cat
'I wonder if the cat is sleeping.'

31
(48) *Que vau véder [las arhagas e son maduras]
DECL I.go see the strawberries if are ripe

E is licensed uniquely within embedded clauses like (49), where se is in clause-initial
position, and e follows an intervening topic phrase. The omission of e is marginal (49b),
and it is not possible for se and e to co-occur without an intervening topic (49c).

(49) a. Que 'm demandi [se lo gat e drom]
DECL me ask if the cat Q sleeps
'I wonder if the cat sleeps'

b. ?Que 'm demandi [se lo gat drom]
DECL me ask if the cat sleeps

c. *Que 'm demandi [se e drom lo gat]
DECL me ask if Q sleeps the cat

Morin notes that this distribution of e in embedded clauses supports an analysis in which
e is the pronunciation of a Fin˚ head with an interrogative specification, and se realizes an
interrogative Force˚ head. The fact that embedded exclamative particles are realized only
in the presence of a topic suggests that there is a default movement chain from Fin˚ to
Force˚ within embedded clauses. However, in the presence of a Topic˚ head that triggers
phrasal movement, Fin-to-Force head movement is blocked; this forces the pronunciation
of the lower exclamative particle, and se is Merged directly in Force˚.
2

To conclude, the Cartographic Approach to the left periphery allows for explanations
of a variety of empirical phenomena involving focus, topicalization, and complementizer
ordering. However, the claim that this series of functional projections is universal, and
that all heads introduce exactly one category feature (Cinque and Rizzi 2010) remain

2
Under this analysis, the pattern appears to reflect a preference within the grammar for Internal
Merge over External Merge, i.e. 'Move over Merge'. For additional discussion, see Motut (2010),
Shima (2000), Boeckx and Grohmann (2007), Chomsky (2013).

32
controversial (Craenenbroeck 2009 and works therein). In chapter 4, I present its
empirical difficulties in accounting for languages where available functional structure
appears to be substantially restricted. In particular, I consider problems in the analysis of
verb second effects, for which a brief summary is given below.

1.3.3 Verb second effects

Informally stated, the verb-second restriction requires the main verb or highest tensed
auxiliary of a clause to be preceded by exactly one phrase at the left edge of the utterance.
Although though there are some language-specific exceptions, the first position in V2
clauses can be occupied by constituents of a variety of grammatical functions or phrasal
categories. Consider the examples from Dutch, as repeated from Haegeman (1996).
Although a variety of arguments can appear in first position (50), it is generally not
possible for multiple phrases to precede the verb simultaneously (51).

(50) a. subject first
Marie zal morgen dit boek kopen
Marie will tomorrow this book buy
'Marie will buy this book tomorrow.'

b. object first
Dit boek zal Marie morgen kopen
this book will Marie tomorrow buy
'Marie will buy this book tomorrow.'

c. adjunct first
Morgen zal Marie dit boek kopen
tomorrow will Marie this book buy
'Marie will buy this book tomorrow'

33
d. wh-word first
Welk boek zal Marie morgen kopen
which book will Marie tomorrow buy
'Which book will Marie buy tomorrow?' (Haegeman 1996)

(51) a. *Morgen dat boek zal Marie kopen
tomorrow that book will Marie buy

b. *Morgen ze zal dat boek kopen
tomorrow she will that book buy (Haegeman 1996)

In addition, first-position elements can carry a variety of information-structure functions,
with some language-specific differences. Non-subject constituents in first position are
generally interpreted as topics or foci (Holmberg 2015), and V2 languages appear to
consistently require wh-movement to first position (Richards 2016). Although V2 is most
consistently attested within the Germanic language family, V2 patterns are also observed
in Breton and historic Brythonic Celtic languages (Stephens 1982; Willis 1998), as well
as a number of Romance languages, including Rhaetoromance dialects (Poletto 2000),
Old French (Foulet 1928; Dupuis 1989; Adams 1987), and Old Italian (Benincà 1983).
Although they are rarely described outside of the Indo-European languages, V2
requirements have additionally been proposed for the Nakh-Dagestanian language Ingush
(Nichols 2011) and the Tupi language Karitiâna (Storto 1999).
There is substantial evidence that V2 patterns require verb movement to a head within
the complementizer domain (den Besten 1983; Travis 1984; Fanselow 2004; Roberts
2004, 2012; Haegeman and Cranenbroeck 2007; among many others). Consider, for
instance, the complementary distribution between V2 and overt complementizers in
German embedded clauses. In embedded clauses with an overt complementizer dass, the
verb appears in a clause-final position, whereas verb-second order is required in

34
complementizer-less clauses. Furthermore, it is not possible for clauses with dass to show
verb-second order.

(52) a. Er sagte [dass er morgen komme]
He said that he tomorrow comes

b. Er sagte [er komme morgen]
He said he comes tomorrow
'He said that he is coming tomorrow.'

c. *Er sagte [dass er komme morgen]
He said that he comes tomorrow

(Holmberg 2015, after den Besten 1983)

Traditional analyses of V2 claim that V2 clauses have two main properties: verb
movement to the C˚ head, and the movement of exactly one phrase to a higher specifier
position (den Besten 1983; Travis 1984; Holmberg 1986; Vikner 1995; Fanselow 2004).
This structure is given in (53).

(53) CP
wp
XP
k
C'
wp
C˚ InflP
3 5
V˚
j
C˚ ... XP
k
... V˚
j
...

However, the "classic" analysis faces several difficulties, chiefly in accounting for V2
languages that allow restricted types of deviations (V1, V3), and languages that permit
V2 to co-occur with overt complementizers. Furthermore, there are few successful
approaches to deriving V2 within a cartographic theory that assumes an extended
inventory of projections. These issues form the main discussion of chapter 4.

35
1.4 Organization of the dissertation

The remainder of the dissertation is organized as follows. Chapters 2 and 3 present a
syntactic and prosodic study of the embedded left periphery of Bangla, aiming to account
for the non-initial placement of the complementizer je in preverbal embedded clauses. By
examining the placement of je relative to embedded topics and foci in different clause
types, I argue that in all embedded clauses, the complementizer has a movement chain in
multiple left-peripheral heads, separated by topics and foci.

(54) [
ForceP
je
i
[
TopicP
(XP) [
FocusP
(XP) [
FinP
je
i
[
InflP
...

A description and analysis of embedded clause prosody reveals that je is obligatorily non-
initial when its embedded clause corresponds to an intonational phrase in prosodic
representation. I propose that non-initial placement of the complementizer results from
the pronunciation of the lower copy in order to prevent the violation of a prosodic
constraint against je in IntP-initial position.
Chapter 4 turns to the problem of deriving verb-second restrictions, with a special
focus on relaxed V2 languages that allow principled deviations from V2 order. It
discusses the challenges that such patterns pose to previous theories of analyzing V2 in
the left periphery, including the Bottleneck Effect and Stacked Head analyses. I propose
that the typology of relaxed V2 systems supports an analysis of V2 as the result of several
parameter settings. In particular, I argue in favor of an approach that permits category
features to vary in whether they are realized on separate heads or if they are bundled into
a single, featurally complex head (i.e. Feature Scattering - Giorgi and Pianesi 1996). To

36
preview the analysis, I claim that strict V2 restrictions result from the bundling of all left-
peripheral features into a single head, whose projection requires one specifier. V2
restrictions are loosened, however, as fewer category features are bundled together.

(55) Force/Top/Foc/FinP
wp
XP
k
Force/Top/Foc/Fin'
wp
V-Force/Top/Foc/Fin˚ InflP
[Force] 5
[Topic] ... XP
k
...
[Focus]
[Finiteness]

Chapter 5 proposes that the bundling of category features and various types of
affixation should be analyzed as the result of a single syntactic operation, Coalescence,
that combines adjacent heads into single complex heads. I argue that Coalescence applies
uniquely in contexts where the lower head is either affixal or phonologically null,
indicating that the operation serves to eliminate structures that would be deficient for
interpretation at the PF interface. The bundling of category features that are
phonologically null accounts for bundling effects attributed to Feature Scattering, while
the bundling of affixal heads accounts for effects attributed to Morphological Merger.

(56) XP

X/YP
3 3
X˚ YP > X/Y˚

...
[X
D
] 3 Coalescence [X
D
]
Y˚ ... [Y
R
]

[Y
R
]

37
I propose an implementation of Coalescence within a Minimalist grammar, showing that
attested variation in the strictness of verb-second effects results from differences in the
distribution of deficient category features. Additionally, I argue that Coalescence
provides a superior theory of head-adjunction than previous approaches.
Chapter 6 summarizes the claims and contributions of the dissertation, and closes
with discussion of its implications for future research at the syntax-phonology interface.

38
Chapter 2: Bangla complementizer order and copy pronunciation

2.1 Introduction: The Bangla je puzzle

This chapter examines the placement of the finite complementizer je in Bangla, whose
distribution has long proved challenging for syntactic analyses. The crux of the problem
is that possible orderings of je within an embedded clause depend on the position of the
embedded clause relative to the main verb. Consider the basic pattern below, as presented
by Bhattacharya (2001). While Bangla shows default SOV order in both main and
embedded clauses (Dasgupta 1980; Sengupta 1990; Rahman 2010), discourse-neutral
embedded clauses headed by je are postverbal.
1
In postverbal embedded clauses, je can
only be pronounced in a clause-initial position. The pronunciation of je is optional in the
examples discussed here; what is of interest is the restrictions that are observed when the
complementizer is overt.

(1) a. Jon jane [je ma oSudh khey-eche]

b. *Jon jane [ ma je oSudh khey-eche]

c. *Jon jane [ ma oSudh je khey-eche]
John knows that mother that medicine that eat-PERF
'John knows that mother took medicine.'

When the embedded clause is preverbal, either sentence-initial or following the subject,
je cannot be clause-initial, and only surfaces after at least one other phrase (Dasgupta

1
As noted by Biberauer et al. (2014) and Biberauer and Sheehan (2012), there appears to be a universal
restriction against head-initial clauses embedded under head-final verbs, perhaps as a reflex of the Final-
over-Final Constraint.

39
1980; Bhattacharya 2001; Simpson and Bhattacharya 2003; Bayer and Dasgupta to
appear).

(2) a. *Jon [je ma oSudh khey-eche] jane

b. Jon [ ma je oSudh khey-eche] jane

c. Jon [ ma oSudh je khey-eche] jane
John that mother that medicine that eat-PERF knows
'John knows that mother took medicine.'

(3) a. *[je ma oSudh khey-eche] Jon jane

b. [ mother je oSudh khey-eche] Jon jane

c. [ mother oSudh je khey-eche] Jon jane
that mother that medicine that eat-PERF John knows
'John knows that mother took medicine.'

In the remainder of this chapter, I present a novel account of this asymmetry in the
distribution of Bangla je. I present data on embedded topicalization and focus to show
that both initial and non-initial complementizer orders are derived in the extended CP
domain (Rizzi 1997; Benincà and Poletto 2004; among many others). Furthermore, I
argue that non-initial complementizer orders are derived through lower copy spell-out of
the complementizer (Franks 1998; Pesetsky 1998; Bošković 2001; Bobaljik 2002). I
claim that both preverbal and postverbal embedded clauses follow the same derivation in
which je is first merged in Fin(iteness)˚ and copied in Force˚.

(4) [
ForceP
(je) [
TopicP(s)
(XP
top
) [
FocusP
(XP
foc
) [
FinitenessP
(je) [
InflP
...

The je-initial order of postverbal embedded clauses results from the default pronunciation
of the complementizer's highest copy in Force˚; Word order in preverbal clauses is
derived through pronunciation of je's lower copy in Fin˚, paired with movement of at

40
least one other phrase to TopicP or FocusP. I additionally claim that je's movement chain
is an instance of long head movement (Lema and Rivero 1991; Rivero 1991; Borsley et
al. 1996), in which movement from Fin˚ to Force˚ skips over the structurally intervening
Topic˚ and Focus˚ heads. Lastly, I argue in favor for this approach over an alternative
analysis of complementizer lowering via Local Dislocation (Embick and Noyer 2001;
Kramer 2010) or Prosodic Inversion (Halpern 1992).

2.2 Preliminaries on Bangla clause structure

The proper syntactic analysis of embedded clause placement in Bangla and its Indo-
Aryan relatives has proven controversial, particularly in the context of the general head-
final nature of these languages
2
. In Bangla, the debate has largely been framed in terms of
identifying the base position of embedded clauses. On one hand, the strict head-final
interpretation of Bangla phrase structure holds that embedded CP are first merged in the
preverbal object position, with other orders derived by left or right extraposition (Bayer
1996). This contrasts with the 'antisymmetric' view, in which the post-verbal CP is in its
base position, and preverbal orders are derived through raising (Simpson and
Bhattacharya 2003).

(5) The head-final analysis (Bayer 1996)
Base order: S [CP...] V
Derived orders: S t
k
V [CP...]
[CP...] S t
k
V

2
See Bayer (1996) and Simpson and Bhattacharya (2003) for extended discussion of this issue in Bangla.
Similar considerations of Hindi-Urdu are found in Bhatt and Dayal (2007) and Manetta (2012).

41
(6) The anti-symmetric analysis (Simpson and Bhattacharya 2003)
Base order: S V [CP...]
Derived orders: S [CP ...]
k
V t
k

[CP ...]
k
S V t
k

While this dissertation does not primarily address the debate, the analysis assumes the
anti-symmetric analysis of Simpson and Bhattacharya, arguing that it accounts for
correlations between embedded clause positions and their interpretations and for
differences in their prosodic realization.
The placement of embedded clauses is restricted by their discourse interpretations. In
neutral contexts or broad focus, embedded clauses with je are obligatorily postverbal. On
the other hand, embedded clauses can precede matrix verbs only under restricted
discourse interpretations. Sentence-medial embedded clauses, which follow the main
clause subject but precede the main verb, obligatorily bear emphatic or corrective focus.
Sentence-initial embedded clauses, which precede main clause subjects, are necessarily
topicalized. A clear reflex of this restriction is the fact that focus is disallowed on
constituents within sentence-initial CPs.

(7) Embedded CP position and interpretation
Postverbal CP: informationally neutral. (initial je)
Medial CP: emphatic or corrective focus. (non-initial je)
Sentence-initial CP: topicalized information. (non-initial je)

As a broader generalization, Bangla exhibits a discourse-configurational tendency to
place topics in initial positions, and focus in medial positions (Choudhury 2010; Rahman
2010; Syed 2013).
We consider the placement of topics within main clauses, and Simpson and
Bhattacharya's argument that a high topic position drives the movement of discourse-

42
given phrases to a sentence-initial position. The initial observation behind this claim is
that wh-phrases typically appear in a post-subject position. If wh-movement is obligatory
in Bangla, as argued in greater detail by the authors, this differs from languages like
English where wh-movement targets the left edge of the clause.

(8) ora [ke aS-be] Sun-eche
they who come-FUT hear-PERF
'Who have they heard will come?' (Bayer 1996)

Simpson and Bhattacharya note that subjects that precede wh-words are obligatorily
topicalized, as revealed by the fact that they must be definite or specific. While Bangla
has no overt definite determiners, the definiteness of nominal phrases is indicated by the
orderings of numerals and classifiers relative to nouns (Bhattacharya 1999). When
numeral-classifier sequences precede the noun ([Num-Cl] NP), only an indefinite reading
is available, whereas nouns that precede the numeral and classifier ([NP [Num-Cl]) are
obligatorily definite. Based on this criterion, only definite nominals can precede wh-
words.

(9) chele du-ʈo [kon boi-ʈa] poɽlo?
boy two-CL which book-CL read
'which books did the two boys read?

(10) *du-ʈo chele [kon boi-ʈa] poɽlo?
Two-CL boy which book-CL read
(intended reading: 'which books did two boys read?')
(Simpson and Bhattacharya 2003)

Furthermore, when a non-wh subject is quantified and thus incompatible with a topic
reading, it must occur to the right of the matrix wh-phrase.

43
(11) [ka-ke] kew voʈ daey-ni
who-DAT anyone vote gave-NEG
'Who did no one vote for?'

(12) * kew [ka-ke] voʈ daey-ni
anyone who-DAT vote gave-NEG
(intended reading: 'who did no one vote for?') (Simpson and Bhattacharya 2003)

On the basis of these facts, Simpson and Bhattacharya propose the existence of a topic
projection above wh-phrases, accessible only to definite nominals or other eligible topics.
Their proposed structure is given in (13).

(13) [Topic [
whP
wh
k
[Subject
(QP/nontopic)
... t
k
...]]]

This ordering of constituents at the left edge of the clause receives a straightforward
adaptation in the Cartographic model of the extended clausal periphery (Rizzi 1997;
Benincà and Poletto 2004; a.o.), which splits the traditional CP into a series of functional
projections that encode various discourse properties. Minimally, this consists of two
complementizer projections, ForceP and FinitenessP (abbreviated as FinP), separated by
TopicP and FocusP. Within this basic structure, I propose that pre-subject topics
identified by Simpson and Bhattacharya move to TopicP, with wh-words in FocusP.

(14) [
ForceP
... [
TopicP
(Topic) [
FocusP
(wh) [
FinP
... [
IP
...

In the section that follows, I show that this inventory of left-peripheral positions is also
available in embedded clauses (cf. Rizzi 1997; Bianchi 1999; Ledgeway 2005; Roussou
2000; Krapova and Karastaneva 2002; Roberts 2004; Aboh 2006; Demonte and
Fernández-Soriano 2009; a. o.) and that the focus position is also available to contrastive
focus. Furthermore, these projections are crucially ordered with respect to je; when the

44
CP is postverbal, topic and focus immediately follow je, whereas both precede je in
preverbal CPs.

2.3 Topicalization and focus in embedded clauses

We now consider the possible orderings of je relative to fronted topics and foci within
both postverbal and preverbal embedded clauses. Based on the attested patterns, I argue
that the differences in the position of je depend on which complementizer head it is
pronounced in. Specifically, je surfaces in Force˚ in postverbal embedded clauses, but
Fin˚ in preverbal embedded clauses.

2.3.1 Postverbal embedded clauses

We first turn our attention to postverbal embedded clauses, where je is obligatorily
clause-initial. Recall that indefinite and quantified DPs in a simple matrix clause cannot
move past a wh-phrase into a high topic projection. This particular diagnostic is
unfortunately unavailable for embedded clauses with je, since je as a declarative
complementizer is incompatible with embedded clauses containing [+wh]-phrases (Bayer
2001). However, topic movement can nonetheless be identified by examining the
possible placement of object DPs relative to subjects. In both main and embedded
clauses, the neutral word order for transitive sentences is SOV. However, OSV word
orders are also available, as given in the pair below.

45
(15) Jon jane [(je) ma oSudh khey-eche]
John knows [that mother medicine eat-PERF]
'John knows that mother took medicine.'

(16) Jon jane [(je) oSudh ma khey-eche]
John knows [that medicine mother eat-PERF]
'John knows that mother took medicine.'

Numerous studies have established that relatively free word order of this type is
nonetheless constrained by restrictions related to information structure. For instance,
placement of objects in pre-subject positions is commonly restricted to definite or
specific DPs (Mahajan 1990; Moltmann 1990; Diesing 1992; de Hoop 1992; Choi 1999),
and indefinites with contrastive focus (Moltmann 1990; Choi 1999); In addition to the
arguments of Simpson and Bhattacharya (2003), Rahman (2010) similarly identifies
sentence-initial positions as restricted to topics. Furthermore, he shows that contrastively
focused objects are preferably placed in medial positions (non clause-initial, not verb-
adjacent), while non-focused objects prefer verb-adjacency.
Let us hypothesize that OSV word orders in Bangla are the result of either
topicalization or contrastive focus of the object, which raises the object above the
subject's base position. If this is correct, we predict that objects that are possible topics
(definite or specific) can freely precede subjects, either with or without focal prominence.
On the other hand, objects that are ineligible topics, such as quantified phrases or
indefinites, can only be fronted if they are clearly contrastively focused. Both of these
predictions are borne out.
We first turn our attention to the patterning of definite object DPs. As expected,
definite objects (here, chatro du-to-ke 'the two students-ACC') can appear in the preverbal
object position (17), or be fronted to the left of subjects in the absence of a focus.

46

(17) John bol-lo [je dadubhai chatro du-to-ke dEkh-eche]
John say-pst that grandfather student 2-CL-ACC see-PERF
'John said that grandfather saw the two students'

(18) John bol-lo [je chatro du-to-ke dadubhai dEkh-eche]
John say-pst that student 2-CL-ACC grandfather see-PERF
'John said that grandfather saw the two students'

We compare this to the patterning of quantified DP objects, specifically those that are
obligatorily non-specific.
3
Within postverbal embedded clauses, quantified objects are
most naturally placed after the subject, in what appears to be the base word order (19).
Fronting of a quantified object (with a non-specific reading) to the left of the subject
results in a loss of grammaticality under an intended non-focused reading (20).

(19) Jon bol-lo [je dadubhai kau-ke dEkh-e-ni]
John say- PST that grandfather anyone-ACC see-PERF-NEG
'John said that grandfather didn't see anyone'

(20) # Jon bol-lo [je kau-ke dadubhai dEkh-e-ni]
John say-PST that anyone-ACC grandfather see-PERF-NEG

The same pattern is observed with indefinite DPs. Recall that nominal phrases with
[[Num-Cl] NP] order are obligatorily indefinite, and thus incompatible with topic
readings. This provides evidence for the existence of a high TopicP above the regular

3
The restriction is not one against all quantified expressions. For example, nouns quantified by kichu
'some' can front before subjects in non-focused readings (anonymous personal communication).

(i) [kichu oSudh] ma kal rat-e khey-eche
some medicine mother last night-LOC eat-PERF
'Mother took some medicine last night.'

For speakers that admit such examples, this would appear to reflect a "specificity effect" well-documented
in Germanic-style scrambling (de Hoop 1992, Diesing 1994, Moltmann 1990), whereby quantified phrases
that are interpreted as specific front more liberally than non-specific quantified phrases.

47
subject position in embedded clauses (the following data are based on those of Simpson
and Bhattacharya 2003).

(21) Jon bol-lo [je dadubhai du-to chatro-ke dEkh-eche]
John say-PST that grandfather 2-CL student-ACC see-PERF
'John said that grandfather saw two students'

(22) # Jon bol-lo [je du-to chatro-ke dadubhai dEkh-eche]
John say-PST that 2-CL student-ACC grandfather see-PERF

However, the word order in (20) is acceptable if the fronted quantifier phrase is
contrastively focused (i.e. understood to exclude some contextually defined set of
possible alternatives) and bears focal prominence.
4
In all subsequent examples, focal
prominence will be indicated with capitalization.

(23) Jon bol-lo [je KAU-KE dadubhai dEkh-e-ni]
John said that anyone-acc grandfather see-PERF-NEG
'John said that grandfather didn't see ANYONE.'

(24) Jon bol-lo [je DU-TO CHATRO-KE dadubhai dEkh-eche]
John say-pst that 2-CL student-ACC grandfather see-PERF
'John said that grandfather saw TWO STUDENTS.'

The acceptability of non-specific DPs in this position under contrastive focus indicates
either that contrastive focus is for some reason permitted to front to TopicP, or that it can
front to a distinct FocusP position above the subject position. In the next section, I
present data to support the latter hypothesis.
In summary, we have initial evidence that postverbal embedded clauses, like main
clauses, have a topic (and possible contrastive focus) projections above the non-topic
subject position. Furthermore, these positions must follow an overtly-realized je. The fact

4
There is some inter-speaker variation in the acceptance of sentences with clause-initial contrastive focus
on objects, with some speakers finding such examples marginal.

48
that in postverbal embedded clauses je is obligatorily initial, preceding fronted topics and
foci, suggests that it instantiates the highest complementizer head Force˚, above both
TopicP and FocusP. The relevant part of the clause structure is given below in (25). The
structural position of non-topic subjects, which we can conclude is below FocusP, will be
discussed in more detail in Section 2.4.2.
5

(25) Postverbal embedded clause
V [
ForceP
(je) [
TopicP(s)
(XP
top
) [
FocusP
(XP
foc
) [
FinP
[
IP
...

Taking this representation as a starting point, I will present evidence that je in preverbal
embedded clauses is pronounced not in Force˚, but in the Fin˚ complementizer head.

2.3.2 Preverbal embedded clauses

In this section, we extend the same diagnostics to preverbal embedded clauses, where je
is obligatorily non-initial. Previous descriptions of the pattern have noted that phrases
that precede non-initial je are typically accompanied by a special discourse interpretation:
Dasgupta (1980) analyzes non-initial je as a focus marker; Bhattacharya (2003) treats
preverbal je as a marker of contrast; Bayer and Dasgupta (to appear) analyze je as a

5
With respect to movement to the left periphery, the present data considers only the patterning of subjects
and objects. Adverbs, for instance, are more freely admitted in pre-subject positions without any clear
requirement of definiteness or being non-quantified (anonymous, personal communication).

(i) [kOkhono-i je Ram e-kOtha bolte pare na] Rebeka janto
[never-emp that Ram this-talk spak.inf can.3 neg Rebecca knew
'Rebecca knew that Ram can not ever say this.'

Rather than call into question the proposed restrictions on object fronting, these facts suggest that adverbs
are somehow less restricted in their placement than subjects or objects. This generalization is also reported
in the experimental study by Rahman (2010), who shows that Bangla wh-adjuncts and adjuncts with
contrastive focus display greater freedom of placement than their subject and object counterparts.

49
complementizer that simultaneously marks emphatic topicalization. I will present new
evidence that these previous descriptions are largely on the right track, but in need of
some refinement. I show that phrases preceding je can be either topicalized or
contrastively focused, with restrictions on the orderings of topics and foci. However,
these interpretational constraints are obligatorily applied to only to fronted objects, but
not subjects.
I first consider the interpretational restrictions of fronted objects that precede je. We
will see that objects in this position must be either topicalized or contrastively focused,
suggesting that object fronting targets Topic and Focus projections above non-initial je.
Non-focused definite DP objects can freely occur to the left of je (26), or remain in a
position below je, as long as je is not initial within the clause (27).

(26) John [chatro du-to -ke je dadubhai dEkh-eche] bol-lo
John student 2-CL -ACC that grandfather see-PERF say- PST
'John said that grandfather saw the two students'

(27) John [dadubhai je chatro du-to -ke dEkh-eche] bol-lo
John grandfather that student 2-CL-ACC see-PERF say-PST
'John said that grandfather saw the two students'

On the other hand, quantified or indefinite object DPs are dispreferred to the left of je
under a neutral reading. However, these word orders are possible if the indefinite is
contrastively focused.

(28) # John [du-to chatro-ke je dadubhai dEkh-eche] bol-lo
John 2-CL student-ACC that grandfather see-PERF say-PST

(29) # John [kau-ke je dadubhai dEkh-e-ni] bol-lo
John anyone-ACC that grandfather see-PERF-NEG say-PST

50
(30) John [DU-TO CHATRO-KE je dadubhai dEkh-eche]bol-lo
John 2-CL student-ACC that grandfather see-PERF say-PST
'John said that grandfather saw TWO STUDENTS.'

The patterning of definite and indefinite objects before je parallels what is observed in the
pre-subject position of postverbal embedded clauses, providing evidence that objects that
precede je in preverbal clauses must occupy either a topic or focus position, based on
differences in the patterning of definite and non-definite objects.
The permitted word orders in sentence-initial embedded clauses, where the embedded
clause precedes the matrix subject, are similar to those of sentence-medial clauses. The
complementizer cannot be clause-initial, and can be preceded by more than one
topicalized phrase. The crucial difference is that focus is unavailable within the fronted
clause.

(31) [kal rate oSudh je dadubhai khey-eche] Jon bol-lo
last night medicine that grandfather eat-PERF] John say-PST

(32) * [DADUBHAI je oSudh khey-eche] Jon bol-lo
[grandfather that medicine eat-PERF] John say-PST

This restriction seems to be due to the fact that sentence-initial embedded clauses are
themselves topicalized, precluding the occurrence of focus within the clause.
We now turn back to sentence-medial embedded clauses, and the observation by
Bhattacharya (2001) that the topic and focus positions can be simultaneously filled, but
with several restrictions. Topics are crucially ordered before focused phrases, and it is not
possible for non-topics to precede focus.

51
(33) John [chatro du-to-ke DADUBHAI je dEkh-eche] bol-lo
John student 2-CL-ACC GRANDFATHER that see-PERF say-PST
'John said that GRANDFATHER saw the two students'

(34) # John [du-to chatro-ke DADUBHAI je dEkh-eche] bol-lo
John 2-CL student-ACC GRANDFATHER that see-PERF say-PST

As additionally noted by Simpson and Bhattacharya, only one contrastively focused
phrase is allowed, though it can be preceded by multiple topics.

(35) John [Meri Borders-e kal HÆMLET je kin-eche] jane
John Mary Borders-loc yesterday HAMLET that buy-perf knew
'John knew that it was HAMLET that Mary bought yesterday at Borders'
(Simpson and Bhattacharya 2003)

A highly similar ordering restriction is observed in Italian main clauses, which places
ordering restrictions on leftward-moved phrases (Benincà and Poletto 2004). When both
topics and focus are moved, topics must precede focus. While multiple topics are
permitted, only a single focused constituent can be fronted. This suggests that the
restrictions observed in Bangla and Italian are due to universal properties of the clausal
left periphery.
Within the structure of the extended left periphery, we now have strong evidence that
non-initial je in preverbal CPs instantiates the Fin head, below both TopicP and FocusP.
TopicP and FocusP are strictly ordered in Bangla, which additionally allows for multiple
TopicPs.

(36) Preverbal embedded clause
[
ForceP
[
TopicP(s)
(XP
top
) [
FocusP
(XP
foc
) [
FinP
(je) [
IP
... ]]]]] ... V

52
At this point, we have developed a structural account of variation in the placement of je;
clause-initial je in postverbal clauses occurs in Force˚, while non-initial je in preverbal
clauses is in Fin˚.
Before turning back to properties of the complementizer, it should be noted that the
extended CP structure allows for natural account for variation in the placement of
embedded clauses. Assuming that postverbal embedded clauses are in their base position,
sentence-medial embedded clause placement is derived by movement of the embedded
clause to FocusP, below the subject in TopicP. Due to the dispreference for focus within
sentence-initial embedded clauses, this clause placement appears to be the result of
movement of the embedded clause to TopicP.

(37) Postverbal clause placement
[
ForceP
[
TopP
[
FocP
[
FinP
[
IP
XP
subj
... [
VP
V [
ForceP
... ] ]]]]]]

(38) Sentence-medial clause placement
[
ForceP
[
TopP
XP
subj
[
FocP
[
ForceP
... ] [
FinP
[
IP
XP
subj
... [
VP
V [
ForceP
... ] ]]]]]]

(39) Sentence-initial clause placement
[
ForceP
[
TopP
[
ForceP
... ] [
FocP
[
FinP
[
IP
XP
subj
... [
VP
V [
ForceP
... ] ]]]]]]

2.4 The syntax of two jes

It appears then, that the complementizer je occurs in different positions in the left
periphery depending on the position of its embedded clause within the matrix clause. In
this section, I argue that the distribution of je can be accounted for without having to
propose that initial and non-initial je are simply different morphemes. Specifically, we

53
will see that surface variation in the placement of je receives a principled explanation as
the result of variation in copy pronunciation.

2.4.1 Copy pronunciation and je placement

Recall that as proposed by Rizzi, the two complementizer heads, Fin˚ and Force˚, account
for various functions performed by complementizers. I assume that Force˚ primarily
encodes clause type properties (declarative, interrogative, etc.), which can be selected by
matrix verbs. Fin˚ determines inflectional properties of the embedded clause, minimally
whether the clause is finite or non-finite. While the functions of Force˚ and Fin˚ are
instantiated with separate morphemes in some languages (Roussou 2000; Roberts 2004;
Aboh 2006), it is possible for complementizers to bear properties of both; English that is
specified as both declarative and finite, as is Bangla je.
I argue that the variable positions of je are best accounted for using the mechanism of
lower copy pronunciation (also referred to as lower copy spell-out), proposed by Franks
(1998) and Pesetsky (1998) and further developed by Bobaljik (2002), Bošković (2002)
and Bošković and Nunes (2007). The model assumes the Copy Theory of Movement
(Chomsky 1993), in which the operation Move creates a copy of a previously merged
element. I propose that all embedded clauses with je have the base structure in (40),
where je is first merged in Fin˚ and a copy is merged in Force˚.
6
This accounts for the
fact that regardless of its position, an embedded clause with overt je has identical

6
For an argument from southern Italian dialects that complementizers leave copies in multiple left-
peripheral heads, see Ledgeway (2005). Branigan (2011) argues for subject agreement on complementizers
in Germanic languages as evidence for Fin-to-Force movement.

54
semantic restrictions; je appears only in declarative clauses, a Force specification, and
occurs only in finite clauses, a Finiteness restriction.

(40) ForceP
3
3
Force˚ TopicP
je
k

3
3
Topic˚ FocusP
3
3
Focus˚ FinP
3
3
Fin˚ TP
je
k
6

At some point in the derivation, presumably spell-out of the CP phase, this syntactic
structure is transferred to the PF grammar, which determines which copy of je is
pronounced.
7
Here, I follow numerous analyses in proposing that the determination of
copy pronunciation results from the interplay between syntactic and prosodically-oriented
requirements. While only the highest copy of any movement chain is pronounced by
default, presumably since it has checked the most uninterpretable features (Nunes 1995),
the pronunciation of a lower copy or of multiple copies can take place to prevent the
violation of PF constraints, including requirements for cliticization or affixation (Franks
1998; Bošković 2001; Bobaljik 2002), avoidance of adjacent phonologically identical
words (Bošković 2002; Bošković and Nunes 2007), constraints on sentential stress

7
This work leaves aside the unresolved though important question of which head delimits the 'CP phase' in
a theory that assumes the existence of multiple complementizer heads. For some discussion, see Roberts
(2012) and (Manetta 2011), whose account does not assume the existence of multiple complementizer
heads.

55
placement (Stjepanović 2007), or requirements on the marking of prosodic edges (An
2007; Kandybowicz 2009). Viewed as a type of constraint interaction, lower copy
pronunciation takes place when pronunciation of the highest copy would result in the
violation of a more highly ranked well-formedness condition. It will be argued in chapter
3 that the pronunciation of je's highest copy in Force˚ is prevented when it would place
the complementizer at an intonational phrase edge.
In Bangla, default pronunciation of the highest copy in Force˚ produces the je-initial
order of postverbal embedded clauses. Two operations are necessary to create non-initial
orderings of je: the lower copy of je is pronounced in Fin˚, and at least one XP must be
moved to a position higher than Fin˚. A critical advantage of a lower copy spell-out
account is that it explains the fact that je's formal properties (+finite, +declarative) are
independent of its surface position; the complementizer follows the same movement
chain in all embedded clauses, and is uniformly specified for properties of both Force˚
and Fin˚. Furthermore, it readily accounts for the fact that je is not strictly confined to
second position, since multiple TopicPs are permitted and TopicP and FocusP can be
simultaneously filled.
One might alternatively consider that je has separate merge sites, with post-verbal je
Merged directly in Force˚, and pre-verbal je Merged in Fin˚. While this would account
for surface word orders, the approach comes with the non-trivial task of explaining why
je is only Merged in Force˚ in postverbal embedded clauses and only in Fin˚ in preverbal
ones. An additional difficulty is the fact that regardless of its surface position, je
apparently bears semantic specifications of both Force˚ and Fin˚ (being both finite and
declarative).

56
The existence of this head movement chain within the left periphery finds further
support in the patterning of jodi 'if', which introduces the precondition of a conditional
statement. Unlike je, jodi appears to freely occur in either a clause-initial position or a
post-subject position (Chacón 2014; David 2015). Crucially, these positions appear to
correspond to those in which je is realized, depending on clause position. While it is not
yet clear why jodi allows for variation in its position, it suggests that individual
complementizers can be pronounced in different positions without effects on their
interpretation.

(41) (jodi) Sita (jodi) bharot-e Sofor kor-be tobe taj mohol dekh-be
if Sita if India-LOC travel do-FUT then Taj Mahal see-FUT
If Sita travels to India, she will see the Taj Mahal.' (David 2015)

It is crucial to this proposal that movement of je from Fin˚ to Force˚ can take place
without leaving intermediate copies in intervening TopicP(s). If je were to leave a copy in
each structurally intervening head, one might expect that in preverbal embedded clauses
je would simply be pronounced in its next highest copy. Specifically, we would not
expect je to be pronounced after more than one phrase, as it is in (31), (33), and (35).
Following Borsley et al. (1996) and Roberts (2000, 2010), I propose that head movement
takes place through feature agreement, and need not take place through all structurally
adjacent heads. Thus, it appears that in Bangla, a Probe-Goal relation (Chomsky 2000) is
not established between Force˚ and Topic˚ or Topic˚ and Focus˚ (the next section
presents some evidence for copying in the Focus˚ head).
The possibility for preverbal je to be pronounced after multiple topics and foci also
excludes an alternative account that appeals to the mechanism of Local Dislocation in

57
Distributed Morphology (Halle and Marantz 1993; Embick and Noyer 2001; Embick
2003; Kramer 2010). Local Dislocation is conceived of as a postsyntactic PF operation
that applies to immediately adjacent morphological terminals, creating a complex head
that potentially alters their base ordering.

(42) X * Y > [Y - X] or [X - Y]

To extend a similar analysis to Bangla, one might propose that in preverbal embedded
clauses, Local Dislocation applies to je and an adjacent phrase, assuming that some
phrases can pattern like heads due to phase impenetrability (Kramer 2010):

(43) [
CP
je * XP ... ] > [
CP
XP - je ... ]

The problem with this approach, and a similar mechanism of Prosodic Inversion (Halpern
1992), is that it cannot straightforwardly account for "non-local" reorderings where more
than one phrase precedes je within its CP. To account for patterns where je is preceded by
both a topic and focus, it would be necessary for both the topic and focus to form a single
constituent, an improbable representation. Such orders, however, are readily predicted to
be available by lower copy pronunciation.

2.4.2 Subjects and non-initial je

Having presented the general structure of the analysis, we will turn to the unique
patterning of subjects that precede je in preverbal embedded clauses. Thus far, we have
observed that object DPs can only precede je with certain restrictions on their

58
interpretation; they must either be topicalized or focused. While one could expect the
same information structure restrictions to hold for subjects preceding je, this is not the
case. In addition to definite subjects, indefinites and quantified phrases are freely
admitted preceding je, even in the absence of focus intonation. If positions preceding je
are uniquely available to topics or focus, the grammaticality of (45)-(47) is unexpected.

(44) John [chatro du-to je eshe-che] bol-lo
John student 2-CL that come-PERF say-PST
'John said that the two students came'

(45) John [du-to chatro je eshe-che] bol-lo
John 2-CL student that come-PERF say-PST
'John said that two students came'

(46) John [kew je ase-ni] bol-lo
John [anyone that come-NEG] say-PST
'John said that no one came'

(47) [kew je ase-ni] John bol-lo
[anyone that come-NEG] John say-PST
'John said that no one came'

What is clear from these examples is that subjects that precede je in preverbal embedded
clauses are not subject to the same discourse restrictions as objects. This pattern presents
a clear challenge to analyses of je-second orders as the result of topic or focus features, as
proposed by Bhattacharya (2001) and Bayer and Dasgupta (to appear).
This particular subject-object asymmetry is attested in many second position
phenomena. In verb-second languages, first-position subjects are typically unrestricted in
their discourse interpretation, while first-position objects are compatible with only a
subset of possible readings. For example, sentence-initial objects in Swedish are
restricted to topics (Holmberg 2015); definite DPs are easily fronted (48), while

59
quantified DP objects cannot appear initially (49). Subjects do not follow this restriction,
as seen by the acceptability of quantified or indefinite subjects in sentences uttered with
broad focus (50).

(48) Den filmen får du bara inte missa
That film must you just not miss
'You simply mustn't miss that film.' (Holmberg 2015)

(49) *? Allt åt Johan
Everything ate Johan (Holmberg 2015)

(50) Allt är stängt
Everything is closed
'Everything is closed'

Similar asymmetries between first-position subjects and objects are observed in Kashmiri
(Bhatt 1999) and German (Travis 1984; Fanselow 2004; Mohr 2009; Fanselow and
Lenertová 2010). Given that verb-second effects are widely claimed to arise in the CP
domain (den Besten 1983; Fanselow 2004; Roberts 2004, among others), this subject-
object asymmetry in Bangla is likely due to the same structural property.
The fact that subjects do not need to be topicalized or focused before je suggests that
they occupy a position in the high clausal periphery other than TopicP or FocusP, one
that does not impose any particular discourse interpretation. In the absence of further
diagnostics for their position, I will tentatively claim that subjects fill the specifier of
FinP, as represented in (51).

(51) [
ForceP
(je) [
TopicP
(XP
top
) [
FocusP
(XP
foc
) [
FinP
(XP
subj
) (je) [
IP
...

The analysis of subjects in a low left-peripheral position is additionally supported by
languages in which subjects precede certain complementizers. For example, consider the

60
subject doubling construction observed in certain Italian dialects which permit [COMP
subject COMP] orders in embedded clauses.

(52) Turin Piedmontese
A venta che gnun ch'a fasa bordel
It needs that nobody that-CL do-SUBJUNC noise
'It is necessary that nobody make noise.' (Poletto 2000)

(53) Castellazzo Bormida
l=ɛ miɐi ke vujautʃ k=i=m=la dagi
it=is better that you that =you=me=it give
It is better that you give it to me' (Manzini 2010)

Crucially, Poletto (2000) argues that the position between the two complementizers is a
genuine subject position since it permits quantifier subjects like nobody, whereas
quantifiers in this language cannot be placed in the left periphery by left-dislocation or
topicalization. Such structures would appear to reflect a structure like (51), with the only
difference being that both the high and low copies of the complementizer are pronounced
in these languages, for reasons yet to be understood.
Turning back to Bangla, however, the analysis in which je has copies only in Fin˚ and
Force˚ predicts that [Focus Subject je ...] word orders should be possible in preverbal
CPs. Such examples, however, are ungrammatical.

(54) *John [DU-TO CHATRO-KE dadubhai je dEkh-eche]bol-lo
John 2-CL STUDENT- ACC grandfather that see-PERF say-PST

Recall that previous analyses of je as a focus marker have followed from the fact that if a
preverbal embedded clause contains a fronted focused phrase, je is placed immediately
after it. I propose that this restriction follows from a property of the focus projection.
Namely, movement of phrases to Spec,FocusP takes place only if a copy of je is merged

61
in Focus˚. Thus, the representation of an embedded clause with preverbal focus is given
in (55).

(55) [
ForceP
(je) [
TopicP
(XP
top
) [
FocusP
(XP
foc
) (je) [
FinP
(XP
subj
) (je) [
IP
...

This elaborated structure remains compatible with the lower copy spell-out proposal. In
the event that je can not be pronounced in Force˚, the second-highest copy is pronounced.
This takes place in Focus˚ if the projection is filled; otherwise, je is pronounced in Fin˚.

2.5 Conclusion

This chapter has presented the puzzling distribution of the complementizer je in Bangla,
and argued that variation in the placement of je within embedded clauses follows from
the pronunciation of different links in a uniform movement chain. The question remains
as to what condition forces the pronunciation of je's lower copy only in preverbal
embedded clauses, but not postverbal ones. In the next chapter, I present the prosodic
properties of Bangla je and embedded clauses and show that lower copy pronunciation
occurs uniquely when embedded clauses are prosodified as intonational phrases, taking
place to prevent the alignment of je with an intonational phrase boundary.

62
Chapter 3: Prosodic constraints on Bangla complementizer order

3.1 Introduction

This chapter proposes that the pronunciation of a lower copy of the Bangla
complementizer je is driven by a restriction on prosodic organization in the language. It
first develops a description and analysis of the prosodic structure of Bangla embedded
clauses and je, and argues that lower copy spell-out takes place exclusively when the
embedded clause is prosodified as a distinct intonational phrase. Furthermore, I argue that
the requirement for je to be non-initial in these cases does not follow from the
complementizer's status as an 'obligatory clitic,' contra Chacón (2014) and Bayer and
Dasgupta (to appear). I propose a view of the PF grammar in which the creation of
prosodic structure and the pronunciation of movement copies are determined
simultaneously, modeling the interaction in an Optimality-Theoretic analysis (Prince and
Smolensky 1993). The chapter concludes with a comparison of the proposal with
previous analyses of Bangla je.

3.2 Prosodic properties of je and embedded clauses

To address how the Bangla pattern is influenced by prosodic requirements, there are two
main sets of questions to be answered. We first consider the prosodic status of postverbal
and preverbal embedded clauses; I present a model of intonational prosody in Bangla and
show that preverbal clauses are demarcated by stronger prosodic boundaries than

63
postverbal ones. Only preverbal clauses are intonational phrases, while postverbal clauses
have the status of recursive phonological phrases.
We then consider the prosodic properties of je. Does je have the properties of a
prosodic clitic, or is it prosodically independent? Furthermore, does je's prosodic status
depend on whether or not it is initial within the embedded clause? I show that je is
crucially not a prosodic enclitic when it is initial within a postverbal embedded clause,
arguing against analyses of non-initial placement of je as the result of its specification as
an obligatory enclitic. Rather, I propose that it follows from a more general ban against
prosodically weak elements at intonational phrase boundaries.

3.2.1 Subjects and data collection

The presented data consist of recordings of three native speakers (2 female, 1 male) of
Kolkata Standard Bangla living in the United States, made between March and May of
2014. The recordings were obtained through a reading task where speakers were
presented with a list of pre-constructed sentences in Romanized script. These included
several simple sentences with no embedded clauses, as well as sentences with embedded
clauses in postverbal, clause-medial, and clause-initial positions, both with and without
je. The basic inventory of names and verbs are based on those of the study in Khan
(2008).
They were asked to read at a neutral speech rate with the intonation that would be
most natural for the written sentence. Where necessary, contrastive focus was marked
through bolded text and indicated to the speakers. Three tokens of each utterance were

64
recorded, with occasional repetitions due to speaker disfluencies or misreadings. Each
speaker was recorded in an indoor room at the University of Southern California, using a
built-in microphone on a MacBook Pro laptop computer. All recordings, pitch analysis,
and prosodic segmentation were done in Praat (Boersma and Weenink 2011).
The informants in the present study are speakers of Kolkata Standard Bangla, rather
than the Bangladeshi Standard Bangla that was the primary basis of Khan's (2008)
analysis. However, unless otherwise noted, no differences in the realization of tones or in
prosodic structure have been observed with respect to the relevant data.

3.2.2 The intonational prosody of Bangla

The intonational phonology of Bangla is relatively well-researched, having been
documented by numerous researchers (Hayes and Lahiri 1991; Selkirk 2007; Khan 2008,
2010, 2014; Féry 2010). In this section, I propose a model of syntax-prosody mapping of
Bangla prosody within Match Theory, focusing on phrasing at the level of the
phonological phrase and the intonational phrase.
The principle claim of Match Theory (Selkirk 2009, 2011) is that the mapping from
syntactic to prosodic structure is principally driven by a universal set of violable
constraints that call for a direct correspondence between syntactic and prosodic
constituency. Crucially, these constraints can be violated to satisfy prosodic markedness
constraints, allowing for mismatches between syntactic and prosodic structure, and
restricted cross-linguistic variation in phrasing. Each MATCH constraint has two
arguments, a syntactic constituent and a prosodic category. A universal set of MATCH

65
constraints accounts for the strong tendency for direct syntax-prosody correspondence at
the levels of words, phrases, and clauses. Here, we will consider two proposed constraints
on phonological phrases (PPhs) and intonational phrases (IntPs). MATCH(LexP, PPh),
calls for the correspondence of lexical maximal projections (e.g. NP, AdjP, VP) to PPhs
(Selkirk 1986; Selkirk 1996; Varis 2012; cf. Elfner 2012), while MATCH(ForceP, IntP)
requires clauses to correspond to IntPs. Here, I will adopt the constraint schema of Elfner
(2012) in which correspondence requires a relation of exhaustive dominance. A prosodic
constituent α exhaustively dominates a set of terminal nodes β if and only if α dominates
all and only the terminal nodes in β.

(1) MATCH(LexP, PPh)
For any lexical maximal projection (LexP) in the syntactic representation that
exhaustively dominates one or more terminal nodes α, assign a violation mark if
there is no phonological phrase (PPh) in the corresponding phonological
representation that exhaustively dominates the phonological exponents of α.

(2) MATCH(ForceP, IntP)
For any Force Phrase in the syntactic representation that exhaustively dominates
one or more terminal nodes α, assign a violation mark if there is no intonational
phrase (IntP) in the corresponding phonological representation that exhaustively
dominates the phonological exponents of α.

We will see that these two constraints are sufficient to derive much of the attested
prosodic phrasing in Bangla.
In Bangla, the two prosodic constituent types are most clearly distinguished by their
tonal realizations, which I annotate using a version of the Tones and Break Indices model
(Silverman et al. 1992; Beckman and Hirschberg 1994), based largely on Khan (2008,
2010). Each phonological phrase (PPh), referred to in some works as the Accentual
Phrase (Jun 2005; Khan 2008, 2010), is associated with a pitch accent on its first stressed

66
syllable (T*) and a boundary accent at its right edge (Ta). All tones are specified as either
high (H) or low (L). While there is some dialectal variation in the default pitch
specifications on PPhs, the dominant pattern observed in standard dialects is a low pitch
accent followed by a high boundary tone (L*...Ha). This contour also characterizes the
vast majority of the data elicited for the present study.
The typical realization of Bangla PPhs is shown in the example below, where the
L*...Ha contour tone is identifiable on PPhs corresponding to the nominal constituents
(DP/NP) Kamal, kal, and Meri-ke. Note as well the general downtrend in pitch
prominence that affects PPhs, such that each Ha pitch peak within the utterance is lower
than the previous one.
(3) Kamal kal Meri-ke nie eshe-che
Kamal yesterday Mary-ACC taken come-PERF
'Kamal brought Mary yesterday'

The proposed syntactic structure for the above sentence is given in (4)
1
, followed by the
prosodic phrasing expected if MATCH(LexP,PPh) is always satisfied in Bangla. Putting
aside the pitch characteristics of the verb phrase nie eshe-che, which shows an

1
As we are not directly concerned with the syntactic position of the adverb and direct object, the phrases
that dominate are labeled as generic Functional Projections (FP).

67
intonational phrase boundary tone, the attested prosodic phrasing in the pitch track above
corresponds with the prosodic structure predicted by the satisfaction of MATCH(LexP,
PPh).
(4) a. Syntactic structure
ForceP
wp
NP FP
5 wp
Kamal NP FP
5 wp
kal NP VP
5 5
Meri-ke nie eshe-che

b. Prosodic representation predicted by MATCH(LexP, PPh):
(
PPh
Kamal) (
PPh
kal) (
PPh
Meri-ke) (
PPh
nie eshe-che)

In the absence of evidence to the contrary, it will be maintained in the remainder of the
analysis that simplex lexical phrases correspond directly to PPhs (first argued by Hayes
and Lahiri 1991). In other words, MATCH(LexP,PPh) is always satisfied in Bangla.
2

Recent studies in syntax-prosody correspondence have argued for widespread
recursion of prosodic categories (Wagner 2005; Selkirk 2011; Elfner 2012; Elordieta
2015) of the kind predicted by Match constraints. Furthermore, phonological processes
and aspects of phonetic interpretation can apply specifically within subparts of recursive
prosodic domains (Itô and Mester 2009, 2013; Elfner 2010a; Elordieta 2015; Hsu 2015).
In particular, Itô and Mester (2009, 2013) propose that the following natural classes of
constituent types within a recursive PPh structure can be targeted by domain-specific
processes:

2
As discussed in greater detail by Hayes and Lahiri (1991) and Khan (2008), the organization of PPhs can
significantly vary due to speech rate or the presence of focus. In the present analysis, I will focus on
phrasing attested at a relatively careful speech rate.

68
(5) Projections of PPh
IntP
|
PPh [+max, -min]
| PPh
[-min]
, non-minimal PPhs
... PPh [-max, -min]
| PPh
[-max]
, non-maximal PPhs
... PPh [-max, +min]
|
Pwd

Here, I present evidence for PPh recursion in Bangla, showing that recursive PPhs differ
in their pitch realization from nonrecursive ones. Specifically, I will argue that the
prosodic constituents labeled as intermediate phrases by Khan (2008) correspond to non-
minimal PPhs, i.e. any PPh that dominates another PPh.
Khan (2008) first notes that somewhat larger constituents pattern differently from the
simplex ones discussed. For example, phrases that correspond to post-positional phrases
and complex NPs exhibit a larger variety of boundary tones than items with less syntactic
branching. For example, while PPhs corresponding to simplex NPs generally permit only
a high boundary tone at their right edges, these phrases permit a wider variety of
boundary tones, including a high tone (H-), low tone (L-), rising contour (LH-), and
falling contour (HL-). Furthermore, high tones associated with these constituents can
violate the downstep trend generally observed with successive PPhs.
For example, consider the following utterance, which differs from the previous one in
that it contains a complex adverbial, dupur belae 'early afternoon.' Here, the high tone
realized on the adverbial is noticeably upstepped from the one on the preceding subject,
unlike the Ha boundary tones previously discussed.

69
(6) Kamal dupur bela-e Meri-ke nie eshe-che
Kamal early.afternoon time-LOC Mary-ACC taken come-PERF
'Kamal brought Mary in the early afternoon'

These items are analyzed by Khan as instantiations of a separate prosodic constituent
above the PPh, the intermediate phrase. However, several authors have noted that the
existence of intermediate categories between the Phonological Phrase and Intonational
Phrase is difficult to account for given a constituent-structure based mapping from
syntactic to prosodic structure (Selkirk and Tateishi 1988; Itô and Mester 2013). Indeed,
although Khan does not aim to propose a formal account for their syntax-prosody
mapping, intermediate phrase status is assigned to a large, non-uniform set of syntactic
constituents: "The intermediate phrase is a grouping of [PPhs] ... often corresponding to a
small phrase (e.g. postpositional phrase, topic, adverbial) and occasionally to a clause
(e.g. relative clause, if-clause, because-clause)" (Khan 2008; 139-140).
Here, I propose that the patterning of these items is straightforwardly accounted for
by assuming that Phonological Phrasing in Bangla can be recursive, and that the phonetic
correlates of PPhs depend on their position within a recursive structure. Under the
assumption that Bangla uniformly maps lexical phrases to PPhs to satisfy MATCH(LexP,

70
PPh), several proposed "intermediate phrase" types should in fact correspond to PPhs that
dominate another PPh.
To illustrate, consider the complex adverbial from the previous example, dupur belae
'early.afternoon time-LOC'. In its syntactic representation, the sequence is plausibly
analayzed as a postpositional phrase PP headed by the locative marker e that dominates
an NP whose head bela 'time' is modified by dupur 'early.afternoon.' Abstracting away
from the syntax of noun-noun modification, the complement of the postposition, dupur
bela corresponds to an NP. If both NP and PP are treated as lexical phrases, the prosodic
structure that satisfies MATCH(LexP, PPh) is a recursive PPh structure. This
correspondence is shown in (7).
(7) a. Syntactic structure
PP
3
NP P
5 -e
dupur bela

b. Prosodic representation predicted by MATCH(LexP, PPh)
(
PPh-non-min
(
PPh-min
dupur bela) e)

The same reasoning can be applied to two other types of constituents identified by Khan
to show "intermediate phrase" status, modified NPs and possessive NPs.

(8) Recursive PPh structures predicted by MATCH(LexP, PPh)

Postpositional phrases: [[NP] P]
PP
> (
PPh
(
PPh
NP) P )

Modified NPs: [AdjP [NP]]
NP
> (
PPh
AdjP (
PPh
NP)))

Possessive NPs: [NP [NP]]
NP
> (
PPh
(
PPh
NP) (
PPh
NP))

71
Rather than attribute the special tonal characteristics of "larger" nominals to the existence
of a separate prosodic category like the intermediate phrase, I propose that these
properties are the phonetic realization of the right edge of a non-minimal PPh. This
allows an account for the attested distinctions in the realization of different phrase types
in Bangla, while maintaining a single syntax-prosody mapping constraint.
The largest prosodic constituent is the intonational phrase (IntP), whose boundaries
typically correspond to main clauses or full utterances. IntPs do not bear a pitch accent,
and are associated with a boundary tone realized at their right edge, labeled T%. A
crucial property of Bangla is that boundary tones belonging to different prosodic levels
do not co-occur. When multiple right boundaries are aligned with each other, the
boundary tone of the highest-level category overrides all other tone specifications.
Intonational phrases bear a larger variety of boundary tones than PPhs (both minimal and
non-minimal). They also differ crucially from PPhs in that there is a closer
correspondence between individual boundary tones and discourse interpretations (Hayes
and Lahiri 1991, Khan 2008). This is illustrated in the following example, which contains
the sharp falling contour (HL%) associated with yes/no questions.

72
(9) Kamal kal Meri-ke nie eshe-che ki
Kamal yesterday Mary-ACC taken come-PERF Q
'Did Kamal bring Mary yesterday?'

Five distinct tones are identified by Khan (2008, 2014): The low tone (L%) appears at
the end of declarative clauses, the high tone (H%) appears as a continuation rise or as an
interrogative marker, a rising tone (LH%) is used in wh-questions, and the falling contour
(HL%) indicates yes/no questions. There is, however, a boundary tone contour associated
only with IntPs, a fall and rise (HLH%), described as having a function like that of a
continuation rise. Here, we will mostly be concerned with the H% and HLH% boundary
tones.
Lastly, IntP boundary tones and those of non-minimal PPhs differ crucially in how
local their effects on nearby tones are. While boundary tones of non-minimal PPhs affect
only the final syllables of their domains (described by Khan as the intermediate phrase
boundary tone locality constraint), IntP boundary tones can override expected pitch
accents and boundary tones of several preceding PPhs.
A summary of the prosodic constituent types in Bangla that are relevant to the current
analysis and their boundary tone correspondence is given below.

73
(10) Summary of prosodic constituent types in Bangla

Prosodic constituent Boundary tone inventory Corresponding syntactic
category
Minimal PPh Ha Simplex lexical XPs
Non-minimal PPh H-, L-, LH-, HL- Complex lexical XPs
IntP H%, L%, LH%, HL%,
HLH%
Clauses

3.2.3 Postverbal embedded clauses

We begin by considering the waveform and pitch track in (11), which contains a
postverbal embedded clause in a neutral reading. Several PPhs are readily identified by
their standard L*...Ha pitch contour where their pitch peak is clearly aligned with the end
of the phrase, Monoara, Kamal, and Meri-ke. Crucially, the main verb bol-lo bears a
rising H- contour that deviates from expected PPh
min
downstep, suggesting the presence
of a non-minimal PPh boundary at the embedded clause break.

(11) Monoara bol-lo [Kamal Meri-ke nie es-che]
Monoara say-PST Kamal Mary-ACC taken come-PERF
'Monoara said that Kamal brought Mary'

74
Most known boundary tones associated with PPh
Non-min
boundaries (H-, LH-, L-) are in
fact attested at the embedded clause breaks. In the following example, the main verb
bears a low boundary tone L-, typically preceded by a high pitch accent H*.

(12) Monoara bol-lo [je Romila Meri-ke nie es-che]
Monoara say-PST that Romila Mary-ACC taken come-PERF
'Monoara said that Romila brought Mary'

What aspect of the syntactic representation triggers the realization of these boundary
tones on verbs preceding postverbal embedded clauses? While we have previously shown
that these tones occur on the right edge of a non-minimal PPh, these are not predicted to
occur on the main verb according to the analysis developed so far. Consider the proposed
syntactic representation in (13) for the sentence in (11), and the prosodic structure
predicted by MATCH(LexP, PPh), MATCH(ForceP, IntP). Here, no right boundary is
aligned with the main verb, which only precedes the left edge of an IntP corresponding to
the embedded clause.

75
(13) a. Syntactic structure
ForceP
wp
TopicP
wp
NP ...
5 VP
Monoara wp
V ForceP
bol-lo wp
NP FP
5 wp
Kamal NP VP
5 5
Meri-ke nie eshe-che

b. Prosodic structure predicted by MATCH(LexP, PPh), MATCH(ForceP, IntP):
{
IntP
(
PPh
Kamal)(
PPh
bol-lo {
IntP
(
PPh
Kamal)(
PPh
Meri-ke)(
PPh
nie eshe-che)})}

To account for the appearance of a PPh non-min boundary tone on the main verb, two
new claims are made. First, I propose that postverbal embedded clauses are "demoted" to
PPhs, in violation of MATCH(ForceP, IntP). Since they necessarily dominate PPhs
associated with phrases within the embedded clause, these thus correspond to non-
minimal PPhs, as shown in (14).

(14) {
IntP
(
PPh
Kamal)(
PPh
bol-lo) (
PPh-nonmin
(
PPh
Kamal)(
PPh
Meri-ke)(
PPh
nie eshe-che) ) }

Second, I propose that in addition to a boundary tone at their right edge, non-minimal
PPhs can trigger the appearance of a boundary tone on material immediately preceding
their left edge. While unexpected, this proposal additionally accounts for data in Khan
(2008) showing identical tonal patterns realized on verbs that precede objects modified
by relative clauses, which are straightforwardly predicted to correspond to non-minimal
PPhs. In Section 3.2.5, I propose that the violation of MATCH(ForceP, IntP) takes place in

76
order to prevent the violation of a separate restriction on prosodic organization,
LAYEREDNESS.

3.2.4 Preverbal embedded clauses

We now turn to the prosodic phrasing of preverbal embedded clauses. First, consider the
sentence-medial embedded clause given below. Unlike postverbal ones, medial
embedded clauses are identified as separate intonational phrases by several diagnostics.
The most salient distinction is the presence of a fall and rise contour realized at the end of
the embedded clause. Crucially, the HLH% contour boundary tone is uniquely associated
with IntP boundaries (Khan 2008). In this particular example, the boundary tone also
contains the pitch peak of the utterance.

(15) John [oSudh je dadubhai kal rate khey-eche] bol-echi-lo
John medicine that grandfather last.night eat-PERF] say-PERF-PST
'John said that grandfather took medicine last night'

A similar patterning is observed for sentence-initial embedded clauses. In the
following example, the end of the embedded clause is again aligned with a HLH% fall
and rise contour tone. It is also clearer in this example that pitch movement on all

77
following material is noticeably compressed. This pitch compression is uniformly
observed following preverbal embedded clauses.
3

(16) [Monoara Romila-ke nie es-che] Kamal bol-lo
Monoara Romila-ACC taken come-PERF Kamal say-PST
'Kamal said that Monoara brought Romila'

The fall and rise contour tone is not always attested at the ends of preverbal
embedded clauses, however. In some instances, the end of a preverbal embedded clause
is aligned with a high tone preceded by a relatively gentle rise (H%).

3
This compression (and possibly suppression) of pitch movement following preverbal embedded clauses
resembles the compression that follows focused phrases in Bangla (Hayes and Lahiri 1991; Selkirk 2007;
Khan 2008). In some instances, this seems to reflect the fact that preverbal embedded clauses themselves
are frequently interpreted as focused. However, since preverbal embedded clauses can receive topic
interpretations, it may reflect a more general requirement for IntP boundary tones like HLH% to be
separated by as few tones as possible from the end of the utterance (Selkirk 2007).

78

(17) [Monoara Romila-ke nie es-che] Kamal bol-lo
Monoara Romila-ACC taken come-PERF Kamal say-PST
'Kamal said that Monoara brought Romila'

An additional prosodic distinction between preverbal and postverbal embedded
clauses is that preverbal embedded clauses also allow for a type of intonational contour
that reflects a corrective semantic interpretation, at least for certain speakers. The pattern
is illustrated below for a sentence-initial embedded clause. As in the previous example,
the end of the embedded clause is aligned with a H% high tone. However, there is an
apparent absence of pitch reset in material that follows, and the utterance ends with a
rising contour (LH%).

(18) [Romila je eshe-che] Meri bol-lo
Romila that come-PERF Mary say-PST
'Mary said that Romila came'

79
This pattern is illustrated below for a sentence-medial embedded clause. Here, the high
tone at the end of the embedded clause is preceded by a longer pitch plateau.

(19) Meri [Ram je Romila-ke nie eshe-che] bol-lo
Mary Ram that Romila-acc taken come-PERF say-PST
'Mary said that Ram brought Romila.'

Because preverbal embedded clauses uniformly are phrased as distinct IntPs, their
prosodic representation satisfies MATCH(ForceP, IntP). The proposed syntactic structure
and prosodic representation for (19), which contains a sentence-medial embedded clause,
is given below.

(20) a. Syntactic structure
ForceP
wp
TopicP
wp
NP FocusP
5 wp
Meri ForceP
k
...
5 VP
Ram je Romila-ke nie eshe-che 3
V ForceP
k

bol-lo

b. Prosodic structure predicted by MATCH(LexP, PPh), MATCH(ForceP, IntP):
{
IntP
(
PPh
Meri){
IntP
(
PPh
Ram) je (
PPh
Romila-ke)(
PPh
nie eshe-che)} (
PPh
bol-lo)}

80

3.2.5 Syntax-prosody mapping of embedded clauses

The analysis of postverbal embedded clauses as non-minimal PPhs and preverbal
embedded clauses as IntPs confirms and formalizes the intuitions reported by Simpson
and Bhattacharya (2003) and Bayer and Dasgupta (to appear) that preverbal embedded
clauses are separated by larger prosodic breaks than postverbal ones. But how can this
difference between postverbal and preverbal embedded clauses in terms of their prosodic
constituency be accounted for within Match Theory? In the absence of conflicting
constraints, one would expect all embedded clauses to map to a single type of prosodic
constituent. Given that all embedded clauses are of the same syntactic constituent type
(ForceP), it would be expected for all embedded clauses to map to IntPs due to a
constraint like MATCH(ForceP, IntP), independently needed for main clauses.
However, the mapping of postverbal embedded clauses to a non-minimal PPh
satisfies a general principle on prosodic organization, LAYEREDNESS (Selkirk 1996),
which states that no prosodic constituent dominates another constituent that is of a higher
type within the Prosodic Hierarchy.

(21) LAYEREDNESS
No PCat
i
dominates PCat
j
, where i < j on the Prosodic Hierarchy (i.e. no PCat
dominates a bigger PCat)

If we maintain the claim that postverbal embedded clauses are complements of the main
verb, a syntax-prosody mapping that satisfies both MATCH(ForceP, IntP) and
MATCH(LexP, PPh) would result in a LAYEREDNESS violation. This is because the PPh
associated with VP dominates an IntP associated with the embedded clause, (
PPh
{
IntP
...}).

81
(22) Syntax-prosody mapping of postverbal embedded clause, satisfying
MATCH(ForceP,IntP) and MATCH(LexP, PPh)

a. [
ForceP
... [
InflP
XP
Subj
[
VP
V [
ForceP
... ]]]]
?
b. {
IntP
(
PPh
Subj ) (
PPh
V {
IntP
} )} violates LAYEREDNESS

Although it remains unresolved as to whether or not LAYEREDNESS is a violable principle
of prosodic organization (Selkirk and Lee 2015), it appears that its violation in Bangla is
avoided by "demoting" postverbal embedded clauses to PPhs, violating MATCH(ForceP,
IntP). This creates a PPh that does not exhaustively dominate a lexical XP, additionally
violating the prosody-syntax mapping constraint MATCH(PPh, LexP).

(23) MATCH(PPh, LexP)
Assign a violation mark for each prosodic phrase in the phonological
representation that does not exhaustively dominate the phonological exponents of
a lexical maximal projection in the syntactic representation.

(24)
[
VP
V [
ForceP
... ]] LAYEREDNESS MATCH
(PPh,LexP)
MATCH
(ForceP,IntP)
( V { ... }
IntP
)
PPh
*!
☞ ( V ( ... )
PPh
)
PPh
* *

On the other hand, if preverbal embedded clause placement is derived by movement
out of VP into TopicP or FocusP, the corresponding Intonational Phrase will no longer be
dominated by a PPh, assuming that no phrase in the extended projection of VP is mapped
to a PPh. Consequently, no LAYEREDNESS violation is predicted.

(25) Syntax-prosody mapping of initial embedded clause, satisfying
MATCH(ForceP,IntP) and MATCH(LexP,PPh)

a. [
ForceP
... [
TopicP
[
ForceP
... ] ... [
InflP
XP
Subj
[
VP
V [
ForceP
... ]]]]
?
b. {
IntP
{
IntP
} (
PPh
Subj ) (
PPh
V ) }

82
(26) Syntax-prosody mapping of medial embedded clause, satisfying
MATCH(ForceP,IntP) and MATCH(LexP,PPh)

a. [
ForceP
... [
TopicP
XP
Subj
[
FocP
[
ForceP
... ] [
VP
V [
ForceP
... ]]]]
?
b. {
IntP
(
PPh
Subj ) {
IntP
} (
PPh
V ) }

The difference in the prosodic size of embedded clauses according to their position
within the clause is thus straightforwardly explained as a means to avoid violation of a
principle of prosodic structure organization. The avoidance of LAYEREDNESS violations
may also provide a principled explanation for why "normal" embedded clauses and
restrictive relative clauses are seemingly never mapped to Intonational Phrases across
languages.

3.3 The prosodic representation of je

At this point, constraints on the distribution of je can be summarized as follows: je is
clause-initial except when its embedded clause forms an Intonational Phrase. In this
section, I consider the prosodic status of the complementizer itself, and argue that the
markedness condition that prevents it from surfacing IntP-initially is one within the
STRONGSTART constraint family (Elfner 2010b, 2012; Selkirk 2011). In addition, I show
that the restriction against IntP-initial je cannot be due to a requirement for je to be an
enclitic to a preceding item, as previously suggested by Chacón (2014) and Bayer and
Dasgupta (to appear), who have proposed that je is uniformly a prosodic enclitic.

83
3.3.1 je in postverbal embedded clauses

We first look at the prosodification of je in postverbal embedded clauses, where the
complementizer must be clause-initial. Consider the typical realization of je in a
postverbal clause preceded by a H- PPh
non-min
boundary tone. Here, the complementizer's
pitch realization falls in between that of the preceding high tone and the low pitch accent
of the PPh that follows. This argues conclusively against the enclisis of je to the main
verb, as one would expect its associated boundary tone to extend through the
complementizer.

(27) John bol-echi-lo [je dadubhai kal rate oSudh khey-eche]
John say-PERF-PST that grandfather last.night medicine eat-PERF
'John had said that grandfather took medicine last night.'

Given that je is obligatorily initial within a postverbal embedded clause and that it is
preceded by a PPh
non-min
boundary, we can admit several hypothetical representations of
the prosodic phrasing of je, following the classification of Selkirk (1996). Potentially, je
is a proclitic to a PPh, either to a minimal PPh (28) or the non-minimal PPh
corresponding to the embedded clause (29). One might also consider the possibility that it
is an independent PPh (30).

84
(28) PPh
min
proclitic
( )
PPhmax
( )
PPhmin
je XP

(29) PPh
non-min
proclitic
( )
PPhnon-min
( )
PPhmin
je XP
(30) Independent PPh (nonclitic)
( )
PPhnon-min
( )( )
PPhmin
je XP

We can rule out the representation of je as an independent PPh by considering its
Prosodic Word (PWd) status. While the PWd plays a minimal role in the Bangla
intonational system, PWd organization is relevant to restrictions on stress and vowel
length. Here, I will focus on its relation to vowel lengthening, as described by
Fitzpatrick-Cole (1991, 1994). Although vowel length in Bangla is not phonemic,
unaffixed monosyllabic words are uniformly realized with long vowels (31a). The same
monosyllabic stems are not lengthened if followed by a lexical affix (31b), yet are
lengthened before postlexical affixes such as focus markers (31c).

(31) a. /rag/ → [ra:g] 'anger'
b. /rag-i/ → [ragi] 'anger-ADJ'
c. /rag-i/ → [ra:gi] 'anger-FOC' (Fitzpatrick-Cole 1991; 157)

To account for these patterns, Fitzpatrick-Cole argues that Bangla requires each PWd to
be minimally bimoraic. Vowel lengthening lengthening thus applies to any PWd that
contains only one vowel. Lexical affixes are adjoined within the stem's PWd, thus
satisfying the minimal word requirement without the need for lengthening. Because
postlexical affixes are adjoined outside of the PWd, lengthening applies normally to
monosyllabic stems.

85
Based on this criterion, the complementizer je does not form its own PWd, as it is
always realized with a short vowel that never undergoes lengthening. Furthermore, je has
no effect on the realization of adjacent monosyllabic lexical words, which uniformly
surface bimoraically, indicating that je is not phrased with PWds of adjacent lexical
items. The fact that je is not a PWd crucially rules out a representation where je forms an
independent PPh, as this would violate HEADEDNESS, which appears to be an inviolable
principle of prosodic organization (Selkirk 1981, 1984, 1996; Nespor and Vogel 1986).
I will therefore analyze je as a syllable that is undominated by a PWd, a
representation consistent with the expectation that small function words do not form
PWds (Selkirk 1996). It remains indeterminate as to whether clause-initial je is
dominated by the minimal PPh of an adjacent word (33), or if it is directly dominated by
the non-minimal PPh that corresponds to the embedded clause (32). It should be noted,
however, that the PPh internal clitic structure is dispreferred given the proposed
inventory of MATCH constraints; The structure incurs a violation of MATCH(LexP, PPh)
because the lexical phrase following je is not exhaustively dominated by a PPh.
(32) PPh Affixal Clitic
PPh
nonmin

PPh
min

|
σ ...
|
je
(33) PPh Internal Clitic
PPh
min

PWd
|
σ ...
|
je

The somewhat indeterminate prosodic status of je in postverbal embedded clauses is
unproblematic for the analysis that follows. The crucial conclusions of this section are
that je is not a prosodic enclitic in postverbal clauses, and that it is not dominated by a

86
PWd node.

3.3.2 je in preverbal embedded clauses

In preverbal embedded clauses where je is non-initial, the pitch behavior of the
complementizer suggests that it is parsed as a type of enclitic, contained within a PPh that
also dominates its preceding syntactic phrase. Initial evidence for this is seen in the pitch
behavior of non-initial je, which is quite systematically aligned with a PPh pitch peak.
Consider the pitch track in (18), repeated below as (34), where the non-initial
complementizer bears a higher pitch than the final syllable of its preceding word, Romila.
The fact that je is aligned with a pitch peak indicates clearly that the two words share a
single PPh, and that je is at its right edge.

(34) [Romila je eshe-che] Meri bol-lo
Romila that come-PERF Mary say-PST
'Mary said that Romila came'

Assuming that the pitch peak realized on je corresponds to a high PPh boundary tone,
this is compatible with two possible clitic configurations, illustrated below with
hypothetical structures for an XP-je sequence. Je is either an internal clitic phrased with

87
its preceding word in a single PPh
min
(35) or an affixal clitic within a recursive PPh
structure (36).

(35) Internal clitic
( )
PPh

XP je

(36) Affixal clitic

(( )
PPh
)
PPh

XP je

A segmental diagnostic for prosodic phrasing, /r/-assimilation (Hayes and Lahiri 1991),
argues in favor of an affixal clitic analysis. In colloquial registers, /r/ is optionally
assimilated by some speakers to any following coronal consonant within the same PPh,
regardless of whether they are separated by a morpheme or word boundary. Crucially, /r/-
assimilation cannot apply if the two consonants are separated by a PPh boundary.

(37) bɔrʃa ~ bɔʃʃa 'rainy season'
bɔrdi ~ bɔddi 'elder sister'
kɔr-ʧ
h
e ~ kɔʧ-ʧ
h
e 'do-3PERS.PRES'
kɔr-lo ~ kol-lo 'do-3PERS.FUT' (Hayes and Lahiri 1991)

Palato-alveolar affricates are also susceptible to /r/ assimilation. Consequently, if je (
[ʤe]) is contained within the same PPh as a preceding word, /r/ assimilation is expected
to be possible. In this context, however, /r/-assimilation is blocked.
(38) [r ʤ] (*[ʤʤ])
| |
Jon [kukur je dadubhai-ke kamr-eche] bol-lo
John dog that grandfather-ACC bite-PRES.PERF say-PST
'John said that a dog bit grandfather.'

(39) [r ʤ] (*[ʤʤ])
| |
[Jon khabar je khey-eche] ami jani
John food that eat-PRES.PERF 1.PERS know
'I know that John ate food.'

88
The impossibility of /r/-assimilation with je thus leads us to posit an affixal enclitic
structure when je is non-initial within its embedded clause, as in (40).
(40) Prosodic structure of non-initial je in preverbal embedded clauses
{
IntP
... ((XP)
PPh
je )
PPh
... } ... V

3.4 Syntax-prosody interaction in phrasing and copy pronunciation

This section extends the Optimality-Theoretic analysis of syntax-prosody mapping to
account for both the prosodic phrasing of initial and non-initial je, and the effects of
prosodic restrictions on copy pronunciation. The analysis assumes that the phrasing of
function words depends on the interaction of MATCH constraints with other markedness
constraints on prosodic organization.

3.4.1 Constraints on prosodic organization and STRONGSTART

We first consider the phrasing of je in preverbal embedded clauses, where it surfaces as
an affixal enclitic to a preceding phrase (the derivation of lower copy spell-out will be
returned to later in this section). Assuming only the direct syntax-prosody
correspondence predicted by MATCH(ForceP, IntP) and MATCH(LexP, PPh), an instance
of je that is not initial within its embedded clause is predicted to be sandwiched between
two PPhs in a free clitic structure: [
ForceP
...

XP [
FinP
je [ XP ... ]]] → {(
PPh
XP) je (
PPh
XP)
...}. However, since je surfaces as an affixal enclitic in this context, this representation
must be ruled out by a separate markedness constraint. One such constraint violated by

89
free clitic representations is EQUALSISTERS (Myrberg 2013), which requires prosodic
sister nodes to be instances of identical prosodic categories. Here, I will consider an
indexed version of EQUALSISTERS that is evaluated among nodes immediately dominated
by an Intonational Phrase, defined below.
4
As long as EQUALSISTERS(IntP) outranks the
prosody-syntax correspondence constraint MATCH(PPh, LexP), je will be phrased within
some PPh.
5

(41) EQUALSISTERS(IntP)
Assign a violation mark if the immediate daughter nodes of an IntP are not
instances of the same prosodic category.

(42) EQUALSISTERS(INTP) >> MATCH(PPH, LexP)
[
ForceP
...

XP [
FinP
je [ XP ... ]]]
EQUAL
SISTERS
(INTP)
MATCH
(PPh, LexP)
{(
PPh
XP) je (
PPh
XP) ...}
free clitic
*!
☞ {((
PPh
XP) je) (
PPh
XP)...}
affixal enclitic
*
☞ {(
PPh
XP) (
PPh
je (
PPh
XP)...}
affixal proclitic
*

While this constraint ranking properly predicts a recursive, affixal clitic structure, it
does not determine whether function words like je are phrased as enclitics or proclitics.
Following Selkirk (2011) and Elfner (2012), I propose that the preference for enclisis is
derived by a STRONGSTART constraint.

4
As defined by Myrberg, EQUALSISTERS makes no reference to specific categories of mother or daughter
nodes. For the present analysis, it is necessary to propose an IntP-specific version of this constraint in order
to permit prosodic representations where je is an affixal enclitic within a PPh, as argued in Section 3.3.2.
5
For presentational simplicity, I will not consider candidates where je is dominated by a minimal prosodic
word associated with an adjacent XP, which additionally violate MATCH(LexP, PPh).

90
(43) STRONGSTART
Assign a violation mark for any prosodic constituent whose leftmost daughter
constituent is lower in the Prosodic Hierarchy than its sister constituent
immediately to the right: * ( π
n
π
n+1
….

In its general form, STRONGSTART amounts to a restriction against proclitic structures.
Schematically, STRONGSTART is violated by any prosodic representation like (44), where
i ≥ j and j > k in the Prosodic Hierarchy.
6

(44) PCat
i

PCat
j

PCat
k

Following Elfner (2012), I assume that STRONGSTART is in fact a family of
constraints, where individual STRONGSTART constraints are indexed to a prosodic
category κ. STRONGSTART(κ) is violated for each prosodic structure in which an instance
of κ is the leftmost immediate daughter of its mother node, and the immediately
following sister node of κ is a larger prosodic category.

(45) STRONGSTART(κ)
Assign a violation mark for every prosodic constituent whose leftmost daughter
constituent is of type κ and is lower in the Prosodic Hierarchy than its sister
constituent immediately to the right.

6
STRONGSTART differs in several important respects from constraints of the form Align-L(PCat
i
,PCat
i-1
),
which have previously been used to disprefer proclitics in the Alignment theory of syntax-prosody
correspondence (Selkirk 1986, 1996). STRONGSTART does not evaluate differences in category size
between prosodic constituents and their leftmost daughters, and instead depends crucially on the relative
sizes of the two leftmost constituents in some domain.

91
Given this constraint schema, any proclitic representation of je violates STRONGSTART(σ)
because je, a syllable not dominated by a PWd, has a sister constituent immediately to its
right that is either a PWd or PPh.

(46) PPh Internal clitic
PPh
min

PWd
|
σ ...
|
je
(47) PPh Affixal clitic
PPh
non-min

PPh
min

|
σ ...
|
je

The addition of STRONGSTART(σ) to the constraint set derives the selection of the affixal
enclitic representation.

(48)
[

XP [
FinP
je [ XP ... ]]] V
EQUAL
SISTERS
(INTP)
MATCH
(PPH,LEXP)
STRONG
START
(σ)
{(
PPh
XP) je (
PPh
XP) ...}
free clitic
*!
☞ {((
PPh
XP) je) (
PPh
XP)...}
affixal enclitic
*
{(
PPh
XP) (
PPh
je (
PPh
XP)...}
affixal proclitic
* *

While STRONGSTART(σ) is not crucially ranked relative to EQUALSISTERS or
MATCH(PPH,LEXP), it is ranked above other markedness constraints that would prefer
proclisis in this context, such as WEAKSTART (Sabbagh 2014) or a hypothetical
STRONGEND. However, these rankings are not crucial to the main analysis, and will be
omitted in the remainder of the chapter.

92
3.4.2 HIGHESTCOPY and lower copy pronunciation

We now turn to the question of deriving lower copy pronunciation in preverbal embedded
clauses. Here, I propose that the input syntactic structure to the component of PF that
creates a prosodic representation includes movement copies. In addition, we consider a
larger candidate set, where je can be spelled out in either Force˚ or Fin˚, generating either
initial or non-initial complementizer orders. The task of the grammar is to disprefer any
possible parse of je in IntP-initial position over one where je is spelled out in its lower
copy and phrased as an affixal enclitic.

(49) Input syntactic structure of preverbal embedded clause with je
[
ForceP
je [
Topic/FocusP
XP [
FinP
je ...]]] ... V

The potential for phonological constraints to determine copy pronunciation is
implemented in an Optimality-Theoretic grammar by allowing GEN, the component that
defines possible outputs, to both access movement chains and to generate output
candidates with different copies pronounced. I propose that the preference to pronounce
only the highest copy of a movement chain, the 'unmarked' option, is implemented in the
grammar by a constraint HIGHESTCOPY.

(50) HIGHESTCOPY
Assign a violation mark if the highest copy of a movement chain is not
pronounced.

Assuming that je is a syllable undominated by a PWd, as argued in Section 3.3, we
can consider several hypothetical phrasings of je in the initial position of a preverbal

93
embedded clause, varying in the smallest prosodic constituent that dominates je's syllable
node.

(51) PPh
min
Clitic
PPh
min

PWd
|
σ ...
|
je
(52) PPh
non-min
Clitic
PPh
nonmin

PPh
min

|
σ ...
|
je

(53) IntP Clitic
IntP

PPh
min/non-min

|
σ ...
|
je

Crucially, each of these representations violates STRONGSTART(σ) in its current
formulation, as the syllable that dominates je is has a sister node that is of a greater
prosodic category, either a PWd or PPh. Because every possible phrasing of je in its
highest copy would violate this constraint, the ranking of STRONGSTART(σ) over
HIGHESTCOPY prefers the pronunciation of a lower copy.

(54)
[
ForceP
je [

XP [
FinP
je ...]]] V STRONGSTART(σ) HIGHESTCOPY
{
IntP
je (
PPh
XP) ... }
higher copy, IntP clitic
*!
{
IntP
(
PPh
je (
PPh
XP)) ... }
higher copy, PPh
non-min
proclitic
*!
{
IntP
(
PPh
je XP) ... }
higher copy, PP
min
proclitic
*!
☞ {
IntP
(
PPh
(
PPh
XP) je)) ... }
lower copy, PPh
non-min
enclitic
*

94
Further evidence that STRONGSTART(σ) is the relevant markedness constraint driving
lower copy pronunciation comes from the fact that bisyllabic complementizers like jodi
'if,' substantial enough to form independent PWds, are able to surface in utterance-initial
positions (Section 2.4.1).
However, the current constraint ranking creates the problematic prediction that lower
copy spell-out should also apply in a postverbal embedded clause that corresponds to a
non-minimal phonological phrase.

(55)
V [
ForceP
je [

XP [
FinP
je ...]]] STRONGSTART(σ) HIGHESTCOPY
(
PPh
je (
PPh
XP) ... )
higher copy, PP
non-min
proclitic
*!
(
PPh
(
PPh
je XP) ... )
higher copy, PP
min
proclitic
*!
M (
PPh
(
PPh
(
PPh
XP) je)) ... )
lower copy, affixal enclitic
*

The root of the problem is that the evaluation of STRONGSTART(σ) is indifferent to any
additional prosodic structure outside of the configurations in (51)-(53). The emergent
generalization is that a violation of STRONGSTART is repaired by lower copy spell-out
when the violating syllable is aligned with the edge of an IntP, while STRONGSTART
violations at a PPh boundary are insufficient to trigger this repair.
7

To account for this asymmetry, I propose that STRONGSTART constraints are indexed
to an additional prosodic constituent, such that the constraints are sensitive to larger
prosodic boundaries that the STRONGSTART-violating structures are aligned with.

7
This resembles implicational cutoff patterns in which markedness restrictions on syllable onsets and
codas are enforced only at prosodic boundaries above a certain level (Flack 2009; Hsu and Jesney 2016).
From a syntax-phonology interface perspective, An (2007) argues that English does not permit both the
Comp head and Spec,CP to be phonetically null if the clause is parsed as an IntP, while relaxing the
restriction for smaller phrase types.

95
(56) STRONGSTART(κ/π)
Assign a violation mark for every prosodic constituent whose leftmost daughter
constituent is of type κ and is lower in the Prosodic Hierarchy than its sister
constituent immediately to the right, where κ is at the left edge of a prosodic
constituent π.

The relevant notion of 'left edge' is defined as follows:

(57) A prosodic constituent κ is at the left edge of prosodic constituent π iff.
a. π dominates κ, and
b. no prosodic constituent that both dominates κ and is dominated by π has a
leftmost daughter constituent that does not contain κ.

Given this expanded constraint schema, we can define two separate constraints,
STRONGSTART(σ/INTP) and STRONGSTART(σ/PPh). The ranking STRONGSTART(σ/INTP)
>> HIGHESTCOPY >> STRONGSTART(σ/PPh) now accounts the fact that lower copy
pronunciation applies in preverbal embedded clauses phrased as IntPs, but not in
postverbal embedded clauses phrased as non-minimal PPhs. In the previous part of the
analysis, the preference for enclisis over proclisis within PPhs is now attributed to
STRONGSTART(σ/PPh).

(58) Postverbal embedded clause: Lower copy pronunciation does not apply.
V [
ForceP
je [

XP [
FinP
je ...]]]
STRONGSTART
(σ/IntP)
HIGHEST
COPY
STRONGSTART
(σ/PPh)
☞ (
PPh
je (
PPh
XP) ... )
free clitic
*
☞ (
PPh
(
PPh
je (
PPh
XP)) ... )
affixal proclitic
*
☞ (
PPh
(
PPh
je XP) ... )
internal proclitic
*
(
PPh
(
PPh
(
PPh
XP) je)) ... )
lower copy, affixal enclitic
*!

96
(59) Preverbal embedded clause: Lower copy pronunciation applies.
[
ForceP
je [

XP [
FinP
je ...]]] V
STRONGSTART
(σ/IntP)
HIGHEST
COPY
STRONGSTART
(σ/PPh)
{
IP
je (
PPh
XP) ... }
free clitic
*!
{
IP
(
PPh
je (
PPh
XP)) ... }
affixal proclitic
*! *
{
IP
(
PPh
je XP) ... }
internal proclitic
*! *
☞ {
IP
(
PPh
(
PPh
XP) je)) ... }
lower copy, affixal enclitic
*

Thus far, we have assumed that at least one phrase occurs in the specifier of the
embedded TopicP or FocusP. Suppose, however that the syntactic derivation creates a
preverbal embedded clause in which no constituents have moved to TopicP or FocusP.
Assuming that string-vacuous movement (i.e. movement that could have no effect on
word order) is possible, pronouncing either the higher or lower copy of je would leave the
complementizer in the clause-initial position. Because all possible prosodic
representations with both options of copy pronunciation violate STRONGSTART(σ/IntP),
selection of the optimal output is determined by lower-ranked constraints. Given the
current ranking of constraints, a free clitic structure created by pronunciation of the
higher copy is selected as the optimal output. This is an unwanted result, however, as je is
not able to surface clause-initially in this context.

97
(60) Empty TopicP and FocusP: string-vacuous higher copy pronunciation predicted
[
ForceP
je
1
[∅ [
FinP
je
2
... XP]]] V
STRONGSTART
(σ/IntP)
HIGHEST
COPY
STRONGSTART
(σ/PPh)
M{
IP
je
1
(
PPh
XP) ... }
free clitic, higher copy
pronounced
*
{
IP
(
PPh
je
1
(
PPh
XP)) ... }
affixal proclitic, higher copy
pronounced
* *!
{
IP
(
PPh
je
1
XP) ... }
internal proclitic, higher copy
pronounced
* *!
{
IP
je
2
(
PPh
XP) ... }
free clitic, lower copy
pronounced
* *!
{
IP
(
PPh
je
2
(
PPh
XP)) ... }
affixal proclitic, lower copy
pronounced
* *! *
{
IP
(
PPh
je
2
XP) ... }
internal proclitic, lower copy
pronounced
* *! *

We are thus led to the conclusion that je must be left unpronounced if a
STRONGSTART(σ/IntP) violation can not be avoided by pronouncing a lower copy, an
apparent instance of prosodically-motivated non-pronunciation (Zec and Inkelas 1990,
Selkirk 2002, McCarthy 2012). The preference for objects in the syntactic input to be
given a phonological realization can be implemented in the OT grammar in the form of a
REALIZE MORPHEME constraint (Samek-Lodovici 1993; Gnanadesikan 1996; Rose 1997;
Walker 1998; Henderson 2012; among others).

(61) REALIZE MORPHEME
A morpheme must have some phonological exponent in the output. (Walker 1998)

The realization of je in preverbal clauses requires REALIZE MORPHEME to be ranked
below STRONGSTART(σ/IntP) and above HIGHESTCOPY. This ensures that lower copy

98
pronunciation is preferred over non-realization as a means to avoid violations of
STRONGSTART(σ/IntP), as shown in (62). The obligatory non-realization of je occurs
uniquely in situations where STRONGSTART(σ/IntP) violations cannot be avoided by
lower copy pronunciation because no overt material intervenes between copies of je (63).

(62) Preverbal embedded with overt TopicP or FocusP: lower copy of je pronounced
[
ForceP
je [

XP [
FinP
je ...]]] V
SSTART
(σ/IntP)
REALIZE-
MORPH
HIGHEST
COPY
SSTART
(σ/PPh)
{
IP
je (
PPh
XP) ... }
free clitic
*!

{
IP
(
PPh
je (
PPh
XP)) ... }
affixal proclitic
*!

*
{
IP
(
PPh
je XP) ... }
internal proclitic
*!

*
☞ {
IP
(
PPh
(
PPh
XP) je)) ... }
lower copy, affixal enclitic

*
{
IP
(
PPh
XP) ... }
no copy pronounced
*!

(63) Preverbal clause with empty TopicP and FocusP: je is unpronounced
[
ForceP
je
1
[∅ [
FinP
je
2
... XP]]] V
SSTART
(σ/IntP)
REALIZE-
MORPH
HIGHEST
COPY
SSTART
(σ/PPh)
{
IP
je
1
(
PPh
XP) ... }
free clitic, higher copy
pronounced
*!

{
IP
(
PPh
je
1
(
PPh
XP)) ... }
affixal proclitic, higher copy
pronounced
*!

*!
{
IP
(
PPh
je
1
XP) ... }
internal proclitic, higher copy
pronounced
*!

*!
{
IP
je
2
(
PPh
XP) ... }
free clitic, lower copy
pronounced
*!

*!
{
IP
(
PPh
je
2
(
PPh
XP)) ... }
affixal proclitic, lower copy
pronounced
*!

*! *
{
IP
(
PPh
je
2
XP) ... }
internal proclitic, lower copy
pronounced
*!

*! *
☞{
IP
(
PPh
XP) ... }
no copy pronounced
*

99
To conclude this section, I have presented a model of the PF grammar where word order
is determined by the interaction of rankable syntactic and phonological well-formedness
conditions. While this model of syntax-phonology interaction is fundamentally quite
similar to the one proposed by López (2009) and Elfner (2010, 2012), the displacements
that can be phonologically conditioned in the present account are restricted by the
availability of copies.

3.4.3 The architecture of the PF grammar

The present proposal has several implications for the structure of the PF grammar. I have
argued that prosodic structure formation and the determination of surface word order take
place in parallel. The grammar determines the pronunciation of a movement chain by
evaluating multiple output candidates that differ in the (non)pronunciation of movement
copies and in their prosodic organization. The surface linearization of a syntactic
representation and its corresponding prosodic structure are determined simultaneously,
rather than serially. The proposal builds on previous works arguing that prosodic
structure formation plays a crucial role in the realization of morpho-syntactic structure.
For instance, parallel interaction prosodic structure formation and lexical insertion is
revealed by cases where contextual allomorphy or morpheme realization are determined
by prosodic context (Zec and Inkelas 1990; Selkirk 2002; McCarthy 2012; Henderson
2012). Together with the present study, this provides evidence against the view that
prosodic structure creation happens after linearization and lexical insertion have taken
place (Nespor and Vogel 1986, p. 302; Embick and Noyer 2010, p. 566).

100
There appear to be limits, however, to interactions between the subparts of the PF
module. In addition to linearization and morpheme realization, there is also evidence for
parallel interaction between prosodic structure (at least at the level of the foot) and
segment-level features of tone, sonority, and syllable weight/quantity (see Blumenfeld
2006 for an overview). It appears, however, that the phonological conditions that
influence word order and morpheme realization appear to be limited to constraints on
prosodic organization (Zec and Inkelas 1990, Selkirk 2002), the realization of prosodic
edges (An 2007; Kandybowicz 2009), and co-occurrence restrictions on morphologically
or phonologically identical elements (Selkirk 2002; Bošković and Nunes 2007; Richards
2010). Rather than allowing full interaction between prosodic structure formation, copy
pronunciation, lexical insertion, and segmental processes, I propose that prosodic
structure formation mediates between these processes, which do not directly interact.

(64) Interaction among components of the PF grammar

Copy Pronunciation

Quantity, tone, sonority Lexical Insertion

There are challenges, however, for the implementation of this type of indirect interaction
in an OT grammar. Because all constraints should in principle be able to interact, there is
no way to prevent the selection of outputs to be determined by direct interaction between
constraints on copy pronunciation, lexical insertion, and markedness constraints on
Prosodic Structure
Formation

101
segments (a similar 'too-many-solutions' problem is discussed by Blumenfeld 2006). A
satisfactory solution to this problem will have to await a more comprehensive
understanding of the interactions that are possible between these various processes.

3.5 Alternative accounts of je placement

In this section, I consider two recent competing approaches to deriving the Bangla
pattern, as proposed by Chacón (2014) and Bayer and Dasgupta (to appear). I argue that
the present analysis has several empirical and theoretical advantages to these alternatives.
It will be shown that neither of the alternative proposals can satisfactorily account for the
placement of je after more than one phrase in preverbal embedded clauses, or for the full
range of patterns involving topicalization and focus in embedded clauses.

3.5.1 The verbal clitic analysis

To account for the non-initial ordering of je within pre-verbal embedded clauses, Chacón
(2014) suggests that je is a verbal clitic that cliticizes to its selecting main verb by a
postsyntactic operation of Morphological Merger (Halle and Marantz 1993).
Schematically, the operation is shown in (65). The approach is similar to the analysis of
the English null complementizer given by Pesetsky (1991) and Bošković and Lasnik
(2003), also applied to Tagalog -ng by Richards (1999).

(65) [
VP
V + je [
CP
je ...]]

102
To derive non-initial complementizer order, Chacón proposes that if linear adjacency is
unavailable between the complementizer and its selecting verb, je is reordered as a last
resort with respect to some phrase within its CP.
There are several empirical issues, both prosodic and syntactic, with this approach. It
was shown in section 3.3.1 that the complementizer is precisely not a prosodic clitic
when it follows the main verb, which is unexpected if this is the preferred context for
cliticization. Furthermore, although it is preferred to maintain adjacency between the
main verb and an embedded clause, it is possible but somewhat marked to separate post-
verbal CPs from their selecting verbs by a dative PP or adverbial. In these contexts, je
nonetheless remains initial within its clause despite the lack of linear adjacency between
je and the main verb.

(66) ?John bol-lo meri-ke [(je) chatro du-to es-che]
John say-PST Mary-acc that student 2-CL come-PERF
'John said to Mary that the two students came.'

(67) ?John bol-lo rege-giye [(je) chatro du-to es-che]
John say-PST angrily that student 2-CL come-PERF
'John said angrily that the two students came.'

Crucially, this is not predicted to be possible if Morphological Merger requires linear
adjacency. Such orderings are in fact impossible for the English null complementizer and
the Tagalog clitic -ng. In these languages, embedded CPs separated from the main verb
by an adverbial modifier can only be realized with the complementizer's full form.

(68) English
Michael acknowledged reluctantly [(that/*Ø) he had made a huge mistake]

103
(69) Tagalog
Hindi niya sinabi nang masaya [(na/*-ng) kinain niya ang tambakol]
Not he said happily COMP ate he T mackerel
'He didn't say happily that he ate the mackerel' (Richards 1999)

The verbal clitic approach faces additional difficulty in accounting for the possibility
for preverbal je to be preceded by more than one phrase within the clause. If cliticization
to a non-verb takes place as a Last Resort, it is not obvious why dislocation can apply
across multiple phrases or why these phrases must be topicalized or focused if they are
object DPs. The possibility is more straightforwardly accounted for using the proposed
lower copy spell-out mechanism, since the two complementizer heads can be separated
by any number of intervening phrases.

3.5.2 Emphatic topicalization

In a forthcoming paper, Josef Bayer and Probal Dasgupta argue that the preverbal
placement of je is the reflex of a process termed emphatic topicalization (ET), a set of
movements that applies to phrases which bear emphasis and constrastivity, but not
information-structure focality. The analysis is in part motivated by the existence of a very
similar, though non-identical pattern in Bavarian (Bayer 1984; Grewendorf 2015). While
the proposal covers some main relevant data and relates the patterning of je to that of
other emphatic particles in Bangla, it has important shortcomings in accounting for the
full range of non-initial je orders; by proposing a highly limited number of projections in
the clausal periphery, their analysis does not account for the possible occurrence of
multiple topicalized phrases before je. Furthermore, not all instances of non-initial je can

104
be reduced to emphatic topicalization: only objects, but not subjects, that precede je have
this discourse restriction; je is also obligatorily non-initial in sentential subject clauses,
which have no apparent relation to emphatic topicalization.
According to Bayer and Dasgupta, emphatic topicalization in Bangla follows a
derivation that involves two crucial movements; movement of an XP to Spec,CP (70) ,
and movement of a CP to the specifier of a functional projection above VP, call it PrtP
(71). It is assumed that the base position for embedded clauses in both languages is
postverbal; the two movement operations place the embedded clause in a preverbal
position, and ensure that the complementizer is preceded by one phrase.
(70) [
PrtP
[
Prt
[
VP
[
V
[
CP
XP [
C
je [
VP
... XP ... ]]]

(71) [
PrtP
[
CP
XP [
C
je [
VP
... XP ... ]]] [
Prt
[
VP
[
V
[
CP
... ]

While this analysis creates essentially the same constituent structure as what I have
proposed, it differs with respect to the nature of the positions targeted by movement, and
to the availability of certain projections. The authors posit the existence of only one
complementizer head, C˚, and do not assume the uniform availability of higher TopicP
and FocusP projections in embedded clauses.
Contra my own analysis, Bayer and Dasgupta argue that preverbal je functions
simultaneously as both a complementizer and an emphatic particle. The authors note the
existence of a separate morpheme je, which functions as an exclamatory emphatic

105
particle in main clauses, primarily used to indicate frustration. In my own analysis, any
relation between the complementizer je and this main clause particle is incidental.
(72) ami toma-ke kOto -bar je bol-lam
I you-OBJ how.many-times je tell-PST
‘I told you this so many times!’

Bayer and Dasgupta propose that the different patterns of postverbal and preverbal je
are the result of a difference in their featural composition. Postverbal je is purely a
complementizer with no featural specification aside from its category label (73), whereas
preverbal je is additionally endowed with an emphatic topicalization feature [iEmp] (74),
presumably the feature shared with the main clause particle.

(73) je [C]
(74) je [C, iEmp]

In the case that the complementizer je is merged during the derivation, the embedded
clause is terminated, assuring initial complementizer order. Selection of je [C, iEmp]
drives the movement of some XP, bearing an uninterpretable [uEmp] feature, to its
specifier.
At this point, it needs to be explained why je[C] can only be postverbal and je[C,
iEmp] must be preverbal. Let us first consider why emphatic je's CP must be raised to a
preverbal position. The authors propose that the [Emp] feature cannot be interpretable in
an embedded clause, and must be raised into a left peripheral projection of the main
clause. The authors note a similarity with wh-scope facts; embedded wh-words in Bangla
can only take main clause scope if the embedded clause raises to a preverbal position.
They thus propose that emphasis, like wide wh-scope, is interpretable exclusively if the

106
embedded clause is raised to a left-peripheral main clause projection that encodes
illocutionary force.
The analysis runs into some difficulty in explaining why je[C] must be postverbal,
and cannot be fronted. The authors propose that the cause is prosodic, namely that je is an
enclitic that requires a host. They argue that this requirement is satisfied when the CP
remains postverbal, as je cliticizes in this context to the main verb. However, fronting of
the CP creates larger prosodic boundaries around the embedded clause, as I have also
claimed; this in turn would prevent clause-initial je from cliticizing onto a host. In other
words, if je is at the upper edge of its clause, it needs to remain in the base postverbal
position in order for it to have access to a clitic host. The issue with this analysis is that it
is not motivated by actual prosodic data. As I have shown in Section 3.3.1, je is precisely
not a prosodic enclitic when it is postverbal, as it immediately follows an intermediate
phrase boundary. If this is the correct prosodic analysis, then a requirement for clitic-
hood cannot be what prevents an initial-je CP from fronting. This misconception,
however, does not pose a major issue for the authors' account. As an alternative
explanation, it is sufficient to say that there is no motivation for the pure complementizer
je[C] to raise because it bears no information structure features that need to be checked
higher in the main clause.
The more serious problem for Bayer and Dasgupta's analysis is its inability to account
for sentences like (75), where more than one phrase precedes je in a preverbal embedded
clause. Furthermore, it is unclear why these additional phrases must be given information
topics, and not emphatic ones.

107
(75) John [chatro du-to-ke DADUBHAI je dekh-eche] bol-lo
John student 2-CL-ACC GRANDFATHER that see-PERF say-PST
'John said that GRANDFATHER saw the two students'

In the emphatic topicalization account, once movement within the embedded clause to
Spec,CP has taken place to check je's [iEmp] feature, there is nothing to allow
topicalization above this position. Bayer and Dasgupta leave this question unresolved,
simply stipulating that additional TopicPs can be formed above an emphatic topic, but not
a plain complementizer, due to "principles of information packaging." The issue is less
problematic for the analysis of Bavarian, which disallows the presence of more than one
constituent before non-initial complementizers. However, this difference between Bangla
and Bavarian suggests that there is a language-specific factor at play, and that general
constraints on information structure do not provide an adequate explanation of the Bangla
data.
As compared with my own account, the Bayer and Dasgupta proposal adheres more
closely to the notion of "what you see is what you get," since je always occupies a single
position in C˚. This follows in part from different assumptions regarding the clausal
periphery. For one, the authors deny (footnote 45) the existence of ForceP in embedded
clauses, arguing that illocutionary force is to be distinguished from clause typing and that
illocutionary force exists only in main clauses. While this distinction may well be a valid
one, it does not necessarily entail the non-existence of a clause-typing complementizer
position in embedded clauses. TopicPs are licensed above the CP projection in restricted,
but unexplained circumstances; they can be present when je is both a complementizer and
emphatic particle, but not when je is only a complementizer.

108
In the account that I have developed, the possibility for je to be preceded by multiple
topics is accounted for by positing the existence of an articulated clause structure in all
embedded clauses. All embedded clauses can contain Focus and Topic phrases; however,
their placement relative to the complementizer is dependent on which copy of je is
pronounced. While topic and focus fronting can take place in postverbal clauses, the
obligatory pronunciation of je in Force ensures the complementizer-initial order. In
preverbal embedded clauses, the placement of topics and foci before je follows from its
pronunciation in Fin˚, below the high TopicP(s) and FocusP.
Another important aspect of the data not approached by Bayer and Dasgupta's
analysis is the subject/object asymmetry with respect to the interpretation of phrases that
precede preverbal je. Recall that while objects that precede je must be either topicalized
or focused, subjects are apparently unrestricted in their interpretation. Non-focused
quantified DPs can precede je in preverbal embedded clauses only if they are subjects.
(76) * Jon [kau-ke je dadubhai dekh-e-ni] bol-lo
John anyone-ACC that grandfather see-PERF-NEG say-PST
Intended reading: 'John said that grandfather didn't see anyone.'
(77) Jon [kew je ase-ni] bol-lo
John [anyone that come-NEG] say-PST
'John said that no one came'

The fact that subjects are not subject to the same interpretational restrictions minimally
suggests that emphatic topicalization does not explain all instances of non-initial je. The
subject/object asymmetry with respect to discourse interpretation in initial position is an
important generalization, as it clearly links the Bangla pattern to verb-second phenomena.

109
There is preliminary evidence as well that this asymmetry is operative in Slavic second-
position cliticization (Diesing and Zec 2011).
There are other reasons to believe that non-initial je orders can not be fully explained
as the result of emphatic topicalization. For instance, consider sentential subject clauses,
which share the same requirement that je be non-initial.
(78) [Ram je boka] eta birokti-kor
Ram that stupid that(DEM) annoyance-ADJ
'That Ram is stupid is annoying'

It should be noted that this construction in Bangla differs from sentential subjects in
languages like English. In Bangla, the sentence-initial clause is obligatorily followed by a
resumptive pronoun eta, which itself appears to occupy a subject position. The crucial
point is that there is no clear reason why a discourse-driven operation like emphatic
topicalization must take place within Bangla sentential subjects, as Bayer and Dasgupta's
analysis would suggest. However, the requirement that je be non-initial would still be the
expected result of a prohibition against je at the left edge of an intonational phrase.
Looking beyond Bangla, it is also difficult to extend the emphatic topicalization
analysis to non-initial complementizer orders in embedded clauses that apparently remain
in their base postverbal positions, as attested in Bulgarian (Rudin 1982; Krapova and
Karastaneva 2002) and Occitan (Sauzet 1989; Oliviéri and Faure 2013). If the discourse-
related features that drive fronting to the left of complementizers are interpretable only in
the main clause, cases like (79) and (80) are unexpected.

110
(79) Bulgarian
Mislja [tvojata sestra če vidjah]
I.think your sister that I.saw
'As for your sister, I think that I saw her.' (Rudin 1982)

(80) Occitan
Sabi [ton paire que vindrà]
I.know your father that will.come
'I know that your father will come.' (Sauzet 1989)

On the other hand, if we assume the existence of an extended clausal periphery in all
embedded clauses, both language-internal and cross-linguistic variation in the placement
of complementizers could receive a promising explanation in terms of the featural
makeup of left-peripheral functional heads and their interaction with various prosodic
constraints.
To conclude, I have argued that Bayer and Dasgupta's proposal does not adequately
account for several important generalizations about the non-initial je pattern, specifically
the subject-object asymmetry in the interpretation of fronted phrases and the possibility
of multiple topicalizations above preverbal je. While they could be correct that a
discourse factor like emphatic topicalization explains portions of the derivations
involved, the proposed mechanism is unable to account for the same range of attested
facts as the proposed lower copy spell-out analysis.

3.6 Conclusion

In this and the previous chapters, I have provided an account for the distribution of the
Bangla complementizer je, arguing that non-initial orderings are derived by lower copy
pronunciation, which applies to prevent the realization of the complementizer at an

111
intonational phrase edge. Furthermore, I have argued that the analysis in terms of lower
copy pronunciation provides the strongest explanatory account versus prior approaches.
The data and analysis contribute to a growing body of evidence that the well-formedness
of certain word orders depends crucially on restrictions on prosodic organization,
indicating that linearization and prosodic structure formation are determined
simultaneously within the PF grammar.

112
Chapter 4: Verb second and its deviations: Bundling in the left periphery

4.1 Introduction

In the analysis of Bangla complementizer order in the previous chapters, I discussed
several empirical advantages of adopting the extended left peripheral structure developed
within the Cartographic Program. This chapter takes a broader cross-linguistic view, and
argues that positing a universal inventory of left-peripheral projections has undesirable
consequences in other cases. I focus on the problem of accounting for verb-second
effects, in which the instantiation of left-peripheral positions appears to be substantially
restricted. I argue that variation in the number of functional category features that are
realized on individual heads, i.e. Feature Scattering (Giorgi & Pianesi 1996), is vital to
accounting for attested cross-linguistic differences. This sets the stage for Chapter 5,
which develops the proposal that this bundling process is the result of an operation called
Coalescence that takes place in the syntax.

4.1.1 Cartography and the problem of restricted instantiation

As discussed in Chapter 1, certain patterns of complementary distribution between V2
and overt complementizers suggest that V2 effects involve movement of the verb to a
position within the complementizer domain (den Besten 1983; Travis 1984; Fanselow
2004; Roberts 2004; Haegeman and Cranenbroeck 2007; Roberts 2012; among many
others). Furthermore, V2 generally interacts with the packaging of information structure,

113
as certain items in first position are obligatorily topic- or focus-marked (Fanselow 2004;
Mohr 2009; Jouitteau 2010; Holmberg 2015). To illustrate, consider the interpretational
restrictions on first-position objects in German, as described by Mohr (2009). Objects can
be licensed in first position only if they are interpreted as given information topics (1a),
or contrastively focused (1b); Objects that are neither topicalized nor focused are not
accepted (1c). Notice that the restrictions are identical to those that apply to objects that
precede je in preverbal embedded clauses in Bangla, as discussed in Chapters 2 and 3.
For presentational clarity, the verbs relevant to the second-position requirement are
shown in bold text in the remainder of the chapter.

(1) a. Diesen minister hat die Presse schon lange kritisiert
This-ACC minister has the press already long criticized
'This minister has long been criticized by the press.'

b. Einen MINISTER hat die Presse schon lange kritisiert,
A minister has the press already long criticized
(aber nicht den Kanzler)
(but not the chancellor)
'The press has already criticized a minister for a long time, not the chancellor.'

c. *Einen Minister hat die Presse schon lange kritisiert
A minister has the press already long criticized
(as broad focus) (Mohr 2009)

It appears natural to pursue an analysis in which the features driving movement to
first positions in such examples are the same ones associated with TopicP and FocusP in
theories that assume an extended left periphery. However, this raises the question of why
in most V2 languages, these positions cannot be simultaneously filled. If it is assumed
that the full inventory of left-peripheral projections is universally present in all languages,
strict V2 must amount to a requirement that phrasal movement target exactly one position

114
above the landing site of the verb. However, this comes with the challenging task of
explaining [1] why is only one position available before the second-position element, and
[2] which functional projection(s) do the first-position and second-position items occupy?
I first review the previously proposed Bottleneck Effect and Stacked Head
approaches to V2, and show that they are generally unable to account for a variety of
"relaxed" V2 systems that allow V3 or V4 in some contexts. I argue for an account of
variation in terms of the Feature Scattering Hypothesis (Giorgi and Pianesi 1996), whose
core claim is that functional categories can be realized as distinct heads, or bundled with
other categories. Finally, I propose that variation in bundling interacts with several other
parameter settings to determine the strictness of V2 effects.

4.1.2 V2 as a movement restriction: The Bottleneck Effect

One way of deriving second-position restrictions within an expanded left periphery
proposes that while left-peripheral positions are universally present, fronting in some
languages is restricted to exactly one left-peripheral position due to a "Bottleneck Effect"
(Haegeman 1996; Roberts 2004; Cardinaletti 2010). The proposal claims that in
languages with strict second-position restrictions, all left-peripheral fronting passes
through Spec, FinP, perhaps due to a general EPP property of the Fin˚ head (Roberts
2004). Once one phrase has moved through FinP, all further movement through this
position is blocked, thereby restricting fronting to one constituent. This blocking has been
proposed to be an effect of Relativized Minimality (Rizzi 1990); whatever phrase is
attracted to Spec, FinP contains features that are sufficiently general to be the closest goal

115
for any attractor in a higher projection, blocking all movement across FinP (Roberts
2004).

(2) a. [
FinP
XP
i
V-Fin [... t
i
... XP
j
...]

b. *[
FP
XP
j
[
FinP
XP
i
V-Fin [... t
i
... t
j
...]

The Bottleneck Effect approach posits that V2 requirements are a restriction on possible
movement. As such, it does not need to propose variation in the available inventory of
functional projections, keeping intact a key tenet of the Cartographic Approach.
Exceptions to V2 in which more than one phrase precedes the verb can nonetheless be
admitted within the Bottleneck Effect Hypothesis, if certain XPs can be first-Merged
(a.k.a. base-generated) in the left-periphery and not placed there by movement. Consider
for example the construction common to non-English Germanic languages known as
Contrastive Left-Dislocation (Thráinsson 1979), in which an XP appears in an utterance-
initial position with contrastive topicalization or focus (3). A related construction,
Hanging Topic Left-Dislocation, permits DPs with nominative case in first-position to
correspond with a resumptive pronoun whose case is determined by the structure of the
V2 clause (4). Crucially, these XPs appear to be invisible with respect to the second-
position requirement. If left-dislocated constituents are base-generated in the left
periphery, as argued by several authors (Anagnostopoulou 1997; Zaenen 1997; Wiltschko

116
1997; Frey 2004), this exception to V2 is unproblematic because the Bottleneck Effect is
a restriction on movement.
1

(3) Den Peter, den habe ich gesehen
3.ACC Peter, him have I seen
'I saw Peter.' (Ott 2014)

(4) Dieser Frosch, den hat die Prinzessin gestern geküßt.
this.NOM frog 3.ACC has the princess yesterday kissed
'This frog, the princess kissed it yesterday.' (Boeckx and Grohmann 2004)

In a similar vein, Poletto (2002) suggests that the left periphery can be further divided
into two domains. Specifier positions of FocusP and below can only be filled by
movement, and are thus subject to the bottleneck restriction. Positions above FocusP can
be filled by first-Merged phrases, making them "invisible" to the verb-second restriction
that holds below FocusP.

(5) [
ForceP
(XP) ... [
TopicP
(XP) ... *[
FocP
XP
k
[
FinP
XP
i
V-Fin [... t
i
... t
k
...]]

Base-generation possible Bottleneck Effect restriction holds

4.2 The problem of relaxed V2

Although certain relaxed V2 patterns can be reconciled with the Bottleneck Effect
Hypothesis, in this section I discuss two common types of V3 that cannot be adequately
accounted for by the approach. These are patterns in which subjects appear in a preverbal
position in addition to another phrase [XP Subject V ...], and patterns in which topics co-

1
It remains controversial, however, whether Contrastive Left-Dislocation and related constructions, Clitic
Left-Dislocation, and Hanging Topic Left-Dislocation are derived by base-generation in the left periphery,
movement, or ellipsis (Alexiadou 2005; Ott 2014).

117
occur with either a focus or wh-phrase that precedes the verb [Topic Focus/Wh V ...]. I
argue that these cases can not be attributed to base-generation in the left periphery,
indicating that Bottleneck Effect analyses of V2 as a movement restriction are on the
wrong track.

4.2.1 [XP Subject V] V3

Canonical cases of V2 exhibit a property known as subject inversion: While subjects can
precede the main verb, if the first position phrase is a non-subject, the subject obligatorily
follows the verb, as illustrated below for Yiddish (Diesing 1990). Subject inversion in the
presence of a non-subject XP in first position is generally taken to be a clear diagnostic
for V2 requirements.

(6) a. Max shikt avek dos bukh
Max sends away the book
'Max sends away the book.'

b. Dos bukh hot Max geleyent
The book has Max read
'Max has read the book.'

c. Vuhin geyt ir
where go you
'Where are you going?' (Diesing 1990)

Subject inversion, however, is not obligatory in all languages that generally show V2
order. We first consider the case of Old English. Like other Germanic languages, Old
English had a general V2 requirement within main clauses (van Kemenade 1987; Pintzuk

118
1993; Kroch et al. 1995; Trips 2002; a.o.). As shown in the following examples, the first
position could be occupied by a variety of phrasal constituents.

(7) a. His mynster ys Hwiterne on Sanctus Martines naman gehalgod
his minster is Hwitern on Saint Martin's name consecrated
'His mynster, Whitern, is consecrated in Saint Martin's name.'

b. Þæt hus hæfdon Romane to ðæm anum tacne geworht ...
that house had Romans to the one sign made
'The Romans had made that house to their sole sign...'

c. On his dagum sende Gregorius us fulluht
On this day sends Gregorius us Christianity
'On this day, Gregorius sends us Christianity.' (Trips 2002)

Of particular interest is the fact that Old English allowed certain exceptions to V2 not
attested in other Germanic languages. As noted in particular by Haeberli (2002a,b), a type
of V3 order is permitted in which a full DP subject immediately precedes the main verb.
In the corpus examined by Haeberli (2002b), [XP Subject V ...] orders were found to
occur in 28.7% of main clauses in which the first XP is neither the subject nor an operator
that triggers strict V2 (i.e. a wh-phrase).

(8) a. ...& fela ðinga swagerad man sceal don
and many things so.wise man must do
'...and such a wise man must do many things.'

b. Sumum monnum God seleð ægðer ge good ge yfel gemenged, ...
Some persons God gives both good and bad mixed
'God gives some people both good and bad things.' (Haeberli 2002b)

Based on their distribution, Bech (1998), Westergaard (2005), and Hinterhölzl and
Petrova (2010) propose that [XP Subject V] orders are possible (though not apparently
obligatory) only for subjects that are given information topics. Such V3 orders, however,

119
are not attested for non-subjects in the immediately preverbal position, indicating that
grammatical subjecthood is crucial to this position.
2
Similar exceptions to V2 have also
been noted for
historic varieties of other West Germanic languages, including Old Saxon
and Middle Low German (Petrova 2012; Walkden 2015).

Similar patternings of subjects in V3 structures in contemporary Germanic languages
are documented for Cimbrian (Bidese 2008; Grewendorf and Poletto 2011; Bidese,
Cognola, and Padovan 2012; Bidese, Tomaselli, and Padovan 2016) and French Flemish
(Ryckeboer 2004; Haegeman and Greco 2016). In Cimbrian, a heritage Germanic
language spoken in the Italian town of Luserna, subjects appear obligatorily in a
preverbal position; The subject-inversion pattern of 'classic' V2 is in fact reported as
ungrammatical in declarative clauses.

(9) a. Haüt dar nano iz gerift atz Lusérn
today the grandfather is around in Lusérn
'Today the grandfather is in Lusérn'

b. Gestern dar puce hott gisekk in has
Yesterday the boy has seen a hare

c. *Gestern hott dar puce gisekk in has
Yesterday has the boy seen a hare
'Yesterday the boy saw a hare' (Bidese et al. 2016)

2
Subject-second V3 order in Old English is generally discussed alongside the fact that subject and object
pronouns also appear in a preverbal position (Pintzuk 1991; Cardinaletti and Roberts 1991; Kroch and
Taylor 1997; Haeberli 2002a; Trips 2002).

(a) Þa hie gefengon micle herehyð, ...
then they took much plunder
'Then they took much plunder ...' (ChronA,:894.20.1027)

(b) Þe biscopes & lered men heom cursed œure
the bishops and learned men them cursed always
'The bishops and learned men always cursed them. (ChronE,ChronE_[Plummer]:1137.54.4063)

However, this pattern is not as clearly problematic for the Bottleneck Effect Hypothesis, as these pronouns
are potentially clitics to the finite verb and thus not relevant to the bottleneck restriction (Cardinaletti and
Roberts 1991).

120

As in Old English, the placement of subjects is sensitive to information structure. In
contrast with definite DP subjects, subject quantifiers like niamat 'nobody' can follow the
second-position verb.

(10) 'Z hat niamat telefonaart
It has nobody telephoned
'Nobody has telephoned.' (Grewendorf and Poletto 2011)

[XP Subject V] orders are also common in the variety of Flemish spoken in northern
France (Vanacker 1977; Marteel 1992; Ryckeboer 2004; Saelens 2014; Haegeman and
Greco 2016). In contrast to standard Dutch, which shows typical subject inversion in V2,
both full DP subjects and subject pronouns can appear in the preverbal position,
following a non-subject phrase. While adverbials are most commonly accepted in first
position, direct and indirect objects are acceptable as well, at least for some speakers.

(11) a. Morren 't komt ten langen leste schooen were
tomorrow it comes at long last nice weather
'Tomorrow it will finally be good weather.'

b. Alle vuuf voet den triek is veg
each five foot the electricity is gone
'The electricity is down every so often' (Ryckeboer 2004)

(12) De nieuwe wagens we makten he
the new cars we made PART
'We made the new cars.' (Haegeman and Greco 2016)

More significantly, the absence of subject inversion appears to be the least marked word
order in declaratives; In clauses that begin with a non-subject XP, the subject precedes

121
the verb in 85.4% and 89.4% of tokens in the corpora of Saelens (2014) and Vanacker
(1977), respectively.
This pattern of exceptional V3 reveals a tendency for subjects, particularly those that
are also given information topics, to move to a position above the main verb. While the
pattern has been taken to suggest that verb movement targets a position below the
complementizer domain like AgrSubjP (Haeberli 2002a), if we take V2 to uniformly
require verb movement to a complementizer-domain head (Vikner and Schwartz 1996;
Roberts 2004; Haegeman and Craenenbroeck 2007; Poletto 2013; Cognola 2013, a.o.),
this suggests that subjects occupy Spec, FinP or a specialized subject position in its
vicinity, as independently argued by Poletto (2000), Aboh (2006), and Ledgeway (2010).
Under this view, [XP Subject V ...] V3 reflects the structure in (13).

(13) [
FP
(XP
top/foc
) [
SubjP
(XP
subj
) [
FinP
V˚ [
InflP
... V˚

Given the above structure, the subject exception to V2 is difficult to handle in a
Bottleneck Effect approach to V2. If subject DPs move to FinP, one expects the
movement of any other phrase above the subject to be blocked. One might admit the
possibility that the first phrase in these V3 examples is a topic that is base-generated in
the left periphery. However, it would remain unexplained as to why this sort of base-
generation is possible only when the main verb is preceded by a subject.

122
4.2.2 [Topic Focus/Wh V] V3

In addition, there are several V2 languages that permit V3 orders where the main verb is
preceded by both topicalized and focused phrases. The Nakh-Dagestanian language
Ingush, as described by Nichols (2011), exhibits a V2 pattern quite similar to the German
pattern. In main clauses, the verb or highest tensed auxiliary generally follows exactly
one phrase.

(14) Xii mol=ii wa?
water drink=Q 2S.ERG
'Would you like a drink of water?

(15) Muusaa vy hwuona telefon jettazh
Musa V.PROG 2S.DAT telephone strike.CVsim
'Musa's calling you / Musa's on the phone for you'

(16) Cwa mealxara jar hwo, Tawaibat, ...
some sad J.be.PST 2S Tawaibat
'You're so sad, Taiwaibat, ...'

In Ingush, phrases in the preverbal position can be interpreted either as topicalized or
focused. Crucially, if a main clause includes both a topic and a focus, both phrases
precede the main verb in the order [Topic Focus V]. Both information focus and wh-
words are analyzed as types of foci by Nichols.

(17) a. Jurta jistie joaqqa sag ull cymogazh jolazh
Town.GEN nearby J.old person lie.PRS sick.CVsim J.PROG.CVsim
(topic) (focus)
'In the next town an old woman is sick'

b. Mista xudar myshta duora?
sour porridge how D.make.IMPF
(topic) (focus)
'How did they make sour porridge?' (Nichols 2011)

123
While such data is troubling for the Bottleneck Effect Hypothesis in its original form, it
conforms to Poletto's proposal that projections above TopicP permit base-generated
phrases. However, it raises the issue of explaining why the co-occurrence of topics and
foci is prevented in V2 languages like Modern German.
More problematic cases are found in languages where this type of V3 is possible only
when the immediately preverbal phrase is a wh-phrase. Consider for instance the Badiotto
variety of Rhaetoromance as discussed by Poletto (2002). In declarative sentences, the
second-position verb is preceded by a focused phrase, and it is not possible for left-
dislocated elements to precede the focused phrase. However, if the verb is preceded by a
wh-phrase it is possible for left-dislocation to target the clause-initial position.

(18) a. *De Giani, CUN PIERO a-i bel baié
of Giani WITH PIERO have-I already spoken

b. De Giani, con che bai-la pa?
of Giani with whom speak-she INTERR.PRT
‘With whom did you talk about John?’ (Poletto 2002)

Similarly, while Yiddish is generally V2, it permits the optional topicalization of exactly
one phrase before wh-words (Diesing 2005).

(19) mit di kinder vos tut men?
with the children what does one
'What does one do with the chidren?' (Diesing 2005)

Minimally, the fact that certain types of pre-verbal phrases are permitted only when they
co-occur with wh-words indicates that it would be overly simplistic to claim that items in
the topic field are uniformly exempt from the bottleneck restriction. This would require

124
the apparent stipulation that the base-generation of topics is in some instances possible
only if wh-movement has already taken place.
Perhaps the most curious case of this type is found in Kashmiri, which we will
consider in greater detail. Kashmiri is relatively unique among Indo-Aryan languages in
that it has a robust V2 requirement in main clauses (Bhatt 1999).

(20) a. rameshan dyut raath laRk-as kalam
Ramesh gave yesterday boy pen
'It was Ramesh who gave a pen to the boy yesterday.'

b. laRk-as dyut rameshan raath kalam
boy gave Ramesh yesterday pen
'It was a boy to whom Ramesh gave a pen yesterday.'

c. *tem raath dyut akh laRk-as kalam
he yesterday gave one boy pen (Bhatt 1999)

The language additionally places an interpretational restriction on non-subjects fronted to
first position, requiring them to be focused, not topicalized. Quantified objects that are
ineligible topics are freely fronted to first position; Furthermore, only phrases in first
position can be suffixed by the focus-sensitive particles -ti and -yioot, which resemble
even and only respectively.

(21) a. sooruyikeNh khyav rameshan
everything ate Ramesh
'Ramesh ate everything'

b. huun-ti chu behna broNh panin jaay goD saaf karaan
dog-even is seat before self's place first clean do-perf
'Even the dog cleans his place before sitting.' (Bhatt 1999)

One type of V3 order is admitted under a fairly restricted circumstance. As typical of V2
languages, wh-phrases move to an immediately preverbal position in interrogative main

125
clauses. However, if the clause includes a topicalized phrase, it is preferably placed in a
clause-initial position preceding the wh-phrase, yielding [XP
top
Wh V ...] orders. This
pattern is particularly unexpected due to the fact that topics do not front to first-position
in declarative V2 main clauses (Bhatt 1999, Manetta 2011).
3

(22) a. tse kyaa dyutnay Rameshan
you what gave Ramesh
'As for you, what is it that Ramesh gave?

b. ?kyaa dyutnay Rameshan tse
what gave Ramesh you (Bhatt 1999)

If the wh-word is placed into a left-peripheral FocusP or InterrogativeP (Rizzi 2001)
by movement from the lower part of the clause, as commonly assumed, the clause-initial
topic must be placed above the wh-phrase by base-generation. Although the proposal by
Poletto (2002) permits base-generation above FocusP, it would have to be stipulated that
base-generated topics are permitted only when followed by a wh-phrase, a rather unusual
restriction. Furthermore, Holmberg (2015) notes that a base-generation analysis of initial
topics is unlikely, as the fact that they are case-marked suggests that they are first Merged
lower in the clause, rather than directly in the left periphery.

3
My analysis of Kashmiri interrogatives is based on the convergent descriptions of Bhatt and Manetta.
However, there appears to be additional variation across speakers with respect to word order in
interrogatives. Koul and Wali (2015) observe that wh-phrases can stay in situ for some speakers, and
speakers vary in whether they permit multiple wh-movement to the preverbal position (p.c. Constantin
Freitag)

126
4.2.3 V>3 and multiple bottlenecks

In a recent corpus study, Wolfe (to appear) identifies substantial word order variation
within main clauses of medieval Romance languages, many of which have previously
been described as exhibiting relaxed V2 restrictions (Benincà 1983, 2006; Adams 1987;
Roberts 1993; Vance 1997; Franco 2009, a.o.). Several generalizations are revealed in
this reproduced summary table. First, V2 is the most common word order in all of the
languages considered. There is variation however, in the types and statistical frequency of
permitted non-V2 orders. For instance, while it is highly rare for more than two items to
precede the verb (V>3) in Old French, Old Spanish, and Old Venetian, such orders, at
least for V4, are more common in Old Occitan and Old Sicilian. There is also a range of
variation with respect to the acceptability of verb-initial (V1) orders, from being
unattested in Old French to consisting of 24.37% of the Old Venetian corpus.

(23) Verb placement in Medieval Romance Main Clauses (Source: Wolfe, to appear)
Old French Old Occitan Old Sicilian Old Venetian Old Spanish
N % N % N % N % N %
V1 0 0.00 48 7.52 52 8.21 154 24.37 11 2.29
V2 475 75.16 340 53.29 318 50.24 371 58.70 436 90.83
V3 155 24.53 188 29.47 189 29.86 103 16.30 32 6.67
V4 2 0.32 50 7.84 61 9.64 4 0.63 1 0.21
V5 0 0.00 8 1.25 11 1.74 0 0.00 0 0.00
V6 0 0.00 4 0.63 1 0.16 0 0.00 0 0.00
V7 0 0.00 0 0.00 1 0.16 0 0.00 0 0.00
Total 632 100 638 100 633 100 632 100 480 100

What is clearly revealed by Wolfe's corpus and the previously discussed examples is the
generalization that there is a continuum of structural restrictions on verb placement, of
which strict V2 appears to be at one extreme. However, as languages permit an increasing

127
number of deviations from linear V2, there is likely a tipping point at which such
languages would not be described by researchers as having any sort of V2 requirement.
The availability of V4 structures has important implications for the Bottleneck Effect
Hypothesis. Wolfe notes that attested V4 orders consist of sequences of a frame-setting
adverbial, a topic, and a focus preceding the verb [Frame-setter Topic Focus V], as
exemplified in this following example from 12th-century Sicilian. The same analysis is
given for V4 orders in Old Italian (Florentine) by Poletto (2014).

(24) Old Sicilian
[tamen poy di la morti loru] [li ossa loru] [pir virtuti divina] operannu
then after of the death their the bones their by virtue divine perform.3PL
miraculi
miracles
'Then after their death, their bones perform miracles through divine virtue.'
(Wolfe, to appear)

(25) Old Italian
[e per volontà de le Virtudi] [tutta questa roba] [tra' poveri] dispense
and by will of the virtues all this stuff among poor-PL distribute.3SG.PST
'And according to the will of the virtues, distributed all these goods among the
poor.' (Poletto 2014)

As noted by Wolfe, such data appear to be compatible with the proposed bottleneck in
FinP, as long as both the frame-setting adverbial and the following topic are base-
generated in the left periphery. A base-generation approach to frame-setting adverbs is
plausible, given that they are adjunct-like in nature and independent of argument
structure within the clause.
However, this analysis raises the important question of why such V4 orders are not
available, or substantially more marginal, in other Old Romance languages or the stricter
modern day Germanic languages. To address this issue of variation, Poletto (2002,

128
2013), Walkden (2015), and Wolfe (2015, to appear) propose that languages can vary in
the height of the bottleneck restriction. Specifically, they propose that the locus of the
bottleneck can be either in FinP or ForceP.

(26) FinP bottleneck
[
FrameP
(XP
Frame-setter
) [
ForceP
... [
TopicP
(XP
Top
) [
FocusP
XP
Foc
[
Fin
XP
Foc
V [
InflP
... V]]]]]

(27) ForceP bottleneck
[
FrameP
(XP
Frame-setter
) [
ForceP
XP
Top/Foc
V [... V]]]

A language with a ForceP bottleneck shows stricter V2 restrictions, since fewer positions
are available for base-generated constituents. This structure is argued to account for
languages with relatively strict patterns, like Modern German and Modern Dutch.
Languages with a bottleneck in FinP permit more relaxed V2 patterns, such as those of
Old Romance, due to the availability of more positions where base-generation is possible.
While the proposal does allow for a means to account for cross-linguistic variation in
the strictness of V2 requirements, the ForceP bottleneck structure presents a number of
theoretical problems. For instance, how can we explain how phrases in Spec, ForceP can
be interpreted as topics, foci, or wh-elements? Within the Cartographic Approach, we
expect such items to be attracted only by the left-peripheral Topic˚ and Focus˚ heads.
One possible analysis is to say that these items first move through TopicP or FocusP
before landing in Spec, ForceP. This movement path is illustrated in (28), where a
focused XP moves first to Spec, FocusP, then to Spec, ForceP. However, this type of
two-step movement would violate Criterial Freezing (Rizzi 2006, 2010), the
generalization that items moved to a position that licenses some discourse property
cannot undergo further movement.

129
(28) [
ForceP
XP
Foc
V˚ [
TopicP
... [
FocusP
XP
Foc
[
Fin
V˚ [
InflP
... V˚ ... XP
foc
]]]]]

An alternative analysis that avoids this issue is to claim that in languages with a ForceP
bottleneck, Force˚ inherits certain features of the lower heads, allowing topics and foci to
move directly to Spec, ForceP. However, this approach requires additional claims about
feature inheritance within the left periphery.
In addition, there are additional empirical limitations to the types of variation that can
be accounted for by varying the height of the bottleneck. This can be illustrated by
comparing Standard Dutch to West Flemish with respect to the availability of frame-
setting adverbials (Haegeman and Greco 2016). Standard Dutch prohibits frame-setting
adverbials to precede V2 declarative clauses, whereas this option is available in West
Flemish dialects.

(29) a. *StD/√WF
Voor da-j dat weet, dat kind is weg me je geld
Before that-you that know that child is away with your money
'Before you know, the child is off with your money.'

b. *StD/√WF
Als mijn tekst klaar is, ik zal je hem opsturen
When my text ready is I will you him send
'When my text is ready, I will send it to you.' (Haegeman and Greco 2016)

To account for the dialectal difference as variation in the height of the bottleneck
restriction, one could propose that the bottleneck is in a higher position in Standard Dutch
than in West Flemish, high enough to prevent anything from being Merged in Spec,
FrameP. Potentially, then, Standard Dutch has a bottleneck in FrameP or higher.
However, this would again raise the question of why the interpretation of first-position
items in Standard Dutch is not restricted to frame-setting functions.

130
Lastly, it should be noted that this approach to the Bottleneck Effect predicts certain
entailment relations in the possible movement types to the left periphery. For instance, if
subjects move to a dedicated position in the preverbal field (as in Old English, Cimbrian,
French Flemish), this indicates that the bottleneck is in a low position like Fin˚. We
predict, then, that all positions above TopicP should be able to be simultaneously filled,
permitting for instance the V4 patterns discussed for Old Sicilian and Old Italian. In the
absence of more detailed data to test this prediction, however, this issue will have to be
left for further research.
Thus far, I have presented an overview of various relaxed V2 patterns, and shown that
they pose numerous challenges for the Bottleneck Effect Hypothesis. In particular, the
existence of highly relaxed V2 languages must be taken to indicate that numerous phrases
can be base-generated in the left periphery for some languages. This, however, requires a
revised analysis of more strict V2 systems. More significantly, the approach fails to
account for patterns of dependence, in which the availability of a high left-peripheral
projection depends on movements that have occurred in a lower projection.

4.2.4 The Stacked Head Hypothesis

Some of the aforementioned difficulties of capturing relaxed second-position patterns
within the Cartographic Program motivated the 'stacked head' theory separately proposed
by Lahne (2009) and Manetta (2011). Significantly, this approach denies the cartographic
assumption that separate left-peripheral features occur in separate heads. The main claim

131
is that rather than a series of functional projections, there is only one C˚ head that
contains a 'stack' of features.

(30) CP
3
3
C˚
[F
1
]
[F
2
]
[F
3
]
...

The features in the stack are crucially ordered such that features at the top of the stack
must be checked before those lower down. The theory further assumes the availability of
multiple specifiers (Chomsky 1995) for the unique CP projection. Consequently, the
relative ordering of phrases moved to the left periphery reflects the order in which
individual features of C˚ are checked. The approach is argued to account for the
descriptive generalizations about word order in the left periphery, while obviating the
question in cartography of which functional head determines the complementizer-domain
phase in the sense of Chomsky (2000, et seq.) an issue that remains largely unresolved
(but see Roberts 2012 for one approach to FinP as the phase).
Consider the derivation of the V3 example in Kashmiri as given by Manetta (2011)
with slightly adapted notation. What is crucial is that two sets of features contain an EPP
feature. The first set [iQ,uFoc, EPP] triggers movement of the wh-word. The second set
[uTop, EPP]
2
then triggers movement of the topic.

132
(31) CP
3
rajan 3
Raj k'm' 3
'who' C˚ ...
[iQ,uFoc, EPP]
1

[uTop, EPP]
2

[uTense]
3

he:v
'showed'

As this is the only deviation from V2 in Kashmiri, this is the only possible stacking of
features that has more than one EPP feature. The standard V2 patterns are derived by
selecting a head with only one EPP feature that is associated with [uFoc].

(32) CP
3
rajan 3
Raj C˚ ...
[uFoc, EPP]
1

[uTense]
2

he:v
'showed'

Although the question is not considered in detail by Lahne or Manetta, the stacked
head approach creates a way to account for cross-linguistic variation; The number and
types of constituents that can front to the left periphery depend simply on the distribution
of EPP features within the C˚ head. A strict second-position requirement is predicted
where the single C˚ head is restricted to bearing one EPP feature. As more feature stacks
with more than one EPP features are admitted, second position requirements become
increasingly relaxed. This allows for a straightforward way to account for the continuum
of V>2 patterns.

133
However, the theory is not without substantial complications. If we assume a direct
mapping from the above structures to word order in terms of the Linear Correspondence
Axiom (Kayne 1994), we predict all fronted left-peripheral phrases to precede a unique
complementizer head. Complementizers are not expected to precede fronted items, and
the occurrence of multiple complementizer heads in distinct portions of the periphery is
unexpected (Roberts 2004 on Welsh; Aboh 2006 on Saramaccan; Demonte and
Fernández-Soriano 2009 on Spanish). Examples of these patterns are given below.

(33) Colloquial Spanish
Dijo [que a ese tío que no podía ni ver-lo]
said [that to that guy that NEG could even see-CL
'S/he said that s/he could not stand that guy.'
(Demonte and Fernández-Soriano 2009)

(34) Saramaccan
Mi táki [táa dí bakúba dɛ́ Amato bói en]
1.SG say that DET banana TOP Amato cook 3SG
'I said that, as for that banana, Amato cooked it.' (Aboh 2006)

To account for these cases, Lahne and Manetta propose that the appearance of such heads
is the result of post-syntactic operations. Complementizer-initial word orders with left-
peripheral topics or foci require a postsyntactic reordering of the C˚ head so that it
precedes the fronted phrases. The apparent realization of multiple complementizer heads
is proposed to result from the postsyntactic insertion of morpho-phonological markers.
This is incompatible, for instance, with the theory developed in this dissertation, since
these positions do not correspond to any copies present in the syntactic structure.
Putting aside this issue of theoretical incompatibility, a more serious empirical
concern is that there is no way to straightforwardly account for attested patterns of
dependence between different types of fronting. Recall that in Kashmiri and San

134
Leonardo di Badia Rhaetoromance, topicalization or left-dislocation are available only if
they are followed by a moved wh-word. However, within stacked head theory in its
current form, there are no inherent constraints on the possible distribution of EPP features
within a single language. One can predict, for instance, the existence of a language that
permits exactly two patterns. In declaratives, the first position is focused, and strict V2 is
observed. In interrogatives however, the first-position wh-word must be followed by a
subject.

(35) CP
3
XP
foc
3
C˚ ...
[uD]
1

[uFoc, EPP]
2
[uTop]
3

[uTense]
4

(36) CP
3
XP
wh
3
XP
subj
3
C˚ ...
[uD, EPP]
1

[iQ, uFoc, EPP]
2
[uTop]
3

[uTense]
4

Thus, while a stacked head approach provides an account of second-position patterns
and deviations from them, it must appeal to mechanisms outside of the proper syntactic
derivation in order to explain patterns that involve complementizer-initial orders and the
simultaneous realization of more than one C-domain head. The approach comes with a
substantial theoretical cost; By permitting reordering and morpheme-insertion within a
postsyntactic module, the predictive power of a system of constrained syntactic

135
operations is greatly weakened. Furthermore, the proposal appears to overgenerate
possible exceptions to verb second.
To summarize the discussion so far, the existence of various relaxed V2 systems
poses significant challenges to the Bottleneck Effect and Stacked Head theories. The
Bottleneck Effect Hypothesis is overly restrictive in the permitted types of relaxed V2
patterns, while the Stacked Head approach overgenerates possible exceptions and is
difficult to reconcile with other patterns that are accounted for within the Cartographic
Program.

4.3 V2 as variation in head bundling

What is clear from the variety of relaxed V2 patterns is that analyses of V2 as the result
of a restriction on movement, as proposed in all instantiations of the Bottleneck Effect
Hypothesis, or a uniform restriction on the number of heads in the left periphery, as in
Stacked Head Theory, are empirically inadequate. If this is the case, it indicates that
alternative parameters are necessary to account for variation in the strictness of V2
requirements.
The existence of a wide range of relaxed V2 patterns indicate that V2 does not arise
from a uniform restriction on possible movement or the number of heads in the left
periphery, but from the confluence of multiple parameter settings. In this section, I argue
that the aforementioned variation in the strictness of V2 systems is the result of variation
in the number of left-peripheral projections, as predicted by the Feature Scattering

136
Hypothesis. Lastly, I consider several additional parameters that can result in relaxed V2
patterns, including the number of permitted specifiers, and the height of verb movement.

4.3.1 The Feature Scattering Hypothesis

The Feature Scattering Hypothesis (Giorgi and Pianesi 1996; Bianchi 1999) proposes that
languages can permit certain category features to head their own projections or allow
them to be bundled on single heads. In the following schematic example, the features X
and Y can either head their own phrases (37) or be grouped together in a single X/YP
(38). In the remainder of the work, I will refer to features that share a head as being
bundled.

(37) XP
3
3
X˚ YP
3
3
Y˚ ...
(38) X/YP
3
3
X/Y˚ ...

Proposed by Georgi & Pianesi to account for variation in the realization of Tense,
Aspect, and Mood, this general approach is adopted in Poletto's (2000) account of cross-
linguistic variation in the realization of subject positions in the IP domain, and given
some application to the realization of complementizers by Bianchi (1999). While not
couched in the same terms, the approach to variation that allows some functional heads to
be bundled has also been proposed for Infl/Agr˚ heads (Iatridou 1990; Speas 1991;
Bobaljik 1995; Thráinsson 1996; Bobaljik and Thráinsson 1998; Ouhalla 1991) and for

137
Voice˚ and Causative˚ heads (Pylkkänen 2002). Although it predates the Stacked Head
theory and much of the Cartographic Approach, it amounts in some respects to a middle
ground between the two programs.
Although the Feature Scattering Hypothesis loosens the requirement for each feature
to be realized on a distinct head, assumed in many Cartographic works (Cinque and Rizzi
2010), it is intended to be compatible with the claim that functional features are strictly
ordered across languages. Possible variation in how multiple features can be realized on
individual heads is substantially constrained by a universal ordering constraint (Giorgi &
Pianesi 1996) on the checking of features, which presumably mirrors the feature-
checking orders proposed in the Cartographic Program.

(39) UNIVERSAL ORDERING CONSTRAINT: The features are ordered so that given
F
1
>F
2
, the checking of F
1
does not follow the checking of F
2
. (Giorgi and Pianesi
1996)

One consequence of the constraint in (39) is that category features bundled onto single
heads must be those that would otherwise be structurally adjacent if realized as distinct
heads (similar arguments are given by Caha (2013) for syncretism in case paradigms).

4.3.2 Bundled heads and relaxed V2

We now consider the application of the Feature Scattering Hypothesis to V2 and relaxed
V2 patterns. I propose that the number of apparently realized projections in the left
periphery varies in accordance with the number of bundled features. The strictest second
position requirement emerges if all left-peripheral category features are bundled into a

138
single head that also attracts verb movement. If this head contains exactly one feature that
triggers movement to its specifier (e.g. an EPP feature), regardless of how many category
features it contains, this ensures that only one phrase precedes the second-position head.

(40) Force/Top/Foc/FinP
wp
XP
k
Force/Top/Foc/Fin'
wp
Force/Top/Foc/Fin˚ InflP
[Force] 5
[Topic] ... XP
k
...
[Focus]
[Finiteness]

Second position restrictions become increasingly relaxed as the number of features in
distinct projections increases. For example, the Old English-style V3 pattern is generated
if subjects are attracted to the specifier of FinP (Aboh 2006), instantiated separately from
the higher Force/Topic/FocusP that houses first position non-subjects. As long as the verb
remains in Fin˚, Subj˚, or a bundled Fin/Subj˚ head, this allows the verb to be preceded
by both a topicalized or focused phrase and the subject.

(41) Old English V3

Force/Top/FocP
wp
XP Force/Top/Foc'
wp
Force/Top/Foc˚ FinP
[Force, EPP] wp
[Topic] (DP) Fin'
[Focus] wp
Fin˚ InflP
[Fin
,
uD, (EPP)] 5
...

139
The case of Ingush, where the main verb can be simultaneously preceded by both a topic
and focus, arises if Topic˚ is realized on its own, while all category features below Focus
are bundled together.

(42) Ingush V3

Force/TopP

wp
XP Force/Top'
wp
Force/Top˚ Foc/FinP
[Force] wp
[Topic, uTop, EPP] XP Foc'

wp
Foc/Fin˚ InflP
[Focus, uFoc, EPP] 5
[Fin] ...

A similar structure accounts for the [Topic Wh V] pattern of V3 attested in Kashmiri,
Yiddish, and Badiotto, under the assumption that wh-phrases in these languages are
attracted by [Focus].

(43) Kashmiri, Yiddish V3

Force/TopP

wp
XP Force/Top'
wp
Force/Top˚ Foc/FinP
[Force] wp
[Topic, uTop, EPP] XP Foc'

wp
Foc/Fin˚ InflP
[Focus, uWh, EPP] 5
[Fin] ...

140
It remains to be explained, however, why in these languages V3 is permitted only in
interrogative clauses with a preverbal wh-phrase, whereas non-interrogative clauses
observe stricter V2. To account for this, I propose that bundling is sensitive to properties
of [Focus]; A [Focus] feature associated with a [uWh] probe does not need to be bundled
with features higher in the left periphery, whereas a non-interrogative [Focus] feature
must undergo bundling. A fuller account of this restriction will be given in Chapter 5.
Further splitting of the [Force] and [Topic] features into separate heads gives a
straightforward way to account for V4 patterns of the type discussed in Old Sicilian,
[Frame-setter Topic Focus V], under the assumption that frame-setting adverbs occupy
Spec, ForceP.
4

(44) Old Sicilian, Old Italian V4

ForceP

wp
XP Force'
wp
Force˚ TopP
[Force, uFrame, EPP] wp
XP Top'

wp
Topic˚ Foc/FinP
[Topic, uTop, EPP] wp
XP Foc/Fin'
wp
Foc/Fin˚ InflP
[Focus, uFoc, EPP] 5
[Fin] ...

4
There are many proposals that hanging topics and scene-setting adverbials in fact occupy projections that
are above and distinct from ForceP, sometimes known as the FrameP field (Poletto 2002; Benincà and
Poletto 2004; Giorgi 2010; Wolfe 2015, a.o.). For presentational simplicity, I will make the simplifying
assumption that these elements are in ForceP, a shorthand for a bundled Force/FrameP.

141
Although there have been proposals that the CP can be either "split," containing multiple
functional projections, or "un-split," consisting of a single C˚ head, (Rizzi 1997; Poletto
and Tomaselli 1999; Shlonsky and Rizzi 2007; Biberauer and Roberts 2015; Douglas
2015), the present system predicts that a language can instantiate any number of left-
peripheral heads between one and the maximum number of left-peripheral features,
whatever it turns out to be. Furthermore, the system does not predict many of the
entailment relations among available positions required by the Bottleneck Effect
Hypothesis, discussed in Section 4.2. Specifically, the availability of a low left-peripheral
projection does not imply that all higher left-peripheral features can trigger movement or
base-generation.

4.3.3 Variation in active features within bundled heads

I propose that a crucial consequence of adopting the Feature Scattering Hypothesis is the
prediction that certain features, even if universally present, can appear to be inactive in a
given language under certain circumstances. Consider the behavior of features that are
bundled with other features on some head, and not associated with an EPP feature, and do
not show overt agreement. Consider for example two bundled heads that contain category
features X and Y. Each head contains one pair of features that triggers movement, [uF,
EPP]. However, the EPP property is associated with X in (45) and Y in (46). In terms of
surface word order, this will give the appearance of the first language having only head
X˚ while the second has only Y˚.

142
(45) X/YP
3
3
X/Y˚ ...
[X, uF, EPP]
[Y]
(46) X/YP
3
3
X/Y˚ ...
[X]
[Y, uF, EPP]

This possibility can explain, for instance, variation in the possible interpretations of
first-position phrases in second position patterns. Consider the difference in
interpretational restrictions on first-position elements in Swedish, Kashmiri, and German.
Recall that in Kashmiri V2, objects fronted to first position are obligatorily focused, and
cannot be topicalized. In contrast, Swedish V2 requires sentence-initial objects to be
aboutness topics or contrastive topics (Holmberg 2015). Possible topics like definite
object DPs can precede the second-position verb, but ineligible topics like bare quantified
DPs can not; The restriction against non-topics in first position does not apply to subjects.
Both the Swedish and Kashmiri patterns can be contrasted with German, which permits
either focused or topicalized readings for first-position objects (1).

(47) a. Den filmen får du bara inte missa
That film must you just not miss
'You simply mustn't miss that film.'

b. *? Allt åt Johan
Everything ate Johan (Holmberg 2015)

c. Allt är stängt
Everything is closed
'Everything is closed'

Given that there is only one available position preceding the verb in German, Swedish,
and Kashmiri, we maintain that declarative clauses contain only a single head with
multiple bundled features. The relevant difference between the three languages is in

143
which category feature(s) the EPP is associated with. Assuming still that subjects are
attracted by a property of [Fin], whereas foci and topics are attracted by [Foc] and [Top]
respectively, the three patterns can be analyzed as follows: Swedish permits a single EPP
feature to be associated with either [Fin] or [Top], Kashmiri permits the EPP feature to be
associated with either [Fin] or [Foc], while German allows the EPP to be associated with
[Fin], [Top], or [Foc].

(48) Swedish: [Fin] or [Top] has an EPP feature

Top/Foc/FinP
wp
XP
k
Top/Foc/Fin'
wp
Top/Foc/Fin˚ InflP
[Topic, (EPP)] 5
[Focus] ... XP
k
...
[Finiteness, (EPP)]

(49) Kashmiri: [Fin] or [Foc] has an EPP feature

Top/Foc/FinP
wp
XP
k
Top/Foc/Fin'
wp
Top/Foc/Fin˚ InflP
[Topic] 5
[Focus, (EPP)] ... XP
k
...
[Finiteness, (EPP)]

(50) German: [Fin], [Top] or [Foc] has an EPP feature

Top/Foc/FinP
wp
XP
k
Top/Foc/Fin'
wp
Top/Foc/Fin˚ InflP
[Topic, (EPP)] 5
[Focus, (EPP)] ... XP
k
...
[Finiteness, (EPP)]

144
Much work remains to be done in understanding the factors that determine variation in
the possible association of the EPP feature with discourse features within bundled heads.
What is crucial is that the Feature Scattering approach allows for a straightforward way to
account for why multiple discourse features can in some languages be associated with an
apparently unique syntactic position.
In summary, adopting the Feature Scattering Hypothesis in the analysis of V2 allows
us to maintain the key insights of the Cartographic approach with respect to the possible
orderings of heads and the checking of features, while accounting for cross-linguistic
variation in the number of realized positions. Furthermore, it allows for an appropriate
amount of flexibility in the number of positions available for movement, relative to the
overly restrictive Bottleneck Effect Hypothesis.

4.4 Additional ingredients of V2

Recall that the idealized, "strict" V2 system arises under conditions where [1] all left-
peripheral category features are bundled on one head, [2] the bundled head attracts
exactly one specifier, and [3] verb movement is triggered by some left-peripheral feature.
So far, we have discussed deviations from V2 that appear to reflect the presence of more
unbundled heads. Here, I briefly discuss other deviations from V2 that result from
parameters [2] and [3]: variation in the number of specifiers permitted in left-peripheral
projections and variation in the target of verb movement.

145
4.4.1 Single versus multiple specifiers

Pesetsky (2000) argues against theories of uniform restrictions on the internal structure of
XPs in syntax, and proposes that individual heads can vary in the number of specifiers
that they permit or require. Heads are individually specified in whether they attract
exactly one phrase, multiple phrases, or none at all.
Under this view, even if all category features in the left periphery are bundled into a
single head, V3 or V4 orders can occur if this projection permits multiple specifiers. This
appears to be the case in San Leonardo di Badia Rhaetoromance (a.k.a. Northern Ladin),
a relaxed V2 language that permits multiple topics to front in declarative main clauses,
creating a certain type of V3 (Casalicchio and Cognola 2015). Subjects and indirect
objects can both precede the verb and are not apparently restricted in their ordering,
which suggests that the subject here does not occupy a dedicated position as it does in
Old English, Cimbrian, and French Flemish.
5

(51) a. Luca ala mama ti-à-l cumprè n liber
Luca to.the mother 3DAT-has-3NOM bought a book

b. ala mama Luca ti-à-l cumprè n liber
to.the mother Luca 3DAT-has-3NOM bought a book
"Luca bought a book for his mother" (Casalicchio & Cognola 2015)

It appears then that the language allows its [Topic] feature to be associated with multiple
EPP features [Topic, EPP*]. Crucially, it is not the case that the bundled complementizer

5
Crucially, while indirect objects and direct objects can be simultaneously fronted, topicalized subjects and
direct objects cannot co-occur in the preverbal position. Casalicchio and Cognola argue that this is due to a
ban on the co-occurrence of certain types of case features within a single projection, potentially a type of
Distinctness restriction (Richards 2010).

146
projection allows multiple specifiers generally. For instance, it is not possible for both a
topic and focus to be fronted before the verb.

(52) *La mama LA ORDÖRA l' a la cumprè
The mother the fruit 3.ACC-has-3.NOM bought

4.4.2 The targets of verb movement

There is evidence that some cases of V1 are the result of verb-movement to a high left-
peripheral position with no specifier. This possibility seems to be most clearly
exemplified by Early Old French, which while generally V2 (Foulet 1928; Dupuis 1989;
Roberts 1993; Vance 1997; Labelle 2007) permitted both V2 and V1 orders in declarative
main clauses (Labelle and Hirschbühler 2012).
6

(53) Cunuit Brendans a l'air pluius que li tens ert mult
annüus
knew Brendan from the air rainy that the weather was very
worrisome
'Brendan knew from the wet wind that the weather was worrisome.'
(1120; Brendan, 56:675)

Labelle (2007) argues that V1 orders are derived by verb-movement to a higher landing
position than in V2 clauses, rather than simply having an empty specifier. This analysis is
supported by the placement of object clitics (Labelle 2007). In V2 clauses, object clitics
appear in an immediately preverbal position (54). In V1 clauses, however, object clitics
appear immediately following the verb (55).

6
The possibility of V1 orders appears to have been lost entirely in later Old French, as seen in the corpus
data of Wolfe (to appear) discussed in section 4.2.3.

147
(54) Et sa seror li fist il esposer
And his sister him made he marry
'and he had him marry his sister' (Labelle 2007)

(55) Vait s'en Brandan vers le grant mer
Go REFL-LOC Brendan towards the big sea
'Brendan goes away twoards the sea' (Labelle 2007)

Initially, the pattern resembles a classic Tobler-Mussafia effect (Tobler 1875; Mussafia
1886) in which object clitics are postposed (i.e. through lower copy pronunciation or
Prosodic Inversion) in order to avoid appearing in a clause-initial position. However,
Labelle (2007) presents evidence that early Old French had no restriction against
Intonational Phrase-initial object clitics. In particular, object clitics can appear
immediately after parentheticals, which are expected to produce intonational phrase
breaks (56). Furthermore, they are permitted to occur following sentence-initial particles
that are themselves phonologically reduced (57). Given that both the initial particle and
object clitic appear to be phonologically clitic-like, we would not expect such examples
to be acceptabile if early Old French simply banned clitics in intonational phrase-initial
position.

(56) a. Jo, qui voldreie parler a tei, le receverái
I, who would.like to.talk to you him will.receive
'I, who would like to talk to you, will receive him.'

b. Tout ainsi, fet il, le ferons
Just so, says he, it will.do
'We will do it just so, says he'

(57) a. N'i ad castel, ki devant lui remaigne
NEG-there have castle that before him stay
'No castle could resist him.'

148
b. S'en volt ostages, e vos l'en enveiez
if-GEN want hostages, and you him-GEN send..
'If he wants hostages, and if you send some to him ...'

Labelle thus argues that object clitics occur in a fixed position in the lower
complementizer domain
7
. This position is above the landing site of the verb in V2
clauses; In V1 clauses, the verb moves to a higher complementizer head above the
position of object clitics.

(58) V1: [
FP
V˚ [... Cl
obj
[
FinP
V˚ [
InflP
... V˚ ]]]]

(59) V2: [
FP
(XP) [... Cl
obj
[
FinP
V˚ [
InflP
... V˚ ]]]]

4.5 Conclusion

This section has discussed several types of relaxed V2 patterns, and the challenges that
they pose for the Bottleneck Effect and Stacked Head approaches to V2. I have proposed
that variation in the distribution of functional features among left-peripheral heads is
essential to accounting for a variety of relaxed V2 patterns. Variation in the bundling of
functional features, in concert with variation in the targets of verb movement and the
number of specifiers permitted per projection, determines the "looseness" of V2
requirements in a given language. In the next chapter, I turn to the nature of the bundling
operation involved, and its place within the grammar. I propose that rather than a type of
variation that occurs in the lexicon, it should be understood as an operation of the narrow

7
For similar arguments that clitic-like items in the left periphery occupy a fixed position in a low
complementizer head, see Roberts (1996) on Old English and Haegeman and Craenenbroeck (2007) on
Belgian Flemish dialects.

149
syntax that is sensitive to PF well-formedness, and that this further allows a unified
account of the effects of Feature Scattering with certain types of affixation.

150
Chapter 5: Coalescence

5.1 Introduction

This section considers the status of bundling operations within the architecture of the
grammar, focusing in particular on Feature Scattering and Morphological Merger. While
these mechanisms have been situated in the presyntactic lexicon and a postsyntactic
module, respectively, I argue that both can be unified as a single operation that applies
during the syntactic derivation, Coalescence. I propose that Coalescence is properly
motivated and constrained by a phonological requirement. Specifically, it applies to
eliminate heads that would be deficient for PF interpretation due to being either affixal or
null in phonological representation. The proposal is illustrated in the analysis of V2
examples discussed in the previous chapter, as well as several cases of head movement.

5.2 The place of bundling in the grammar

5.2.1 Feature Scattering: Bundling before syntax?

The basic claim of the Feature Scattering Hypothesis is that languages can vary in the
distribution of category features across functional heads. Specifically, category features
can either head their own projections or be bundled with other features. Crucially, since
the featural composition of heads is determined before they enter the Numeration, this
type of bundling takes place in the lexicon. Therefore, in order to account for the

151
generalization that features can only be bundled if they would otherwise be realized on
adjacent heads, it is necessary to propose a metacondition on the sets of features that can
be bundled, such as Giorgi and Pianesi's (1996) UNIVERSAL ORDERING CONSTRAINT on
feature checking.
Let us contrast this with how the universal order of feature checking is accounted for
within a strictly cartographic theory, in which each category feature is realized on a
separate head. I will adopt the proposal based on several works (Julien 2002; Adger
2003; Di Sciullo and Isac 2008) that the order of feature checking reflects a type of
category selection, by which heads select the category of their complements. In this
approach, each head contains an interpretable category feature, and optionally an
uninterpretable category feature. Merge is triggered in order to check uninterpretable
category features present on a head, in the configuration in (1). Uninterpretable category
features permit heads to select the categories of their complements. Consequently, the
universal order of feature checking is encoded in the atoms of the syntactic derivation.

(1) XP
[X]
3 -> 3
X Y X Y
[X, uY] [Y] [X, uY] [Y]

On the other hand, if some categories can be bundled onto single heads in the lexicon
prior to numeration of the syntactic derivation, as one interpretation of the Feature
Scattering Hypothesis might suggest, the restriction on the order of feature checking has
to exist independently of lexical entries as a stipulated metacondition on bundling in the
lexicon, such as the Universal Ordering Constraint. Restrictions on the order of feature

152
checking are thus duplicated both in the syntax and in the lexicon, a redundant
specification in separate modules of the grammar.

5.2.2 Morphological Merger: Bundling after syntax?

Much research on the interface between syntactic structures and their morphological
exponence has proposed the necessity of an operation that converts structurally adjacent
heads into a single complex head. Consider Morphological Merger (Marantz 1988; Halle
and Marantz 1993), an operation proposed to account for mismatches between syntactic
and phonological constituency created by affixation.

(2) YP YP
3 3
Y˚ XP > X˚ ...
3 Morphological Merger 3
X˚ ... Y˚ X˚

To illustrate the motivation for such an operation, consider the case of contractions in
English (e.g. should not > shouldn't, will have > will've). While their syntactic
distribution suggests that English negation morphemes instantiate the head of NegP
(Pollock 1989), n't is clearly a phonological clitic of some kind, as it lacks a full vowel
and cannot bear stress. In other words, while shouldn't does not correspond to a
constituent in the syntactic representation, it does form one at some level of prosodic
representation.

(3) a. Syntactic constituent structure: [
TP
should [
NegP
not [ ... ] ]
b. Phonological constituent structure: (
Pwd
shouldn't)

153
Marantz (1984, 1988) proposes that the mismatch in constituency is derived by an
operation that alters the structural relations established by the narrow syntax,
Morphological Merger, whose output is interpreted by the phonological grammar.

(4) TP TP
3 3
T˚ NegP > T' ...
should 3 Morphological Merger 3
Neg˚ ... T˚ Neg˚
not should 'nt

Morphological Merger is restricted to apply to heads that are adjacent at linearization (i.e.
with no intervening specifier), replacing them with an affixal structure.
1
This locality
restriction is referred to as structural adjacency. Subsequent work in the research
tradition of Distributed Morphology (DM) regards Morphological Merger as a
postsyntactic operation that applies on the PF branch of the grammar (Embick and Noyer
2001). Although the question is not generally directly addressed, the operation appears to
be consistently driven by a need for prosodically weak elements to have a host.
Working from a somewhat different perspective, Matushansky (2006) proposes that a
similar operation, M-Merger, is responsible for adjunction in head movement. In her
proposal, traditional head movement takes place in two steps. First, a lower head is
attracted to the specifier position of the highest head in the derivation. At this point, M-
Merger applies to adjacent heads, combining them into a complex head. The complex
head, but not its subparts, can undergo further head movement. In terms of the structural

1
Here, we are concerned with Local Dislocation in the terminology of Embick and Noyer (2001), distinct
from a proposed variety of Morphological Merger that does not require linear adjacency, Lowering,
proposed to account for Affix Hopping in English.

154
conditions on its application and the structure that it produces, M-Merger is identical to
Morphological Merger.

(5) XP XP
3 3
Y˚
i
X' X˚ YP
3 3 3
X˚ YP > Y˚
i
X˚ ... Y'
3 M-Merger 3
... Y' Y˚
i
...
3
Y˚
i
...

Matushansky argues that M-Merger should be analyzed as a morphological operation,
rather than a syntactic one. The key motivation for this distinction is the fact that the
movement of subparts of a complex head formed by head adjunction (a.k.a.
excorporation) appears to be impossible (Baker 1988; Julien 2002). In other words, the
internal structure of the complex head is opaque to later syntactic operations.
Matushansky proposes that this opacity reflects the fact that M-Merged heads have
undergone Spell-Out to PF. Crucially, this requires an architecture of the grammar in
which syntactic and morphological operations are interleaved throughout the derivation.
Later in the chapter, I will argue that the inaccessibility of subparts of the complex heads
need not require an analysis of bundling as an operation that takes place after Spell-Out.
Rather, this inaccessibility is accounted for if no head-internal branching structure is
created when heads are bundled.

155
5.2.3 Bundling before and after syntax?

Thus far, we have reviewed various properties of Feature Scattering and Morphological
Merger, which posit similar bundling mechanisms in the presyntactic lexicon and
postsyntactic morphology, respectively. While it may turn out to be the case that similar
operations can take place in various components of the grammar, as generally claimed in
DM, the pursuit of a theory that minimizes the complexity of grammatical derivations
should motivate a more unified analysis. Here, I will discuss structural similarities
between Feature Scattering and Morphological Merger that suggest that both types of
bundling can be attributed to a single operation that applies within a single component of
the grammar.
First, consider the locality conditions that constrain bundling in both Feature
Scattering and Morphological Merger. Under the Feature Scattering Hypothesis, category
features can be bundled only if they would otherwise be realized on adjacent heads.
Similarly, Morphological Merger applies to adjacent heads in an asymmetric c-command
relation, with no intervening specifier. In this sense, the bundled heads permitted within
the Feature Scattering Hypothesis are restricted by the same condition of structural
adjacency as those proposed for Morphological Merger.
We can also note a similarity in the motivation for both types of bundling. In the case
of Morphological Merger, it is clear that the bundling of adjacent heads takes place if one
of them is realized as a morphological affix (a definition to be clarified in the next
section), in order to satisfy a condition like the Stray Affix Filter (Lasnik 1981). While it
may not be initially obvious to look at Feature Scattering in these terms, it nonetheless

156
appears that category features that are bundled into featurally complex heads are those
that have no independent phonological exponent. These categories either lack an audible
head, or cannot attract another constituent in overt syntax, as either head movement or
phrasal movement. In other words, bundled features are those that are phonologically
null, another type of phonological deficiency.
Given the identical structural constraints on bundling proposed for Feature Scattering
and Morphological Merger, I argue in the remainder of the chapter that their effects can
be unified as a single syntactic operation. In this proposal, all category features enter the
derivation as distinct heads, but can be concatenated into complex heads as the result of a
syntactic operation that I will call Coalescence.

5.3 Defining Coalescence

We first consider the structural definition of this bundling operation, which I will refer to
as Coalescence to distinguish it from previously proposed types of bundling. I propose
that Coalescence applies to structurally adjacent heads to create a single node that
contains all features associated with the individual heads.
2

2
The output of Coalescence bears many similarities to Matching Projections in the theory of Haider (1988).
Haider proposes that functional projections whose heads have no phonetic realization must be
superimposed onto another projection within the same extended projection; the resulting Matching
Projection carries features of both projections. My proposal differs from Haider's in when bundling is
triggered: Unlike Haider's proposal, I propose that projections with null heads are licensed if they have an
overt specifier, and that Coalescence also takes place if a lower head is affixal.

(i) X/ZP
3
Spec,X/Z X/Z'
3
X/Z ...

157

(6) XP

X/YP
3 3
X˚ YP > X/Y˚

...
[X] 3 Coalescence [X]
Y˚ ... [Y]

[Y]

Departing from prior proposals on Morphological Merger and M-Merger, I do not
assume the existence of internal branching structure within the newly formed head. Note
that this collapses a distinction made in DM between Morphological Merger, a variant
operation that alters the linear ordering between the heads (Local Dislocation), and
Fusion, which collapses adjacent heads into a single terminal (Halle and Marantz 1993).
3

Nonetheless, there are several motives for this move. One primary benefit of positing no
internal structure in the complex head is that it directly accounts for the impossibility of
excorporation from bundled heads, without having to assume that bundling triggers Spell-
Out to PF, as required for Matushansky's M-Merger proposal. In Section 5.5, I show that
the proposed representation avoids incompatibilities of traditional adjunction structures
with considerations of Bare Phrase Structure (Chomsky 1994).

5.3.1 Dominance and Recession

Although Morphological Merger and M-Merger are assumed to take place when some
head requires phonological support, no concrete mechanism or structural condition has

3
Fusion is argued to produce portmanteau forms, in which multiple morphemes are realized in a syncretic
exponent.

158
been proposed to determine when bundling is triggered.
4
Recall the suggestion made in
the previous section that bundling uniformly depends on the presence of an element
whose phonological realization is null or affixal. Here, I propose that Coalescence takes
place when a category feature is 'deficient' in some way, such that it must be bundled
with another feature. This is formalized as a distinction between dominant versus
recessive features. I propose that all category features are specified as being either
dominant or recessive, and that this distinction in a given language is determined by the
criteria in (7). Although dominance and recession will be specified only on category
features in this thesis, the possibility that other feature types can carry such distinctions is
left open to future research. A more detailed discussion of these criteria is returned to in
section 5.5.6.

(7) A category feature [F] is dominant if:
a. A head containing only [F] has a non-null exponent in its phonological
representation.
OR
b. A head containing only [F] can trigger phrasal movement to its specifier.
Otherwise, [F] is recessive.

In addition to the distinction between dominant and recessive category features, I propose
a distinction between dominant versus recessive heads. While features are lexically
specified as to whether they are dominant or recessive, the status of a head is determined
by its featural composition. A head that contains at least one dominant feature is
dominant (8), whereas a head that contains only recessive features is recessive (9). In all

4
Roberts (2005), discussing a proposed syntactic bundling operation Incorporation, similar to M-Merger,
attributes it to an [affix] feature specified on items that must undergo bundling. In the analysis that follows,
we will see that this is on the right track, with some elaboration.

159
following examples, dominance is indicated with subscript
D
, recessiveness with
subscript
R
.

(8) X/Y˚
D

[X
D
]
[Y
R
]
(9) X˚
R

[X
R
]

Following Chomsky (1993, 1995, 2000), I assume that the grammaticality of surface
forms is determined not only by principles of the syntax proper, but by well-formedness
requirements of the Conceptual-Intentional (LF) and Articulatory-Perceptual (PF)
interfaces.
5
I propose that recession amounts to a deficiency for interpretation at PF, and
that Coalescence is motivated by a need to eliminate deficient features before syntactic
structures are transferred to the PF interface. I will refer to this restriction as an inviolable
Dominance Condition.

(10) DOMINANCE CONDITION
At Spell-Out, all terminal nodes of the syntactic representation contain a dominant
feature.

This is the motivation for Coalescence, which is restricted to apply in the configuration
(11) where a dominant head c-commands a recessive one. Two important consequences
of this restriction are that recessive heads cannot trigger Coalescence with lower adjacent
heads, and pairs of dominant heads can not be bundled.

5
For instance, Chomsky (1993) proposes that overt movement takes place to check certain "strong"
features, which would otherwise cause the derivation to crash at the PF interface. There is as of yet little
consensus on whether crashes are induced by deficiencies for interpretation only at PF, only at LF, or at
both interfaces (Lasnik 1999).

160

(11) XP

X/YP
3 3
X˚ YP > X/Y˚

...
[X
D
] 3 Coalescence [X
D
]
Y˚ ... [Y
R
]

[Y
R
]

Here, I illustrate the derivation of a bundled X/Y˚ head. I will maintain the
assumption that the universal order of projections and feature-checking is implemented
through the distribution of c-selectional uninterpretable features. Crucially, I propose that
the inventory of category features and their c-selectional properties are identical
regardless of whether or not they are dominant.

(12) Numeration:
X [X
D
, uY]
Y [Y
R
, uZ]
Z [Z
D
]

In the first step, we manipulate the Z˚ head and the Y˚ head. Z˚ is dominant, as it contains
a dominant category feature. Since Y˚ contains only recessive category features, the head
is recessive. The next step is identical to the derivation without a dominant/recessive
distinction; Y˚ Merges with Z˚ to check the [uZ] feature on Y˚, and Y˚ projects its
category feature onto the newly formed node.

(13) YP
[Y
R
]
3 > 3
Y˚
R
Z˚
D
Y˚
R
Z˚
D

[Y
R
, uZ] [Z
D
] [Y
R
, uZ] [Z
D
]

161
We now introduce X˚, whose category feature is dominant. Merge applies to check its
[uY] feature and creates a new phrase headed by X.

(14) XP
D

[X
D
]
3 3
X˚
D
YP
R

X˚
D
YP
R

[X
D
, uY] [Y
R
] > [X
D
, uY] [Y
R
]
3 3
Y˚
R
Z˚
D
Y˚
R
Z
D

[Y
R
, uZ] [Z
D
] [Y
R
, uZ] [Z
D
]

Because this step creates two heads in structural adjacency where the lower head is
recessive, Coalescence creates a bundled head. Since the head formed by X and Y
contains one dominant category feature ([X
D
]), the head is dominant and no recessive
heads remain in the workspace. At this point, the question arises of how the label of the
highest node is to be determined, as the projecting head now contains multiple features. I
will adopt the claim of Di Sciullo and Isac (2008) that all features of the projecting head
project to the root node.

(15) XP
D
X/YP
D

[X
D
] [X
D
]
3 [Y
R
]
X˚
D
YP
R
> 3
[X
D
, uY] [Y
R
] Coalescence X˚
D
Z˚
D

3 [X
D
, uY] [Z
D
]

Y˚
R
Z˚
D
[Y
R
, uZ]
[Y
R
, uZ] [Z
D
]

To generate more complex heads from more than two recessive heads, it is necessary
for Coalescence to take place iteratively from the top down. At each step in (16) and (17),
the topmost dominant head in the workspace undergoes Coalescence with the

162
immediately lower recessive head. For representational simplicity, checked features and
features associated with non-head nodes will be omitted in the remainder of the chapter.

(16) XP
D
X/YP
3 3
X˚
D
YP
R
> X/Y˚
D
ZP
3 Coalescence [X
D
] 3
Y˚
R
ZP
R
[Y
R
]

Z˚
R
...
3

Z˚
R
...

(17) X/YP
D
X/Y/ZP
3 3
X/Y˚
D
ZP
R
> X/Y/Z˚
D
...
[X
D
] 3 Coalescence [X
D
]
[Y
R
] Z˚
R
...

[Y
R
]

[Z
R
]

5.3.2 Labels of bundled heads

Although the possibility that phrasal nodes can have labels determined by multiple
features is rarely considered within Minimalist analyses, it permits an explanation of
certain properties of extended projections (van Riemsdijk 1988, 1998; Grimshaw 1991,
2000; Keine 2015). For concreteness, I adopt Keine's (2015) proposal of Category
Percolation, which claims that all heads inherit the category features of all lower heads
within the same extended projection.

(18) CATEGORY PERCOLATION
Given an extended projection Φ = {Π
n
> Π
n-1
> ... Π
1
}, the categorial features of
Π
m
percolate to Π
m+1
. (Keine 2015)

163
This is schematically represented in (19), for a structure in which the features [X], [Y],
and [Z] are in the same extended projection.

(19) [X, Y, Z] "XP"
3
X˚ [Y, Z] "YP"
[X] 3
Y˚ [Z] "ZP"
[Y] 3
Z˚ ...
[Z]

Category Percolation allows for an explanation of the generalization that items that select
for a complement of a given size can also impose requirements on properties associated
with lower heads within the complement constituent. For instance, Grimshaw (1991,
2000) notes that verbs like request can select CP complements that are inflected for
subjunctive mood, a property of InflP: we requested that he leave/?left a note. Because
category-selection is strictly local, one possible interpretation of this is that tense
specifications on Infl˚ percolate to the higher C˚ head.
6
Keine (2015) also shows that
Category Percolation accounts for his observation that if a constituent is inaccessible to
an agreement operation (A or A-bar), all larger constituents within the same extended
projection are also inaccessible to the same type of agreement.
Crucially, if phrasal nodes contain all features of lower heads within the same
extended projection, Coalescence will not create phrasal nodes whose labels are distinct
from those that would exist if no bundling applied. This is shown in (20), in which the
highest phrasal node contains category features [X, Y, Z]; The highest node contains the

6
An alternative explanation, however, holds that verbs like request simply select for a CP headed by
that
subjunctive
, which in English happens to be homophonous with that
indicative
, and that the subjunctive
complementizer C-selects for a subjunctive InflP complement (Bruening 2009).

164
same features even if Coalescence applies to X˚ and Y˚, because the set of features in
"YP", [Y, Z], is a subset of those contained in "XP", [X, Y, Z].

(20) [X, Y, Z] [X, Y, Z]
3 3
X˚ [Y, Z] X/Y˚ [Z]
[X
D
] 3 > [X
D
] 3
Y˚ [Z] Coalescence [Y
R
] Z˚ ...
[Y
R
] 3 [Z
D
]
Z˚ ...
[Z
D
]

5.3.3 Coalescence and head movement

We now consider the integration of Coalescence within a theory of head movement and
head adjunction. This is straightforward if we adopt Matushansky's proposal that head
movement involves movement of a lower head into the specifier of its probe (2006). In
contexts where the probe is a recessive head and the goal is dominant, this creates the
structural environment in which Coalescence applies.

(21) ZP

ZP
3 3
Z˚
D
YP > Z/Y˚
D
...
[Z
D
] 3 Coalescence [Z
D
] 5
Y˚
R
... [Y
R
]

... Z
D
...
[Y
R
] 5
... Z
D
...

How is the label of the new root node determined after movement? Given the principle of
Category Percolation, the label of a root node remains unchanged, as long as the moved
object originates within the same extended projection as its attractor. To illustrate this

165
with a highly simplified V-T-C extended projection, the basic featural makeup of the CP
projection is shown in (22), in which each phrasal node contains the features of all lower
heads within the extended projection.

(22) [C, T, V] "CP"
3
C˚ [T, V] "TP"
[C] 3
T˚ [V] "VP"
[T] 3
V˚ ...
[V]

Note that if the T˚ head moves, and the labeling of the topmost phrasal node still respects
Category Percolation, its label will be identical to that of the CP node prior to movement,
because [T] is already contained within the CP node.

(23) [C, T, V] "CP" [C, T, V] "CP"
3 3
T
D
˚ [C, T, V] "CP" T/C
D
˚ [T, V] "TP"
[T
D
] 3 > [T
D
] 3
C
R
˚ [T, V] "TP" Coalescence [C
R
] T
D
˚ [V] "VP"
[C
R
] 3 [T
D
] 3
T
D
˚ [V] "VP" V
D
˚ ...
[T
D
] 3 [V
D
]
V
D
˚ ...
[V
D
]

As long as head-movement takes place only within extended projections, Category
Percolation ensures that neither head movement nor Coalescence can create new phrasal
node labels that would not be present in the absence of either operation.
We now turn to the question of what motivates head movement. In the system
developed thus far, this would have to be independent of the EPP property, given the

166
claim that recessive heads precisely lack this ability to attract phrasal movement. While
Matushansky (2006) proposes that head movement simply takes place as a consequence
of c-selection between a head and its complement, we nonetheless must appeal to an
additional factor to explain why languages vary in the possible head-movement paths
within extended projections. It has been proposed, however, that many instances of head-
movement are apparently phonologically motivated, necessary to provide hosts for affixal
functional heads (Baker 1988; Julien 2002; Harley 2004; Roberts 2005). This
generalization is easily translated into the present proposal: The movement of a dominant
head to the specifier of a recessive one enables the application of Coalescence, satisfying
the Dominance Condition.
There are thus two ways for recessive heads to be eliminated during the derivation.
They can be bundled via Coalescence with an Externally Merged dominant head, as
discussed in previous sections, or they can be bundled with a dominant head that arrives
via Internal Merge. The two options are illustrated below for a derivation in which a
recessive head Y˚
R
has been Merged, in a structure like (24).

(24) YP
3
Y˚
R
ZP

[Y
R,
uZ] 3
Z˚
D
...
[Z
D
]

First, the structural adjacency necessary for Coalescence can be achieved if a dominant
head is Externally Merged with YP.

167

(25) XP XP
3 3
X˚
D
YP X/Y˚
D
ZP
[X
D
] 3 > [X
D
] 3
Y˚
R
ZP

Coalescence [Y
R
] Z˚
D
...
[Y
R
] 3 [Z
D
]
Z˚
D
...
[Z
D
]

Alternatively, suppose that Y˚
R
can trigger the movement of a lower, dominant head Z˚
D

to its specifier. Again, this provides the structural adjacency necessary for [Y
R
] to be
bundled into a dominant head.

(26) YP YP
3 3
Z˚
D
YP Z/Y˚
D
ZP
[Z
D
] 3 > [Z
D
] 3
Y˚
R
ZP

Coalescence [Y
R
] Z˚
D
...
[Y
R
] 3
Z˚
D
...
[Z
D
]

As we will see in the remainder of the chapter, languages often vary in the strategy used
in the elimination of a given recessive head. While we cannot yet provide a deeper
explanation for this variation, I propose that some recessive heads are strong, simply able
to compel the head-movement of a dominant item (Julien 2002; Roberts 2005). Recessive
heads that lack this property will only be bundled after another item has been Externally
Merged.
I will leave it an open question as to whether dominant heads can also have a strong
property that triggers head-movement. Matushansky (2006) suggests that this possibility

168
may account for cases of "long head movement" where head movement appears to skip
over a potential intervener [X˚
i
... [Y˚ ... [X˚
i
...]]] (Rivero 1991; Borsley et al. 1996;
Roberts 2010 and references therein), noting that the structure could be derived by
successive cyclic movement of X˚ to Spec, YP but with no adjunction, allowing X˚ to
continue moving. On the other hand, surface syntactic structures containing a
configuration [
YP
X
D
˚ [Y
D
˚ [ ... X
D
˚ ... ] ], where Y˚ clearly attracts a head X˚ and both
heads are dominant (non-affixal) appear to be rare.
7
This may indicate a deeper
connection between strength and recessive features. This question, however, must be left
to future empirical investigation.
To illustrate this system, head movement and head adjunction in Romance-style V-to-
T head movement is accounted for as follows. I assume that T˚ is externally Merged upon
the completion of the VP, which includes a dominant V˚ head and a subject in Spec, VP.
In the first step, a recessive T˚ head with a probe for phi-features (i.e. person, number) is
Merged.

7
Vicente (2007) discusses a number of apparent cases of head movement with no adjunction. One possible
case of this is the participle fronting construction observed in Slavic languages, found in a compound tense
formed by a be auxiliary and a participle verb (Lema and Rivero 1989; Wilder and Ćavar 1994; Bošković
1995; Embick and Izvorski 1995; Migdalski 2006, a.o.). In clauses with an overt subject, the auxiliary
precedes the participle. If no subject is present, however, the participle precedes the auxiliary, as shown in
Migdalski's (2006) Bulgarian examples.

(i) a. Az sŭm čel kniga-ta
I be.AUX.PRES.1SG read.PART.F.SG book-the

b. Čel sŭm kniga-ta
read.PART.F.SG be.AUX.PRES.1SG book-the
'I have read the book'

Given the strict complementary distribution between overt subjects and fronted participles, and that
participles cannot raise above TP (Bošković 1995), it appears that subjects and fronted participles occupy
the same position in TP (Migdalski 2006). Furthermore, because strict linear adjacency is required between
fronted participles and the auxiliary (i.e.) nothing is pied-piped, participle fronting can then potentially be
analyzed a case of head-movement of V˚ to Spec, TP (Vicente 2007; Harizanov 2016).

169

(27) TP
3
T˚
R
VP

[T
R
*, uPhi]3
DP V'
3
V˚
D
...
[V
D
, EPP]

Because the recessive head additionally has a strong feature (indicated by the asterisk), a
copy of V˚
D
is Merged in Spec, TP (28). Because the dominant V˚
D
head now c-
commands recessive T˚
R
, Coalescence applies to create a complex V/T˚ head (29).

(28) TP
wp
V˚
D
T'

[V
D
, EPP] 3
T˚
R
VP

[T
R
*, uPhi] 3
DP V'
3
V˚
D
...
[V
D
, EPP]

(29) TP
wp
V/T˚
D
VP

[V
D
, EPP] 3
[T
R
*, uPhi] DP V'

3
V ˚
D
...
[V
D
, EPP]

170
Here, I claim the bundling of V˚ and T˚ into a single head V/T˚ enables the [uPhi] probe
on [T] to inherit the use of the EPP feature associated with [V
D
], thus triggering
movement of the subject to Spec, TP.

(30) TP
wp
DP T'
wp
V/T˚
D
VP

[V
D
, EPP] 3
[T
R
*, uPhi] DP V'

3
V ˚
D
...
[V
D
, EPP]

5.3.4 Coalescence by Internal or External Merge

Languages do appear to vary in whether the elimination of recessive heads is fed by
External Merge or Internal Merge. As an initial illustration, consider the role of head
movement in V2 patterns. Since V2 crucially requires verb-movement to a left-peripheral
position, the idealized strict V2 restriction is analyzed as follows. Suppose that all left-
peripheral category features (including Force) are recessive, and a recessive head
corresponding to each feature is Merged into the derivation.

171
(31) ForceP

wp
Force˚
R
TopP
[Force
R
*] 3
Top˚
R
FocP

[Top
R
] 3
Foc˚
R
FinP
[Foc
R
] 3
Fin˚
R
InflP
[Fin
R
] 5
... V˚
D
...
[V
D
, EPP]

If the derivation were to end here, the structure would not be interpretable at PF, as it
contains recessive heads. However, the recessive Force˚
R
head in this case is strong,
triggering the movement of V˚
D
to Spec, ForceP.

(32) ForceP
wp
V˚ Force'
[V
D
, EPP] wp
Force˚
R
TopP
[Force
R
*] 3
Top˚
R
FocP

[Top
R
] 3
Foc˚
R
FinP
[Foc
R
] 3
Fin˚
R
InflP
[Fin
R
] 5
... V˚
D
...

This produces the environment for Coalescence, which applies iteratively until all
recessive features are bundled into the dominant head.

172

(33) Force/Top/Foc/FinP
wp
Force/Top/Foc/Fin/V˚
D
InflP
[V
D
, EPP] 5
[Force
R
*] ... V˚
D
...
[Topic
R
]
[Focus
R
]
[Finiteness
R
]

If Coalescence permits probes associated with recessive category features to use EPP
features associated with the dominant head, this enables an explanation for why in V2
structures an additional XP must precede the verb. In this case, a probe associated with
one of the left-peripheral features uses the verb's EPP property to trigger the movement
of the object that it agrees with.

(34) Force/Top/Foc/FinP
wp
XP
k
Force/Top/Foc/Fin'
wp
Force/Top/Foc/Fin/V˚ InflP
[V
D
, EPP] 5
[Force
R
*] ... XP
k
...
[Topic
R
]
[Focus
R
]
[Finiteness
R
]

Suppose, however, that rather than introducing a strong recessive Force˚ head, a
dominant Force˚ head is Externally Merged (35). This allows Coalescence to bundle all
left-peripheral features into a single head, without the need for verb movement out of
InflP (36).

173
(35) ForceP
3
Force˚
D
TopP
[Force
D
] 3
Top˚
R
FocP

[Top
R
] 3
Foc˚
R
FinP
[Foc
R
] 3
Fin˚
R
InflP
[Fin
R
] 5
... V˚
D
...

(36) Force/Top/Foc/FinP
wp
Force/Top/Foc/Fin˚
D
InflP
[Force
D
] 5
[Topic
R
] ... V˚
D
...
[Focus
R
]
[Finiteness
R
]

The availability of both of these options accounts for 'asymmetric' V2 patterns like
the one in German, in which embedded clauses with an overt complementizer and those
with V2 are in complementary distribution.

(37) a. Er sagte [dass er morgen komme]
He said that he tomorrow comes

b. Er sagte [er komme morgen]
He said he comes tomorrow
'He said that he is coming tomorrow.'

c. *Er sagte [dass er komme morgen]
He said that he comes tomorrow
(Holmberg 2015 after den Besten 1983)

The alternation is accounted for if the overt complementizer is the phonological
realization of a dominant Force˚ head directly Merged in the left periphery that does not
have an EPP property. Verb movement into the left periphery does not take place simply

174
because there is no strong head to trigger it. In clauses without an overt complementizer,
Force˚ is recessive and strong when Merged, and verb movement takes place to permit
the application of Coalescence.

5.4 Coalescence case studies

This section presents several studies of variation in the realization of functional
projections of the inflectional and complementizer domains. I first return to the typology
of relaxed V2 cases discussed in the previous chapter, and show that V2 requirements are
increasingly relaxed as the number of dominant left-peripheral heads increases. We then
consider two instances in which languages permit a given head to have both dominant
and recessive variants, the Kashmiri V2/V3 alternation and the realization of sentential
negation in English. Lastly, within-language variation in the realization of the Catalan
past perfect and the instantiation of the Wolof left periphery are attributed to differences
in the distribution of dominant features and strong recessive features across functional
heads.

5.4.1 Relaxed V2

At this point, we can revisit the analysis of relaxed V2 proposed in chapter 4 within the
framework of Coalescence. Recall that an idealized, strict V2 arises in a situation where
all left-peripheral features are bundled into a single head that attracts one specifier. The
bundling of all left-peripheral categories on one head takes place if all left-peripheral

175
category features are recessive, and verb movement is compelled by a strong, recessive
Force˚ head.

(38) Force/Top/Foc/FinP
wp
XP
k
Force/Top/Foc/Fin'
wp
Force/Top/Foc/Fin/V˚ InflP
[V
D
, EPP] 5
[Force
R
*] ... XP
k
...
[Topic
R
]
[Focus
R
]
[Finiteness
R
]

As the number of dominant category features increases, fewer applications of
Coalescence take place, leaving a greater number of independent heads in the left
periphery. We first consider languages like Old English and Cimbrian, which permit
preverbal subjects to co-occur with one other phrase: [XP Subject V ...]. Because
preverbal subjects occupy a low left-peripheral position, it appears that Fin˚ is the highest
head that attracts the verb. The pattern is thus accounted for if [Finiteness
R
*] is strong,
attracting verb movement. Furthermore, Force˚ is dominant when first Merged,
permitting all higher left-peripheral features to be bundled into a single head.

176
(39) Old English/Cimbrian V3

Force/Top/FocP
wp
XP Force/Top/Foc'
wp
Force/Top/Foc˚ FinP
[Force
D
, EPP] wp
[Topic
R
] (DP) Fin'
[Focus
R
] wp
Fin˚ InflP
[V
D,
(EPP)] 5
[Fin
R
*, uD] ...

The Ingush pattern in which topics and foci can simultaneously precede the verb crucially
requires the [Topic] and [Focus] features to be realized on separate dominant heads. This
occurs if [Focus
D
] is dominant, as it prevents Coalescence from applying between the
Topic˚ and Focus˚ heads, and if a [uTop] probe is associated with a dominant head above
FocusP.

(40) Ingush V3

Force/TopP

wp
XP Force/Top'
wp
Force/Top˚
D
FocP
[Force
D
, EPP] wp
[Topic
R
, uTop] XP Foc'

wp
Focus˚
D
FinP
[Focus
D
, uFoc, EPP] 3
Fin/V˚
D
InflP
. [V˚
D
] 5
[Fin
R
] ... V˚
D
...

177
There is an alternative derivation of this pattern in which Focus˚ is Merged as a strong
recessive head that attracts the verb. In the absence of a clear diagnostic to favor one
derivation over the other, however, I will simply assume a relatively uniform verb-
movement path across V2 structures in which the verb moves only as far as FinP.
The presence of an additional dominant feature on [Topic
D
] permits the generation of
V4 patterns of the sort attested in Old Sicilian and Old Italian.

(41) ForceP

wp
XP Force'
wp
Force
D
˚ TopP
[Force
D
, uFrame, EPP]wp
XP Top'

wp
Topic
D
˚ FocP
[Topic
D
, uTop, EPP]wp
XP Foc'

wp
Focus˚
D
FinP
[Focus
D
, uFoc, EPP] 3
Fin/V˚
D
InflP
. [V˚
D
] 5
[Fin
R
] ... V˚
D
...

5.4.2 Kashmiri V2/V3

Further support for the distinction between dominant and recessive heads is found in
individual languages that contain both dominant and recessive variants of the same
category feature. Here, I argue that the Kashmiri V2/V3 alternation reflects precisely this
possibility.

178
Recall that in Kashmiri V2, phrases in first positions are typically interpreted as foci,
not topics. However, if a wh-phrase is fronted, a topic is optionally permitted in first
position. Under the assumption that foci and wh-phrases occupy the same position in the
left periphery, we can propose that Kashmiri allows the Focus head to contain either a
dominant or recessive category feature. Specifically, the category feature [Focus] is
dominant only if it has a [uWh] probe, otherwise, it is recessive. In contrast, languages
like Ingush, where V3 is available in both declaratives and interrogatives, [Focus] is
always dominant regardless of its other properties.
Consider how the derivation proceeds if the Numeration contains the dominant [uWh]
Focus head. Since both Topic˚ and Focus˚ are dominant as Merged, Coalescence can not
apply, leaving [Topic] and [Focus] realized in separate projections. Because both heads
contain an EPP feature, this creates the V3 word order in interrogative clauses.

(42) Numeration for V3
[Focus
R
*, uWh]
[Topic
D
, uTop, EPP]

(43) TopicP

wp
XP Top'
wp
Topic˚
D
FocusP
[Top
D
, uTop, EPP] wp
XP Focus'
wp
Focus ˚
D
...

[Focus
D
, uWh, EPP]

On the other hand, suppose that a recessive Focus head with a [uContrast] probe is
selected instead. Coalescence now applies once the dominant Topic˚ head is Merged.

179
This bundles the [Topic] and [Focus] features into a single projection, leaving only one
position available for movement.

(44) Numeration for V2
[Focus
R
, uContrast]
[Topic
D
, uTop, EPP]

(45) TopicP

Topic/FocusP
wp wp
Topic˚
D
FocusP > Topic/Focus˚
D
...
[Topic
D
, uTop, EPP] wp [Topic
D
, uTop, EPP]
Focus˚
R
... [Focus
R
, uContrast]

[Focus
R
, uContrast]

An important question that arises in this analysis, however, is why only a focused
constituent can be attracted to first position, despite the fact that [Topic
D
] is associated
with an EPP feature. There does not as of yet appear to be an elegant solution to the
problem. However, one possible way to account for this is to propose that multiple probes
within a head differ in the priority with which they can be associated with the EPP
(Lahne 2010; Georgi and Müller 2010; Manetta 2011; Martinovic 2015). In the case of
Kashmiri, the checking of [uContrast] takes precedence over the checking of [uTop].
Nonetheless, the Coalescence analysis of the Kashmiri V2/V3 alternation allows for
an explanation of the intuition that the realization of the high topic position depends on
the presence of a wh-phrase. Capturing this particular relationship between the
availability of topicalization and wh-movement in the Bottleneck Effect or Stacked Head
approaches requires stipulation about when base-generation is possible, or possible
combinations of features within the C head, respectively. On the other hand, this type of
interaction is predicted in the present theory by the possible distribution of dominance
and recession.

180
5.4.3 English negative contraction

English negation is another case where a head has both dominant and recessive varieties.
English has a 'full' negative morpheme (orthographic not) and an affixal form
(orthographic n't). In many contexts, the two forms appear to be in free variation, with the
affixal form apparently derived by optional phonological reduction.

(46) a. Michael did not make a mistake.
b. Michael didn't make a mistake.

As noted by Zwicky and Pullum (1983) however, the distribution of the two forms is
constrained by syntactic factors, and the use of a particular form is obligatory in certain
contexts. For example, consider the case of negative inversion. In English, auxiliary verbs
raise to a pre-subject position in interrogative contexts. If the negation morpheme raises
along with the auxiliary, use of the affixal form is obligatory (47a). This gives the effect
of contraction feeding raising. On the hand, only the full form is possible if the negative
remains in a post-subject position (47b). Under an approach where the affixal form is
derived by an operation that applies after the syntactic derivation, the obligatory use of
the affixal negative when it raises with the auxiliary is perhaps unexpected.

(47) a. Didn't Lindsay host the gala? (cf. *Did not Lindsay host the gala?)
b. Did Lindsay not host the gala? (cf. *Did Lindsay n't host the gala?)

Matushansky (2006) makes the key observation that the different distribution of the
full and affixal forms is explained if the affixal form is derived by a cliticization process
that takes place during the derivation. Specifically, she proposes that affixation by M-

181
Merger optionally applies to Neg˚ and Aux˚ as soon as the items are merged into the
derivation, and that M-Merged Neg˚ corresponds to n't. If Aux˚ and Neg˚ are M-Merged
prior to movement, both negation and the auxiliary undergo movement together when
Aux˚ is attracted to C˚. If M-Merger does not apply, the auxiliary raises alone.
This analysis has a straightforward equivalent in terms of Coalescence. The different
distribution of the full and affixal forms is accounted for if the full form enters the
derivation with a dominant category feature [Neg
D
], while the affixal form begins with a
recessive feature [Neg
R
]. Let us assume that an Auxiliary head then merges with NegP.
The recessive affixal head must undergo Coalescence with the auxiliary before the
auxiliary raises (48). Thus, when Aux is subsequently attracted to an interrogative
complementizer position (simplified here as CP), the affixal negative must be copied as
well (49). I will assume that the lower copy is left unpronounced at PF.

(48) AuxP

Aux/NegP
3 3
Aux˚
D
NegP > Aux/Neg˚
D
...
[Aux
D
] 3 Coalescence [Aux
D
]
Neg˚
R
... [Neg
R
]
[Neg
R
]

(49) CP

CP

3 3
C˚
R
... Aux/Neg˚
D
C'
[C
R
*] 3 [Aux
D
] 3
Aux/NegP [Neg
R
] C˚
R
...
3

>

[C
R
*] 3
Aux/Neg˚
D
... Aux/NegP
[Aux
D
] 3
[Neg
R
] Aux/Neg
D
...
[Aux
D
]
[Neg
R
]

182
On the other hand, the dominant Neg˚ head that corresponds to the full form does not
undergo Coalescence when the auxiliary is Merged. In inversion contexts, the
interrogative complementizer head thus attracts only Aux
D
˚, leaving Neg
D
˚ in its first-
Merged position (51).

(50) AuxP

3
Aux˚
D
NegP
[Aux
D
] 3 Coalescence does not apply
Neg˚
D
...
[Neg
D
]

(51) CP

CP

3 3
C˚
R
... Aux˚
D
C'
[C
R
*] 3 [Aux
D
] 3
AuxP C˚
R
...
3

[C
R
*] 3
Aux˚
D
NegP > AuxP
[Aux
D
] 3 3
Neg˚
D
... Aux˚
D
NegP
[Neg
D
] [Aux
D
] 3
Neg˚
D
...
[Neg
D
]

It should be noted that there are some additional constraints on the distribution of
contracted negation. Specifically, negative contraction is not possible after non-verbal
elements, non-finite verbs or auxiliaries, as shown in the following examples (based on
Zwicky and Pullum's ex. 11, 14).

(52) a. Buster doesn't try {*n't/not} to pay attention; he just can't help it.
b. For him {*n't/not} to use an alliance-approved assistant is the last straw.
c. Should you have {*n't/not} held the door for the man in the pricy suit?

183
Rather than a constraint on the possible clitic hosts for n't, the unavailability of n't in the
above examples indicates that contraction is sensitive to the contrast between sentential
negation and constituent negation, (Klima 1964). Among other differences in their
syntactic distribution, sentential negation occurs only immediately following a tensed
auxiliary or do, whereas constituent negation immediately precedes negated phrases.
From this, we can conclude that sentential [Neg] is always recessive, whereas constituent
[Neg] is always dominant.
It is important to observe that negative contraction differs in crucial ways from the
phenomena known as Auxiliary Reduction (Zwicky 1970; Anderson 2008), which affects
English auxiliary forms is, has, would, had, have, am, are, and will. First, reduced
auxiliaries do not form syntactic units with their hosts in the manner of contracted
negation. For instance, reduced auxiliaries do not pattern as if they raise with elements
that they are affixed to.

(53) a. Who's going to Phoenix?
b. *Who's do you think who's going to Phoenix?

Furthermore, whereas contracted negation only follows tensed auxiliaries, reduced
auxiliaries is, has, would, had, are generally unrestricted by the category or phrase
structure status of the items that they precede: no touching's allowed by the guards / the
role that he auditioned for's been given to someone else.
8
Given that Coalescence has
been defined as a process that applies between heads, these cases can not be analyzed as
the result of Coalescence.

8
The reduced forms of will, have, are, and am are substantially more restricted, appearing only after
pronouns and wh-words.

184
Lastly, although reduced auxiliaries can follow contracted negation (e.g.
shouldn't've), contracted negation cannot follow reduced auxiliaries (*he'sn't/he isn't).
The fact that auxiliary reduction appears to bleed contracted negation suggests that the
process does not take place during the syntactic derivation, but rather applies after
syntactic structure has been transferred to the PF branch. Another important implication
of this observation is that an affixal representation of a morpheme in surface phonology
does not unambiguously indicate that its corresponding syntactic feature is recessive, as
defined in this thesis.

5.4.4 The Catalan perfect

In the previous section, I argued that recessive heads can be eliminated either by External
Merge or Internal Merge of a dominant item, and that this accounts for the 'asymmetric'
V2 alternation where overt complementizers and V2 are in complementary distribution.
Here, I present another case in which a language permits variation in whether recessive
heads are eliminated by External Merge or movement: the realization of the past perfect
in Catalan.
As described by Oltra-Massuet (2013), some dialects of Catalan express the past
perfect either in a synthetic form, in which the subject and tense and aspect morphemes
are realized as suffixes to a lexical verb, and an analytic form, in which subject
agreement is realized on an auxiliary verb, anar 'to go.'
9

9
Interestingly, Catalan differs from other Romance languages in using anar 'to go' as an auxiliary for past
tense, rather than a form of 'to have.'

185
(54) purific-ares
purify-2SG.PST.PERF
'you purified'
(55) vas purificar
AUX.2SG purify
'you purified'

An unusual and important property about the alternation is that there is no apparent
semantic difference between the two ways of forming the past perfect. In the words of
Oltra-Massuet, "these forms do not express different lexical or truth-conditional
semantics, nor do they show different morpho-syntactic functions, and individual
speakers use some subset of them without distinction." However, it is clarified that both
within and across speakers, variation depends on the lexical items and conjugations used;
speakers do not probabilistically use both forms for any given verb and conjugation pair.
Oltra-Massuet proposes that the form of the syncretic past perfect is derived by head-
movement of the verb through Aspect˚ and Tense˚ heads. Abstracting away from the
distinction between V and little v, not relevant to the present analysis, she proposes the
following structures. In synthetic forms like purificares, the verb moves to a Tense˚ head
that carries specifications for past tense, perfective aspect, and telicity.

(56) TP
3
T˚ ...
3
V˚ T˚
[PAST, PERF, TELIC]

In the analytic past prefect, Oltra-Massuet proposes that [+PAST, +PERF] are associated
with T˚, while the [+TELIC] feature is contained in a separate Aspect˚ head. Furthermore,
verb movement stops at Aspect˚, and the V-Asp˚ head is pronounced as the participle.

186

(57) TP
wp
T˚ AspP
[PAST, PERF] 3
Asp˚ VP
3 ...
V˚ Asp˚
[TELIC]

However, because T˚ is obligatorily a suffix, it is proposed that the realization of this
structure triggers go-support, the insertion of an anar auxiliary verb that supports the
inflection of both tense and subjecthood.
There is a clear parallel between the Catalan pattern and the approach developed here
for V2 and overt complementizer alternations. Again, the distinction between the
synthetic and analytic forms does not result from a difference in the distribution of
dominance; It has to be the case that T˚ and Asp˚ are always recessive, since their
phonological exponents are realized as affixes. However, the target of verb movement
differs. I propose that in the derivation for the syncretic form, V˚ raises to T˚, and is then
bundled with the recessive T˚ and Asp˚ heads.
10

(58) VP

wp
V˚
D
T'
[V
D
] wp
T˚
R
AspP

[T
R
*, PAST, PERF] wp
Asp˚
R
VP
[Asp
R
, TEL] wp
V˚
D
...
[V
D
]

10
We do not exclude the possibility that V˚ moves first through AspP before raising to TP. However, this
does not bear crucially for the purposes of this discussion, since the two possible derivations will result in
the same bundled head structure.

187
(59) T/Asp/VP
wp
T/Asp/V˚
D
VP
[V
D
] wp
[T
R
*, PAST, PERF] V˚
D
...
[Asp
R
,

TEL] [V
D
]

Suppose that in these dialects of Catalan, it is also possible for verb movement to proceed
only as far as Spec, AspP. At this point, Coalescence bundles V˚ and Asp˚.

(60) VP V/AspP
3 wp
V˚
D
AspP

V/Asp˚
D
VP
[V
D
] 3 > [V
D
] 3
Asp˚
R
VP Coalescence [Asp
R
*, TEL] V˚
D
...
[Asp
R
*, TEL] 3
V˚
D
...
[V
D
]

At the next step of this derivation, the recessive T˚
R
head is Merged. However, because
T˚
R
, the recessive head can only be eliminated if a dominant item is externally Merged.

(61) TP

wp
T˚
R
V/AspP
[T
R
,

PST, PERF] wp
V/Asp˚
D
VP

[V
D
] 3
[Asp
R
*, TEL] V˚
D
...
[V
D
]

I propose that the need to eliminate the recessive T˚ head can be satisfied in Catalan via
External Merge of the dominant auxiliary head that corresponds to anar. I will remain
agnostic as to the categorial status of the auxiliary, labeling it simply as [Aux]. After the
auxiliary enters the derivation, Coalescence bundles Aux˚
D
and T˚
R
.

188
(62) AuxP

wp
Aux˚
D
TP
[Aux
D
] wp
T˚
R
V/AspP
[T
R
, PST, PERF] wp
V/Asp˚
D
VP

[V
D
] 3
[Asp
R
*, TEL] V˚
D
...
[V
D
]

(63) Aux/TP

wp
Aux/T˚
D
V/AspP
[Aux
D
] wp
[T
R
, PST, PERF] V/Asp˚
D
VP
[V
D
] 3

[Asp
R
*, TEL] V˚
D
...
[V
D
]

Catalan thus permits two strategies for eliminating the recessive T˚
R
head: It can be
eliminated either by being bundled with a dominant verb that moves to its specifier, or by
being bundled with a dominant auxiliary directly merged above it.

5.4.5 Wolof C/T˚ bundling

Lastly, we consider a case of word order variation in Wolof, in which clause types differ
in the amount of functional structure permitted in the C-T domain. Again, we will see
that the distinction can be accounted for as variation in the distribution of dominant
features.
Wolof requires the presence of a complementizer element in all finite clauses, whose
form varies according to clause type (Dunigan 1994). Martinović (2015, 2016) notes that

189
the language appears to have two clause types that differ in the amount of functional
structure that they instantiate. In particular, they differ with respect to the placement of
verbs relative to the complementizer element. In one type of clause termed V-raising, the
verb occurs immediately following the subject, and is immediately followed by the
complementizer suffix. V-raising is the pattern of neutral declaratives (affirmative and
negative), and sentences characterized as verb or predicate focus.
11

(64) Xale yi lekk-oon-na-ñu ceeb bi
child DEF.PL eat- PERF-C-3PL.SUB cake DEF.SG
'the children ate the rice (a long time ago).' (Martinović 2015)

This pattern contrasts with wh-questions and exhaustive identification focus, which show
what Martinović terms the N-raising pattern. In these sentences, the complementizer
element is realized following a clause-initial wh-word or focused item. When a non-
subject has been extracted, subjects obligatorily appear after the complementizer. Main
verbs follow the subject, and do not show a complementizer suffix.

(65) Lan la xale yi lekk
what CWH child DEF.PL ate
'What did the children eat?' (Martinović 2016)

(66) Musaa la xale yi gis
Moussa CWH child DEF.PL see
'It is the child that Moussa saw.' (Martinović 2015)

On the basis of these differences, Martinović proposes that V-raising structures involve a
bundled complementizer and tense head C/T˚, while in N-raising structures, C˚ and T˚ are
realized as separate positions. Intriguingly, the pattern bears some similarities to the

11
Predicate focus structures differ from neutral declaratives by requiring a do-auxiliary to precede the
complementizer suffix, rather than the main verb. For the purposes of this discussion, I will restrict
attention to neutral declaratives.

190
Kashmiri V2/V3 alternation. Informally stated, the presence of a wh-phrase in the left
periphery of Kashmiri requires the realization of a higher topic position. In Wolof,
however, foci and wh-phrases appear to license a lower subject or tense position.

(67) V-raising: C˚ and T˚ are bundled

C/TP

wp
DP C/T'

xale yi 3
the children C/T˚ ...
3
V˚ C/T˚
lekk na
eat

(68) N-raising: C˚ and T˚ are distinct

CP

wp
DP C'

lan wp
what C ˚ TP
la wp
DP T'
xale yi 3
the children T˚ ...

Martinović takes a distinct approach to bundling, which proposes that functional
features of the complementizer and tense domains enter the derivation in a single head, in
which probes associated with the different features are hierarchically arranged, not unlike
the Stacked Head approach of Manetta and Lahne. However, this head can be compelled
to split if its probes cannot be satisfied - a possibility that explains the N-raising pattern.
In essence, her analysis is as follows: In V-raising sentences, the bundled head enters the
derivation with two probes, one that triggers movement of the subject [EPP*] and one

191
that triggers verb movement [V*]. In this case, both head movement and phrasal
movement are able to proceed, creating the final representation in (70).
12

(69) C/TP

wp
C/T˚ VP
[EPP*] 3
[V*] DP V'
3
V˚

DP

(70) C/TP

wp
DP C/T'
wp
C/T˚ VP
3 3
V˚ C/T˚ DP V'
[EPP*] 3
[V*] V˚

DP

In N-raising structures, the C/T˚ head includes both the subject probe [EPP*] and an
additional probe for a wh-phrase or exhaustive focus. In the first step, the [EPP*] probe
compels the subject to move to its specifier, as in V-raising (72).

(71) C/TP

wp
C/T˚ VP
[EPP*] 3
[Wh*] DP V'
la 3
V˚

DP
wh

12
Martinović's analysis takes EPP movement to take place prior to verb movement, in which case verb
movement does not extend the tree. However, this ordering of operations does not seem to be crucial to the
analysis.

192
(72) C/TP

wp
DP C/T'
wp
C/T˚ XP
[EPP*] 3
[Wh*] DP V'
la 3
V˚

DP
wh

At this point, the [Wh*] probe must be checked. However, due to a ban on multiple
specifiers within the C/T projection, movement of the wh-phrase across the subject is
impossible. Thus, the requirement for [Wh*] to be checked compels a head splitting
operation, in which a portion of the C/T˚ head containing its unchecked features is split
off and Merged again at the root (73). This then permits movement of the wh-word,
yielding the final word order (74).
13

(73) C/TP

wp
C/T˚ C/T'
[Wh*] wp
la DP C/T'
wp
C/T˚ VP
[EPP*] 3
DP V'
3
V˚

DP
wh

Head Splitting

13
Martinović does not propose a probe for verb-movement in N-raising structures. However, surface word
order does not seem to provide a way to determine whether or not the verb raises to the lower C/T˚ head, as
subjects are presumed to move from Spec, VP to Spec, TP.

193

(74) C/TP

wp
DP
Wh
C/T'
wp
C/T˚ C/T'
[Wh*] wp
la DP C/T'
wp
C/T˚ VP
[EPP*] 3
DP V'
3
V˚

DP
wh

We will discuss at the end of the section the relative merits of head-splitting vs. head-
bundling. I will first show, however, how the Wolof alternation can be accounted for in
terms of Coalescence. Rather than beginning the derivation in a bundled state, the C˚ and
T˚ heads are separately Merged into the derivation. We first consider the derivation of an
N-raising wh-question structure, in which verb movement only moves as high as Tense˚.
The relevant portion of the derivation begins when the recessive Tense˚ head is Merged
(75). In the next step, the dominant verb moves to Spec, TP (76).

(75) TP

3
T˚
R
VP
[T
R
*,

uPhi] 3
V˚
D
...
[V
D
, EPP]

(76)

TP
3
V˚
D
TP
[V
D
, EPP] 3
T˚
R
VP
[T
R
*,

uPhi] 5
... V˚
D
...

194
Coalescence then applies to create a bundled V/T˚ head (77). In the next step, the [uD]
probe of the tense head and EPP feature of V˚ trigger movement of the subject to Spec,
TP (78).

(77) TP TP

3 3
V˚
D
TP > V/T˚
D
VP
[V
D
, EPP]3 Coalescence [V
D
, EPP] 5
T˚
R
VP [T
R
*,

uPhi] ... V˚
D
...
[T
R
*,

uPhi] 5
... V˚
D
...

(78) TP

3
DP T'
3
V/T˚
D
VP
[V
D
, EPP] 5
[T
R
*,

uPhi] ... V˚
D
...

This is the point at which complementizer heads enter the derivation. I propose that
the distinguishing property of N-raising clauses is that they contain a C˚ head that is both
dominant when first Merged, and endowed with an EPP feature. No verb movement is
required to eliminate a recessive head, and the derivation ends after the probes on C˚ have
triggered movement to Spec, CP (80).

(79) CP

wp
C
D
˚ TP
[C
D
, uwh, EPP] 3
DP T'
3
V/T˚
D
VP
[V
D
, EPP] 5
[T
R
*,

uPhi] ... V˚
D
...

195
(80) CP

wp
XP C'
wp
C
D
˚ TP
[C
D
, uwh, EPP] 3
DP T'
3
V/T˚
D
VP
[V
D
, EPP] 5
[T
R
*,

uPhi] ... V˚
D
...

We now turn to the derivation of V-raising clauses. Here, I will suppose that the
derivation proceeds identically through TP, picking up at (78). In this case, however, I
propose that the C˚ head is recessive and strong, and further endowed with a [uD] feature,
to be justified shortly.

(81) CP

3
C
R
˚ TP
[C
R
*, uD] 3
DP T'
3
V/T˚
D
VP
[V
D
, EPP] 5
[T
R
*,

uPhi] ... V˚
D
...

After the bundled V/T˚
D
head moves to Spec, CP (82), Coalescence bundles the topmost
heads (83).

196
(82) CP

wp
V/T˚
D
C'
[V
D
, EPP] 3
[T
R
,

uPhi] C
R
˚ TP
[C
R
*, uD]3
DP T'
3
V/T˚
D
VP
[V
D
, EPP] 5
[T
R
*,

uPhi] ... V˚
D
...

(83) CP

3
V/T/C
D
˚ TP
[V
D
, EPP] 3
[T
R
*,

uPhi] DP T'
[C
R
*, uD] 3
V/T˚
D
VP
[V
D
, EPP] 5
[T
R
*,

uPhi] ... V˚
D
...

If the derivation were to stop at this point, we would incorrectly predict that only verb-
initial orders are possible. To allow subjects to raise to the specifier, the bundled head
requires a relevant probe, in this case the [uD] probe on C˚.

(84) CP

wp
DP C'

3
V/T/C˚
D
TP
[V
D
, EPP] 3
[T
R
*,

uPhi] DP T'
[C
R
*, uD] 3
V/T˚
D
VP
[V
D
, EPP] 5
[T
R
*,

uPhi] ... V˚
D
...

197
On what basis should we prefer a splitting approach, in which extended projections
begin bundled and split during the derivation, or a bundling approach in which all heads
are Merged separately and bundled during the derivation? Choosing between the two
options likely depends on whether bundling or splitting can receive more plausible
functional motivations. Crucially, not all of the case studies in this chapter in which
functional heads are realized on separate heads can be accounted for as the result of head-
splitting motivated by feature checking, particularly the alternations in English negation
and the Catalan past perfect. This is particularly true of the alternation between the
synthetic and analytic forms of the Catalan past perfect. Because the two options do not
differ in their semantic functions, there would appear to be no purely syntactic motivation
for tense features to split off in order to generate the analytic form. On the other hand, the
synthetic and analytic forms straightforwardly reflect two strategies (External Merge and
Internal Merge) used to eliminate the recessive Tense˚ head, whose realization is
obligatorily affixal.

5.5 Complex heads as minimal projections

In Section 5.3, it was proposed that Coalescence leaves no internal branching structure
within the complex head, accounting for the inability for later syntactic operations to
affect subparts of bundled heads. This section discusses additional theoretical benefits
and implications of this proposal. First, I show that the elimination of branching structure
within complex heads obviates persistent concerns about head-movement in Minimalist
Syntax, particularly within the considerations of Bare Phrase Structure. Lastly, I consider

198
implications of the proposal for the identification of copy chains, and for the
determination of affix ordering.

5.5.1 Coalescence and Phrase Structure Uniformity

The non-existence of internal branching structure within heads created by Coalescence
provides a solution to conceptual problems raised by traditional head-adjunction
structures within Minimalism, specifically given the proposal of Bare Phrase Structure
(Chomsky 1994). The key claim of the proposal is that the phrase structure status of a
syntactic object (head vs. phrasal) is determined uniquely by its position within the
syntactic tree. Specifically, an object whose label does not project is a maximal
projection (notationally represented as XP or X
max
), while an object whose label is not
identical to that of a node that it dominates is a minimal projection (X˚ or X
min
). An
object with both properties is simultaneously maximal and minimal. The three options are
schematized in (85).

(85) X
max

3
X
min
Y
max

3
Y
min
Z
max/min

In this system, a head that does not project is both a minimal and maximal projection.
Intermediate projections that are neither minimal nor maximal, corresponding to the X'
levels in X-bar Theory, are proposed to be inaccessible to syntactic operations and thus
cannot undergo movement.

199
In order to rule out movement structures deemed to be impossible, such as head-to-
specifier movement or XP adjunction to heads, Chomsky posits a uniformity condition on
movement chains, which I will refer to as the Phrase Structure Uniformity Condition to
distinguish it from Nunes' Featural Uniformity Condition.

(86) PHRASE STRUCTURE UNIFORMITY CONDITION
A chain is uniform with respect to phrase structure status

Chomsky observes that the traditional structure associated with head movement
violates the Phrase Structure Uniformity Condition. To illustrate, consider the structure
that is the predicted result of V˚-to-Infl˚ head movement, where V˚ adjoins to I˚.

(87) I
max

wp
... I
wp
I V
max

3 3
V
min/max
k
I
min
... V
3
V
min
k
...

The problem noted by Chomsky is that the two copies of V in this structure differ in their
phrase structure status. Because the higher copy of V does not project, it is both minimal
and maximal, whereas the lower copy is minimal, since it projects. On these grounds, the
adjunction structure violates the Phrase Structure Uniformity Condition. While Chomsky
suggests that word-like items may simply be immune from this condition, this is a
stipulation that should be avoided if possible.
14
Additional issues arise from the fact that

14
In subsequent work, Chomsky (2001) argues that head movement is an operation that takes place in PF,
not within the narrow syntax, based on the additional consideration that many instances of head movement

200
the moved and adjoined head does not c-command its lower copy, a suspect property that
would distinguish head movement from phrasal movement (Matushansky 2006). Head
movement would further differ from phrasal movement because it fails to extend the tree,
unlike other forms of Merge (Chomsky 1995).
As noted by Harley (2013), successive-cyclic head movement proves even more
problematic for the Phrase Structure Uniformity Condition. Consider the representation
that results from movement and adjunction of the complex V-T head created in (87) to a
higher complementizer head.

(88) C
max

wp
C I
max

3 3
I
max
k
C
min
... I
3 3
V
min/max
I
min
I
k
...
3
V
min/max
I
min

The more serious problem raised by this case is that the lower copy in this structure is
neither minimal nor maximal. Since intermediate projections are proposed to be invisible
to syntactic operations, movement of the complex head is predicted to be impossible.
We now consider how violations of the Phrase Structure Uniformity Condition and
reference to intermediate projections are avoided in the Coalescence proposal put forth.
This rests on two claims made in this chapter: First, that moved heads project their labels;
Second, that complex heads formed by Coalescence include no internal branching

seem to lack semantic effects. However, there is substantial evidence of head-movement having effects on
LF interpretation, (Lechner 2006; Matushansky 2006; Roberts 2010; Hartman 2011), suggesting that head
movement must take place in syntax.

201
structure. Consider the derivation of a bundled V/Infl head. In the first step, the dominant
V
min
moves up to Spec, Infl. The labeling of phrasal nodes at each step is determined by
Category Percolation. Because the highest node contains the features of both [V] and
[Infl], it is a maximal projection of both categories (I
max
and V
max
). After Coalescence
applies, this featural specification remains unchanged. Crucially, however, the bundled
head is now a minimal projection of both [V] and [Infl]. The two links of the verb's
movement chain are thus minimal projections, in accordance with Phrase Structure
Uniformity, and the bundled head c-commands the lower copy of the verb.

(89) V/I
max

V/I
max

[Infl
R
] [Infl
R
]
[V
D
] [V
D
]
3 wp
V
min
D
I/V > V/I
min
D
V
[V
D
] [Infl
R
] Coalescence [V
D
] [V
D
]
[V
D
] [Infl
R
] 5
3 ... V
min
...

I
min
R
V

[V
D
]
[Infl
R
] [V
D
]
5
... V
min
...
[V
D
]

Later head movement of the complex head will similarly produce only minimal
projections, and there is no need to posit movement of an intermediate-level projection.
Coalescence and Category Percolation therefore provide a means to derive head
movement within syntax in accordance with Minimalist considerations.

202
5.5.2 Coalescence and the identification of copy chains

The account of head adjunction in terms of Coalescence raises some questions about its
compatibility with the Copy Theory of Movement assumed thus far within the
dissertation. To illustrate, recall that V-to-T movement is analyzed as the result of two
operations: Merge of a copy of V˚ in Spec, TP followed by Coalescence, which bundles
T˚ and V˚ into a complex head. The two steps are shown in (90) and (91).

(90) TP

wp
V˚
D
TP
[V
D
] wp
T˚
R
VP

[T
R
*] wp
... V'
wp
V˚
D
...
[V
D
]

(91) TP

wp
V/T˚
D
VP
[V
D
] wp
[T
R
*] ... V'
wp
V˚
D
...
[V
D
]

After the higher copy of the verb has been bundled with T, it is not clear that V/T˚
and V˚ should be treated as copies of a chain without additional stipulation about
linearization. The problem is that V/T˚ is non-distinct from the lower copy of the verb in
the sense that it contains more category features. If the two items that contain [V
D
] cannot
be recognized as copies by the grammar, one might expect them to be pronounced

203
separately, in a case of multiple copy pronunciation (Nunes 1999, 2004; Dourado 2002;
Kandybowicz 2007). It is crucial for the proposals of this dissertation, however, that the
PF component be able to identify copy chains in order to determine their optimal
pronunciation.
I propose that the identification of a movement chain does not require chain members
to be identical in terms of features, but rather a subset relationship between the features
contained in each chain link. This notion is essential to Roberts' (2010) proposal that head
movement takes place only when a head enters an agreement relation with a goal that
contains a proper subset of its formal features, a relation characterized as defectivity.

(92) A goal G is defective iff G's formal features are a proper subset of those of G's
probe P.

As an illustration, T˚ attracts V˚ in a given language if T˚ contains the features [T, uV],
while V˚ contains only its interpretable category feature [V] (abstracting away from case
and phi-features). In languages like English which lack V-to-T movement, Roberts
proposes that T˚ lacks a [uV] feature. Since V˚ has a [V] feature but T˚ does not, it can
not be a defective goal with respect to T˚, making V-to-T movement impossible.
I propose that the identification of items in the derivation as copies depends not on
featural identity, but on the relation of defectivity. Movement chains are thus recognized
on the basis of (93).

(93) CHAIN RECOGNITION
α and β form a movement chain iff α c-commands β and the formal features of β
are a proper subset of those of α.

204
This method of chain identification permits the two copies of the verb in (91) to be
identified as a chain, since the formal features of the lower verb V˚ are a subset of those
in the bundled V/T˚ head. Thus, even though Coalescence has bundled a verb copy with a
recessive T˚ head, it does not prevent the two verbs from being identified as a chain.
When the derivation is transferred to the interface, the PF grammar identifies two
copies as belonging to a chain if the condition in (93) is met. It is important to note that
while all heads transferred to the PF component must be dominant, they need not be
pronounced in the output of the grammar. Rather, dominance simply reflects the
eligibility of a head to be transferred to the PF interface. Whether or not a dominant head
is pronounced in the final surface representation depends on the interaction of constraints
on copy pronunciation, morpheme realization, and and prosodic structure organization, as
argued in Chapter 3.

5.5.3 Coalescence and affix ordering

Here, I consider the implications of Coalescence with respect to theories of affix
ordering, particularly in contrast with "classic" views of head adjunction that assume an
internal branching structure. To illustrate, consider the standard analysis of the head
adjunction structure formed by verb movement through the Aspect˚ and Tense˚
inflectional heads. The structure in (94) conforms with both the Right Hand Head Rule
(Williams 1981), which places the projecting head at each level on the right, and the
Mirror Principle (Baker 1985), which requires the hierarchical organization of
morphemes to reflect the orderings of their corresponding heads within the clause.

205

(94) T˚
3
Asp˚ T˚
3
V˚ Asp˚

On the other hand, the Coalescence approach to head-bundling produces a single terminal
node that contains the features of each of the bundled heads. Because this structure makes
no predictions as to the orderings of the morphological exponents of these features, it
requires all constraints on ordering to be done after Spell-Out, within the PF branch.

(95) V/Asp/T˚
[V
D
]
[Asp
R
]
[Tense
R
]

If there are empirical advantages to the existence of branching structure, this would be a
shortcoming of the Coalescence proposal put forth. However, I will present evidence
from existing literature that affix ordering is determined by a variety of competing
preferences (McCarthy and Prince 1993; Hyman 2003; Ussishkin 2007; Rice 2011), and
that a satisfactory theory of affix ordering need not require featurally complex heads to be
branching constituents.
Although complex heads in many languages do obey the Mirror Principle and Right
Hand Head Rule, there are many attested exceptions to these principles. For instance,
while there appears to be a general preference for suffixation across languages, certain
affixes must be specified as prefixes, contra the Right Hand Head Rule. However, it is
possible to maintain the proposal that complex heads do have internal branching structure
in their syntactic representation, by positing that affixes are reordered in the postsyntactic

206
component. This is the approach used in DM, in which phonological restrictions such as
constraints on the attachment direction of affixes can trigger the operation Local
Dislocation, which alters the linear ordering of sister nodes prior to their pronunciation.

(96) Asp˚
3 Local Dislocation [Asp - V]
V˚ Asp˚ >

If Local Dislocation applies between sister nodes of complex heads from the bottom up,
this V/Asp/T˚ head can be given the following orderings: [[V-Asp]-T], [[Asp-V]-T], [T-
[V-Asp]], [T-[Asp-V]]. However, it is not possible to generate orders where the Mirror
Principle ordering is violated among the affixes on a single side of the root (Harley
2013), in particular *[Asp-T-V], *[V-T-Asp]. To the extent that this generalization is
true, this would support the existence of a branching head-adjunction structure within the
syntax.
However, the Mirror Principle of affix ordering is also not without exceptions, and
languages can place restrictions on the orderings of morphemes that contradict what is
expected based on their scopal or derivational properties. The subsequent examples in
this section are taken from Rice's (2011) overview of factors influencing affix ordering.
A well-known Mirror Principle violation is described for Bantu languages by Hyman
(2003) and Good (2003), in which derivational suffixes largely occur in a fixed 'CARP'
order. Notably, this results in surface affix orderings that violate the Mirror Principle. In
Chichewa, for example, both causativized applicatives and applicativized causatives
occur with the same suffix order V-CAUS-APPL, creating a mismatch in scope and linear
ordering for causativized applicatives.

207

(97) 'CARP' template: Verb stem - Causative - Applicative - Reciprocal - Passive

(98) a. lil-its-il b. takas-its-il
cry-CAUS-APPL stir.with-CAUS-APPL
'cause to cry with' 'cause to stir with'
[[cause cry] with] [[stir with] cause] (Hyman 2003)

Furthermore, Mirror-Principle preferences on affix ordering can be violated in favor of
more apparently phonological restrictions. For instance, suffix order in Pulaar (a.k.a.
Fula) generally follows a fixed 'TDNR' order in which more sonorous suffixes occur
further from the root (Arnott 1970; Paster 2005, 2006). While this ordering typically
obeys the Mirror Principle by default, and some ordering reversals are permitted to
directly reflect semantic scope (Paster 2005, 2006), certain fixed orders do lead to Mirror
Principle violations. For instance, the modal suffix ir, which introduces a
manner/instrumental argument, and the repetitive aspect suffix it occur in a fixed order,
even if repetitive aspect takes scope over the modal (99b).

(99) a. o udd-it-ir-ii baafal ŋgal sawru woɗndu
3.SG close-REPET-MODAL-PST door the stick different
'He closed the door again with a different stick'

b. mi udd-it-ir-ii baafal ŋgal sawru
3.SG close-REPET-MODAL-PST door the stick
'I closed the door with a stick again.' (same stick) (Paster 2005)

Lastly, affix ordering and placement in some instances show sensitivity to prosodic
properties of stems (Ussishkin 2007; McCarthy and Prince 1993). If prosodic structure is
built on the PF branch, this indicates clearly that processes that affect affix ordering apply
after syntactic structure undergoes Spell-Out.

208
The generalization that emerges is that affix ordering is determined by a number of
competing functional preferences (cf. Manova and Aronoff 2010, Rice 2011 for an
overview), including those based on parsimony with syntactic derivations or semantic
scope, phonological well-formedness, or morpheme-specific restrictions. While the
Mirror Principle and Right Hand Head Rule do appear to be functionally grounded
preferences for affix linearization, they can be overridden by competing factors. This
provides support for constraint-based theories of affix ordering, in which affix order is
determined by the interaction of constraints grounded in both syntactic and phonological
principles (Hyman 2003; Paster 2009; McPherson and Paster 2009; Caballero 2010, a.o.)
similar to this thesis' proposal on copy pronunciation made in chapters 2 and 3. To
conclude, there is no empirical necessity based on patterns of morphological exponence
for internal branching structure for heads in their syntactic representation.

5.6 Coalescence and the architecture of the grammar

We now return to issues of how dominance and recession are defined, how these
specifications are acquired, and the implications of the proposals for the structure of the
grammar. First, consider the proposed criteria for distinguishing between dominant and
recessive category features, repeated in (100).

(100) A category feature [F] is dominant if:
a. A head containing only [F] has a non-null exponent in its phonological
representation.
OR
b. A head containing only [F] can trigger phrasal movement to its specifier.
Otherwise, [F] is recessive.

209
Condition (100a) ensures that a category feature whose exponents are obligatorily affixal
will be recessive, and thus subject to Coalescence. An obligatorily affixal item can be
understood as one that requires linear adjacency to the exponents of a different feature,
and is associated with some preference for left or right attachment (McCarthy & Prince
1993). Crucially, this is to be distinguished from items that are affixal in terms of their
prosodic representation, but are free of such requirements of adjacency. As an example,
the exponent of the declarative Force˚ head in Bangla, je, does not form a PWd, and is
directly dominated by the PPh headed by an adjacent item, in an affixal clitic structure.
Nonetheless, its corresponding head is dominant because its realization does not depend
on linear adjacency with the exponents of a particular morpheme.
The language learner can thus determine whether or not a category feature is
dominant or recessive based on properties of its phonological exponent. Once a learner
has identified a series of segments as being the exponents of a relevant syntactic feature,
he needs only to determine whether or those segments systematically require linear
adjacency with the exponents of a different feature. If so, the feature is identified as
recessive, otherwise it is analyzed as dominant.
Because dominance and recession are proposed to drive an operation that applies
within syntax, these properties must be present within lexical entries. This runs counter to
the principle of Late Insertion in the form proposed by Halle and Marantz (1993), which
states that all phonological information enters the derivation after syntactic structures are
transferred to the PF interface. There are indeed many cases where Late Insertion seems
to be a correct generalization. First, it accounts for the observation that syntax appears to
be generally blind to phonological properties, particularly at the segmental level. More

210
importantly, it accounts for patterns in which morpheme realization is sensitive to some
syntactic context, which are unexpected for traditional lexicalist theories in which
morpheme insertion takes place prior to syntactic operations.
The present account can be reconciled with the above generalizations by proposing
that lexical insertion takes place in multiple steps. A limited amount of information about
phonological realization (e.g. dominance vs. recession) is supplied within lexical entries,
which allows syntactic operations to build a hiearchical structure that serves as a
"scaffold" for the full range of phonological properties (segmental and prosodic structure)
introduced in the PF interface.

5.7 Conclusion

In this chapter, I have proposed that a variety of bundling processes that affect heads
should be understood as the result of a single syntactic operation, Coalescence. Although
Coalescence presents a non-trivial addition to the set of syntactic operations permitted in
standard Minimalism, it reduces the need to specify separate bundling operations in both
the lexicon and in the postsyntactic PF branch. In addition, it allows for a unified analysis
of phenomena that received disparate analyses under previous theories.

211
Chapter 6: Conclusion

6.1 Overview of results

This dissertation has presented theoretical analyses of several cases of word order
variation involving heads in the clausal left periphery and the inflectional domain. In
particular, it has argued that an understanding of a variety of phonological properties is
crucial to the analyses of these patterns.
In the first section of the work, I presented an empirical study of syntactic and
prosodic properties of embedded clauses and the complementizer je in Bangla. I have
argued that the ordering restrictions on je receive a straightforward account if [1] the
complementizer head-moves through the expanded left periphery and [2] the
pronunciation of its movement copies is determined by the interaction of violable
syntactic and prosodic well-formedness constraints. The analysis integrates the insights
of multiple theoretical frameworks, including the Copy Theory of Movement, the
Cartographic Approach to syntactic structures, and Prosodic Phonology.
The second half of the dissertation focused on the analysis of "bundling" processes
that affect heads in syntax. I have proposed that a variety of bundling effects are best
attributed to an operation that applies during the syntactic derivation, Coalescence.
Furthermore, the application of Coalescence is conditioned by a requirement to eliminate
heads that would be affixal or null in theirphonological representation, prior to the
transfer of syntactic structures to the PF interface. A crucial implication of this approach
is that a limited amount of phonological information, implemented as the distinction

212
between dominance and recession, is visible to the syntax. This allows for a unified
account of apparent variation in the inventory of instantiated functional projections and
head-adjunction structures, previously attributed to separate components of the grammar.
Lastly, I have shown that Coalescence resolves longstanding problems that traditional
head-adjunction structures have posed within Minimalist syntax.

6.2 Theoretical Implications

The dissertation concludes with a discussion of the primary theoretical implications of the
presented analyses with respect to theories of syntax and the syntax-phonology interface.

6.2.1 Constraint interaction in copy linearization

On the basis of the analysis of Bangla complementizer order, I have argued that syntactic
preferences on linearization can be violated in order to satisfy phonological requirements,
and that this interaction within the PF component can be formalized as an Optimality-
Theoretic grammar. While OT analyses of the influence of prosody on word order have
been previously proposed (Samek-Lodovici 2005; Vogel 2006; Anttila et al. 2010; Elfner
2012; Clemens 2014; Bennett et al. 2016), the present model differs from these analyses
by proposing that the PF grammar only manipulates word order by determining the
pronunciation of movement copies. The model is thus provides a more restrictive theory
of possible phonologically-driven displacements. On the other hand, the lower copy
pronunciation proposal allows for an account of "long-distance" displacements in which

213
the postposed item is separated from its expected position by more than two phrasal
constituents. Such cases do not receive an explanation in terms of the operations of
Distributed Morphology, typically assumed to require structural adjacency. The analysis
further supports a view of the PF grammar in which linearization, prosodic structure
formation, and morpheme insertion take place in parallel.

6.2.2 The effects of phonological defectivity on syntactic operations

In this dissertation, I have examined two stages of the derivation at which different types
of phonological properties can influence word order. In Chapters 2 and 3, I presented
Bangla data to argue that the optimal pronunciation of the complementizer je's movement
copies is determined in part by constraints on surface prosody. Lower copy pronunciation
is argued to take place to prevent the violation of a STRONGSTART constraint at
intonational phrase edges. Furthermore, prosodic data is used to argue against previous
analyses of je as an "obligatory enclitic." This suggests also that similar reassessments
may be possible for other analyses that stipulate the attachment directionality of
prosodically weak items.
In chapter 5, I argued that a more abstract notion of defectivity, recession, triggers the
head-bundling syntactic operation Coalescence. This allows a unified analysis of
phenomena that were previously attributed to separate components of the grammar,
Feature Scattering and Morphological Merger. Furthermore, Coalescence allows for a
new account of head movement in syntax, avoiding the many problems posed by
traditional head-adjunction models. This analysis, however, requires some information

214
about phonological exponence to be accessible to syntactic operations, contra traditional
interpretations of Late Insertion (Halle and Marantz 1993) and Phonology-Free Syntax
(Zwicky and Pullum 1986; Pullum and Zwicky 1989). After all, if features related to
semantic interpretations at LF are accessible to syntactic operations, there is no principled
reason to prevent properties of phonological form from being accessible as well, since the
output of syntax must be convergent at both the LF and PF interfaces (for similar
arguments, see Roberts 2005).

6.2.3 Variation in the instantiation of functional projections

While research in the Cartographic Program has identified a growing number of
functional projections needed to account for extensive cross-linguistic variation in the left
periphery, it has proven difficult to reconcile the articulated structure with languages that
restrictively instantiate left-peripheral positions. Coalescence provides a straightforward
means to account for varying degrees of articulation both across and within languages,
while maintaining the principle of "one head, one feature" on items in the numeration.
The application of a bundling operation like Coalescence leads to the elimination of
many phonologically null projections from proposed syntactic representations. This in
itself may have independent theoretical benefits. One area in which bundled projections
may be crucial is in accounting for movement restrictions attributed to Spec-to-Spec
Anti-Locality, a proposed constraint that prevents A-bar movement from displacing a
constituent from a specifier position to the specifier of the projection that immediately
dominates it (Erlewine 2014).

215
(1) [
CP
XP
k
C˚ [
TP
XP
k
T˚ [ ... ]]] Violates anti-locality

(2) [
CP
XP
k
C˚ [intervening material [
TP
XP
k
T˚ [ ... ]]] Satisfies anti-locality

Generally, Spec-to-Spec Anti-Locality accounts for bans against movements that are "too
short," typically cases of subject movement into CP (Erlewine 2014; Douglas 2015).
Crucially, some of these restrictions may be impossible or more difficult to formulate in
terms of Anti-Locality if all clauses contain a fully articulated Cartographic structure with
many null functional projections. Of course, it remains open as to whether Anti-locality is
the best characterization of these movement constraints. Nonetheless, it appears that the
relevant generalization could be difficult to account for given a proliferation of
phonologically null projections.

6.3 Directions for future research

While it is outside the goal of this work to provide a comprehensive account of the
influence of prosodic requirements on word order, it provides new predictions to be
tested. A major goal of future work is to determine the precise extent to which
phonological restrictions can affect word order. It has been argued, for instance, that
constraining prosodically motivated displacements to those that can be created by
manipulating copy pronunciation may be overly restrictive if displacements that violate
syntactic constraints on movement are systematically observed (Agbayani and Golston
2010; Agbayani et al. 2011, 2014; Bennett et al. 2016). In addition, the validity and

216
generalizability of the proposals in Chapters 2 and 3 should be tested within a broader
study of the ordering and prosodic phrasing of complementizers, in particular within
other languages that show variable complementizer order.
With respect to Coalescence, several questions remain open. While I have argued that
Coalescence is motivated by a requirement to eliminate recessive heads from the final
representation, no conclusion has been reached as to the specific points in this derivation
at which this requirement holds. In a theory that assumes cyclic Spell-Out (Chomsky
2000, 2001, 2008, among many others), we expect recessive heads to be eliminated at the
point at which they are spelled out. For instance, if phasal heads trigger the Spell-Out of
their complements, we expect the highest head in this complement domain to be
obligatorily dominant. Furthermore, it should not be possible for Coalescence to bundle a
phasal head with the head of its complement. However, given the unresolved issues
regarding what constitutes syntactic phasehood and the timing which with syntactic
structures undergo Spell-Out, the verification of these predictions is left to future
research.
Similarly, the analysis can potentially be extended to account for languages that
compel certain projections to have an overtly realized head, or a filled specifier. For
instance, An (2007) argues that in English, CPs whose left edges are aligned with an
intonational phrase edge must contain an overt head or specifier. In a more complex case,
Kandybowicz (2009) shows that subject extraction out of embedded clauses is possible
only if the embedded TP is realized with an overt head or filled specifier. Martins (2004)
argues that emphatic affirmative clauses in European Portuguese and Brazilian
Portuguese require the overt realization of both a complementizer head and a Polarity

217
head, and that this can be satisfied by a variety of strategies, including verb movement
and the direct Merge of a polarity-encoding adverbial. Potentially, such cases can be
explained in terms of requirements for certain heads to be dominant. However, much
work needs to be done in order to explain how such requirements come about.
It also remains an open question as to whether there are additional constraints on the
distribution of dominance and recession across different syntactic categories. For
instance, given the understanding that dominant heads are those that are either non-affixal
or able to have a specifier, we might expect lexical heads to be universally dominant. In
addition, we have yet to investigate whether suprasegmental properties like stress or
prosodic phrasing can introduce additional constraints on the distribution of dominance
and recession.
Lastly, if we are correct that the bundling of heads is performed during the syntactic
derivation and triggered by requirements on PF well-formedness, we can wonder if other
syntactic operations like movement or agreement can have prosodic motivations.
Interestingly, Richards (2010, 2016) provides numerous arguments to suggest that certain
requirements of prosodic prominence like stress placement and prosodic boundary
marking, can trigger syntactic movement, suggesting as well that some lexical insertion
takes place prior to spell-out at PF.
To conclude, the claims of this dissertation point the way to numerous avenues for
further exploration, and it is my hope that the investigation of these issues will lead to a
greater understanding of the organization of the language faculty.

218
References
ABOH, ENOCH OLADÉ. 2006. Complementation in Saramaccan and Gungbe: The case of
C-type modal particles. Natural Language & Linguistic Theory 24.1–55.
ADAMS, MARIANNE PATALINO. 1987. Old French, null subjects, and verb second
phenomena. Los Angeles: University of California, Los Angeles dissertation.
ADGER, DAVID. 2003. Core Syntax : A Minimalist approach. Oxford: Oxford University
Press.
AGBAYANI, BRIAN; and CHRIS GOLSTON. 2010. Phonological movement in Classical
Greek. Language 86.133–167.
AGBAYANI, BRIAN; CHRIS GOLSTON; and DASHA HENDERER. 2011. Prosodic movement.
Proceedings of the 28th West Coast Conference on Formal Linguistics, ed. by Mary
Byram Washburn, Katherine McKinney-Bock, Erika Varis, Ann Sawyer, and
Barbara Tomaszewicz, 231–239. Somerville, MA: Cascadilla Proceedings Project.
AGBAYANI, BRIAN; CHRIS GOLSTON; and TORU ISHII. 2014. Syntactic and prosodic
scrambling in Japanese. Natural Language & Linguistic Theory 33.47–77.
ALEXIADOU, ARTEMIS. 2006. Left Dislocation (including CLLD). The Blackwell
Companion to Syntax, ed. by Martin Everaert and Henk van Riemsdijk, 668-699.
Malden, MA: Blackwell.
AN, DUK-HO. 2007. Syntax at the PF Interface: Prosodic mapping, linear order, and
deletion. Storrs, CT: University of Connecticut dissertation.
ANAGNOSTOPOULOU, ELENA. 1997. Clitic left-dislocation and contrastive left-dislocation.
Materials on left-dislocation, ed. by Elena Anagnostopoulou, Henk van Riemsdijk,
and Frans Zwarts, 151–192. Amsterdam: John Benjamins.
ANDERSON, STEPHEN R. 2008. English reduced auxiliaries really are simple clitics.
Lingue e Linguaggio 7.169–186.
ANTTILA, ARTO; MATTHEW ADAMS; and MICHAEL SPERIOSU. 2010. The role of prosody
in the English dative alternation. Language and Cognitive Processes 25.946–981.
ARNOTT, D. W. 1970. The nominal and verbal systems of Fula. Oxford: Clarendon Press.
BAKER, MARK. 1985. The mirror principle and morphosyntactic explanation. Linguistic
Inquiry 16.373–415.
BAKER, MARK. 1988. Incorporation: A theory of grammatical function changing.
Chicago: University of Chicago Press.

219
BAYER, JOSEF. 1984. Comp in Bavarian syntax. The Linguistic Review 3.209–274.
BAYER, JOSEF. 1996. Directionality and Logical Form. Boston: Kluwer.
BAYER, JOSEF. 2001. Two grammars in one: sentential complements and
complementizers in Bengali and other South-Asian languages. The Yearbook of
South Asian Languages: Tokyo Symposium on South Asian Languages - Contact,
Convergence, and Typology, ed. by P Bhaskarorao and K. V. Subbarao, 11–36. New
Delhi: Sage Publications.
BAYER, JOSEF; and PROBAL DASGUPTA. to appear. Emphatic Topicalization and the
structure of the left periphery : Evidence from German and Bangla. Syntax.
BECH, KRISTIN. 1998. Pragmatic factors in language change: XVS and XSV clauses in
Old and Middle English. Folia Historica Linguistica 19.79–102.
BECKMAN, MARY; and JULIA HIRSCHBERG. 1994. The ToBI Annotation Conventions.
Columbus, OH: Ohio State University manuscript.
BENINCÀ, PAOLA. 1983. Un’ipotesi sulla sintassi delle lingue Romanze medievali.
Quaderni Patavini Di Linguistica 4.3–19.
BENINCÀ, PAOLA. 2006. A detailed map of the left periphery of medieval Romance.
Negation, Tense and Clausal Architecture: Cross-linguistic Investigations, ed. by
Raffaella Zanuttini, Héctor Campos, Elena Herburger, and Paul H. Portner, 53–86.
Washington, D.C.: Georgetown University Press.
BENINCÀ, PAOLA; and CECILIA POLETTO. 2004. Topic, Focus, and V2. The Structure of
CP and IP, ed. by Luigi Rizzi, 52–75. Oxford: Oxford University Press.
BENNETT, RYAN; EMILY ELFNER; and JAMES MCCLOSKEY. 2016. Lightest to the right:
An apparently anomalous displacement in Irish. Linguistic Inquiry 47.169–234.
BERMÚDEZ-OTERO, RICARDO. 1999. Constraint interaction in language change: quantity
in English and Germanic. Manchester and Santiago de Compostela: University of
Manchester and University of Santiago de Compostela dissertation.
BERMÚDEZ-OTERO, RICARDO. 2012. The architecture of grammar and the division of
labour in exponence. The morphology and phonology of exponence, ed. by Jochen
Trommer, 8–83. Oxford: Oxford University Press.f
BHATT, RAJESH; and VENEETA DAYAL. 2007. Rightward scrambling as rightward
remnant movement. Linguistic Inquiry 38.287–301.
BHATT, RAKESH. 1999. Verb movement and the syntax of Kashmiri. Dordrecht: Kluwer.

220
BHATTACHARYA, TANMOY. 1999. The structure of the Bangla DP. London: University of
London dissertation.
BHATTACHARYA, TANMOY. 2001. Peripheral and clause-internal complementizers in
Bangla : A case for remnant movement. Proceedings of WECOL 2000 12.1–14.
BIANCHI, VALENTINA. 1999. Consequences of Antisymmetry: Headed relative clauses.
Berlin: Mouton de Gruyter.
BIBERAUER, THERESA; ANDERS HOLMBERG; and IAN ROBERTS. 2014. A syntactic
universal and its consequences. Linguistic Inquiry 45.169–225.
BIBERAUER, THERESA; and IAN G. ROBERTS. 2015. The clausal hierarchy, features, and
parameters. Beyond functional sequence: The cartography of syntactic structures,
Volume 10, 295–313. Oxford: Oxford University Press.
BIBERAUER, THERESA; and MICHELLE SHEEHAN. 2012. Disharmony, antisymmetry, and
the Final-over-Final Constraint. Ways of Structure Building, ed. by Myriam Uribe-
Etxebarria and Vidal Valmala, 206–244. Oxford: Oxford University Press.
BICKMORE, LEE S. 1990. Branching nodes and prosodic categories: Evidence from
Kinyambo. The Phonology-Syntax Connection, 1–18. Chicago: University of
Chicago Press.
BIDESE, ERMENEGILDO. 2008. Die diachronische Syntax des Zimbrischen. Tübingen:
Günter Narr.
BIDESE, ERMENEGILDO; FEDERICA COGNOLA; and ANDREA PADOVAN. 2012. Zu einer
neuen Verb-Zweit-Typologie in den germanischen Sprachen : der Fall des
Zimbrischen und des Fersentalerischen. In simplicitate complexitas. Festgabe für
Barbara Stefan zum 70, ed. by Peter Anreiter, Ivi Hajnal, and Manfred
Kienpointner, 69–86. Vienna: Praesens.
BIDESE, ERMENEGILDO; ALESSANDRA TOMASELLI; and ANDREA PADOVAN. 2016. V2 and
Nominative Case: The view from a German-based language-contact dialect in Italy.
Paper presented at Rethinking Verb Second, University of Cambridge.
BLUMENFELD, LEV A. 2006. Constraints on phonological interactions. Stanford: Stanford
University dissertation.
BOBALJIK, JONATHAN DAVID. 1995. Morphosyntax: The syntax of verbal inflection.
Cambridge, MA: Massachusetts Institute of Technology dissertation.
BOBALJIK, JONATHAN DAVID. 2002. A-Chains at the PF-Interface : Copies and 'covert'
movement. Natural Language & Linguistic Theory 20.197–267.

221
BOBALJIK, JONATHAN DAVID; and HÖSKULDUR THRÁINSSON. 1998. Two heads aren’t
always better than one. Syntax.37–71.
BOECKX, CEDRIC; and KLEANTHES GROHMANN. 2007. Putting phases in perspective.
Syntax 10.204–222.
BOECKX, CEDRIC; and KLEANTHES K. GROHMANN. 2004. Left dislocation in Germanic.
Focus on Germanic typology, ed. by Werner Abraham, 131–144. Berlin: Akademie-
Verlag.
BOECKX, CEDRIC; and SANDRA STJEPANOVIC. 2001. Head-ing toward PF. Linguistic
Inquiry 32.345–355.
BOERSMA, PAUL; and DAVID WEENINK. 2011. Praat: doing phonetics by computer
(version 5.2.23). Computer Program. http://www.praat.org/.
BORSLEY, ROBERT D.; MARIA-LUISA RIVERO; and JANIG STEPHENS. 1996. Long head
movement in Breton. The syntax of the Celtic languages: a comparative perspective,
ed. by Robert Borsley and Ian G. Roberts, 53–74. Cambridge: Cambridge University
Press.
BOS ̌KOVIĆ, ŽELJKO. 1995. Participle movement and second position cliticization in
Serbo-Croatian. Lingua 96.245–266.
BOŠKOVIĆ, ŽELJKO. 2001. On the nature of the syntax-phonology interface: Cliticization
and related phenomena. Oxford: Elsevier.
BOŠKOVIĆ, ŽELJKO. 2002. On multiple Wh-fronting. Linguistic Inquiry 33.351–383.
BOŠKOVIĆ, ŽELJKO; and LASNIK. 2003. On the distribution of null complementizers.
Linguistic Inquiry 34.527–546.
BOŠKOVIĆ, ŽELJKO; and JAIRO NUNES. 2007. The copy theory of movement: A view
from PF. The Copy Theory of Movement, ed. by Norbert Corver and Jairo Nunes,
13–74. Amsterdam: John Benjamins.
BOUZET, JEAN. 1932. Les particules enonciatives du beamais. Oc 8.40–53.
BRANIGAN, PHIL. 2011. Provocative syntax. Cambridge, MA: MIT Press.
BRESNAN, JOAN. 1972. Theory of complementation in English syntax. Cambridge, MA:
Massachusetts Institute of Technology dissertation.
BRODY, MICHAEL. 1995. Lexico-Logical Form - A radically Minimalist theory.
Cambridge, MA: MIT Press.

222
BRUENING, BENJAMIN. 2009. Selectional asymmetries between CP and DP suggest that
the DP Hypothesis is wrong. University of Pennsylvania Working Papers in
Linguistics 15.26–35.
CABALLERO, GABRIELLA. 2010. Scope, phonology and morphology in an agglutinating
language: Choguita Rarámuri (Tarahumara) variable suffix ordering. Morphology
20.165-204.
CAHA, PAVEL. 2013. Explaining the structure of case paradigms by the mechanisms of
Nanosyntax: The Classical Armenian nominal declension. Natural Language and
Linguistic Theory 31.1015–1066.
CAMPOS, HÉCTOR. 1992. Enunciative elements in Gascon. Linguistics 30.911–940.
CARDINALETTI, ANNA. 2010. On a (wh-)moved topic in Italian, compared to Germanic.
Advances in comparative Germanic syntax, ed. by Artemis Alexiadou, Jorge
Hankamer, Thomas McFadden, Justin Nuger, and Florian Schaeffer, 3–40.
Amsterdam: John Benjamins.
CARDINALETTI, ANNA; and IAN G. ROBERTS. 1991. Clause structure and X-second.
Venice and Geneva: University of Venice and University of Geneva manuscript.
CASALICCHIO, JAN; and FEDERICA COGNOLA. 2015. On the left periphery of relaxed V2
languages: A comparison between Rhaetoromance and Mòcheno. Trento:
University of Trento manuscript.
CHACÓN, DUSTIN ALFONSO. 2014. The antisymmetric analysis of Bangla
complementation. College Park, MA: University of Maryland manuscript.
CHOI, HYE-WON. 1999. Optimizing structure in context: Scrambling nd information
structure. Stanford: CSLI Publications.
CHOMSKY, NOAM. 1965. Aspects of the theory of syntax. Cambridge, MA: MIT Press.
CHOMSKY, NOAM. 1973. Conditions on transformations. A festschrift for Morris Halle,
ed. by Stephen R. Anderson and Paul Kiparsky, 232–286. New York: Holt, Rinehart
and Winston.
CHOMSKY, NOAM. 1981. Lectures on government and binding: The Pisa lectures.
Dordrecht: Foris.
CHOMSKY, NOAM. 1993. A minimalist program for linguistic theory. The view from
Building 20, ed. by Kenneth Hale and Samuel Jay Keyser, 1–52. Cambridge, MA:
MIT Press.

223
CHOMSKY, NOAM. 1994. Bare Phrase Structure. MIT Occasional Papers in Linguistics
5.1–15.
CHOMSKY, NOAM. 1995a. The Minimalist Program. Cambridge, MA: MIT Press.
CHOMSKY, NOAM. 2000. Minimalist inquiries: the framework. Step by Step: essays on
Minimalist syntax in honour of Howard Lasnik, ed. by Roger Martin, David
Michaels, Juan Uriagereka, and Samuel Jay Keyser, 89–115. Cambridge, MA: MIT
Press.
CHOMSKY, NOAM. 2001. Derivation by phase. Ken Hale: A life in language, ed. by
Michael Kenstowicz, 1–52. Cambridge, MA: MIT Press.
CHOMSKY, NOAM. 2008. On phases. Foundational Issues in Linguistic Theory, ed. by
Robert Freidin, Carlos P. Otero, and Maria Luisa Zubizarreta, 133–166. Cambridge,
MA: MIT Press.
CHOMSKY, NOAM. 2013. Problems of projection. Lingua 130.33–49.
CHOMSKY, NOAM; and HOWARD LASNIK. 1977. Filters and control. Linguistic Inquiry
8.425–504.
CHOUDHURY, ARUNIMA. 2010. A study of focus in Bangla. Los Angeles: University of
Southern California masters thesis.
CINQUE, GUGLIELMO. 1999. Adverbs and functional heads: a cross-linguistic perspective.
Oxford: Oxford University Press.
CINQUE, GUGLIELMO; and LUIGI RIZZI. 2010. The cartography of syntactic structures.
The Oxford handbook of linguistic analysis, ed. by Bernd Heine and Heiko Harrog,
51–65. Oxford: Oxford University Press.
CLEMENS, LAUREN EBY. 2014. Prosodic noun incorporation and verb-initial syntax.
Cambridge, MA: Harvard University dissertation.
COGNOLA, FEDERICA. 2013. Syntactic variation and verb second: A German dialect in
northern Italy. Amsterdam/Philadelphia: John Benjamins.
CRAENENBROECK, JEROEN VAN (ed.) 2009. Alternatives to Cartography. Berlin: De
Gruyter Mouton.
DASGUPTA, PROBAL. 1980. Questions and relative and complement clauses in a Bangla
grammar. New York: New York University dissertation.
DAVID, ANNE BOYLE. 2015. Descriptive grammar of Bangla. Berlin: De Gruyter Mouton.

224
DE HOOP, HELEN. 1992. Case configuration and noun phrase interpretation. Groningen:
University of Groningen dissertation.
DE LACY, PAUL. 2002. The formal expression of markedness. Phonology 21.145–199.
Amherst, MA: University of Massachusetts, Amherst dissertation.
DE LACY, PAUL. 2004. Markedness conflation in Optimality Theory. Phonology 21.145–
199.
DEMONTE, VIOLETA; and OLGA FERNÁNDEZ-SORIANO. 2009. Force and finiteness in the
Spanish complementizer system. Probus 21.23–49.
DEN BESTEN, HANS. 1983. On the interaction of root transformations and lexical deletive
rules. On the formal syntax of the Westgermania, ed. by Werner Abraham, 47–131.
Amsterdam: John Benjamins.
DI SCIULLO, ANNA MARIA; and DANIELA ISAC. 2008. The asymmetry of Merge.
Biolinguistics 2.260–290.
DIESING, MOLLY. 1992. Indefinites. Cambridge, MA: MIT Press.
DIESING, MOLLY. 2005. The upper functional domain in Yiddish. Focus on Germanic
typology, ed. by Werner Abraham, 195–209. Berlin: Akademierverlag.
DIESING, MOLLY; and ZEC. 2011. Interface effects : Serbian clitics. Experiments at the
interfaces, syntax and semantics volume 37, ed. by Jeffrey Runner, 1–30. Leiden:
Brill.
DOUGLAS, JAMIE. 2015. The that-trace and anti-that-trace Effects: Unification and
theoretical implications. Paper presented at the 46th Meeting of the North East
Linguistic Society, Concordia University.
DOURADO, LUCIANA. 2002. Construções aplicativas em Panará. D.E.L.T.A 18.67–86.
DUNIGAN, MELYNDA B. 1994. On the clausal structure of Wolof. Chapel Hill, N.C.:
University of North Carolina at Chapel Hill dissertation.
DUPUIS, FERNANDE. 1989. L’Expression du sujet dans les subordonnées en ancien
Français. Montréal: Université de Montréal dissertation.
ELFNER, EMILY. 2010a. The interaction of linearization and prosody: Evidence from
pronoun postposing in Irish. Formal approaches to Celtic linguistics, ed. by Andrew
Carnie, 1–34. Cambridge: Cambridge Scholars Publishing.

225
ELFNER, EMILY. 2010b. Recursivity in prosodic phrasing: Evidence from Conamara Irish.
NELS 40: Proceedings of the 40th Meeting of The North East Linguistic
Society.191-204.
ELFNER, EMILY. 2012. Syntax-prosody interactions in Irish. Amherst, MA: University of
Massachusetts, Amherst dissertation
ELORDIETA, GORKA. 2015. Recursive phonological phrasing in Basque. Phonology 32
.49–78.
EMBICK, DAVID. 2003. Linearization and local dislocation: derivational mechanics and
interactions. Linguistic Analysis 33.303–336.
EMBICK, DAVID. 2010. Localism versus globalism in morphology and phonology.
Cambridge, MA: MIT Press.
EMBICK, DAVID; and ROUMYANA IZVORSKI. 1995. On long head movement in Bulgarian.
ESCOL ’94, ed. by Janet M. Fuller, Ho Han, and David Parkinson, 104–115. Ithaca,
NY: CLC Publications.
EMBICK, DAVID; and ROLF NOYER. 2001. Movement operations after syntax. Linguistic
Inquiry 32.555–595.
ERLEWINE, MICHAEL YOSHITAKA. 2014. Anti-locality and optimality in Kaqchikel Agent
Focus. Natural Language & Linguistic Theory 34.429-479.
FANSELOW, GISBERT. 2004. Cyclic phonology-syntax interaction: Movement to first
position in German. Working papers of the SFB 632. Interdisciplinary studies on
information structure 1, ed. by Shinichiro Ishihara and Michaela Schmitz:1–42.
Potsdam: Universitätsverlag.
FANSELOW, GISBERT; and DENISA LENERTOVÁ. 2010. Left peripheral focus: mismatches
between syntax and information structure. Natural Language & Linguistic Theory
29.169–209.
FÉRY, CAROLINE. 2010. Indian languages as intonational “phrase languages.” Festschrift
to honour Ramakant Agnihotri, ed. by I. Hasnian and S. Chaudhury. Delhi: Aakar
Publisher.
FITZPATRICK-COLE, JENNIFER. 1991. The minimal word in Bengali. Proceedings of the
Ninth West Coast Conference on Formal Linguistics, ed. by Aaron Halpern. 157–
170. Stanford: CSLI.
FITZPATRICK-COLE, JENNIFER. 1994. The Prosodic Domain Hierarchy in reduplication.
Stanford: Stanford University dissertation.

226
FLACK, KATHRYN. 2009. Constraints on onsets and codas of words and phrases.
Phonology 26.269–302.
FOULET, LUCIEN. 1928. Petite syntaxe de l’ancien Français. Paris: Editions Champion.
FRANCO, IRENE. 2009. Verbs, subjects and stylistic fronting. Siena: University of Siena
dissertation.
FRANKS, STEVEN. 1998. Clitics in Slavic. Paper presented at the Comparative Slavic
Morphosyntax Workshop, Spencer Creek, Indiana
FREY, WERNER. 2004. Notes on the syntax and the pragmatics of German Left
Dislocation. The syntax and semantics of the left sentence periphery, ed. by Horst
Lohnstein and Susanne Trissler, 203–233. Berlin: Mouton de Gruyter.
GEORGI, DOREEN; and GEREON MÜLLER. 2010. Noun phrase structure by reprojection.
Syntax 13.1–36.
GIORGI, ALESSANDRA. 2010. About the speaker: Towards a syntax of indexicality.
Oxford: Oxford University Press.
GIORGI, ALESSANDRA; and FABIO PIANESI. 1996. Tense and aspect: From semantics to
morphosyntax. New York: Oxford University Press.
GNANADESIKAN AMALIA. 1996. Phonology with ternary scales. Amherst, MA: University
of Massachusetts, Amherst dissertation.
GOOD, JEFFREY C. 2003. Strong linearity: Three cases studies towards a theory of
morphosyntactic templatic constructions. Berkeley: University of California,
Berkeley dissertation.
GREWENDORF, GÜNTHER. 2015. Double fronting in Bavarian left periphery. Beyond
functional sequence: The cartography of syntactic structures, volume 10, 232–254.
Oxford: Oxford University Press.
GREWENDORF, GÜNTHER; and CECILIA POLETTO. 2011. Hidden verb second: The case of
Cimbrian, 2011. Studies on German-language islands, ed. by Michael T. Putnam.
Philadelphia: John Benjamins.
GRIMSHAW, JANE. 1979. Complement selection and the lexicon. Linguistic Inquiry
10.279–326.
GRIMSHAW, JANE. 1991. Extended projections. Waltham, MA: Brandeis University
manuscript.

227
GRIMSHAW, JANE. 2000. Extended projection and locality. Lexical specification and
insertion, ed. by Peter Coopmans, Martin Everaert, and Jane Grimshaw, 115–133.
Amsterdam: John Benjamins.
GROAT, ERICH; and JOHN O’NEIL. 1996. Spell-Out at the LF interface. Minimal Ideas:
Syntactic Studies in the Minimalist Framework, ed. by Werner Abraham, Samuel
David Epstein, Höskuldur Thráinsson, and C. Jan-Wouter Zwart, 113–139.
Amsterdam/Philadelphia: John Benjamins.
HAEBERLI, ERIC. 2002a. Inflectional morphology and the loss of V2 in English. Syntactic
effects of morphological change, ed. by David L. Lightfoot, 88–106. Oxford
University Press.
HAEBERLI, ERIC. 2002b. Observations on the loss of Verb Second in the history of
English. Studies in comparative Germanic syntax: Proceedings from the 15th
Workshop on Comparative Germanic Syntax, ed. by C. Jan-Wouter Zwart and
Werner Abraham, 245–272. Amsterdam/Philadelphia: John Benjamins.
HAEGEMAN, LILIANE. 1996. Verb second, the split CP, and null subjects in early Dutch
finite clauses. Geneva Generative Papers 4.135–175.
HAEGEMAN, LILIANE; and JEROEN VAN CRAENENBROECK. 2007. The derivation of
subject-initial V2. Linguistic Inquiry.167–178.
HAEGEMAN, LILIANE; and CIRO GRECO. 2016. V>2. Paper presented at Rethinking Verb
Second, University of Cambridge.
HAIDER, HUBERT. 1988. Matching Projections. Constituent structure: Papers from the
1987 GLOW Conference, ed. by Anna Cardinaletti, Guglielmo Cinque and Giuliana
Giusti, 101-121. Dordrecht: Foris.
HALLE, MORRIS; and ALEC MARANTZ. 1993. Distributed morphology and the pieces of
inflection. The view from Building 20, ed. by Kenneth Hale and Samuel K. Keyser,
111–176. Cambridge, MA: MIT Press.
HALPERN, AARON. 1992. Topics in the placement and morphology of clitics. Stanford:
Stanford University dissertation.
HARGUS, SHARON, and ELLEN KAISSE (eds.). 1993. Studies in Lexical Phonology
(Phonetics and Phonology 4). San Diego: Academic Press.
HARIZANOV, BORIS. 2016. Head movement to specifier positions in Bulgarian participle
fronting. Paper presented at the 90th Annual Meeting of the Linguistic Society of
America, Washington, D.C.

228
HARLEY, HEIDI. 2004. Head movement and conflation. NELS 34: Proceedings of the 34th
Meeting of the North East Linguistic Society, ed. by Keir Moulton and Matthew
Wolf, 239–254. Amherst, MA: GLSA.
HARLEY, HEIDI. 2013. Getting morphemes in order: Merger, affixation and head
movement. Diagnosing Syntax, ed. by Lisa L. Cheng and Norbert Corver, 44–74.
Oxford: Oxford University Press.
HARTMAN, JEREMY. 2011. The semantic uniformity of traces: Evidence from ellipsis
parallelism. Linguistic Inquiry 42.367–388.
HAYES, BRUCE; and ADITI LAHIRI. 1991. Bengali intonational phonology. Natural
Language & Linguistic Theory 9.47–96.
HENDERSON, ROBERT. 2012. Morphological alternations at the intonational phrase edge:
The case of K'ichee'. Natural Language & Linguistic Theory 30.741–787.
HINTERHÖLZL, ROLAND; and SVETLANA PETROVA. 2010. From V1 to V2 in West
Germanic. Lingua 120.315–328.
HOLMBERG, ANDERS. 1986. Word order and syntactic features in the Scandinavian
languages and English. Stockholm: University of Stockholm dissertation.
HOLMBERG, ANDERS. 2015. Verb second. Syntax – an international handbook of
contemporary syntactic research, ed. by Tibor Kiss and Artemis Alexiadou, 342–
383. Berlin: Walter de Gruyter Verlag.
HSU, BRIAN. 2014. Unifying phonotactics and derived environment blocking through
prosodic constraint indexation. Supplemental proceedings of Phonology 2013, ed. by
John Kingston, Claire Moore-Cantwell, Joe Pater and Robert Staubs. Online:
http://journals.linguisticsociety.org/proceedings/index.php/amphonology/article/vie
w/19.
HSU, BRIAN. 2015. Constraining exceptionality as prosody-morphology mismatch: a
study of French nasal vowels. Proceedings of the 49th Annual Meeting of the
Chicago Linguistic Society. Chicago: Chicago Linguistic Society.
HSU, BRIAN; and KAREN JESNEY. 2016. Scalar positional markedness and faithfulness in
Harmonic Grammar. Proceedings of the 51st Annual Meeting of the Chicago
Linguistic Society, 241-255. Chicago: Chicago Linguistic Society.
HYMAN, LARRY M. 2003. Suffix Ordering in Bantu: A morphocentric approach.
Yearbook of Morphology 2002, ed. by Geert Booij and Jaap van Marle, 245–281.
Dordrecht: Springer.
IATRIDOU, SABINE. 1990. About Agr(P). Linguistic Inquiry 21.551–577.

229
ITÔ, JUNKO; and ARMIN MESTER. 2009. The onset of the prosodic word. Phonological
argumentation: Essays on evidence and motivation, ed. by Steve Parker, 227–260.
London: Equinox.
ITÔ, JUNKO; and ARMIN MESTER. 2013. Prosodic subcategories in Japanese. Lingua
124.20–40.
JOUITTEAU, MÉLANIE. 2010. A typology of V2 with regard to V1 and second position
phenomena: An introduction to the V1/V2 volume. Lingua 120.197–209.
JULIEN, MARIT. 2002. Syntactic heads and word formation: A study of verbal inflection.
New York: Oxford University Press, USA.
JUN, SUN-AH (ED.) (ed.) 2005. Prosodic typology: The phonology of intonation and
phrasing. Oxford: Oxford University Press.
KANDYBOWICZ, JASON. 2007. On fusion and multiple copy spell-out: The case of verbal
repetition. The Copy Theory of Movement, ed. by Norbert Corver and Jairo Nunes,
119–150. Amsterdam/Philadelphia: John Benjamins.
KANDYBOWICZ, JASON. 2009. Embracing edges : Syntactic and phono-syntactic edge
sensitivity in Nupe. Natural Language & Linguistic Theory 27.305–344.
KANDYBOWICZ, JASON. 2015. On prosodic vacuity and verbal resumption in Asante Twi.
Linguistic Inquiry 46.243–272
KAYNE, RICHARD. 1994. The antisymmetry of syntax. Cambridge, MA: MIT Press.
KEINE, STEFAN. 2015. Selective opacity. Amherst, MA: University of Massachussetts,
Amherst manuscript.
KENSTOWICZ, MICHAEL. 1997. Quality-sensitive stress. Rivista di Linguistica 9.157–188.
KHAN, SAMEER UD DOWLA. 2008. Intonational phonology and focus prosody of Bengali.
Los Angeles: University of California, Los Angeles dissertation.
KHAN, SAMEER UD DOWLA. 2010. Bengali (Bangladeshi Standard). Journal of the
International Phonetic Association 40.221–225.
KHAN, SAMEER UD DOWLA. 2014. The intonational phonology of Bangladeshi Standard
Bengali. Prosodic Typology II: the Phonology of Intonation and Phrasing, ed. by
Sun-Ah Jun. Oxford: Oxford University Press.
KIPARSKY, PAUL. 1985. Some consequences of Lexical Phonology. Phonology 2.85.
KIPARSKY, PAUL. 2000. Opacity and cyclicity. The Linguistic Review 17.1–18.

230
KIPARSKY, PAUL. 2008. Fenno-Swedish quantity: Contrast in Stratal OT. Rules,
constraints, and phonological phenomena, ed. by Bert Vaux and Andrew Nevins,
185–219. Oxford: Oxford University Press.
KISSEBERTH, CHARLES W. 1994. On domains. Perspectives in Phonology, ed. by Jennifer
Cole and Charles W. Kisseberth, 133–166. Stanford: CSLI Publications.
KLIMA, EDWARD S. 1964. Negation in English. The structure of language: Readings in
the philosophy of language, ed. by Jerry A. Fodor and Jerrold J. Katz, 246–323.
Englewood Cliffs, NJ: Prentice Hall.
KOOPMAN, HILDA. 1983. The syntax of verbs: From verb movement in the Kru languages
to Universal Grammar. Dordrecht: Foris.
KOUL, OMKAR N.; and KASHI WALI. 2015. Kashmiri. Munich: Lincom Europa.
KRAMER, RUTH. 2010. The Amharic definite marker and the syntax-morphology
interface. Syntax 13.196–240.
KRAPOVA, ILIYANA; and TSENA KARASTANEVA. 2002. On the structure of the CP field in
Bulgarian. Balkanistica 15.293–321.
KROCH, ANTHONY S.; ANN TAYLOR; and DON RINGE. 1995. The Middle English verb-
second constraint: A case study in language contact and language change. Textual
parameters in older languages, ed. by Susan C. Herring, Pieter van Reenen, and
Lene Schøsler, 353-391. Philadelphia: John Benjamins.
KROCH, ANTHONY; and ANN TAYLOR. 1997. Verb movement in Old and Middle English:
dialect variation and language contact. Parameters of morphosyntactic change, ed.
by Ans van Kemenade and Nigel Vincent, 297–325. Cambridge: Cambridge
University Press.
LABELLE, MARIE. 2007. Clausal architecture in Early Old French. Lingua 117.289–316.
LABELLE, MARIE; and PAUL HIRSCHBÜHLER. 2012. Topic and focus in Old French V1
and V2 structures. Paper presented at DiGS 13, University of Pennsylvania.
LAHNE, ANTJE. 2010. A multiple specifier approach to left-peripheral architecture.
Linguistic Analysis 35.73–108.
LASNIK, HOWARD. 1981. Restricting the theory of transformations: A case study.
Explanation in linguistics, ed. by Norbert Hornstein and David Lightfoot, 152–173.
London: Longman.
LASNIK, HOWARD. 1999. On feature strength: Three Minimalist approaches to overt
movement. Linguistic Inquiry 30.197–217.

231
LECHNER, WINFRIED. 2006. An interpretive effect of head movement. Phases of
interpretation, ed. by Mara Frascarelli, 45–71. Berlin: Mouton de Gruyter.
LEDGEWAY, ADAM. 2005. Moving through the left periphery: the dual complementiser
system in the dialects of southern Italy. Transactions of the Philological Society
103.339–396.
LEMA, JOSÉ; and MARÍA LUISA RIVERO. 1989. Long head movement: ECP vs. HMC.
NELS 30: Proceedings of the 30th Meeting of the North East Linguistic Society, ed.
by Masako Hirotani, Andries Coetzee, Nancy Hall and Ji-yung Kim, 333–347.
Amherst, MA: GLSA.
LEMA, JOSÉ; and MARIA-LUISA RIVERO. 1991. Types of verbal movement in Old
Spanish: Modals, futures, and perfects. Probus 3.237–278.
LÓPEZ, LUIS. 2009. Ranking the Linear Correspondence Axiom. Linguistic Inquiry
40.239–276.
MAHAJAN, ANOOP KUMAR. 1990. The A/A-bar distinction and movement theory.
Cambridge, MA: Massachusetts Institute of Technology dissertation.
MANETTA, EMILY. 2011. Peripheries in Kashmiri and Hindi-Urdu: The syntax of
discourse-driven movement. Amsterdam/Philadelphia: John Benjamins.
MANETTA, EMILY. 2012. Reconsidering rightward scrambling: Postverbal constituents in
Hindi-Urdu. Linguistic Inquiry 43.43–74.
MANOVA, STELA; and MARK ARONOFF. 2010. Modeling affix order. Morphology
20.109–131.
MARANTZ, ALEC. 1988. Clitics, Morphological Merger, and the mapping to phonological
structure. Theoretical Morphology, ed. by Michael Hammond and Michael Noonan,
253–270. San Diego: Academic Press.
MARTEEL, J. 1992. Cours de Flamand. Het Vlaams dan men oudders klappen. Lille:
Miroirs Editions.
MARTINOVIC, MARTINA. 2015. Feature geometry and head-splitting: Evidence from the
morpho-syntax of the Wolof clausal periphery. Chicago: University of Chicago
dissertation.
MARTINOVIC, MARTINA. 2016. Head-splitting at the Wolof clausal periphery. Poster
presented at the 34th West Coast Conference on Formal Linguistics, University of
Utah.

232
MARTINS, ANA MARIA. 2004. Emphatic affirmation and polarity. Romance Languages
and Linguistic Theory 2004: Selected papers from ‘Going Romance’, Leiden, 9–11
December 2004, ed. by Jenny Doetjes and Paz González, 197-223.
Amsterdam/Philadelphia: John Benjamins.
MATUSHANSKY, ORA. 2006. Head movement in linguistic theory. Linguistic Inquiry
37.69–109.
MCCARTHY, JOHN J. 2012. Pausal phonology and morpheme realization. Prosody
Matters: Essays in honor of Elisabeth Selkirk, ed. by Toni Borowsky, Shigeto
Kawahara, Takahito Shinya, and Mariko Sugahara, 341-373. London: Equinox.
MCCARTHY, JOHN J; and ALAN PRINCE. 1993. Generalized alignment. Yearbook of
Morphology.79–153.
MCPHERSON, LAURA; and MARY PASTER. 2009. Evidence for the mirror principle and
morphological templates in Luganda affix ordering. Selected proceedings of the 39th
Annual Conference on African Linguistics: Linguistic Research and Languages in
Africa, ed. by Akinloye Ojo and Lioba Moshi, 56-66.
MIGDALSKI, KRZYSZTOF MAREK. 2006. The syntax of compound tenses in Slavic.
Tilburg: Tilburg University dissertation.
MOHANAN, K. P. 1986. Lexical Phonology. Dordrecht: Kluwer.
MOHANAN, K. P.; and TARA MOHANAN. 1984. Lexical phonology of the consonant
system in Malayalam. Linguistic Inquiry 15.575–602.
MOHR, SABINE. 2009. V2 as a single-edge phenomenon. Selected Papers from the 2006
Cyprus syntaxfest, ed. by Kleanthes K. Grohmann and Phoevos Panagiotidis, 141–
159. Cambridge: Cambridge Scholars Publishing.
MOLTMANN, FRIEDERIKE. 1990. Scrambling in German and the specificity effect.
Cambrdige, MA: Massachusetts Institute of Technology manuscript.
MORIN, ANNICK. 2006. On the syntax of clause type particles: Evidence from Gascon,
Innu and Quebec French. Montréal: Concordia University masters thesis.
MOTUT, ALEXANDRA. 2010. Merge over Move and the empirical force of economy in
Minimalism. Toronto Working Papers in Linguistics 33.1–54.
MUSSAFIA, ALFREDO. 1886. Una particolarità sinttatica della lingua italiana dei primi
secoli. Misccellanea di filologia e linguistica in memoria di Napoleone Caix e Ugo
Angelello Canello, ed. by G.I. Ascoli and Et Al., 255–261. Florence: LeMonnier.
MYRBERG, SARA. 2013. Sisterhood in prosodic branching. Phonology 30.73–124.

233
NESPOR, MARINA; and IRENE VOGEL. 1986. Prosodic Phonology. Dordrecht: Foris.
NICHOLS, JOHANNA. 2011. Ingush Grammar. Berkeley and Los Angeles, California:
University of California Press.
NUNES, JAIRO. 1995. The Copy Theory of Movement and linearization of chains in the
Minimalist Program. College Park, MD: University of Maryland, College Park
dissertation.
NUNES, JAIRO. 1999. Linearization of chains and phonetic realization of chain links.
Working Minimalism, ed. by Samuel David Epstein and Norbert Hornstein, 217–
249. Cambridge, MA: MIT Press.
NUNES, JAIRO. 2004. Linearization of chains and sideward movement. Cambridge, MA:
MIT Press.
OLIVIÉRI, MICHÈLE; and RICHARD FAURE. 2013. Stratégies de topicalisation en occitan.
Corpus 12.231-270.
OLTRA-MASSUET, ISABEL. 2013. Variability and allomorphy in the morphosyntax of
Catalan past perfective. Distributed Morphology Today: Morphemes for Morris
Halle, ed. by Ora Matushansky and Alec Marantz, 1–20. Cambridge, MA: MIT
Press.
OTT, DENNIS. 2014. An ellipsis approach to Contrastive Left-Dislocation. Linguistic
Inquiry 45.269–303.
OUHALLA, JAMAL. 1991. Functional categories and parametric variation. London:
Routledge.
PASTER, MARY. 2005. Pulaar verbal extensions and phonologically driven affix ordering.
Yearbook of Morphology 2005.155–199.
PASTER, MARY. 2009. Explaining phonological conditions on affixation : Evidence from
suppletive allomorphy and affix ordering. Word Structure 2.18–37.
PASTER, MARY ELIZABETH. 2006. Phonological conditions on affixation. Berkeley:
University of California, Berkeley dissertation.
PESETSKY, DAVID. 1982. Paths and categories. Cambridge, MA: Massachusetts Institute
of Technology dissertation
PESETSKY, DAVID. 1991. Zero Syntax II: An essay on infinitives. Cambridge, MA:
Massachusetts Institute of Techonology manuscript..

234
PESETSKY, DAVID. 1998. Some optimality principles of sentence pronunciation. Is the
best good enough?, ed. by Pilar Barbosa, Daniel Fox, Paul Hagstrom, Martha
McGinnis, and David Pesetsky, 337-383. Cambridge, MA: MIT Press.
PESETSKY, DAVID. 2000. Phrasal movement and its kin. Cambridge, MA: MIT Press.
PETROVA, SVETLANA. 2012. Multiple XP-fronting in Middle Low German root clauses.
Journal of Comparative Germanic Linguistics 15.157–188.
PINTZUK, SUSAN. 1991. Phrase structures in competition: variation and change in Old
English word order. Philadelphia: University of Pennsylvania dissertation.
PINTZUK, SUSAN. 1993. Verb seconding in Old English: verb movement to Infl. The
Linguistic Review 10.5–35.
PLATZACK, CHRISTER. 2013. Head movement as a phonological operation. Diagnosing
Syntax, ed. by Lisa Lai-Shen Cheng and Norbert Corver, 21-43. Oxford: Oxford
University Press.
POLETTO, CECILIA. 2000. The higher functional field. New York: Oxford University
Press, USA.
POLETTO, CECILIA. 2002. The left-periphery of V2-Rhaetoromance dialects: a new view
on V2 and V3. Syntactic microvariation, ed. by Sjef Barbiers, Leonie Cornips, and
Susanne van der Kleij, 214–242. Amsterdam: Meertens.
POLETTO, CECILIA. 2013. On V2 types. The Bloomsbury Companion to Syntax, ed. by
Silvia Luraghi and Claudia Parodi, 154–164. London: Bloomsbury.
POLETTO, CECILIA. 2014. Word order in Old Italian. Oxford: Oxford University Press.
POLETTO, CECILIA; and ALESSANDRA TOMASELLI. 1999. Towards a typology of V2.
Paper presented at the XV IGG Siena.
POLLOCK, JEAN-YVES. 1989. Verb movement, Universal Grammar, and the structure of
IP. Linguistic Inquiry 20.365–424.
PRINCE, ALAN; and PAUL SMOLENSKY. 1993. Optimality Theory: Constraint interaction
in generative grammar. Malden, MA: Blackwell.
PULLUM, GEOFFREY K.; and ARNOLD M. ZWICKY. 1989. The syntax-phonology interface.
Linguistics: The Cambridge Survey, Vol. 1. Linguistic Theory: Foundations, ed. by
Frederick J. Newmeyer, 255–280. Cambridge: Cambridge University Press.
PUSCH, CLAUS. 2002. Preverbal modal particles in Gascony Occitan. Belgian Journal of
Linguistics 16.105–118.

235
PYLKKÄNEN, LIINA. 2002. Introducing arguments. Cambridge, MA: Massachusetts
Institute of Technology dissertation.
RAHMAN, SYED SHAHRIER. 2010. Wh-Question and focus construction in Bengali: Word
order variation and acquisition. Tromsø: University of Tromsø masters thesis.
RICE, KEREN. 2011. Principles of affix ordering: An overview. Word Structure 4.169–
200.
RICHARDS, NORVIN. 1999. Complementizer cliticization in Tagalog and English.
Proceedings of the 6th Meeting of the Austronesian Formal Linguistics Association.
RICHARDS, NORVIN. 2010. Uttering trees. Cambridge, MA: MIT Press.
RICHARDS, NORVIN. 2016. Contiguity Theory. Cambridge, MA: MIT Press.
RIVERO, MARIA-LUISA. 1991. Long head movement and negation: Serbo-Croatian vs.
Slovak and Czech. The Linguistic Review 8.319–351.
RIZZI, LUIGI. 1990. Relativized Minimality. Cambridge, MA: MIT Press.
RIZZI, LUIGI. 1997. The fine structure of the left periphery. Elements of Grammar, ed. by
Liliane Haegeman, 281–337. Dordrecht: Kluwer.
RIZZI, LUIGI. 2001. On the position Int(errogative) in the left periphery of the clause.
Current studies in Italian Syntax: Essays offered to Lorenzo Rizzi, ed. by Guglielmo
Cinque and Giampaolo Salvi, 287–96. Oxford: Elsevier.
RIZZI, LUIGI. 2006. On the form of chains: Criterial positions and ECP effects. On Wh
Movement, ed. by Lisa Cheng and Norbert Corver, 97–133. Cambridge, MA: MIT
Press.
RIZZI, LUIGI. 2010. On some properties of Criterial Freezing. The complementizer phase:
Subjects and operators, ed. by E. Phoevos Panagiotidis, 17–32. Oxford: Oxford
University Press.
ROBERTS, IAN G. 1993. Verbs and diachronic syntax. A comparative history of English
and French. Dordrecht: Kluwer.
ROBERTS, IAN G. 1996. Remarks on the Old English C-system and the diachrony of V2.
Linguistische Berichte, Sonderheft 7: Language Change and Generative Grammar,
ed. by Ellen Brandner and Gisella Ferraresi, 154–167.
ROBERTS, IAN G. 2000. Head movement. Handbook of Syntactic Theory, ed. by Mark
Baltin and Chris Collins, 113–147. Oxford: Blackwell.

236
ROBERTS, IAN G. 2004. The C-System in Brythonic Celtic Languages, V2, and the EPP.
The Structure of CP and IP, ed. by Luigi Rizzi, 297–328. Oxford: Oxford University
Press.
ROBERTS, IAN G. 2005. Principles and parameters in a VSO Language: A case study in
Welsh. Oxford: Oxford University Press.
ROBERTS, IAN G. 2010. Agreement and head mmovement: Clitics, incorporation, and
defective goals. Cambridge, MA: MIT Press.
ROBERTS, IAN G. 2012. Phases, head movement and second-position effects. Phases:
Developing the framework, ed. by Angel Gallego, 385–440. Berlin: De Gruyter
Mouton.
ROSE, SHARON. 1997. Theoretical issues in comparative Ethio-Semitic phonology and
morphology. Montréal: McGill University dissertation.
ROUSSOU, ANNA. 2000. On the left periphery. Journal of Greek Linguistics 1.65–94.
RUDIN, CATHERINE. 1982. Complementizers and WH constructions. Bloomington, IN:
Indiana University dissertation.
RYCKEBOER, HUGO. 2004. Fransvlaams. Lannoo: Tielt.
SABBAGH, JOSEPH. 2014. Word order and prosodic structure constraints in Tagalog.
Syntax 17. 130-189.
SAELENS, J. 2014. Topicalisering zonder inversie: een ingveonisme? Ghent: Ghent
University masters thesis.
SAMEK-LODOVICI, VIERI. 2005. Prosody-syntax unteraction in the expression of focus.
Natural Language & Linguistic Theory 23.687–755.
SAMEK-LODOVICI, VIERI. 1993. Morphological gemination. Paper presented at Rutgers
Optimality Workshop 1.
SAUZET, PATRICK. 1989. Topicalisation et prolepse en occitan. Revue des langues
romanes 93.235–273.
SELKIRK, ELISABETH. 1980. Prosodic domains in phonology: Sanskrit revisited. Juncture,
ed. by Mark Aronoff and Marie-Louise Kean, 107–129. Saratoga, CA: Anma Libri.
SELKIRK, ELISABETH. 1981a. On the nature of phonological representation. The Cognitive
Representation of Speech, ed. by John Anderson, John Laver, and Terry Meyers.
Amsterdam: North Holland.

237
SELKIRK, ELISABETH. 1981b. On prosodic structure and its relation to syntactic structure.
Nordic Prosody II: Papers from a symposium, ed. by Thorstein Fretheim, 111–140.
Trondheim: TAPIR.
SELKIRK, ELISABETH. 1984. Phonology and syntax: The relation between sound and
structure. Cambridge, MA: MIT Press.
SELKIRK, ELISABETH. 1986. On derived domains in sentence phonology. Phonology
Yearbook 3.371–405.
SELKIRK, ELISABETH. 1996. The prosodic structure of function words. Signal to syntax:
Bootstrapping from speech to grammar in early acquisition, ed. by James L.
Morgan and Katherine Demuth, 187–214. Mahwah, NJ: Lawrence Erlbaum
Associates.
SELKIRK, ELISABETH. 2002. On the phonologically driven non-realization of function
words. Proceedings of the Twenty-Seventh Annual Meeting of the Berkeley
Linguistics Society, ed. by Charles Chang, Michael J. Houser, Yuni Kim, David
Mortensen, Mischa Park-Doob and Maziar Toosarvandani, 257--270. Berkeley:
Berkeley Linguistics Society.
SELKIRK, ELISABETH. 2007. Bengali intonation revisited: An optimality theoretic analysis
in which Focus stress prominence drives Focus phrasing. Topic and Focus: Cross-
Linguistic Perspectives on Meaning and Intonation, ed. by Chungmin Lee and
Matthew Gordon, 6:215–244. Springer.
SELKIRK, ELISABETH. 2009. On clause and intonational phrase in Japanese : The syntactic
grounding of prosodic constituent structure. Gengo Kenkyu.1–39.
SELKIRK, ELISABETH. 2011. The Syntax-phonology interface. The handbook of
Phonological Theory, ed. by John Goldsmith, Jason Riggle, and Alan Yu, 1–66. 2nd
ed. Oxford: Blackwell.
SELKIRK, ELISABETH; and SEUNGHUN J LEE. 2015. Constituency in sentence phonology :
an introduction. Phonology 32.1–18.
SELKIRK, ELISABETH; and KOICHI TATEISHI. 1988. Constraints on minor phrase formation
in Japanese. Papers from the 24th Annual Regional Meeting of the Chicago
Linguistic Society, 316–336. Chicago: Chicago Linguistic Society.
SENGUPTA, GAUTAM. 1990. Binding and scrambling in Bangla. Amherst, MA: University
of Massachusetts, Amherst dissertation.
SHIMA, ETSURO. 2000. A preference for Move over Merge. Linguistic Inquiry 31.375–
385.

238
SHLONSKY, UR; and LUIGI RIZZI. 2007. Strategies of subject extraction. Interfaces +
Recursion = Language? Chomsky’s minimalism and the view from syntax-
semantics, ed. by Hans-Martin Gärtner and Uli Sauerland, 115–160. Berlin: Mouton
de Gruyter.
SILVERMAN, KIM; MARY BECKMAN; JOHN PITRELLI; MORI OSTENDORF; COLIN W.
WIGHTMAN; PATTI PRICE; JANET PIERREHUMBERT; and JULIA HIRSCHBERG. 1992.
ToBI: A standard scheme for labeling prosody. Proceedings of the Second
International Conference on Spoken Language Processing.867–870.
SIMPSON, ANDREW; and TANMOY BHATTACHARYA. 2003. Obligatory overt Wh-
movement in a Wh-in-situ language. Linguistic Inquiry 34.127–142.
SPEAS, MARGARET. 1991. Phrase structure in natural language. Dordrecht: Kluwer.
STEPHENS, JANIG. 1982. Word order in Breton. London: School of Oriental and African
Studies, University of London dissertation.
STJEPANOVIĆ, SANDRA. 2007. Free word order and copy theory of movement. The Copy
Theory of Movement, ed. by Norbert Corver and Jairo Nunes, 219–248.
Amsterdam/Philadelphia: John Benjamins.
STORTO, LUCIANA. 1999. Aspects of Karitiana grammar. Cambridge, MA:
Massachussetts Institute of Technology dissertation.
SVENONIUS, PETER. 1994. C-selection as feature checking. Studia Linguistica 48.133–
155.
SYED, SAUROV. 2013. DP-Internal focus and topic in Bangla. Online proceedings of
GLOW in Asia IX, ed. by Nobu Goto, Koichi Otaki, Atsushi Sato, and Kensuke
Takita. Mie University.
THRÁINSSON, HÖSKULDUR. 1979. On complementation in Icelandic. New York: Garland.
THRÁINSSON, HÖSKULDUR. 1996. On the (non)-universality of functional projections.
Minimal ideas, ed. by Werner Abraham, Samuel David Epstein, Hoskuldur
Thrainsson, and C. Jan-Wouter Zwart, 253–81. Amsterdam: John Benjamins.
TOBLER, ADOLF. 1875. Besprechung von J. Le Coultre, De l’ordre des mots dans
Chrétien de Troyes. Leipzig: Vermische Beiträge zur französischen Grammatik 5.
TRAVIS, LISA. 1984. Parameters and effects of word order variation. Cambrdige, MA:
Massachusetts Institute of Technology dissertation.
TRIPS, CAROLA. 2002. From OV to VO in Early Middle English. Philadelphia: John
Benjamins.

239
USSISHKIN, ADAM. 2007. Morpheme position. The Cambridge Handbook of Phonology,
ed. by Paul de Lacy, 457–472. Cambridge: Cambridge University Press.
VAN KEMENADE, ANS. 1987. Syntactic case and morphological case in the history of
English. Dordrecht: Foris.
VAN RIEMSDIJK, HENK. 1988. The representation of syntactic categories. Proceedings of
the Conference on the Basque Language, Basque World Congress 1.104–116.
VAN RIEMSDIJK, HENK. 1998. Categorial feature magnetism: The endocentricity and
distribution of projections. Journal of Comparative Germanic Linguistics 2.1–48.
VANACKER, V. F. 1977. Syntactische overeenkomsten tusen Frans-Vlaamse en
Westvlaamse dialekten. De Franse Nederlanden. Les Pays Bas Français. Jaarboek,
206–216. Rekkem: Ons Erfdeel.
VANCE, BARBARA. 1997. Syntactic change in Medieval French. Dordrecht: Kluwer.
VARIS, ERIKA ELIZABETH. 2012. The Spanish feminine el at the syntax-phonology
interface. Los Angeles: University of Southern California dissertation.
VICENTE, LUIS. 2007. The syntax of heads and phrases: A study of verb (phrase) fronting.
Leiden: Leiden University dissertation.
VIKNER, STEN. 1995. Verb movement and expletive subjects in the Germanic languages.
New York and Oxford: Oxford University Press.
VIKNER, STEN; and BONNIE SCHWARTZ. 1996. The verb always leaves IP in V2 clauses.
Parameters and functional heads, ed. by Adriana Belletti and Luigi Rizzi, 11–72.
Oxford: Oxford University Press.
VOGEL, RALF. 2006. Weak function word shift. Linguistics 44.1059–1093.
WAGNER, MICHAEL. 2005. Prosody and recursion. Cambridge, MA: Massachusetts
Institute of Technology dissertation.
WALKDEN, GEORGE. 2015. Verb-third in early West Germanic: a comparative
perspective. Syntax over time: lexical, morphological, and information-structural
interactions, ed. by Theresa Biberauer and George Walkden, 236–248. Oxford:
Oxford University Press.
WALKER, RACHEL. 1998. Nasalization, neutral segments, and opacity effects. Santa Cruz:
University of California, Santa Cruz dissertation.

240
WATANABE, AKIRA. 1993. Larsonian CP recursion, factive complements and selection.
NELS 23: Proceedings of the 23rd meeting of the North East Linguistics Society, ed.
by Amy Schafer, 523–537. Amherst, MA: GLSA.
WESTERGAARD, M. 2005. Norwegian child language and the history of English: the
interaction of syntax and information structure in the development of word order.
Contexts—Historical, Social, Linguistic: Studies in Celebration of Toril Swan., ed.
by K. McCafferty, T. Bull, and K. Killie, 293–410. Bern: Peter Lang.
WILDER, CHRIS; and DAMIR ĆAVAR. 1994. Long head movement? Verb movement and
cliticization in Croatian. Lingua 93.1–58.
WILLIAMS, EDWIN. 1981. On the notions “lexically related” and “head of a word.”
Linguistic Inquiry 12.245–274.
WILLIS, DAVID. 1998. Syntactic change in Welsh: A study of the loss of verb-second.
Oxford: Clarendon Press.
WILTSCHKO, MARTINA. 1997. Parasitic operators in German left-dislocations. Materials
on left-dislocation, ed. by Elena Anagnostopoulou, Henk van Riemsdijk, and Frans
Zwarts, 307–339. Amsterdam: John Benjamins.
WOLF, MATTHEW. 2008. Optimal Interleaving: Serial phonology-morphology interaction
in a constraint-based model. Amherst, MA: University of Massachusetts, Amherst
dissertation.
WOLF, MATTHEW. 2011. Candidate chains, unfaithful spell-out, and outwards-looking
phonologically-conditioned allomorphy. Morphology 23.145–178.
WOLF, MATTHEW. 2015. Lexical Insertion occurs in the phonological component.
Understanding allomorphy: Perspectives from Optimality Theory, ed. by On
Microvariation in V2 Systems in Old Romance and Beyond. Linguistic Variation.
WOLFE, SAM. to appear. On microvariation in V2 Systems in Old Romance and beyond.
Linguistic Variation.
WOLFE, SAM. 2015a. Microvariation in Medieval Romance Syntax: A comparative study.
Cambridge, MA: University of Cambridge dissertation.
WOLFE, SAM. 2015b. The nature of Old Spanish verb second reconsidered. Lingua
164.132–155.
ZAENEN, ANNIE. 1997. Contrastive dislocation in Dutch and Icelandic. Materials on left-
dislocation, ed. by Elena Anagnostopoulou, Henk van Riemsdijk, and Frans Zwarts,
110–148. Amsterdam: John Benjamins.

241
ZEC, DRAGA; and SHARON INKELAS. 1990. Prosodically constrained syntax. The
Phonology-Syntax Connection, ed. by Sharon Inkelas and Draga Zec, 365–378.
Chicago: University of Chicago Press.
ZUBIZARRETA, MARIA LUISA. 1998. Prosody, focus, and word order. Cambridge, MA:
MIT Press.
ZWICKY, ARNOLD M. 1970. Auxiliary reduction in English. Linguistic Inquiry 1.323–336.
ZWICKY, ARNOLD M.; and GEOFFREY K. PULLUM. 1983. Cliticization vs. inflection:
English N’T. Language 59.502–513.
ZWICKY, ARNOLD M.; and GEOFFREY K. PULLUM. 1986. The principle of phonology-free
syntax. Ohio State University Working Papers in Linguistics 32.63–91.

Abstract (if available)

Abstract This dissertation addresses two core questions in linguistic theory. [1] What are the sources of variation in word order, both within and across languages? [2] What information is shared between the syntactic and phonological modules of natural language grammar? It makes several proposals regarding possible mappings of syntactic structures to surface word order, and the role that phonological information plays in syntactic derivations. It argues that phonological considerations can influence word order by determining the optimal pronunciation of movement copies, and that some syntactic operations are driven by requirements to produce phonologically licit structures. ❧ The first few chapters examine the role of the Phonetic Form (PF) grammar in the linearization of movement copies and in the creation of prosodic structure. It presents an analysis of a puzzling pattern in Bangla (a.k.a. Bengali) that involves variation in the placement of the subordinating complementizer je. It is shown that previous analyses, which rely purely on grammatical or discourse properties, do not adequately account for the pattern. New data on Bangla prosody is presented to argue that variation in complementizer placement is driven by a grounded phonological constraint against the placement of je in the initial position of an Intonational Phrase. This leads to an implementation of the PF component as an Optimality-Theoretic grammar in which phonological and syntactic well-formedness conditions compete to determine the optimal pronunciation of movement copies. ❧ The remainder of the dissertation turns to the accessibility of phonological information to syntactic operations. It proposes the existence of a syntactic operation, Coalescence, which bundles structurally adjacent heads into a single, featurally complex head. This operation derives variation in how category features are realized on heads, as well as certain types of head-adjunction and cliticization. It is argued that the application of Coalescence is properly motivated and constrained by a requirement to eliminate heads that would be deficient in their phonological realization, suggesting that a restricted amount of information about phonological exponence is visible to syntactic operations. The approach is illustrated in analyses of cross-linguistic variation in the realization of positions within the clausal left periphery and the inflectional domain, with a focus on verb second effects and verb movement.

Linked assets

University of Southern California Dissertations and Theses

Conceptually similar

PDF

Towards the unity of movement: implications from verb movement in Cantonese

PDF

Silence in answers: a study of ellipsis in Hindi

PDF

The Spanish feminine el at the syntax-phonology interface

PDF

Perspective in Turkish complementation

PDF

Prosodic recursion and syntactic cyclicity inside the word

PDF

Harmony in gestural phonology

PDF

Comparative iIlusions at the syntax-semantics interface

PDF

Copy theory of movement and PF conditions on spell-out

PDF

The morphosyntax of states: deriving aspect and event roles from argument structure

PDF

Soft biases in phonology: learnability meets grammar

PDF

Functional categories: the syntax of DP and DegP

PDF

Building adjectival meaning without adjectives

PDF

Tone gestures and constraint interaction in Sierra Juarez Zapotec

PDF

The grammar of correction

PDF

The case of a person: The person case constraint in German

PDF

Building phrase structure from items and contexts

PDF

The grammar of individuation, number and measurement

PDF

Articulatory dynamics and stability in multi-gesture complexes

PDF

A reduplicative analysis of sentence modal adverbs in Spanish

PDF

Decomposing Slavic aspect: the role of aspectual morphology in Polish and other Slavic languages

Asset Metadata

Creator Hsu, Brian (author)

Core Title Syntax-prosody interactions in the clausal domain: head movement and coalescence

School College of Letters, Arts and Sciences

Degree Doctor of Philosophy

Degree Program Linguistics

Publication Date 07/25/2016

Defense Date 05/31/2016

Publisher University of Southern California (original), University of Southern California. Libraries (digital)

Tag complementizers,head movement,OAI-PMH Harvest,phonology,syntax,verb second,word formation

Format application/pdf (imt)

Language English

Contributor Electronically uploaded by the author (provenance)

Advisor Simpson, Andrew (committee chair), Jesney, Karen (committee member), Nayak, Krishna S. (committee member), Pancheva, Roumyana (committee member), Walker, Rachel (committee member)

Creator Email brianhhsu@gmail.com,hsub@usc.edu

Permanent Link (DOI) https://doi.org/10.25549/usctheses-c40-278325

Unique identifier UC11280323

Identifier etd-HsuBrian-4609.pdf (filename),usctheses-c40-278325 (legacy record id)

Legacy Identifier etd-HsuBrian-4609.pdf

Dmrecord 278325

Document Type Dissertation

Format application/pdf (imt)

Rights Hsu, Brian

Type texts

Source University of Southern California (contributing entity), University of Southern California Dissertations and Theses (collection)

Access Conditions The author retains rights to his/her dissertation, thesis or other graduate work according to U.S. copyright law. Electronic access is being provided by the USC Libraries in agreement with the a...

Repository Name University of Southern California Digital Library

Repository Location USC Digital Library, University of Southern California, University Park Campus MC 2810, 3434 South Grand Avenue, 2nd Floor, Los Angeles, California 90089-2810, USA