• unlimited access with print and download
    $ 37 00
  • read full document, no print or download, expires after 72 hours
    $ 4 99
More info
Unlimited access including download and printing, plus availability for reading and annotating in your in your Udini library.
  • Access to this article in your Udini library for 72 hours from purchase.
  • The article will not be available for download or print.
  • Upgrade to the full version of this document at a reduced price.
  • Your trial access payment is credited when purchasing the full version.
Buy
Continue searching

Radical discontinuity: Syntax at the interface

ProQuest Dissertations and Theses, 2009
Dissertation
Author: Natalia Kariaeva
Abstract:
In this dissertation, I examine discontinuous constituents in Ukrainian and Modern Greek and advance the Radical Discontinuity Hypothesis. I argue that discontinuous constituents in which an adjectival modifier surfaces in a distance from the noun it modifies do not result from splitting a single phrasal constituent by way of movement but are the product of long-distance concord. An adjectival modifier is base-generated at a distance from the noun and is licensed by agreement. I demonstrate that the Radical Discontinuity Hypothesis explains freedom of lexical item ordering in discontinuous constituents, complex cases of discontinuity that involve tripartitioning of the noun phrase, and contrasts in the availability of movement and discontinuity in similar syntactic environments. While the order between lexical categories that form an agreement-based discontinuous constituent is free, functional categories always linearly precede the left-most lexical item associated with the discontinuous constituent. I claim that the surface distribution of functional categories in discontinuous constituents is determined not in narrow syntax but in the course of linearization of syntactic structure at the interface with PF. I argue that the F-value ordering principle responsible for extended projection formation in narrow syntax (Grimshaw (2005)) also guides linearization of the hierarchical structure at PF. By establishing a correlation between the asymmetric c-command and the F-value based ordering of heads of an extended projection, I formulate the Mapping Constraint on Linearization (MCL) and offer a linearization algorithm that implements it. The MCL algorithm incorporates the F-value ordering principle into the LCA of Kayne (1994) and ensures that the F-value ordering relations are enforced in a linear string. The application of the MCL algorithm extends beyond the phenomenon of constituent discontinuity and is examined in relation to various movement transformations. The MLC algorithm restricts the output of both XP and head movement, accounts for the distribution of PP-modifiers and Genitive possessors in movement-based discontinuous constructions, and derives Determiner Spreading in Modern Greek. I also examine locality restriction on movement and agreement in Ukrainian and demonstrate that agreement is constrained not in terms of intervention effects but in terms of agreement domains, which are distinct from PIC.

TABLE OF CONTENTS

Abstract ............................................................................................................................... ii Acknowledgement ............................................................................................................. iv Table of Contents ............................................................................................................... vi Introduction ..........................................................................................................................1 Part One: Linearization Algorithm ...................................................................................22 Chapter 1. Extended Projections at PF .............................................................................22 1.0 Introduction .............................................................................................................22 1.1 Mapping Constraint on Linearization .....................................................................25 1.2 The MCL Algorithm and Syntactic Structure .........................................................38 1.2.1 XP Movement ...................................................................................................38 1.2.2 Head Movement ...............................................................................................45 1.2.3 Multiple Adjunction .........................................................................................49 1.3 Linearization at PF and Agreement .........................................................................54 1.4 Summary .................................................................................................................65 Part Two: Discontinuous Constituents ..............................................................................68 Chapter 2. Constituent Discontinuity as Long-distance Concord ....................................68 2.0 Introduction .............................................................................................................68 2.1 Typology of Discontinuous Constituents ................................................................70 2.2 Long-distance Concord: Syntactic Considerations .................................................76 2.3 Linear Discontinuous Noun Phrases .......................................................................82

vii

2.4 Discontinuous Constituents and Information Structure ..........................................88 2.5 NP Discontinuity and Inversion ..............................................................................97 2.6 Summary ...............................................................................................................107 Chapter 3. Discontinuous PPs ..........................................................................................110 3.0 Introduction ...........................................................................................................110 3.1 Structural Considerations ......................................................................................113 3.2 Linear Discontinuous PPs .....................................................................................120 3.3 Inverse Discontinuous PPs ....................................................................................124 3.4 Preposition Distribution and Distributed Deletion ...............................................131 3.5 Summary ...............................................................................................................138 Chapter 4. Discontinuous DPs ........................................................................................141 4.0 Introduction ...........................................................................................................141 4.1 Linear Discontinuous DPs .....................................................................................146 4.2 Inverse Discontinuous DPs ...................................................................................149 4.3 Preposition-Determiner Stacking ..........................................................................151 4.4 DP Discontinuity and Determiner Spreading ........................................................155 4.5 Summary ...............................................................................................................173 Chapter 5. Degree Adverbs and Tripartite Discontinuity ...............................................176 5.0 Introduction ...........................................................................................................176 5.1 Degree Expressions ...............................................................................................179 5.2 Degree Adverbs and Agreement ...........................................................................188 5.3 Degree Adverbs and Linearization of NPs ............................................................198 5.4 Degree Adverbs and Linearization of VPs ............................................................203

viii

5.5 Multiple Modification and Discontinuity ..............................................................206 5.6 Summary ...............................................................................................................212 Chapter 6. PP-modifiers and Possessors in Discontinuous Constructions .....................215 6.0 Introduction ...........................................................................................................215 6.1 PP-modifiers ..........................................................................................................218 6.2 Genitive Possessors ...............................................................................................227 6.3 Summary ...............................................................................................................238 Chapter 7. Agreement and Locality ................................................................................241 7.0 Introduction ...........................................................................................................241 7.1 The Locality of Move and Agree ..........................................................................248 7.2 Intervention Effects ...............................................................................................253 7.3 Agreement Domains ...............................................................................................264 7.4 Adjective-Headed Domains ..................................................................................278 7.5 Constituent Discontinuity and Clausal Boundaries ...............................................290 7.6 Agreement Domains and Phases ...........................................................................305 7.7 Summary ...............................................................................................................314 Chapter 8. The Semantics of Discontinuous Constituents ..............................................319 8.0 Introduction ...........................................................................................................319 8.1 Discourse Representation Theory .........................................................................320 8.2 Feature Sharing and Discourse Referents .............................................................326 8.3 Summary ...............................................................................................................330

ix

Chapter 9. Movement-Based Approaches to Constituent Discontinuity ........................331 9.0 Introduction ...........................................................................................................331 9.1 Distributed Deletion ..............................................................................................332 9.2 Remnant Movement ..............................................................................................336 9.3 Left Branch Extraction ..........................................................................................342 9.4 Summary ...............................................................................................................346 Conclusion .......................................................................................................................348 References ........................................................................................................................363 Bibliography ....................................................................................................................367 Curriculum Vita ...............................................................................................................395

1

Introduction

A basic operation Merge has been posited as the null hypothesis in the recent implementations of transformational grammar. Merge is minimally defined as “a primitive operation that takes n objects already constructed, and constructs from them a new object” (Chomsky (2005): 11). Two sub-cases of the operation Merge are distinguished. Given object A, External Merge combines object B that is not part of A with A. Given object A, Internal Merge, or Move, combines object B that constitutes part of A with A itself: 1. a) External Merge: [B [A]] b) Internal Merge: [B [A B]]

Ideally, to maintain the generality of the linguistic computational mechanism, Merge should be left unconstrained. The well-formedness of linguistic expressions should be defined not as the result of specific restrictions on Merge but as the result of external interface conditions: the conditions imposed by the sensorimotor system and the conditions imposed by the conceptual-intentional system.

Basic syntactic objects that are used in the process of linguistic computation are of two types: lexical categories and functional categories. Given the unconstrained nature of Merge, Merge should not differentiate between lexical and functional items and apply equally to both. Lexical and functional items, however, exhibit various asymmetries in relation to Merge. In particular, a complement of a lexical item can undergo Internal Merge (2a) while a complement of a functional item cannot (2b):

2

2. a) [B … X … [A lexical B]] b) * [B … X … [A functional B]]

A complement of the verb, for instance, can be topicalized by being raised to the left periphery of the clause while a complement of the determiner cannot. If the operation Merge is unconstrained, such differences should not exist.

The structural asymmetry in (2) is judged to be responsible for the fact that while the linear representation in (3a) is common cross-linguistically, the linear representation in (3b) is not: 3. a) , where B is the complement of A lexical

b) * , where B is the complement of A functional

It is taken for granted that the linear representation in (3a) is derived from the underlying structural representation in (2a) and the linear representation in (3b) is derived from the underlying structural representation in (2b). This correspondence is the only possible option if the linearization process is purely structure driven. It presupposes, however, that ordering of all items takes place in the domain of structure building, thus placing the burden of ruling out (3b) on Merge. If, on the other hand, item ordering is determined in the course of the transfer of the syntactic object to the PF component of grammar, the operation Merge can be left simple and unconstrained. In the latter case, one can consider a possibility that the linear order in (3b) is not available while the structure in (2b) is indeed possible. This can occur if the structure in (2b) is mapped into Spell-Out not as (3b) but as (4): 4.

3

The linear order in (4) has been considered the output of obligatory pied-piping of functional categories in narrow syntax as in (5): 5. [[A functional B] … X … [A functional B]] In other words, (4) has been derived by imposing a specific constraint on Merge.

In this dissertation, I set out to explore the hypothesis that unavailability of (3b) and availability of (4) are a consequence of an interface condition rather than a constraining condition on Merge. In particular, the relation between the functional categories and their complements has been argued to involve a principle of item ordering. Grimshaw’s (2005) theory of extended projection centers around the idea that lexical categories stand in a special relation to the functional categories of relevant type, and this relation can be defined in terms of functional value (F-value) ordering. The label of each syntactic item is identified with two constants: categorial value and functional value. The first constant defines the domain to which the second constant applies. Thus, if A and B share the same categorial value and the F-value of A is higher than the F-value of B, A and B are ordered. Grimshaw (2005) argues that item ordering of this type underlies phrase structure: 6. a) [ DP D [ NP N]] b) * [ NP N [ DP D]], given that D: {categorial value [nominal], functional value [5]} N: {categorial value [nominal], functional value [0]}

I extend this argument to claim that item ordering of this type is a more general condition. The F-value based ordering of items has to be enforced during the transfer of items from one linguistic component to the other, including linearization of items at PF.

4

More specifically, I offer a linearization algorithm that combines the structure driven principle of the Linear Correspondence Axiom (LCA) of Kayne (1994) with the F-value ordering principle of Grimshaw (2005). The LCA associates a hierarchical representation with the linear ordering by making use of the relation of asymmetric c-command between non-terminal nodes in a phrase marker. By establishing a correlation between the asymmetric c-command and the ordering of F-values inside an extended projection, I formulate a well-formedness condition on the linearization of syntactic structure: the Mapping Constraint on Linearization for Head-initial Languages (MCL-HI). The linearization algorithm based on the MCL-HI relies on the idea that extended projections can be identified uniquely in terms of feature sharing (see Pesetsky and Torrego (2007)). The F-value ordering condition applies only to those heads that share the categorial feature. Like the LCA, the MCL algorithm generates a maximal set of pairs of non- terminals that are characterized by the relation of asymmetric c-command. After this set is generated, the F-value ordering mechanism is activated. It reverses the order between those items in a pair that violate the F-value ordering condition. The output of this mechanism is then used to produce the linear ordering of items. As a result, the MCL algorithm ensures that the linear representation is always faithful to the F-value ordering principle. Consequently, the MCL algorithm rules out (3b) at the interface while leaving Merge unconstrained. Indeed, if in (2b), the F-value of A functional is higher than the F- value of B, as is the case with functional vs. lexical categories, and A functional and B share a categorial feature, the MCL algorithm maps (2b) at the PF interface not as (3b) but as (4).

5

In the case of complement fronting, the output of the Merge operation that pied-pipes functional categories in narrow syntax and the output of the Merge operation that does not pied-pipe those categories in narrow syntax but relies on the MCL linearization mechanism overlap. The role of the F-value based linearization is, therefore, obscured by an alternate derivational path that is available, given fully unconstrained nature of Merge. It does, however, become uniquely visible in the phenomenon known as constituent discontinuity. In a number of languages, an adjectival attributive modifier can surface not only immediately adjacent to the noun it modifies but also in a distance from the noun, separated from it by other linguistic material, thus forming a discontinuous constituent: Ukrainian:

7. Velyku Ivan kupyv kvartyru

big.F.SG.ACC John.NOM bought apartment.F.SG.ACC 1

“John bought a BIG apartment.”

Crucially, functional categories that belong to the discontinuous constituent in question surface either preceding the modifier or preceding the noun, depending on the order between the two: Ukrainian: 8. a) V novomu Ivan žyve budynku

in new. M.SG.LOC John.NOM lives building.M.SG.LOC “John lives in a NEW building.”

b) V budynku Ivan žyve novomu

in building. M.SG.LOC John.NOM lives new.M.SG.LOC “As for the building John lives in, it is a NEW one.” 2

6

Modern Greek:

9. a) Ena megalo agorase aftokinito

a.N.SG.ACC big.N.SG.ACC bought.3.SG car.N.SG.ACC “He bought a BIG car.”

b) Ena aftokinito agorase megalo

a.N.SG.ACC car.N.SG.ACC bought.3.SG big.N.SG.ACC “He bought a big CAR.”

There have been several attempts made to account for the distribution of functional categories in discontinuous constituents in terms of movement, resorting to various forms of syntactic pied-piping (see, for instance, Zabrocki (1984), Borsley and Jaworska (1988), Van Riemsdijk (1989), Corver (1990, 1992), Yearley (1993), Franks and Progovac (1994), Junghanns and Zybatow (1995), Androutsopoulou (1997, 1998), Sekerina (1997, 1999), Müller (1998), Fanselow & Čavar (2001, 2002), Bašić (2004), Bošković (2005), Franks (2007), Kučerova (2007), Pereltsvaig (2008); an overview of movement-based approaches to constituent discontinuity is provided in chapter 9). Discontinuous constituents, however, exhibit a wide range of rather complex distribution patterns that do not lend themselves easily to an account based on syntactic pied-piping and require rather complex and construction specific constraints. Alternatively, discontinuous constituents in (8)-(9) can be derived by applying Merge to lexical items, as long as the placement of functional categories in these constituents is the result of the interface linearization mechanism rather than syntactic pied-piping.

Indeed, discontinuous constituents in (8)-(9) fall under the general pattern in (2b), repeated below with a slight modification as (10): 10. [A lexical … X … [A functional ( A lexical )]…]

7

The fronted lexical item A lexical is merged at the edge of the domain to receive discourse prominence. The functional item A functional that belongs to the same extended projection as the fronted lexical item is mapped preceding this lexical item at PF driven by the F- value ordering mechanism, as in (11): 11. Consequently, the theory of discontinuous constituents put forward in this dissertation rests on the idea that the surface distribution of lexical and functional items in discontinuous constituents is determined by two distinct mechanisms. The surface realization of lexical items is determined by Merge while the surface realization of functional items is determined by the interface linearization mechanism. Lexical items can be either Internally or Externally Merged at the edge of a given domain for the sake of pragmatic prominence. Their structural location carries conceptual-intentional import and is, therefore, preserved at the interface. Functional items, however, are linearized according to the F-value ordering constraint.

I illustrate the role linearization plays in determining the distribution of functional and lexical categories by examining discontinuous constituents in Ukrainian—a Slavic language that exhibits a diverse array of discontinuous constructions and combines rich overt agreement morphology with functionally simple nominal domain. I supplement my analysis of discontinuous constituents in Ukrainian with the data from Modern Greek. 3

Modern Greek shares with Ukrainian the diversity of discontinuous structures while also having multiple overt nominal functional categories. I focus exclusively on those discontinuous constituents the parts of which are related to each other in terms of

8

agreement. I examine agreement-based discontinuous constituents with the Radical Discontinuity Hypothesis (RDH) in mind: I claim that agreement-based discontinuous constituents do not map onto a phrasal constituent at any point in the derivation. The Radical Discontinuity Hypothesis follows from the postulate that agreement can be established between two items that are not adjacent to each other (Chomsky (2000, 2001)). As long as feature sharing is allowed to occur without structural adjacency, the items that can form a single phrasal constituent may be generated at a distance from each other, forming an abstract constituent through long-distance agreement rather than structural adjacency. Agreement-based discontinuous constituents are, therefore, analyzed in this dissertation not as the result of splitting a single phrasal constituent into several parts but as the result of long-distance concord. A modifier is base-generated at a distance from the noun it modifies and is licensed by agreement with the noun. I claim that the Radical Discontinuity Hypothesis can be maintained given the concept of agreement as feature sharing and the MCL linearization algorithm.

I start my analysis of discontinuous constituents by looking at the simplest form of discontinuity: a discontinuous bare NP that contains a single adjectival modifier separated from the noun it modifies. I, then, proceed to more complex discontinuous structures that involve reversal of basic adjective-noun order, discontinuity of constituents other than bare NPs, three-part constituent discontinuity, and various forms of embedding. I demonstrate at each step that more complex structures arise on the basis of the same priciples that are responsible for the simplest form of discontinuity in the

9

language and comply with the linearization mechanism based on F-value ordering. In this dissertation, I account for the following facts: (1) Discontinuous constituents are characterized by an asymmetry in the distribution of functional and lexical categories. The distribution of functional categories is much more restricted than the distribution of lexical categories. Unlike the adjectival modifier, the preposition and the determiner cannot be separated from the associated noun: Ukrainian:

12. a) Ivan kupyv velyku kvartyru

John.NOM bought big.F.SG.ACC apartment.F.SG.ACC “John bought a big apartment.”

b) Velyku Ivan kupyv kvartyru

big.F.SG.ACC John.NOM bought apartment.F.SG.ACC “John bought a BIG apartment.” 13. a) Ivan žyve bilja školy

John.NOM lives next-to school.F.SG.GEN “John lives next to a school.”

b) * Bilja Ivan žyve školy

Next-to John.NOM lives school.F.SG.GEN (“John lives NEXT to a school.”)

Modern Greek: 14. a) Agorazi megala aftokinita

buy.3.SG big.N.PL.ACC car.N.PL.ACC “He buys big cars.”

b) Megala agorazi aftokinita

big.N.PL.ACC buy.3.SG car.N.PL.ACC “He buys BIG cars.” 15. a) Agorase to aftokinito

bought.3.SG the.N.SG.ACC car.N.SG.ACC “He bought the car.”

10

b) * To agorase aftokinito

the.N.SG.ACC bought.3.SG car.N.SG.ACC

The order between the adjective and the noun can be changed while the order of functional categories is fixed. The preposition must precede the determiner and both must precede all other lexical categories associated with the same noun phrase: Ukrainian: 16. a) Ivan žyve bilja velykoho budynku . John.NOM lives next-to big.M.SG.GEN building.M.SG.GEN “John lives next to a big building.”

b) Ivan bilja velykoho žyve budynku . John.NOM next-to big.M.SG.GEN lives building.M.SG.GEN “John lives next to a BIG building.”

c) Ivan bilja budynku žyve velykoho . John.NOM next-to building.M.SG.GEN lives big.M.SG.GEN “As for the building John lives next to, it is a BIG one.”

d) * Ivan velykoho žyve bilja budynku

John.NOM big.M.SG.GEN lives next-to building. M.SG.GEN

e) * Ivan budynku žyve bilja velykoho

John.NOM building.M.SG.GEN lives next-to big.M.SG.GEN

Modern Greek: 17. a) Se ena megalo meni spiti

in a.N.SG.ACC big.N.SG.ACC lives.3.SG house.N.SG.ACC “He lives in a BIG house.”

b) * Ena se megalo meni spiti

a.N.SG.ACC in big.N.SG.ACC lives.3.SG house.N.SG.ACC

c) * Se megalo meni ena spiti

in big.N.SG.ACC lives.3.SG a.N.SG.ACC house.N.SG.ACC

d) * Ena megalo meni se spiti

a.N.SG.ACC big.N.SG.ACC lives.3.SG in house.N.SG.ACC

11

(2) Not only can an adjectival modifier be separated from the noun it modifies, but a degree word can also be separated from its adjective. This can result in a tripartitioning of the noun phrase. The order of lexical items in a tripartite discontinuous constituent is free: Ukrainian: 18. a) Ivan kupyv duže velyku kvartyru

John.NOM bought very big.F.SG.ACC apartment.F.SG.ACC “John bought a very big apartment.”

b) Duže Ivan velyku kupyv kvartyru

very John.NOM big.F.SG.ACC bought apartment.F.SG.ACC “John bought a VERY BIG apartment.”

c) Kvartyru Ivan duže kupyv velyku

Apartment.F.SG.ACC John.NOM very bought big.F.SG.ACC “As for the apartment John bought, it is a VERY BIG one.”

d) Duže Ivan kvartyru kupyv velyku

very John.NOM apartment.F.SG.ACC bought big.F.SG.ACC “As for an apartment John bought, it is a VERY BIG one.”

Nominal functional categories, however, are required to surface before all lexical items, including a degree adverb: 19. a) Ivan žyve bilja duže velykoji školy

John.NOM lives next-to very big.F.SG.GEN school.F.SG.GEN “John lives next to a very big school.”

b) * Duže Ivan velykoji žyve bilja školy

very John.NOM big.F.SG.GEN lives next-to school.F.SG.GEN

c) * Duže Ivan bilja velykoji žyve školy

very John.NOM next-to big.F.SG.GEN lives school.F.SG.GEN

d) Bilja duže Ivan velykoji žyve školy

next-to very John.NOM big.F.SG.GEN lives school.F.SG.GEN “John lives next to a VERY BIG school.”

12

Modern Greek:

20. a) O Janis meni the.M.SG.NOM John.NOM live.3.SG se ena poli megalo diamerisma

in a.N.SG.ACC very big.N.SG.ACC apartment.N.SG.ACC “John lives in a very big apartment.”

b) * Poli o Janis very the.M.SG.NOM John.NOM megalo meni se ena diamerisma

big.N.SG.ACC live.3.SG in a.N.SG.ACC apartment.N.SG.ACC

c) * Poli o Janis very the.M.SG.NOM John.NOM se ena megalo meni diamerisma

in a.N.SG.ACC big.N.SG.ACC live.3.SG apartment.N.SG.ACC

d) Se ena poli o Janis in a.N.SG.ACC very the.M.SG.NOM John.NOM megalo meni diamerisma

big.N.SG.ACC live.3.SG apartment.N.SG.ACC “John lives in a VERY BIG apartment.”

(3) Nouns can be modified not only by adjectives but also by PPs. However, in discontinuous noun phrases, nominal functional categories have different distribution in relation to adjectival and PP-modifiers. The determiner is required to precede the adjectival modifier but cannot precede the PP-modifier: Modern Greek: 21. a) Sinandise tin psili jineka

met.3.SG the.F.SG.ACC tall.F.SG.ACC woman.F.SG.ACC “He met the tall woman.”

b) * Psili sinandise tin jineka

tall.F.SG.ACC met.3.SG the.F.SG.ACC woman.F.SG.ACC

c) Tin psili sinandise jineka

the.F.SG.ACC tall.F.SG.ACC met.3.SG woman.F.SG.ACC “He met the TALL woman.”

13

22. a) Sinandise met.3.SG tin jineka apo to sxolio

the.F.SG.ACC woman.F.SG.ACC from the.N.SG.ACC school.N.SG.ACC “He met the woman from the school.”

b) Apo to sxolio sinandise from the.N.SG.ACC school.N.SG.ACC met.3.SG tin jineka

the.F.SG.ACC woman.F.SG.ACC “He met the woman FROM THE SCHOOL.”

c) * Tin apo to sxolio sinandise the.F.SG.ACC from the.N.SG.ACC school.N.SG.ACC met.3.SG jineka

woman.F.SG.ACC

Like the determiner, the preposition must precede the adjectival modifier which forms a linear discontinuous constituent with the noun embedded inside a prepositional phrase. However, a PP-modifier cannot be fronted when it modifies a noun phrase embedded inside another prepositional phrase: Modern Greek: 23. a) Me tin psili stathike jineka

with the.F.SG.ACC tall.F.SG.ACC stood.3.SG woman.F.SG.ACC “He stood with the TALL woman.”

b) * Tin psili stathike me jineka

the.F.SG.ACC tall.F.SG.ACC stood.3.SG with woman.F.SG.ACC

c) * Psili stathike me tin jineka

tall.F.SG.ACC stood.3.SG with the.F.SG.ACC tall.F.SG.ACC

24. a) * Me tin apo to sxolio

with the.F.SG.ACC from the.N.SG.ACC school.N.SG.ACC stathike jineka

stood.3.SG woman.F.SG.ACC (“He stood with the woman FROM THE SCHOOL.”)

14

b) * Me apo to sxolio stathike with from the.N.SG.ACC school.N.SG.ACC stood.3.SG tin jineka

the.F.SG.ACC woman.F.SG.ACC

c) * Apo to sxolio stathike from the.N.SG.ACC school.N.SG.ACC stood.3.SG me tin jineka

with the.F.SG.ACC woman.F.SG.ACC

(4) Unlike PP-modifiers, Genitive possessors can be fronted when embedded under a PP. The preposition has to precede the possessor when the possessor is fronted: Ukrainian: 25. a) Ivan žyve bilja školy brata

John.NOM lives next-to school.F.SG.GEN brother.M.SG.GEN “John lives next to his brother’s school.”

b) * Ivan brata žyve bilja školy

John.NOM brother.M.SG.GEN lives next-to school.F.SG.GEN

c) Ivan bilja brata žyve školy

John.NOM next-to brother.M.SG.GEN lives school.F.SG.GEN “John lives next to his BROTHER’s school.”

(5) Locality restrictions on movement and on discontinuity are not always the same. In some cases, extraction is allowed even though a similar discontinuity is blocked: Ukrainian:

26. a) Ivan zahubyv knyžku John.NOM lost book.F.SG.ACC pro sučasnu arxitekturu

about contemporary.F.SG.ACC architecture.F.SG.ACC “John lost a book about contemporary architecture.”

b) Pro sučasnu arxitekturu

about contemporary.F.SG.ACC architecture.F.SG.ACC Ivan zahubyv knyžku John.NOM lost book.F.SG.ACC “John lost a book ABOUT CONTEMPORARY ARCHITECTURE.”

15

c) * Pro sučasnu Ivan zahubyv about contemporary.F.SG.ACC John.NOM lost knyžku arxitekturu

book.F.SG.ACC architecture.F.SG.ACC (“John lost a book about CONTEMPORARY architecture.”)

d) Pro sučasnu Ivan arxitekturu zahubyv knyžku about contemporary John.NOM architecture lost book.F.SG.ACC “John lost a book about CONTEMPORARY ARCHITECTURE.”

27. a) Mykola znaje jaka divčyna Michael.NOM knows which.F.SG.NOM young-woman.F.SG.NOM kupyla červonu mašynu

bought red.F.SG.ACC car.F.SG.ACC “Michael knows which young woman bought the red car.”

b) Červonu mašynu Mykola znaje red.F.SG.ACC car. F.SG.ACC Michael.NOM knows jaka divčyna kupyla which.F.SG.NOM girl.F.SG.NOM bought “As for the red car, Michael knows which young woman bought it.”

c) * Červonu Mykola znaje red.F.SG.ACC Michael.NOM knows jaka divčyna kupyla mašynu

which.F.SG.NOM young-woman.F.SG.NOM bought car.F.SG.ACC (“As for a/the RED car, Michael knows which young woman bought one/it.” or “Michael knows which young woman bought a/the RED car.”) 4

d) Červonu Mykola mašynu znaje red.F.SG.ACC Michael.NOM car.F.SG.ACC knows jaka divčyna kupyla which.F.SG.NOM girl.F.SG.NOM bought “As for the RED car, Michael knows which young woman bought it.”

Whether discontinuity is possible depends on the lexical category of the item that contains the lower part of the discontinuous constituent: Ukrainian:

28. a) Novu kupyv Ivan knyžku

new.F.SG.ACC bought John.NOM book.F.SG.ACC “John bought a NEW book.”

16

b) Novoju Ivan buv zadovolenyj robotoju

new.F.SG.INST John.NOM was satisfied.M.SG.NOM work.F.SG.INST “John was happy with his NEW job.”

c) * Novoho Ivan zahubyv knyžku profesora

new.M.SG.GEN John.NOM lost book.F.SG.ACC professor.M.SG.GEN (“John lost a book that belongs to the NEW professor.”)

d) * Ostannjoju Ivan zadovolenoho kontrol’noju

last.F.SG.INST John.NOM satisfied.M.SG.ACC test.F.SG.INST zustriv studenta met student.M.SG.ACC (“John met a student happy with the LAST test.”)

Availability of constituent discontinuity across the clausal boundary correlates with the presence and the categorial status of the complementizer in the language: Ukrainian:

29. a) * Červonu Mykola znaje ščo Ivan kupyv mašynu

red.F.SG.ACC Michael.NOM knows that John.NOM bought car.F.SG.ACC (“Michael knows that John bought a/the RED car.” or “As for a/the RED car, Michael knows that John bought one/it.”)

a) Červonu Mykola xoče kupyty mašynu

red.F.SG.ACC Michael.NOM wants to-buy car.F.SG.ACC “Michael wants to buy a RED car.” or “As for a RED car, Michael wants to buy one.”

I also account for Determiner Spreading in definite discontinuous DPs in Modern Greek, preposition stranding, and preposition doubling.

I provide the following explanation for the facts listed above. (1) Nominal functional categories are required to be base-generated next to the noun by the theory of extended projection. Adjectives, however, are lexical categories and are not tied to the nominal extended projection at the outset of the derivation. I adopt Baker’s (2003) treatment of adjectives as a default lexical category that can be generated in any syntactic position in

17

the clause as long as that position permits free adjunction. I further claim that an adjectival modifier base-generated at a distance from the noun it modifies is licensed by agreement with the noun in the same way as the adjectival modifier base-generated inside the noun phrase is. The agreement relation established between the noun and the adjectival modifier associates the adjective with the nominal extended projection through feature sharing, thus generating an agreement-based discontinuous constituent. Since the MCL linearization algorithm linearizes all the items according to their F-value based on their association with a particular extended projection, the agreement-based association of the long-distance adjectival modifier with the nominal extended projection determines the linearization of the nominal functional categories before the adjectival modifier. The F-value of all lexical items, however, is the same; therefore, these items can be re-ordered in relation to each other in the hierarchical structure and this re-ordering is maintained in the course of their linearization at PF.

Full document contains 406 pages
Abstract: In this dissertation, I examine discontinuous constituents in Ukrainian and Modern Greek and advance the Radical Discontinuity Hypothesis. I argue that discontinuous constituents in which an adjectival modifier surfaces in a distance from the noun it modifies do not result from splitting a single phrasal constituent by way of movement but are the product of long-distance concord. An adjectival modifier is base-generated at a distance from the noun and is licensed by agreement. I demonstrate that the Radical Discontinuity Hypothesis explains freedom of lexical item ordering in discontinuous constituents, complex cases of discontinuity that involve tripartitioning of the noun phrase, and contrasts in the availability of movement and discontinuity in similar syntactic environments. While the order between lexical categories that form an agreement-based discontinuous constituent is free, functional categories always linearly precede the left-most lexical item associated with the discontinuous constituent. I claim that the surface distribution of functional categories in discontinuous constituents is determined not in narrow syntax but in the course of linearization of syntactic structure at the interface with PF. I argue that the F-value ordering principle responsible for extended projection formation in narrow syntax (Grimshaw (2005)) also guides linearization of the hierarchical structure at PF. By establishing a correlation between the asymmetric c-command and the F-value based ordering of heads of an extended projection, I formulate the Mapping Constraint on Linearization (MCL) and offer a linearization algorithm that implements it. The MCL algorithm incorporates the F-value ordering principle into the LCA of Kayne (1994) and ensures that the F-value ordering relations are enforced in a linear string. The application of the MCL algorithm extends beyond the phenomenon of constituent discontinuity and is examined in relation to various movement transformations. The MLC algorithm restricts the output of both XP and head movement, accounts for the distribution of PP-modifiers and Genitive possessors in movement-based discontinuous constructions, and derives Determiner Spreading in Modern Greek. I also examine locality restriction on movement and agreement in Ukrainian and demonstrate that agreement is constrained not in terms of intervention effects but in terms of agreement domains, which are distinct from PIC.