WH-QUANTIFICATION IN VIETNAMESE

by

Thuan Tran

A dissertation submitted to the Faculty of the University of Delaware in partial fulfillment of the requirements for the degree of Doctor of Philosophy in Linguistics

Winter 2009

Copyright 2009 Thuan Tran All Rights Reserved

UMI Number: 3344072

INFORMATION TO USERS

The quality of this reproduction is dependent upon the quality of the copy

submitted. Broken or indistinct print, colored or poor quality illustrations and

photographs, print bleed-through, substandard margins, and improper

alignment can adversely affect reproduction.

In the unlikely event that the author did not send a complete manuscript

and there are missing pages, these will be noted. Also, if unauthorized

copyright material had to be removed, a note will indicate the deletion.

______________________________________________________________

UMI Microform 3344072Copyright 2009 by ProQuest LLC

All rights reserved. This microform edition is protected against unauthorized copying under Title 17, United States Code.

_______________________________________________________________

ProQuest LLC 789 East Eisenhower Parkway

P.O. Box 1346 Ann Arbor, MI 48106-1346

WH-QUANTIFICATION IN VIETNAMESE

by

Thuan Tran

Approved: __________________________________________________________ Frederick Adams, Ph.D. Chair of the Department of Linguistics and Cognitive Science Approved: __________________________________________________________ Thomas M. Apple, Ph.D. Dean of the College of Arts and Sciences Approved: __________________________________________________________ Debra Hess Norris, M.S. Vice Provost for Graduate and Professional Education

I certify that I have read this dissertation and that in my opinion it meets the

academic and professional standard required by the University as a dissertation for the degree of Doctor of Philosophy.

Signed: __________________________________________________________ Benjamin Brueing, Ph.D. Professor in charge of dissertation I certify that I have read this dissertation and that in my opinion it meets the

academic and professional standard required by the University as a dissertation for the degree of Doctor of Philosophy.

Signed: __________________________________________________________ Satoshi Tomioka, Ph.D. Member of dissertation committee I certify that I have read this dissertation and that in my opinion it meets the

academic and professional standard required by the University as a dissertation for the degree of Doctor of Philosophy.

Signed: __________________________________________________________ Gabriell Hermon, Ph.D. Member of dissertation committee I certify that I have read this dissertation and that in my opinion it meets the

academic and professional standard required by the University as a dissertation for the degree of Doctor of Philosophy.

Signed: __________________________________________________________ Howard Lasnik, Ph.D. Member of dissertation committee

iv

ACKNOWLEDGMENTS

I am grateful to my high school teacher, Cao Xuân Duẫn, and to my college

teachers, Nguyễn Tư Trừng and Trần Văn Phương, who awoke in me the love of

language. I thank my college teacher, Trần Văn Phước, who provided me with a basic

knowledge of linguistics and advised me to take it seriously in my academic life.

I am deeply indebted to Edward Baker for having confidence in me and

making my graduate studies in the U.S. possible. I give my thanks to Susan Alpert for her

special commitment in my scholarly career and Nguyễn Nam for his administrative

assistance. I am thankful to the Harvard Yenching Institute for their generous funding of

my studies at the University of Massachusetts and the University of Delaware.

I am grateful to my first linguistics teachers in the U.S.: Esther Torrego for

her enthusiastic lectures on generative linguistics and Susumu Kuno for his skeptical

lectures, which showed me how to look at things from another perspective and why it is

important to do so.

I owe this dissertation to my professors at the University of Delaware. I am

grateful to Satoshi Tomioka for welcoming me as a lodger at his Newark home during the

summer, and for his Zen-flavored questions, which brought me a sudden enlightenment

so that I could come up with a coherent story to tell. I thank my supervisor, Benjamin

Bruening, for being patient and supportive. His questions and suggestions help me

transform my thinking and utilize theoretical linguistic tools effectively to navigate safely

in the uncharted ocean of Vietnamese language. My thanks also go to Jane Creswell for

her administrative support.

v

I am honored and proud to have Howard Lasnik as an external member of the

dissertation committee. I am very grateful to him for his proposals and commentary. I am

thankful to Norvin Richard, Andrew Simpson, Nigel Duffield, Zeljko Boskovic, Ivano

Caponigro, Maria Polinsky, Peter Cole and Gabriella Hermon for their suggestions and

comments on my work. I thank the Linguistic Society of America for the grant that

funded my attendance at the Linguistic Society of America Summer Institute at MIT and

Harvard in 2005.

I thank Robert Kluender for getting me involved in teaching Vietnamese

linguistics analysis at the University of California, San Diego. I also pay my thanks to

Catherine Alioto and John Moore for their timely response to administrative issues.

I am grateful to my colleagues in the department: Chang-yong Sim, Chun-

Chieh Hsu, Yassir Nassanius Tjung, Durvasula Karthik, Min-Jeong Son, Sean Madigan,

Karl M. Chandler and many others, for making my graduate life at the University of

Delaware less stressful and more memorable. My special thanks go to MaryEllen

Cathcart and Nadya Pincus for having proofread my dissertation. My years in Newark

have been enriched thanks to my neighbors in Conover Apartments. I thank Shin Fukuda

and KimLoan Hill for helping me in my first days in San Diego, and my neighbors at La

Jolla Del Sol for sharing with me new experiences in California. Thanks also go to my

friends in Vietnam, Nguyễn Văn Hiệp and Phạm Anh Tú, and Vietnamese students in

California for providing me with judgments on Vietnamese grammaticality. I am grateful

to Kevin Bowen, Phan Thị Ngọc Chấn, Nguyễn Bá Chung, Vũ Xuân Sơn, Trương Như

Thông, Bùi Hanh and Đàm Quang Pháp for their advice and friendship.

My thanks and love go to my wife, my daughter, and my son, whose

presence helped make those white cold winters less severe, and whose laughs drove away

my nervousness. I owe them an apology for making them undergo long distance

vi

movement and trace deletion. I promise I will do reconstruction after finishing this

dissertation.

My parents are always in my mind wherever I am. I long for a reunion, as do

they.

vii

TABLE OF CONTENTS

ABSTRACT………………………………………………………………………….......xi

Chapter

1 INTRODUCTION…………………………………………………………...........1

1.1 Introduction………………………………………………………………..1

1.2 Outline of the Thesis………………………………………………............8

2 VIETNAMESE SYNTAX……………………………………………………….10

2.1 Introduction…………………………………………………………........10

2.2 Clausal Structures…………………………………………………..........10

2.2.1 Matrix Clauses…………………………………………………...10

2.2.2 Embedded Clauses………………………………………….........15

2.2.3 Yes-no Questions………………………………………………...18

2.2.4 Subject Position………………………………………………….22

2.3 Preverbal Particles: Tense/Aspect………………………………………..25

2.4 Negation………………………………………………………………….30

2.4.1 Negative Polarity Items………………..........................................38

viii

2.5 Sentence Final Particles….........................................................................41

2.6 Prepositional Phrases.................................................................................43

2.7 Adjective Phrases and Adverb Phrases......................................................45

2.8 Verb Phrases..............................................................................................47

2.9 Noun Phrases.............................................................................................50

2.10 Relativization.............................................................................................56

2.11 Pronouns....................................................................................................58

2.12 Complex Sentences with Adverb Clauses.................................................71

2.13 Island Effects.............................................................................................75

3 UNIVERSAL QUANTIFICATION......................................................................82

3.1 Introduction................................................................................................82

3.2 Previous Accounts……………………………………………………….87

3.2.1 Nishigauchi (1986, 1991)…………………………………………..87

3.2.2 Cheng (1991, 1995), Huang and Cheng (1996)………………........94

3.2.3 Shimoyama (2001, 2006), Kratzer and Shimoyama (2002)….......100

3.3 Wh-universals in Vietnamese …………………………………………..106

3.4 Proposal…………………………………………………………………112

3.4.1 Cũng: Focus Sensitive Particle…………………………………112

3.4.2 Cũng as a Universal Quantifier……………………………........114

3.5 A Special Case of Wh-universals………………………………………123

3.6 Conclusion……………………………………………………………...128

4 EXISTENTIAL QUANTIFICATION……………………………………….....131

ix

4.1 Introduction……………………………………………………………..131

4.2 Literature Review……………………………………………………….132

4.2.1 Cheng (1991)……………………………………………….......132

4.2.2 Lin (1998, 2004)………………………………………………..135

4.3 Bare Wh-indefinites…………………………………………………….139

4.3.1 Licensing Condition………………………………………………141

4.3.2 Intermediate Scope………………………………………………..143

4.3.3 Analysis…………………………………………….......................144

4.3.3.1 Locality Condition and Narrow Scope…………145

4.3.3.2 Intermediate Scope: Not Always Available……153

4.4 Non-bare Wh-indefinites……………………………………………….159

4.5 Conclusion……………………………………………………………...166

5 A NEO-HAMBLIN ANALYSIS OF WH-QUESTIONS……………………...168

5.1 Introduction……………………………………………………………..168

5.2 Data……………………………………………………………………..169

5.2.1 Argument Wh-phrases………………………………………….170

5.2.2 Island Effects……………………………………………….......174

5.2.3 Intervention Effects……………………………………………..177

5.2.4 D-linking and Question Particles……………………………….183

5.2.5 D-linked Wh-phrases…………………………………………...189

5.2.6 A Matrix-Embedded Contrast…………………………………..190

5.3 Previous Analysis……………………………………………………….192

x

5.4 A Neo-Hamblin Analysis……………………………………………….195

5.5 Wh-islands...............................................................................................210

5.6 Adjunct Wh-phrases.................................................................................216

5.6.1 Long Distance Readings………………………………………..220

5.6.2 Subjacency……………………………………………………...223

5.7 A Complex NP Island Experiment……………………………………..227

5.7.1 The Experiment…………………………………………………228

5.7.1.1 Control Items……………………………………….......229

5.7.1.2 Test Items……………………………………………….231

5.7.1.3 Results…………………………………………………..233

5.7.1.4 Analysis……………………………………………........236

5.7.2 Implications……………………………………………………..237

5.8 Conclusion……………………………………………………………...238

6 CONCLUSION……………………………………………………………........240

6.1 Summary………………………………………………………………..240

6.2 Typological Claims……………………………………………………..244

6.2.1 Wh-indefinites…………………………………………………..245

6.2.2 Wh-universal……………………………………………………247

6.2.3 Wh-questions…………………………………………………...247

6.3 The Acquisition of Wh-questions………………………………………248

REFERENCES………………………………………………………………………....250

xi

ABSTRACT

This dissertation looks at the dual lives of wh-phrases in Vietnamese, namely

their syntax-semantic behavior in interrogative and non-interrogative contexts. It is a

common idea, since Nishigauchi (1986), and Cheng (1991), that wh-phrases are variables

without any quantificational force, but obtain their interpretations from their binders.

However, I argue that Hamblin (1973) semantics as revived by Kratzer &Shimoyama

(2002) (henceforth neo-Hamblin semantics) is more plausible in that it allows us to

maintain a uniform denotation of wh-phrases, and, at the same time to account for their

apparent various quantificational interpretations.

Chapter I presents a brief review of quantification theories and wh-quantification

in Vietnamese.

Chapter II is concerned with Vietnamese syntax. In this chapter, I investigate

Vietnamese nominal and clausal structures and various topics in the nominal and clausal

domains. Particularly, I am concerned with nominal and clausal structures of

Vietnamese, aspect and tense particles, and sentential final particles. I also examine

various operations such as topicalization, relativization, and scopal interaction between

negation and quantification operators. This chapter serves as a basic preparation for the

discussion in the following chapters.

xii

Chapter III explores one aspect of the non-interrogative life of wh-phrases,

namely when they obtain universal interpretations. When a wh-phrase associates locally

or non-locally with cũng, a focus sensitive particle, a universal interpretation arises. I

propose that in local cases this focus sensitive particle is head of a Focus Phrase in overt

syntax that requires a focused element in its Spec. The focus particle is a universal

quantifier that takes a propositional alternative set as its argument, which yields a

universal interpretation. I argue that the non-local association between a wh-phrase and

this particle is only apparent given that this particle is not obligatory and its presence

adds a concessive flavor. I therefore propose that a universal interpretation in non-local

contexts derives from a covert universal quantifier over situations.

Chapter IV continues with the non-interrogative life of wh-phrases with a

discussion of existential interpretations of wh-phrases. I hold that existential

interpretations arise from the existential closure by alternative set-taking operators. The

availability of these operators is subject to licensing. The requirement that a licensor must

c-command its licensee, but does not need to be clause-mate with it, fits into the picture

of neo-Hamblin semantics. In addition, a wh-phrase can form a constituent with a

particle. This constituent does not need licensing and can take any scope. I propose that

the existential interpretation of this constituent is derived from the existential closure of a

choice function variable introduced by the particle.

Chapter V looks into the interrogative life of wh-phrases. Wh-questions in

Vietnamese employ two strategies: LF movement and non-movement. LF movement is

required by the Hamblin interpretative mechanism: A wh-phrase must undergo LF

xiii

movement to Spec, CP to stop the alternative set introduced by the wh-phrase from

expanding. When a question operator is available, namely, when it is licensed by a wh-

particle, no movement is needed: The propositional alternative set is captured by this

operator, through which a question reading obtains.

Chapter VI summarizes the main points in the five chapters and indicates what

this dissertation contributes to the field. I hold that it is universal that wh-phrases denote

sets of in individual alternatives. Their quantificational forces derive from the operators

that capture alternative sets. Languages differ with respect to what type of operators is

used. With respect to wh-questions we propose that overt movement is purely syntactic,

and it is universal that wh-in-situ languages do not employ covert movement. The covert

movement in Vietnamese is language specific. Finally, I offer an account for the island

effects in Vietnamese from the perspective of language acquisition.

1

Chapter 1

INTRODUCTION

1.1 Introduction

This dissertation looks at the dual lives of wh-phrases in Vietnamese: As

interrogative and non-interrogative expressions. For example, the wh-phrase meaning

‘who’ in (1a) is interrogative in that its occurrence in the sentence gives rise to a wh-

question reading, while its presence in (1b) is non-interrogative because the sentence in

which it appears is not construed as a wh-question. Instead, sentence (1b) asserts that Tan

just met someone. Thus, the wh-phrase in (1b) is interpreted as an indefinite NP.

(1) a. Tân mới gặp ai (thế)?

Tan just meet who PRT

‘Who did Tan just meet?’

b. Chắc Tân mới gặp ai.

maybe Tan just meet who

‘Maybe Tan just met someone.’

2

c. Chắc Tân mới gặp ai đó.

maybe Tan just meet who DEM

‘Maybe Tan just met someone.’

The ability of a wh-phrase to switch between interrogative and non-interrogative life is

observed cross-linguistically. Haspelmath (1997) reports that 63 out of 100 languages

have wh-phrase-based indefinite NPs and that a wh-phrase-based indefinite NP

(henceforth wh-indefinites) can be identical to a wh-phrase as in (1b) or can be formed by

attaching additional morphology to a wh-phrase, as in (1c).

The essential difference between the wh-indefinite in (1b) and that in (1c) is that

while the former requires a licensor (the epistemic operator ‘maybe’ in this case) and

takes scope lower than its licensor, the latter does not need licensing and can take widest

scope.

Vietnamese wh-indefinites display a property similar to that of lexical indefinite

NPs (see Heim 1982). That is, they can vary their quantificational interpretations

depending on contexts. For illustration, consider the examples below.

(2) a. Ai Tân cũng gặp.

who Tan CUNG meet

‘Tan met everyone.’

3

b. Tân cũng gặp ai ?

Tan CUNG meet who

‘Who did Tan also meet?’

As illustrated in (2a), the wh-phrase ‘who’ obtains a universal interpretation when it

associates with the particle cũng. This association requires the wh-phrase to move overtly

from its base position as object of the verb ‘meet’ to the sentence-initial position: If a wh-

phrase stays in its base position, the sentence is interpreted as a question, as in (2b), with

the particle cũng construed as an additive particle.

Accounts for this quantificational variability of wh-phrases abound. One of the

most influential theories is the open predicate analysis of Nishigauchi (1986, 1991),

based on Heim’s (1982) unselective binding theory. Originally, Heim’s (1982) theory

aims to account for the quantificational variability of lexical indefinite NPs. For example,

consider the following sentences.

(3) a. If a man owns a donkey, he always beats it. → always: ∀

a’. ∀x, y [ man (x) & donkey (y) & own (x, y) → x beats y]

b. Sometimes if a cat falls from the fifth floor, it survives. → sometimes: ∃

b’. ∃ x [cat (x) & fall (x) & (survive (x)]

As shown in (3a’), which is the logical form of (3a), the indefinite NPs ‘a man’ and ‘a

donkey’ in (3a) are construed as universally quantified expressions. In contrast, the

4

indefinite NP ‘a cat’ in (3b) is interpreted as an existentially quantified expression, as

illustrated in the logical form (3b’).

Heim (1982) proposes that indefinite NPs are open predicates with variables. For

example, the denotation of a student is student (x). Indefinite NPs are, therefore, non-

quantificational. Their quantificational interpretations come from the quantifiers that bind

them. The binding is unselective. In this case, the quantifiers are the adverbs of

frequency, namely, ‘always’, and ‘sometimes’.

Extending Heim’s (1982) theory to wh-indefinites in Japanese, Nishigauchi

(1986, 1991) holds that Japanese wh-phrases are like lexical indefinite NPs in that they

denote open predicates with variables. For example, the denotation of the wh-phrase

dare, ‘who’ is person (x), and that of nani, ‘what’, is thing(x), but in (4), they are

construed as quantificational expressions.

(4) Dare- mo ga nani-ka o tabe-te-iru.

who MO NOM what KA ACC eating-be

‘Everyone is eating something.’

The quantificational forces of these wh-phrases, according to Nishigauchi (1986,

1991), come from their binders, namely, from mo, a universal quantifier, and ka, an

existential quantifier. Therefore, dare-mo, ‘who-mo’, is equivalent to its English

counterpart ‘everyone’, and nani-ka, ‘who-ka’, means ‘someone’.

5

Nishigauchi’s analysis accounts nicely for the quantificational variability of wh-

phrases. However, if a wh-phrase and a lexical NP have the same denotation, namely, if

both who and a person denote an open predicate with variable: person (x), then it would

be difficult to account for why (5b) is ungrammatical as a question as opposed to (5a).

(5) a. Who came?

b. *A person came?

Cheng (1991), based on Nishigauchi’s theory, provides the same denotation for

wh-indefinites in Chinese, namely, a wh-phrase like shei, ‘who’, denotes an open

predicate with variable. This analysis faces a similar problem as shown in the following

examples.

(6) Shei xian lai, shei jiu xian chi.

who first come who then first eat

‘If X comes first, X eats first.’ (Cheng and Huang 1996)

(7) *You ren xian jinlai, you ren xian chi.

there’s a person first come there’s person first eat

‘If a person 1 first comes, a person 1 first eats.’ (Bruening and Tran 2006)

Cheng and Huang (1996) refer to sentences like (6) as bare conditionals because

they lack the subordinating element ‘if’. Bruening and Tran (2006) call it wh-conditionals

6

to refer to the defining property of these conditionals: They must contain a matched pair

of wh-phrases, one in the antecedent and one in the consequent. If a wh-indefinite and a

lexical indefinite have the same denotation, replacing shei, ‘who’, in (6) with ren,

‘person’, as in (7) would result in grammaticality, contrary to fact.

Further issues that challenge these analyses will be discussed in chapters 3 and 4.

In what follows, I briefly introduce the neo-Hamblin analysis, revived by Shimoyama

(2001) and Kratzer & Shimoyama (2002), based on Hamblin (1973) semantics of wh-

questions, and argue that it fares better than the open predicate theory.

Basically, in their theory, wh-phrases are non-quantificational, but denote sets of

individual alternatives, while other lexical items denote singleton sets of their standard

denotations. Via Hamblin functional application, given in (8), wh-phrases form sets of

alternatives of higher type. These sets then must be captured by an alternative set-taking

operator in order not to crash. Their theory does not need to posit a local relation between

quantifiers and wh-phrases.

(8) Hamblin Functional Application

If α is a branching node with daughters β and γ, and [[β]]w,g ⊆ Dσ and [[γ]]w,g⊆ D<στ>,

then [[α]]w,g = {a∈ Dτ :∃b ∃c [ b∈[[β]]w,g & c∈ [[γ]]w,g & a = c (b)]}.

For illustration, consider the sentence in (9).

7

(9) Tân gặp ai.

Tan meet who

(10) [[ TP]] w, g = [[Tân meet who ]] w, g

= {p: human (x) (w) & p = λw’. meet (Tan) (x) (w’)}

As illustrated in the diagram (11), the wh-phrase ‘who’ denotes a set of individual

alternatives, while the verb ‘meet’ denotes a singleton set. The individual alternative set

introduced by ‘who’ combines with other elements, namely, the NP ‘Tan’ and the verb

‘meet’ and forms a propositional alternative set, as in (10).

(11) CP [[TP]] w, g ={p: human(x) (w)& p = λw’. meet (Tan)(x) (w’)} Op TP Tan

VP = [[ VP ]] w, g = {λyλw’. meet (y) ([[ who]]w, g ) (w’)} V NP meet who = [[ who]]w, g = { x: human (x) (w)} [[meet]]w, g = {λx λyλw’. meet (y) (x) (w’)}

If the set in (10) is captured by a question operator, then by definition the question

operator returns the same set in (10), which is the denotation of a question in Hamblin

(1973) semantics. If the set in (10) is taken by an existential operator, the set will be

existentially closed as in (12).

(12) [[ ∃α ]]w,g = {λw’. ∃p [p∈ [[ TP ]]w,g & p(w’)=1]}

8

= {λw. ∃x [human (x) (w) & met (Tan) (x) (w)]}

= that Tan met at least one person

1.1 Outline of the Thesis

The thesis consists of 6 chapters. The second chapter introduces basic Vietnamese

syntax, investigating issues related to the nominal and clausal domains. This chapter

serves as background for the discussion in the chapters to follow. Chapter 3 looks at wh-

universal interpretations in Vietnamese. It is argued that Vietnamese wh-universals are

best analyzed within the framework of neo-Hamblin semantics proposed by Shimoyama

(2001) and Kratzer and Shimoyama (2002). Part of this chapter is a revision, from a neo-

Hamblin semantics perspective, of Bruening and Tran’s (2006) analysis of wh-

conditionals in Vietnamese. Chapter 4 is on wh-existential quantifications. In this chapter

I investigate the contrast in syntactic and semantic behavior of bare wh-phrases and non-

bare wh-phrases in existential contexts and propose a neo-Hamblin analysis for these two

types of wh-phrases. Chapter 5 is concerned with wh-questions. Based on Bruening and

Tran’s (2006) analysis of wh-questions in Vietnamese, I employ the neo-Hamblin

semantics to reanalyze wh-questions in Vietnamese with a view to providing a uniform

analysis for wh-phrases and wh-indefinites, namely, they all denote sets of individual

alternatives. The last chapter, chapter 6, concludes the thesis with typological claims on

quantification in natural languages and spells out the theoretical and descriptive

contribution of the thesis to the understanding of human language. Specifically, it is

concluded that cross-linguistically wh-phrases denote sets of individual alternatives and

9

that their interpretations come from neighboring operators. The interpretative mechanism

requires neither covert nor overt movement in wh-questions. Covert movement in

Vietnamese wh-questions is language-specific.

10

Chapter 2

VIETNAMESE SYNTAX

2.1 Introduction

In this chapter I provide a basic and essential description of Vietnamese syntax, as

a background for the discussions in the next chapters. Vietnamese is an isolating

language where grammatical relations such as subject, object are manifested through

word order, not through case or agreement morphology. For example, a nominal

constituent is a subject when it precedes the verb, but an object if it follows the verb.

Also, a lexical item can perform cross-categorical functions, namely, as a preposition or a

verb or a noun, without any change in form.

2.2 Clausal Structures

2.2.1 Matrix Clauses

The basic word order of a declarative clause in a pragmatically neutral context is

S(ubject) V(erb) O(bject) as illustrated in (1a), where the subject-agent is represented by

11

the bare NP ‘elephant’, and the object-theme by the bare NP ‘sugarcane’.1

(1) Voi ăn mía.

elephant eat sugarcane

‘Elephants eat sugarcane.’

There is no morphological case and no agreement in Vietnamese. For example, in

(2a, b) the plural third person pronoun ‘they’ and the singular third person pronoun ‘she’

remain unchanged irrespective of whether they function as subject or object.2 Tense

interpretation is realized by a preverbal particle and the verb does not change its form:

The past tense in (2a) is derived from the particle đã, and the future tense in (2b) from the

particle sẽ. Also, verbs and subjects do not show any morphological agreement: The verb

‘help’ is morphologically the same irrespective of the φ-features of the subject pronouns,

namely, plural in (2a), singular female in (2b), and singular male in (2c).

1 Abbreviations used: ASP = Aspect, FUT = Future, PST = Past, DEM= Demonstrative, CL = Classifier, TOP = Topic, NEG = Negation, PART = Particle, REL = Relative, COMP = Complementizer, PLU = Plural, Q = Question 2 The pronouns glossed as ‘her’ and ‘she’ in (2a, b) are composed of two words with the first word cô being a lexical item, meaning ‘aunt’ and ấy a demonstrative meaning ‘that’. Similarly, the pronoun glossed as ‘he’ in (2c) consists of a lexical item meaning ‘brother’ and ấy, a demonstrative meaning ‘that’. We will discuss these types of pronouns in the subsection on pronouns.

12

(2) a. Họ đã giúp cô ấy.

they PST help her

‘They helped/have helped her.’

b. Cô ấy sẽ giúp họ.

she FUT help them

‘She will help them.’

c. Anh ấy sẽ giúp họ.

he FUT help them

‘He will help them.’

A nominal expression can be used as a nominal predicate. The use of a copular

verb is dependent on the types of nominal predicates. For example, a sentence with an

occupation-denoting expression as its predicate requires a copular verb, as in (3a), but a

sentence with an age-denoting expression does not, as in (3b).

(3) a. Anh-Thơ là học-sinh.

Anh-Tho be student

‘Anh-Tho is a student.’

b. Anh-Thơ mười bốn tuổi.

Anh-Tho ten four year

‘Anh-Tho is fourteen years old.’

13

In a pragmatically marked context, that is, in a topic-comment structure, (4a) can

be realized as (4b) with the topicalized object in the initial position optionally marked by

the topic marker thì.3

(4) a. Người thợ-săn này bắt con gấu ấy.

CL hunter DEM catch CL bear DEM

‘This hunter caught that bear.’

b. Con gấu ấy ( thì ) người thợ-săn này bắt.

CL bear DEM TOP CL hunter DEM catch

‘That bear, this hunter caught.’

The topicalized NP ‘that bear’ in (4b) receives the same thematic role as the object NP in

(4a). However, a topic phrase can bear no thematic relation to the predicate, as illustrated

in (5).

(5) Ông Tân (thì) con-cái rất thành-đạt.

Mr Tan TOP children very successful

‘As for Mr. Tan, his children are very successful.’

3 For the discussion of basic word order see Song (2001), Payne (1997). Note that some classifiers are derived directly from lexical items such as người ‘human’ as in (4a, b), or con, ‘child’, as in (4b).

14

The topic ông Tân ‘Mr Tan’ in (5) is unlike the topic con gấu ấy ‘that bear’ in (4b) in that

it bears no thematic relation to the predicate. The property of being successful is

predicated of the children, not of Mr Tan.

A topic can be related to a predicate via a resumptive pronoun as in (6).

(6) Quyển sách này cô ấy sẽ không phải mua nó mà.

CL book DEM she FUT NEG must buy it PART

‘This book, she will not have to buy it (I am sure).’

The pronoun ‘she’ is the subject of the predicate ‘buy’. The topic phrase ‘this book’ in

(6a) is related to the pronoun ‘it’ in object position. The sentence is also accompanied by

a final particle, the meaning of which is roughly represented by the translation in

brackets: ‘I am sure’. When a resumptive pronoun is used, there is no island effect, as

exemplified in the sentences below.

(7) a. *Con gấu ấy tôi đã gặp [ người thợ săn [ mà đã tấn công ]].

CL bear DEM I PST meet CL hunter REL ASP attack

*‘That bear, I met the hunter that attacked.’

b. Con gấu ấy tôi đã gặp [người thợ săn [ mà nó đã tấn công]].

CL bear DEM I PST meet CL hunter REL it PST attack

‘That bear, I met the hunter that it attacked.’

15

The topic phrase ‘that bear’ in (7a) is related to a gap inside a Complex Noun Phrase

island, and the sentence is ungrammatical. The sentence is well-formed if the gap is filled

with a resumptive pronoun as in (7b).

Adverbs modifying verbs must follow objects, not verbs, as shown by the contrast

in the examples below.

(8) a. Các sinh-viên này đã giải bài toán ấy rất nhanh mà.

PLU student DEM PST solve CL math DEM very fast PART

‘These students just solved that math problem very fast.’

b. ?? Các sinh-viên này đã giải rất nhanh bài toán ấy mà.

PLU student DEM PST solve very fast CL math DEM PART

‘These students just solved that math problem very fast.’

In short, the matrix template of a full fledged sentence in Vietnamese is in (9).

(9) Topic-Subject-Tense/Aspect particle-Negation-Modal verb-Main verb-Object-

Adverb- Final particle

2.2.2 Embedded Clauses

In the following an argument position is taken by a clause, not an NP. The

argument clause is embedded in the matrix clause and functions as subject or object of

the matrix clause.

16

(10) a. [ Bắc Hàn vừa thử bom hạt-nhân ] khiến cả thế-giới lo-ngại.

North Korea just test bomb nuclear cause all world worry

‘That North Korea just tested nuclear bombs made the whole world worried.’

b. Hiền biết (là/rằng ) [ Bắc Hàn vừa thử bom hạt-nhân ] .

Hien know COMP North Korea just test bomb nuclear

‘Hien knows that North Korea just tested nuclear bombs.’

c. *[Là/Rằng Bắc Hàn vừa thử bom hạt-nhân ] khiến cả thế-giới lo-ngại.

COMP North Korea just test bomb nuclear cause all world worry

‘That North Korea just tested nuclear bombs made the whole world worried.’

There is no change in word order: The word order of both the subject embedded

clause in (10a) and object embedded clause in (10b) is SVO. The matrix clause word

order is also SVO. The difference between subject and object embedded clause is the

occurrence of the complementizer: là/rằng is not allowed in subject embedded clauses

and is optional in object embedded clauses (as shown by the contrast in (10a) and (10c)),

which is unlike English, where the presence of a complementizer in a subject embedded

clause is obligatory. It is possible to hold that subject clauses do not exist and (10a) in

fact means something like ‘N. Korea tested nuclear bombs, and that made the world

worry’. In this analysis, ‘and that’ would be phonologically null, so (10a) would be

considered a compound sentence. One argument against this analysis is the lack of pause

after the object of the embedded verb ‘test’, orthographically represented by a comma.

Even though (10a) is intuitively interpreted as a single sentence, it is difficult to show

17

clearly how the subject clause is grammatically marked because there is subject-verb

agreement in Vietnamese.

It is more common to nominalize the subject embedded clause with a

‘nominalizer’ deriving from a lexical item, meaning something like ‘thing’, ‘deed’, as

shown in (11).

(11) [Việc một học- sinh lớp 12 tấn-công website Bộ Giáo-dục Đào-tạo ]

deed one student class 12 attack website ministry education training

đã gây-ra nhiều luồng ý-kiến khác nhau.

PST cause many stream opinion different (VN Express, Jan 04, 2007)

‘The fact that a 12 th grade student hacked into the Education Ministry website caused a

heated controversy.’

A topic in embedded clauses occupies a position lower than the complementizer,

as in (12b).

(12) a. Tôi biết là [ ông ấy cất mấy quyển sách đó trong tủ].

I know COMP he keep some CL book DEM in case

‘I know that he keeps those books in the case.’

b. Tôi biết là [ mấy quyển sách đó ông ấy cất trong tủ ].

I know COMP some CL book DEM he keep in case

‘I know that those books he keeps in the case.’

18

In summary, the basic word order of a declarative clause in a pragmatically

unmarked context is subject-verb-object. There is no case morphology and agreement

morphology. A topic in a topic comment structure is optionally marked by a topic

marker. Embedded clauses have the same form as matrix clauses except that object

clauses optionally have a complementizer.

2.2.3 Yes-no Questions

A yes-no question can be derived from a declarative sentence by using rising

intonation. For example, a sentence such as (13) can be construed as a yes-no question,

just like its English counterpart in (13b).

(13) Tân đã gặp Lan.

Tan ASP meet Lan

a. ‘Tan met Lan.’

b. ‘Tan met Lan?’

It is, however, more common in Vietnamese to employ a number of final particles

in yes-no questions. These particles may vary dialectally in form, but they consistently

perform the same function, namely, to change a non-interrogative clause into an

interrogative clause. For instance, in (14a) the final particle à is commonly used in the

central dialect, the particle ư in the northern dialect, and hả in the southern dialect.

19

Pragmatically, these particles express speakers’ surprise and doubt toward the

propositional content of the sentence.

(14) a. Tân đã gặp Lan à/ư/hả?

Tan PST meet Lan Q

‘Did Tan meet Lan?’

b. Anh muốn gặp Lan không?

you want meet Lan Q

‘Do you want to meet Lan?

The final particle không can appear by itself, as in (15b) or it can appear with a preverbal

particle, có as shown in (15a).

(15) a. Em có đói không?

you Q hungry Q

‘Are you hungry?’

b. Em đói không?

you hungry Q

‘Are you hungry?’

Note that the particle có precedes the modal verb, not the main verb as in (16a). It

also precedes other particles: i.e. aspect particles, as in (16b), and negative particles, as in

20

(16c). For ease of exposition, let us use the term ‘preverbal particle’ to refer to this

particle.

(16) a. Tân có nên nộp bài không?

Tan Q should hand.in paper Q

‘Should Tan hand in his paper?’

b. Anh-Thơ có đang học bài không?

AnhTho Q ASP study lesson Q

‘Is AnhTho studying?’

c. Cô ấy có không muốn gặp Tân không?

she Q NEG want meet Tan Q

‘Does she not want to meet Tan?’

Normally, when the event encoded by the sentence is interpreted as taking place

in the past, the pair có ... không is used for yes-no questions, with có occuring preverbally

and không sentence-finally. The disjunctive particle hay ‘or’ preceding the sentence-final

không is optional in matrix questions.

(17) Tân có gặp cô gái ấy ( hay) không?

Tan Q meet CL girl DEM or Q

‘Did Tan meet that girl?’

21

The preverbal particle có marks the scope of the question. If it appears in the matrix

clause, the sentence is a matrix question with illocutionary force as in (18a). If it appears

in the embedded clause, it is an embedded question, and the sentence has no illocutionary

force as in (18b). Note that the disjunctive particle hay is obligatory and that the

complementizer rằng, ‘that’, is bad in this case.

(18) a. Tân có biết [ ( rằng) cô ấy đi San Diego ] (hay) không?

Tan Q know that she go San Diego or Q

‘Did Tan know that she went to San Diego?’

b. Tân biết [(* rằng) cô ấy * (có ) đi San Diego *(hay ) *(không ) ]?

Tan know that she Q go San Diego or Q

‘ Tan knows whether she went to San Diego or not.’

The question particles à, hả, and ư are interpreted in association with matrix

clauses only. Thus (19) is interpreted as a matrix question; the embedded question

reading is not available.

(19) Tân biết [ (rằng) cô ấy đi San Diego ] à/ ư/hả?

Tan know that she go San Diego Q

a. ‘Did Tan know that she went to San Diego?’

b.* ‘Tan knows whether she went to San Diego or not.’

22

As shown in (20), an embedded clause as complement of a question-selecting verb such

as muốn biết is introduced optionally by the complementizer liệu.

(20) Tân muốn biết [ ( liệu ) Lan *( có ) gặp Tân hay *(không )].

Tan want know whether Lan Q meet Tan or Q

‘Tan wondered whether Lan met Tan.’

Note that only the pair có...không is used in this embedded question and the occurrence of

both is obligatory.

2.2.4 Subject Position

The template of a Vietnamese sentence in (9), repeated below, indicates that a

subject appears in a position higher than a tense/aspect particle as illustrated in (21).

(9) Topic - Subject - Tense/Aspect particle - Negation - Modal verb - Main verb -

Object - Adverb - Final particle

(21) Anh ấy sẽ đến.

he FUT arrive

‘He will arrive.’

23

However, it is possible for a subject to appear in a position lower than a

tense/aspect particle as illustrated by the example in (22). The two sentences (21) and

(22) are synonymous. The minimal difference between them is that the latter requires the

occurrence of a particle having the same form as the yes-no particle có.

(22) Sẽ *(có) anh ấy đến.

FUT CO he arrive

‘He will arrive.’

Recall that the yes-no question precedes the negative particle, as shown in (16c)

repeated below. By contrast, this particle follows the negative particle, as in (23a).

(16) c. Cô ấy có không muốn gặp Tân không?

she Q NEG want meet Tan Q

‘Does she not want to meet Tan?’

(23) a. Sẽ không có Tân đến.

FUT NEG CO Tan arrive

‘Tan will not arrive.’

b. Tân sẽ không đến.

Tan FUT NEG arrive

‘Tan will not arrive.’

24

The subjects in the examples above are referential expressions and the meanings

of these sentences remain unchanged irrespective of whether a negative marker is

involved or not. Let us consider the following examples where the subjects are

quantificational expressions.

(24) a. Nhiều người sẽ đến.

many person FUT arrive

‘Many people will arrive.’

b. Sẽ có nhiều người đến.

FUT CO many person arrive

‘Many people will arrive.’

The subjects of the two sentences in (24a, b) are quantificational expressions and

the sentences are synonymous. What happens when a negative particle is used?

(25) a. Nhiều người sẽ không đến.

many person FUT NEG arrive

‘Many people will not arrive.’

b. Sẽ không có nhiều người đến.

FUT NEG CO many person arrive

‘Not many people will arrive.’

25

The two sentences in (25) differ truth-conditionally. While (25a) asserts that the number

of people who will not arrive is many, (25b) holds that the number of people who will

arrive is not many, just a few.

In summary, in affirmative contexts, a subject can appear in a position higher or

lower than the tense/aspect particle without affecting the meaning of the sentence.

However, if a negative particle is involved and a subject is a quantificational expression,

then the sentence varies its meaning according to surface word order of the two scope-

bearing elements.

2.3 Preverbal Particles: Tense/Aspect

We observed in the first subsection that tense/aspect interpretation in Vietnamese

derives from tense/aspect particles. These particles always occur pre-verbally. Consider,

for example, the sentence in (26).

(26) Tân đã / sẽ / sắp gặp May.

Tan PST/ FUT/ FUT meet May

‘Tan met/will meet/ will meet May soon.’

The particle đã encodes a past tense interpretation. The particles sẽ and sắp are future

tense particles. However, there is some subtle difference between them: While the former

is a pure future particle, the latter is a ‘near’ future particle. For illustration, consider the

following examples.

26

(27) a. Ngày-mai trời sẽ mưa.

tomorrow sky FUT rain

‘It will rain tomorrow.’

b.# Ngày-mai trời sắp mưa.

tomorrow sky FUT rain

‘It will soon rain tomorrow.’

c. Trời sắp mưa.

sky FUT rain

‘It will rain soon.’

The future particle sẽ is compatible with the time phrase ‘tomorrow’ as in (27a), but the

future particle sắp is pragmatically odd when it co-occurs with this time phrase as shown

by the contrast between (27b) and (27c). (27c) is appropriately uttered in a context where

a speaker detects a black cloud that he believes will soon bring rain. In a similar context

where a speaker who is cooking rice thinks the rice is going to be edible in a second,

sentence (28a) is pragmatically odd while (28b) is appropriate. To express this subtle

difference, the adverb ‘soon’ is added to the translation of the ‘near’ future particle.

(28) a. # Cơm sẽ chín.

rice FUT edible

‘Rice will be edible.’

27

b. Cơm sắp chín.

rice FUT edible

‘Rice will soon be edible.’

The particle đã also functions as a perfective aspect particle. In (29), Hoa’s being

married has been realized in the past and remains so at the time of speaking. A past tense

interpretation as in (29b) is not available.4

(29) Hoa đã có chồng.

Hoa ASP have husband

a. ‘Hoa has been married.’

b. *‘Hoa was married.’

As illustrated in (30), the progressive aspect is encoded by the particle đang.

(30) Anh-Thơ đang nói chuyện với Nhi.

AnhTho ASP talk story with Nhi

‘Anh-Tho is talking with Nhi.’

4 As suggested by Bruening (p.c.) the perfective reading of the particle đã in (29) as opposed to the absence of this reading in (26) is probably due to the telic-atelic in the predicates. Namely, ‘meet’ is a telic predicate, while ‘be married’ is atelic.

28

Note that the occurrence of a particle is not obligatory because tense interpretation

can derive from adverbial phrases such as ‘yesterday’ or ‘next year’, as shown by the

contrast between the past tense interpretation of (31a) and the future tense interpretation

of (31b).

(31) a. Hôm qua Tân gặp May.

yesterday Tan meet May

‘Tan met May yesterday.’

b. Ngày mai Tân gặp May.

tomorrow Tan meet May

‘Tan will meet May tomorrow.’

When there is no time phrase, a sentence with a stative verb is commonly

construed as present tense, as in (32a), while a sentence with an eventive verb is normally

interpreted as past tense, as in (32b).

(32) a. Tân thích bóng-đá.

Tan like football

‘Tan likes football.’

b. Tân gặp May.

Tan meet May

‘Tan met May.’

29

Past tense interpretation can also be encoded by the pre-verbal particle có. The

difference between đã and có is that while the latter adds some emphatic flavor to the

sentence as shown by the translation in (33), the former does not. For an extensive

discussion of có interested readers can refer to Duffield (2007), who proposes that có is

the overt head of the Assertion Phrase, which is base-generated lower than the Tense

head.

(33) Tân có gặp May.

Tan PST meet May

‘Tan did meet May.’

Note in passing that when a particle having the same form as this particle occurs

pre-verbally in a clause in a complex sentence as in (34), it marks that clause as the

antecedent of a conditional sentence. The subordinating element ‘if’ is optional in this

case. No past tense interpretation arises in this clause.

(34) (nếu) Anh có gặp Tân thì nhắc Tân mua sách.

(if) you CO meet Tan then remind Tan buy book

‘If you meet Tan, remind him to buy books.’

30

In summary, in Vietnamese tense and aspect interpretations are derived from

preverbal particles. When a time phrase is available, the occurrence of a preverbal

particle is optional.

2.4 Negation

Sentential negation is realized by a number of negative particles: không, chẳng,

chưa, đâu. These particles precede modal verbs or main verbs, but follow aspect

particles. Of these particles, only the particles không and chẳng are interchangeable. The

particle chưa provides extra irrealis information, that is, the event denoted by the

sentence has not been realized. The particle đâu implies a past tense interpretation,

namely, the event did not happen.

(35) a. Tân không / chẳng gặp May.

Tan NEG meet May

‘Tan will not/did not meet May.’

b. Tân chưa gặp May.

Tan NEG meet May

‘Tan has not met May.’

c. Tân đâu gặp May.

Tan NEG meet May

‘Tan did not meet May.’

31

As shown in the examples above, the particles không and chẳng are interchangeable and

are neutral with respect to tense interpretation: (35a) can be construed as past tense or

future tense, depending on contexts. By contrast, the particles chưa and đâu are not

neutral: the former encodes an irrealis interpretation (35b), and the latter a past tense

interpretation (35c).

Negation can also be marked by a pair of particles, có...đâu, with có in preverbal

position and đâu in final position.

(36) a. Lan chẳng / không / đâu mua quyển sách đó.

Lan NEG buy CL book DEM

‘Lan did not buy that book.’

b. Lan có mua quyển sách đó đâu.

Lan NEG buy CL book DEM NEG

‘Lan did not buy that book.’

The scope of the negation is determined by the position of the negative particle. With the

pair có...đâu, the scope of negation is dependent on the position of có. As shown in

(37a, b) the sentence final element đâu is structurally ambiguous: It is construed as an

embedded element in (37a), but as a matrix element in (37b).

32

(37) a. Tân biết [ Lan có mua quyển sách đó đâu ].

Tan know Lan NEG buy CL book DEM NEG

‘Tan knew that Lan did not buy that book.’

b. Tân có biết [ Lan mua quyển sách đó] đâu.

Tan NEG know Lan buy CL book DEM NEG

‘Tan did not know that Lan bought that book.’

In (37a) the particle có is in the embedded clause and the embedded clause is negated; in

(37b) it is in the matrix clause and it is the matrix clause that is negated.5

The scope of negation with respect to other scope-bearing elements is not always

decided by their surface word order. In (38a) the vì ‘because’ phrase precedes the

negative particle, and ‘because’ is not in the scope of negation. However, (38b) is

5 For unknown reason, this pair of particles is not allowed in the sentential subject context as in (i) and (ii): Neither keeping the particle đâu in the embedded clause nor leaving it in the matrix clause results in grammaticality.

i. *[Hoa có thi-rớt đâu ] khiến cả nhà rất vui. Hoa NEG fail NEG make all family very happy

‘That Hoa did not fail made the whole family happy.’ ii. *[ Hoa có thi-rớt ] khiến cả nhà rất vui đâu.

Hoa NEG fail make all family very happy NEG ‘That Hoa did not fail made the whole family happy.’

Other particles such as không, chẳng are allowed in this context as in (iii).

iii. [Hoa không/chẳng thi rớt] khiến cả nhà rất vui. Hoa NEG fail make all family very happy

‘That Hoa did not fail made the whole family happy.’

33

scopally ambiguous: vì ‘because’ can be understood as having either narrow or wide

scope with respect to the negative marker không, as illustrated in the translations.

(38) a. Vì tiền Liên không trồng hoa. Because > NEG

because money Lien NEG grow flower

‘Because of money Lien did not grow flowers.’

b. Liên không trồng hoa vì tiền. NEG > Because

Lien NEG grow flower because money Because > NEG

‘Lien did not grow flowers because of money.’

i. ‘Lien did not grow flowers, and the reason for that was money.’

ii. ‘Lien grew flowers for other reasons.’

The data in (38) suggest that when a scope-bearing element c-commands the negative

marker on the surface, it definitely outscopes the negative marker. However, if it is c-

commanded by the negative marker on the surface, it can obtain either narrow or wide

scope with respect to the negative marker.

In what follows we will discuss the interaction between the negative marker and

the quantifiers. Consider the example in (39).

(39) Tất-cả các sinh-viên sẽ không đến. ∀> NEG

all PLU student FUT NEG arrive * NEG > ∀

‘All the students will not arrive.’

34

i. For every x, x student, it holds that x does not arrive.

ii. *It is not the case that it holds for every x, x student, x arrives.

As indicated by the informal representations in (39i) and (39ii), the universal quantifier

takes wider scope than the negative marker. In order for the universal quantifier to take

narrow scope, another form of negation must be used as in (40).

(40) Không phải (là) tất-cả các sinh-viên sẽ đến. ∀> NEG

not true COMP all PLU student FUT arrive NEG > ∀

‘It is not true that all the students will arrive.’

i. For every x, x student, it holds that x will not arrive.

ii. It is not the case that it holds for every x, x student, x will arrive.

(40) is structurally ambiguous: The scope of negation can be over the embedded sentence

or over the universal quantifier. Syntactically, the English translation of (40) renders

exactly the structure of this kind of negation, namely, the negative marker occurs in the

matrix clause, whose subject, similar to the expletive it in English, is empty. The reading

(40ii) entails the reading (40i). We refer to negation of this type biclausal negation,

conveying the fact that the negative marker is located in a higher clause.

35

There is another universal quantifier: mọi.6 As shown in the example below, this

quantifier is similar to the quantifier tất- cả regarding scope interaction with negative

markers.

(41) a. Mọi sinh-viên/ tất-cả các sinh-viên sẽ không đến. ∀> NEG

all student / all PL student FUT not arrive * NEG > ∀

‘Every student will not arrive.’

b. Không phải (là) mọi sinh-viên/tất-cả các sinh-viên sẽ đến. ∀> NEG

not true (COMP)all student / all PLU student FUT arrive NEG > ∀

‘It is not the case that every student will arrive.’

6 The quantifier mọi differs from the quantifier tất-cả in that while the former prefers a bare NP as its argument, the latter take can a non-bare NP as its argument, as illustrated below.

(i) *mọi các sinh viên này all PLU student DEM ‘all these students’

(ii) mọi sinh viên all student

‘all students’ (iii) tất-cả các sinh-viên này

all PLU student DEM ‘all these students’

(iv) tất-cả sinh-viên all student ‘all students’

36

(41a) does not allow wide scope reading of the negative marker. The sentence cannot

mean ‘Not everyone will arrive’. However, this reading is available in the biclausal

negation case (41b).

There is, interestingly, a subject-object asymmetry in scope interaction. While

quantifiers in subject position always take wide scope over the negation, object

quantifiers can obtain either wide or narrow scope. Consider the examples below.

(42) Tân sẽ không gặp tất-cả các sinh-viên. ∀> NEG

Tan FUT not meet all PLU student NEG> ∀

‘Tan will not meet all the students.’

i. For every x, x student, it holds that Tan will not meet x.

ii. It is not the case that it holds for every x, x student, Tan will meet x.

The translation in (42i) indicates that the quantifier takes scope over negation. With this

reading the sentence is truthful in the context where there are ten students and Tan will

not meet these ten students. The reading (42ii) is satisfied in a context where out of ten

students, Tan will meet some of them.

The interaction between negative operators and quantifiers, as discussed above,

can be accounted for if we assume that Q(uantifier) R(aising) applies in Vietnamese.

Scope interaction, thus, follows the generalization in (43).

37

(43) A negative operator must c-command a quantifier at the surface level to take

scope over the quantifier.

(44) Tân sẽ không gặp nhiều sinh-viên.

Tan FUT NEG meet many student

‘Tan will not meet many students.’

a. [TP Q i [TP [vP NEG t i ] ] ]

The LF representation (44a) accounts for the ambiguity of (44). At the surface level, the

negative operator (NEG) c-commands the quantifier (Q); hence the former outscopes the

latter. However, at LF, the Q, after QR as shown in (44a), c-commands NEG, so the

quantifier outscopes the negative operator.

(45) Nhiều sinh-viên sẽ không đến.

many student FUT NEG arrive

‘Many students will not arrive.’

a. [TP Q i [TP t i [vP NEG ] ] ]

The LF representation (45a) illustrates the absence of scopal ambiguity of (45): The Q c-

commands the NEG both at LF and at the surface level. It follows, presumably, that

Vietnamese does not allow quantifier reconstruction: The quantificational phrase in (45)

cannot reconstruct to the thematic position, which is below the negative marker. A

38

subject quantifier instead can stay in the thematic position at the surface level, as shown

in (46).

(46) Sẽ không có nhiều sinh-viên đến.

FUT NEG CO many student arrive

‘Not many students will arrive.’

a. [TP [ [vP NEG Q ] ] ]

As discussed in the subsection of subject position, the quantificational phrase ‘many

students’ in (46) stays in its thematic position. Accordingly, the negative operator c-

commands the quantifier and takes scope over it. Hence, (46) is truthful in the context

where the number of students who will arrive is not many.

In summary, Vietnamese allows a quantifier to undergo QR for scope

interpretation, but it does not make available reconstruction of subject quantifiers.

2.4.1 Negative Polarity Items

It is well known that one of the contexts where negative polarity items are

licensed is in the scope of negation. In Vietnamese negative polarity items (NPIs) such as

chút nào, ‘at all’, bao giờ, ‘when’, hề ,‘ever’, and một giọt nào, ‘a drop’, always require a

39

c-commanding licensor, one of which is the negative marker không.7 Consider the

examples in (47).

(47) a. Cô ấy không hồi-phục chút nào.

she NEG recover little which

‘She did not recover at all.’

b. *Cô ấy hồi-phục chút nào.

she recover little which

‘She recovered at all.’

c. *[ Cô ấy hồi-phục chút nào ] không khiến mọi người rất lo-lắng.

she recover little which NEG make everyone very worry

‘That she recovered at all did not worry everyone very much.’

d. [ Cô ấy không hồi-phục chút nào ] khiến mọi người rất lo-lắng.

she NEG recover little which make everyone very worry

‘That she did not recover at all worried everyone very much.’

The contrast in grammaticality between (47a) and (47b) indicates that a negative operator

is needed to license the NPI chút nào ‘at all’. Moreover, this negative operator must c-

command the NPI in order to license it. This is illustrated by the ill-formedness of (47c),

where the matrix negative marker does not c-command the NPI, as opposed to the

7 Other negative markers such as chẳng, đâu, có...đâu are also NPI licensors and behave syntactically the same as không.

40

grammaticality of (47d), where the negative marker and the NPI are in the embedded

clause and the former c-commands the latter.

More examples of NPIs are given in (48).

(48) a. Tân *(không) uống một giọt nào.

Tan NEG drink one drop which

‘Tan did not drink a drop.’

b. Lan *(không) bao-giờ gặp Hùng.

Lan NEG when meet Hung

‘Lan has not ever met Hung.’

c. Hoàng *(không) hề mua quà cho con.

Hoang NEG ever buy present for child

‘Hoang has not ever bought presents for his kids.’

The licensing of the NPIs by the negative marker is strictly local. The biclausal

negation không phải cannot license an NPI as shown in (49). A negative marker must be

a clause-mate of an NPI in order to license it, as demonstrated in (49-50).

(49) a.* Không phải là Tân uống một giọt nào.

not true COMP Tan drink one drop which

‘It is not true that Tan drinks a drop.’

41

b.* Không phải là Lan bao-giờ gặp Hùng.

not true COMP Lan when meet Hung

‘It is not true that Lan ever met Hung.’

(50) a. *Phú không tin cô ấy vui chút nào.

Phu NEG believe she happy little which

‘Phu did not believe that she was happy at all.’

b. *Phú không tin cô ấy hề mua quà.

Phu NEG believe she ever buy present

‘Phu did not believe that she ever bought presents’

The data above show that like in other languages in Vietnamese NPIs can be

licensed in negation contexts. However, the licensing in Vietnamese is strictly local,

which is unlike English where the licensing can be long distance.

2.5 Sentence Final Particles

A sentence final particle does not affect the propositional content of the sentence

it attaches to. It instead conveys illocutionary speech acts such as questioning,

commanding..., as illustrated in the examples below.8

8 Abbreviations used: COMM = Commanding particle, QUES = Questioning particle, ASSRT = Asserting particle

42

(51) a. Chúng-ta ăn cơm đi.

we eat rice COMM

‘Let’s have dinner.’

a’. # Cô ấy ăn cơm đi.

she eat rice COMM

b. Lan mua quyển sách đó à?

Lan buy CL book DEM QUES

‘Did Lan buy that book?’ (I am surprised)

c. Lan mua quyển sách đó mà!

Lan buy CL book DEM ASSRT

‘Lan bought that book.’ (I am sure! You’ve got to believe me!)

The sentence-final particle đi in (51a) conveys the illocutionary force of commanding,

urging, etc. Therefore, substituting a second person pronoun with a third person pronoun

results in anomaly, as in (51a’), because it is not possible to command in the third person.

The particle à in (51b) expresses the speaker’s surprise as to the truthfulness of the

propositional content and (51b) is commonly construed as a yes-no question. The particle

mà in (51c) involves the speaker’s commitment to the truth of the proposition. In a word,

these particles, besides conveying illocutionary force, are grammatical devices encoding

modality. They express the speaker’s attitude toward the propositional content’s

believability or desirability as well as the speaker’s attitude or emotion toward the state of

affairs expressed in the propositional content.

43

Finally, as illustrated in the table in (52), some final particles vary dialectically in

form, but not in function. For ease of exposition, the final particles used in this

dissertation are consistently from the northern dialect.

(52) Dialect Declarative Wh-question

Y/N question Imperative

Northern mà thế chăng / à/ ư đi Central mà rứa à đi Southern mà vậy sao/hả đi 2.6 Prepositional Phrases

The complement of a preposition in a prepositional phrase can be a non-bare NP

as in (53), where the NP in (53a) is definite and that in (53b) is indefinite.

(53) a. trong căn nhà này

in CL house DEM

‘in this house’

b. trong căn nhà nọ

in CL house DEM

‘in a house’

A preposition can also take a bare noun as its complement. A bare noun in this

context is always interpreted as definite.

44

(54) a. trên bàn b. trong vườn

on table on garden

‘on the table/*on a table’ ‘in the garden/*in a garden’

A prepositional phrase always follows the noun phrase it modifies, as in (55).

(55) a. [ [ Quyển sách ] trên bàn ] còn mới.

CL book on table still new

‘The book on the table is still new.’

b. * [ Trên bàn [ quyển sách ]] còn mới.

on table CL book still new

‘The book on the table is still new.’

As an argument of verbs like ‘put’, ‘keep’, a prepositional phrase follows the

noun phrase argument as shown by the contrast in (56a, b). Fronting a prepositional

phrase as in (56c) is also ungrammatical.

(56) a. Tân cất thư [ trong túi-áo ].

Tan keep letter in shirt pocket

‘Tan keeps letters in his shirt pocket.’

b. *Tân cất [trong túi-áo ] thư.

Tan keep in shirt pocket letter

45

c. *[Trong túi-áo ] Tân cất thư.

in shirt pocket Tan keep letter

A sentence-modifying prepositional phrase can appear sentence-finally or

sentence-initially.

(57) a. Nhiều người đang làm-việc [trong vườn ].

many person ASP work in garden

‘Many people are working in the garden.’

b. [Trong vườn ] nhiều người đang làm-việc.

in garden many person ASP work

2.7 Adjective Phrases and Adverb Phrases

An adjective phrase consists of an adjective and a modifier. The modifier rất

precedes the adjective, while the modifiers lắm, quá, ghê follow it.

(58) a. rất buồn

very sad

‘very sad’

b. buồn lắm/quá/ghê

sad very

‘very sad’

46

An adjective phrase with the modifier rất can be used predicatively as in (59) or

attributively as in (60). An adjective modified by lắm/ quá/ ghê can be used predicatively

only.

(59) Đây là một quyển sách rất buồn / *buồn lắm/ *buồn quá/ *buồn ghê.

this be one CL book very sad sad very

‘This is a very sad book.’

(60) Cô ấy rất buồn / buồn lắm / buồn quá/ buồn ghê.

she very sad sad very

‘She is very sad.’

As illustrated in (61), an adverb phrase is usually composed of a preposition, and

a noun phrase. For example, an adverb phrase of time consists of a preposition lúc, vào

and a time-denoting noun phrase such as ba giờ, ‘three hour’, năm ngoái, ‘last year’. An

adverb of place is a combination of a preposition, tại ‘at/in/on’, and a place-denoting

noun phrase.

(61) a. lúc ba giờ

at three hour

‘at three o’clock’

47

b. tại San Diego

in San Diego

‘in San Diego’

c. AnhThơ là một học-sinh rất giỏi.

AnhTho be one student very good

‘AnhTho is a very good student.’

d. AnhThơ học toán rất giỏi.

AnhTho study math very good

‘AnhTho studies math very well.’

There is no morphological distinction between an adjective phrase and an adverb phrase.

An expression such as rất giỏi ‘very good’, can be used either as an adjective phrase as in

(61c) or as an adverb phrase as in (61d) depending on the context where it appears.

2.8 Verb Phrases

A verb takes a complement on its right, be it a noun phrase or a clause. An NP

complement can be a non-bare noun or bare noun, with the former being commonly

object-referring, and the latter kind-denoting (generic). For example, the object bare noun

chó ‘dog’ in (62a) is generic as opposed to the full NP in (62b), which is object-referring.

48

(62) a. Tân thích chó. Generic/Kind

Tan like dog

‘Tan likes dogs’

b. Tân thích con chó này. Definite

Tan like CL dog DEM

‘Tan likes this dog.’

In sentences with ditransitive verbs, indirect objects precede direct objects as

illustrated in (63).

(63) a. Tân cho cô ấy một quyển sách.

Tan give he one CL book

‘Tan gave her a book.’

b. * Tân cho một quyển sách cô ấy.

Tan give one CL book her

‘Tan gave her a book.’

In summary, a complement of a verb always appears on its right. As shown

previously, a modifying adverb occurs after the object.

Finally, Vietnamese makes available serial verb constructions as illustrated

below.

49

(64) a. Họ mua bánh về cúng.

they buy cake return offer

‘They bought cakes for ancestor offering ceremony.’

b. Ông ấy cuốc đất trồng khoai.

he hoe earth grow potato

‘He hoed the land, growing potatoes.’

The sentences above consist of a string of verbs, and express consecutive actions (64a) or

simultaneous actions (64b). There are no connectives in these sentences. Serial verbs

together can have preverbal elements such as tense particles, which precede the first verb,

but they cannot each have them.

(65) a. Họ sẽ/nên mua bánh về cúng.

they will/should buy cake return offer

‘ They will/should buy cakes for ancestor offering ceremony.’

b.* Họ sẽ/nên mua bánh sẽ/nên về cúng.

they will/should buy cake will/should return offer

‘They will/should buy cakes for ancestor offering ceremony.’

50

2.9 Noun Phrases

A noun phrase can surface bare as in (66), where it can function either as a

nominal predicate or as an argument. That is, the bare noun ‘student’ is a nominal

predicate in (66a), and is an argument of the verb ‘learn’ in (66b).

(66) a. Tân là học-sinh.

Tan be student

‘Tan is a student.’

b. Học-sinh đang học bài.

student ASP learn lesson

‘Students/The students are learning.’

A noun modified by a numeral commonly requires a classifier as illustrated by the

contrast in (67a, b).

(67) a. * ba sách b. ba quyển sách

three book three CL book

‘three books’ ‘three books’

However, most compound nouns such as sinh- viên ‘student’, gia- đình ‘family’,

biệt- thự ‘building’, phi- cơ ‘airplane’ can appear with a numeral without a classifier.

51

(68) a. Năm sinh-viên này đang nói-chuyện trong lớp.

five student DEM ASP talk in class

‘These five students are talking in the class.’

b. Ba gia-đình này rất thân nhau.

three family DEM very close each other

‘These three families are close to each other.’

‘Body’ nouns such as tay ‘hand’, chân ‘foot’, mắt ‘eye’ do not require classifiers.

(69) Tại sao ta có hai tai, hai mắt, mà chỉ có một miệng?

why we have two ear two eye but only have one mouth

‘How come we have two ears, two eyes, but only one mouth?’

In an NP, a quantifier is the initial element, followed by a plural marker or

numeral, then a classifier, a noun, and an adjective, and the final element is a

demonstrative or a relative clause. In short, a noun phrase normally follows the linear

word order in (70).

(70) Quantifier - Plural marker - Classifier - Noun - Adjective - Relative clause/Dem

Illustrative examples are given in (71).

52

(71) a. tất-cả những quyển sách cũ ấy

all PLU Cl book old DEM

‘all these old books’

b. tất cả những quyển sách cũ [ mà tôi mua hôm-qua]

all PLU Cl book old REL I buy yesterday

‘all the books that I bought yesterday’

c.*tất cả những quyển sách tiêng Pháp cũ [ mà Tân mua năm-ngoái ] này

all PLU CL book language French old REL Tan buy last year DEM

‘all these old French language book that Tan bought last year’

d.*tất cả những quyển sách tiếng Pháp cũ này [ mà Tân mua năm-ngoái ]

all PLU CL book language French old DEM REL Tan buy last year

‘all these old French language book that Tan bought last year’

As shown in (71a, b), the relative clause and the demonstrative both occur in final

position. Note that they are in complementary distribution: It is ungrammatical when a

demonstrative follows the relative clause, as in (71c), or precedes the relative clause, as

in (71d).

Vietnamese makes available a count-mass distinction as evidenced in what

follows. First, Vietnamese employs two forms for amount questions, just like in English,

where there are two interrogative phrases for the how-question: how many and how much,

with the former being used with count nouns, and the latter with mass nouns. Let us look

at the following.

53

(72) a. *Anh cần mấy nước/tiền /muối /gạo/đường?

you need how.many water/money/salt/ rice/sugar

*‘How many water/money/salt/sugar do you need?’

b. Anh cần bao nhiêu nước/tiền /muối/gạo/đường?

you need how.much water/money/salt/rice/ sugar

‘How much water/money/salt/rice/ sugar do you need?’

c. Anh cần mấy/ bao nhiêu thùng nước/đô la /kí muối/kí gạo/ kí đường?

you need how.many barrel water/dollar/kilo salt/kilo rice/kilo sugar

‘How many barrels of water/dollars/kilos of salt/kilos of rice / kilos of sugar do

you need?’

The question word mấy as indicated in (72a, c) does not take mass nouns, while the form

bao nhiêu can take both count and mass nouns.

Certain quantifiers occur only with count nouns while others occur only with mass

nouns.

(73) Quantifiers occurring with count nouns

a. mỗi nhà (count) d. *mỗi muối (mass)

each house each salt

b. mọi nhà (count) e. * mọi muối (mass)

every house every salt

54

c. vài nhà (count) f. * vài muối (mass)

some house some salt

(74) Quantifiers occurring with mass nouns

a. một chút muối (mass) c. * một chút nhà (count)

one little salt one little house

b. chút muối (mass) d. * chút nhà (count)

little salt little house

There are quantifiers that appear with both mass and count nouns.

(75) a. nhiều muối (mass) a’. nhiều nhà (count)

much salt many house

b. một số muối (mass) b’. một số nhà (count)

one number salt one number house

Next, let us consider pluralization in Vietnamese. Even though Vietnamese does

not employ plural morphemes like the plural morpheme s in English, and a bare noun is

normally interpreted as either singular or plural, depending on the context, there seem to

be two elements that function as plural markers, những and các. First, whenever they co-

occur with a noun, only a plural interpretation obtains.

55

(76) a. những sinh-viên

PLU student

‘students’

b. các sinh-viên

PLU student

‘the students’

c. những sinh-viên [REL đã tốt-nghiệp năm ngoái ]

PLU student PST graduate year last

‘The students who graduated last year’

d. những sinh-viên này

PLU student DEM

‘Those students’

e. những sinh-viên khoa ngôn-ngữ

PLU student department language

‘linguistics students’

The noun phrases in (76a) and (76b) differ with respect to definiteness: While (76b) is a

definite plural noun phrase, (76a) is a plural indefinite noun phrase and is preferably

modified by an adjective phrase, rather than occurring on its own. The noun phrase in

(76e) is more acceptable than that in (76a). (76a) obtains a definite interpretation only

when modified by a relative clause, or a demonstrative, as in (76c, d).

As expected, these plural markers never appear with mass nouns.

56

(77) a. *những muối b. những bao muối

PLU salt PLU bag salt

‘salts’ ‘bags of salt’

c.* các muối d. các bao muối

PLU salt PLU bag salt

‘the salts’ ‘the bags of salt’

2.10 Relativization

There is a difference between subject relativization and object relativization.

While object relativization allows the optional occurrence of a relative marker, subject

relativization does not, as illustrated in the examples below.

(78) [DP Con khỉ [CP (mà) ông Hoàng đang nuôi ] ] biết đọc tiếng Việt.

CL monkey ( REL) Mr Hoang ASP keep know read Vietnamese

‘The monkey Mr Hoang is keeping can read Vietnamese.’

(79) [DP Con khỉ [CP (*mà ) đang xem tivi]] là của ông Hoàng.

CL monkey REL ASP watch TV be of Mr.Hoang

‘The monkey that is watching TV belongs to Mr. Hoang.’

There is evidence showing that relativization in Vietnamese involves movement. First, as

shown in (80), long distance relativization is allowed.

57

(80) Cô gái [ (mà) Tân nói là [ đã đánh cảnh-sát ] ] vừa đến.

CL girl REL Tan say COMP ASP hit police just arrive

‘The girl who Tan said that has hit the police just arrived.’

Relativization out of a CNP island results in ungrammaticality.

(81) a.*Cô gái [ (mà) [ Lan mới gặp [ viên cảnh-sát [ đã đánh ] vừa đến.

CL girl REL Lan just meet CL police ASP hit just arrive

‘The girl Lan just met the police that hit just arrived.’

b. * Viên cảnh-sát [ Lan mới gặp [ [ đã đánh cô gái ]] vừa đến.

CL police Lan just meet ASP hit CL girl just arrive

‘The police Lan just met that hit the girl just arrived.’

The sentence in (81a) involves object relativization where the head noun ‘girl’ is

separated from its gap by a CNP island ‘the police that hit’ and the sentence is

ungrammatical. Similarly, the sentence in (81b) displays subject relativization and is

ungrammatical for the same reason: The head noun ‘the police’ is separated from its gap

by a CNP island.

I propose that the head noun is base-generated in the head position and the gap in

the restricting clause results from the overt movement of a relative operator, that is, the

overt operator mà in object relativization. With respect to the disallowance of this particle

in subject relativization, I assume that subject relativization employs a covert particle that

58

undergoes the same movement as the overt particle mà in object relativization, namely

from inside the restriction clause to Spec, CP. This assumption is illustrated by the

diagram in (82).

(82) DP

D NP

(Head) NP CP (Restricting clause)

Spec C’

C TP

mà

In summary, a relative clause consists of a head noun and a restricting clause.

Relativization involves overt movement of a relative operator from inside the restricting

clause to its Spec. This movement is sensitive to island effects.

2.11 Pronouns

Vietnamese employs two types of pronouns: In one type, we have a fixed number

of pronouns with fixed φ-features, just like pronouns in English; the other type of

pronouns is lexically based, namely, pronouns of this type are derived from lexical items.

One prominent property of this type of pronouns is that the person feature is not fixed,

59

but varies according to the context. For example, a pronoun such as anh ‘brother’ can be

used as first or second person, or even third person. For ease of exposition, I am

borrowing the terms used by Noguchi (1997): D-pronouns and N-pronouns, where D

stands for Determiner and N for nouns, to refer to pronouns in Vietnamese. Specifically,

the term D-pronoun refers to pronouns with fixed φ-features, while the term N-pronoun

refers to pronouns that derive from lexical items.

Before we move on with the discussion of Vietnamese pronouns, a few words

about the terminology are necessary. The term D-pronoun probably originates from the

work of Postal (1969 cited in Noguchi 1997), who argues that personal pronouns in

English pattern with definite articles as illustrated in (83), where personal pronouns can

be followed by nouns just like determiners.

(83) a. We Americans (cf. the American)

b. Us linguists (cf. some linguists)

c. You communists (cf. many communists)

Noguchi (1997), following Postal (1969) and Abney (1987), assumes that personal

pronouns head a functional rather than a lexical projection as illustrated in (84), where

the head D is represented by the pronoun and its complement is phonetically null.

(84) [ DP[ D I/us/you/he/she/it ]] (Noguchi’s (25), p. 777)

60

In examples like those in (85), by contrast, the complement is a common noun.

(85) [ DP[ D us ] [NP linguists ] (Noguchi’s (26), p. 777)

Noguchi, based on data below in Japanese, argues that Japanese pronouns are lexical

rather than functional because they can be modified by adjectives, preceded by a

possessive pronoun or a demonstrative pronoun. Japanese pronouns are N-pronouns, not

D-pronouns.

As mentioned above, pronouns in Vietnamese can be classified into two groups:

Those with fixed φ-features are D-pronouns and those deriving from lexical items are N-

pronouns.

D-pronouns in Vietnamese only contain the φ-features of person, and number,

with the number feature by default being singular. For example, tao ,‘I’, is the first

person singular pronoun; nó, ‘he, she, it’, is third singular pronoun; mày, ‘you’, is the

second person singular pronoun. The gender feature is unspecified. The number feature is

expressed through the morpheme chúng prefixed to the singular form: chúng tao,‘we’,

chúng nó, ‘they’, chúng mày, ‘you’. Pragmatically, this set of pronouns is popular in

spoken and informal contexts, but not in formal contexts, where it is considered impolite

and vulgar. Syntactically, I propose that just like English pronouns on Noguchi’s

analysis, these pronouns are base-generated at the head D position as in (86).

61

(86) a. [ DP[ D tao / chúng tao ]]

‘ I’ ‘ we’

b. [ DP[ D mày / chúng mày] ]

‘you’ SING ‘you’ PLU

c. . [ DP[ D nó/ chúng nó ]]

‘he/she/it’ ‘they’

On this analysis, the NP complement of the head D is null. The definite interpretation is

realized by the pronoun as head of the DP.

The other set of pronouns, more commonly used than the previous one, consists

of pronouns deriving from lexical items, mostly from kinship terms such as anh, ‘elder

brother’, chị, ‘elder sister’, con, ‘child’, or from occupation terms such as thầy, ‘teacher’,

xếp, ‘boss’, tôi, ‘servant’, religious terms such as sư, ‘monk’, ni, ‘nun’. These pronouns

are syntactically similar to N-pronouns on Noguchi’s analysis of Japanese pronouns in

that they can be modified by an adjective or a demonstrative as illustrated below.

(87) a. anh hai

brother two

‘you’ (to address to the eldest brother or to a person deemed as old as the

speaker’s eldest brother)

62

b. cô ấy

aunt DEM

‘she’

c. sư ông

monk grandfather

‘you/he’ (to address to a venerable monk)

An N- pronoun varies in its reference depending on the context. For example, the N-

pronoun anh ‘brother’ can be used as first singular if the hearer is younger than the

speaker or second singular if the speaker is younger than the hearer. To refer to the third

person, however, a demonstrative is normally added to the N-pronoun as in (87b). The

most commonly used demonstrative is ấy. The paradigm of this type of pronouns is as

follows.9

(88) Second person pronouns

a. anh ‘brother’: second, male, as old as speaker’s elder brother

b. chị ‘sister’ : second, female, as old as speaker’s elder sister

c. chú ‘uncle’ : second, male, as old as speaker’s uncle

d. bác ‘uncle’ : second, male, as old as speaker’s uncle (older than father)

e. cô ‘aunt’ : second, female, as old as speaker’s aunt

9 When the referent is contextually salient, the demonstrative can be deleted; thus an N-pronoun such as ông ‘grandfather’ or chị ‘sister’ can be used as first, second, or third person.

63

e. ông ‘grandfather’: second, male, as old as speaker’s grandfather

h. bà ‘grandmother’: second, female, as old as speaker’s grandmother

(89) Third person pronouns

a. anh ấy: third, male, as old as speaker’s elder brother

b. chị ấy: third, female, as old as speaker’s elder sister

c. chú ấy: third, male, as old as speaker’s uncle

d. bác ấy: third, male, as old as speaker’s uncle (older than father)

e. cô ấy: third, female, as old as speaker’s aunt

f. ông ấy: third, male, as old as speaker’s grandfather

h. bà ấy: third, female, as old as speaker’s grandmother

N-pronouns display the same syntactic behavior as D-pronouns in that they obey

condition B of the Binding Theory (Chomsky 1981), rather than condition C. Let us

consider how binding conditions apply in the sentences with D-pronouns in (90).

(90) a. *Tân i thích nó i.

Tan like him

‘Tan likes himself.’

b. Tân i biết [Mai thích nó i].

Tan know Mai like him

‘Tan knows Mai likes him.’

64

c. *Nó i biết [Mai thích Tân i].

he know Mai like Tan

‘He knows Mai likes Tan.’

The sentence in (90a) is ungrammatical if the D-pronoun ‘him’ is construed as bound by

the R-expression. This is because of a violation of condition B. By contrast, the D-

pronoun ‘him’ in the embedded clause can be bound by the R-expression in (90b)

because they are not in the same domain. The sentence in (90c) is ungrammatical on the

interpretation that the R-expression in the embedded clause is bound by the D-pronoun, a

violation of condition C.

As exemplified in (91), N-pronouns show a parallel pattern with respect to

binding conditions.10

(91) a.* Ông Minh i thích ông ấy i.

Mr. Minh like grandfather DEM

‘Mr. Minh likes himself.’

b. Ông Minh i biết [ bà Liên thích ông ấy i].

Mr Minh know Ms Lien like grandfather DEM

‘Mr. Minh knows that Ms Lien likes him.’

10 Lexical items such as ông ‘grandfather’, bà ‘grandmother’ attached to a proper name as in (91) are glossed as Mr, and Ms to indicate that they are lexically meaningless when used in this context.

65

c. * Ông ấy i biết [bà Liên thích ông Minh i].

grandfather DEM know Ms Lien like Mr. Minh

‘He knows Ms Lien likes Mr. Minh.’

(91a) is ungrammatical if the N-pronoun is bound by the R-expression. The binding of an

embedded N-pronoun by a matrix R-expression in (91b) is grammatical. (91c) is ill-

formed if the matrix N-pronoun binds the embedded R-expression, a violation of

condition C.

In a word, N-pronouns display the same pattern as D-pronouns in that they obey

condition B, namely, a pronoun must be free in its governing category.

Now let us look at reflexive pronouns and reciprocal pronouns. A reflexive

pronoun can be derived from the lexical item mình, meaning ‘body’. The lexical item

mình can be used as a first person pronoun as in (92a) or third person pronoun as in

(92b).11

(92) a. Họ khen mình.

they praise body

‘They praised me.’

b. Họ biết mình cần học nhiều hơn.

they know body need learn many more

‘They know that they need to learn more.’ 11 In husband and wife conversations, this item can also be used as a second person pronoun.

66

When this item is used together with the pre-verbal particle tự it obtains a

reflexive interpretation as in (93).

(93) a. Họ tự giúp mình.

they TU help body

‘They helped themselves.’

b. Tân tự giúp mình.

Tan TU help body

‘Tan helped himself.’

c. Tôi tự giúp mình.

I TU help body

‘I helped myself.’

d. Mày hãy tự giúp mình!

you let TU help body

‘You should help yourself!’

Reflexive pronouns can also derive from proper names (94a) or D-pronouns

(94b). The use of the pre-verbal particle tự is optional.12

12 Note that the use of proper names as reflexive pronouns requires both the subject and object to be proper names. If we replace a proper name with a pronoun, the sentence will be ungrammatical.

(i) John ( tự ) thương John. John TU love John ‘John loves himself.’

67

(94) a. Tân (tự) khen Tân.

Tan TU praise Tan

‘Tan praised himself.’

b. Họ (tự) khen họ.

they TU praise them

‘They praised themselves.’

The pre-verbal particle tự must associate locally with the item mình for the latter

to obtain a reflexive pronoun interpretation. For example, in (95a), where the particle and

mình appear in the embedded clause, mình is construed as a reflexive pronoun, bound by

the embedded pronoun ‘they’. It does not function as a reflexive pronoun, but as a

pronoun in (95b), where the particle tự is in the matrix clause, and mình is in the

embedded clause. The occurrence of the particle tự conveys a meaning similar to the

(ii) *Nó (tự) thương John. he TU love John ‘He loves himself.’

However, Lasnik (1986) holds that a sentence similar to (i) without tự as in (iii) is ungrammatical because it violates the constraint that, according to Lasnik (1986), in Vietnamese an R-expression is free in its governing category. Hence, the sentence in (iii) is ungrammatical on the interpretation where the object ‘John’ is bound by the subject ‘John’.

(iii) *John thương John. John like John ‘John likes John.’ (Lasnik 1986, p.154)

Lasnik (1986) does not discuss the use of proper names as reflexive pronouns. I would like to note that in my judgment the sentence in (iii) is grammatical with the interpretation where the object is bound by the subject.

68

expression ‘by himself’, and (95b) means ‘Tan knows by himself, without the help from

other people, that they helped him’.

(95) a. Tân biết họ tự giúp mình.

Tan know they TU help body

‘Tan knows they helped themselves.’

b. Tân tự biết họ giúp mình.

Tan TU know they help body

‘Tan knows by himself that they helped him.’

The only reciprocal pronoun nhau requires its antecedent to be plural as illustrated

by the contrast in (96a, b).

(96) a. Họ khuyên nhau.

they advise NHAU

‘They advised each other.’

b. *Quang khuyên nhau.

Quang advise NHAU

*‘Quang advised each other.’

In addition, a reflexive pronoun and its antecedent must occur in the same domain as

demonstrated by the ungrammaticality of (97a,c) and the grammaticality of (97b).

69

(97) a. *Họ khuyên [ Quang giúp nhau].

they advise Quang help NHAU

‘They advised Quang to help each other.’

b. Quang khuyên [ họ giúp nhau ].

Quang advise they help NHAU

‘Quang advised them to help each other.’

c. *Họ nghĩ [ nhau là sai].

they think NHAU be wrong

‘They think each other is wrong.’

Even though Vietnamese has no agreement morphology, it allows null pronouns

as shown by the data below. In (98a), the null pronoun is the subject, while in (98b) it is

the object. Especially, when contexts allow, a sentence can employ null pronouns in

subject and object positions as in (98c).

(98) a. Ngày mai e sẽ nhận sách.

tomorrow FUT receive book

‘Tomorrow we/they/he/she/you/I will receive books.’

b. Tân đã gặp e.

Tan PST meet

‘Tan met us/them/him/her/me.’

70

c. e Mua rồi e.

buy already

‘We/They/He/She/I bought it/them.’

Null pronouns are also allowed in embedded contexts as in (99).

(99) a. Tân nói là ngày mai e sẽ nhận sách.

Tan say COMP tomorrow FUT receive book

‘Tan said we/they/he/she/you/I will receive books.’

b. Nhi nghĩ là Tân đã gặp e.

Nhi think COMP Tan PST meet

‘Nhi thought Tan met us/them/him/her/you/me.’

Finally, let us look at interrogative pronouns. These pronouns are dialectally

distinct in form, as shown in the table (100), but not in function; namely, they display the

same syntactic behavior in wh-questions. These pronouns will be the subject of the rest of

the thesis.

(100)

Northern Central Southern who ai / người nào ai / người mô ai /người nào what gì / cái gì chi / cái chi gì / cái gì which + NP CL + nào CL + mô CL + nào why tại sao / vì sao tại răng / vì răng tại sao / vì sao how bằng cách nào bằng cách răng bằng cách nào

71

how như thế nào như răng như thế nào where đâu mô đâu when khi nào khi mô khi nào

In summary, Vietnamese makes available two types of pronouns, D-pronouns and

N-pronouns, with the latter deriving from lexical items. Reflexive pronouns are realized

by a lexical element in local association with a preverbal particle. Null pronouns can

occur freely in subject and object position.

2.12 Complex Sentences with Adverb Clauses

A complex sentence usually employs a subordination element or a connective. For

example, a conditional sentence uses a connective similar to ‘if’ in English in the

antecedent clause and an element, roughly meaning ‘then’ in the consequent clause as in

(101a). The antecedent clause can follow the consequent clause as in (101b), but in that

case the element meaning ‘then’ must be deleted.

(101) a. Nếu ngày mai Tân gặp Liên thì chúng tôi sẽ rất vui.

if tomorrow Tan meet Lien then we FUT very happy

‘If Tan meets Lien tomorrow, we will be very happy.’

b. (*Thì) Chúng tôi sẽ rất vui nếu ngày mai Tân gặp Liên.

then we FUT very happy if tomorrow Tan meet Lien

‘We will be very happy if Tan meets Lien tomorrow.’

72

Alternatively, a preverbal particle mà in the antecedent and thì in the consequent can be

used as in (102a, b). Note that mà can follow or precede the subject so long as it is not in

sentence-initial position as shown in the ungrammaticality of (102c). Note further that mà

induces a focus effect on the element it follows. This is demonstrated in the minimal

difference between (102a) and (102b): The focus in the antecedent in (102a) is on is Tan;

by contrast, the focus in (102b) is on the time, namely, tomorrow.

(102) a. Tân mà gặp Liên thì chúng tôi sẽ rất vui.

Tan MA meet Lien then we FUT very happy

‘We will be very happy if Tan meets Lien tomorrow.’

b. Ngày mai mà Tân gặp Liên thì chúng tôi sẽ rất vui.

tomorrow MA Tan meet Lien then we FUT very happy

‘We will be very happy if Tan meets Lien tomorrow.’

c. * Mà Tân gặp Liên thì chúng tôi sẽ rất vui.

MA Tan meet Lien then we FUT very happy

‘We will be very happy if Tan meets Lien tomorrow.’

Unlike English where counterfactual conditionals are expressed via morphological

means, namely, via changes in the verb forms, Vietnamese employs a different

subordinating morpheme, giá như, as a counterfactual conditional, as shown in (103).

73

(103) a. Giá như hôm qua Tân gặp Liên thì chúng tôi rất vui.

if yesterday Tan meet Lien then we very happy

‘If Tan had met Lien, we would have been very happy.’

b. (*Thì) Chúng tôi rất vui giá như hôm qua Tân gặp Liên.

then we very happy if yesterday Tan meet Lien

‘We would have been very happy if Tan had met Lien yesterday.’

Note that the order of the clauses can vary, namely, the consequent clause precedes the

antecedent clause, but in that case, thì must be deleted.

Complex sentences with other adjunct clauses are given in what follows. An

adjunct clause of reason is introduced by vì or bởi vì, and an adjunct clause of concessive

by tuy or mặc dù, an adjunct clause of time by khi, an adjunct clause of purpose by để,

and an adjunct clause of manner by như. Note that the main clause can either precede the

subordinating clause as in (104a), or follow it, as in (104b).

(104) a. Hoàng không mua xe vì/ bởi vì e không vay được tiền ngân-hàng.

Hoang NEG buy car because NEG borrow obtain money bank

‘Hoang did not buy a car because he could not borrow money from the bank.’

b. Tuy/Mặc dù e rất thành-công, Hoàng vẫn không hạnh-phúc.

although very successful Hoang still NEG happy

‘Although he is very successful, Hoang is not happy.’

74

As shown in (104a, b), the null pronoun is allowed in the adjunct clause with the proper

name in the main clause as the antecedent. However, having a null pronoun in the main

clause is ungrammatical, as illustrated in (105a, b).

(105) a. * e Không mua xe vì/ bởi vì Hoàng không vay được tiền ngân-hàng.

NEG buy car because Hoang NEG borrow obtain money bank

‘Hoang did not buy a car because he could not borrow money from the bank.’

b. * e Vẫn không hạnh-phúc mặc dù Hoàng rất thành-công.

still NEG happy although Hoang very successful

‘Although he is very successful, Hoang is not happy.’

Adjunct clauses can occur without subordinating morphemes, where the relation between

the clauses is dependent on context. Consider, for example, the sentence in (106).

(106) Trời mưa, chúng tôi ở nhà đọc sách.

sky rain we stay home read book

a. ‘If it rains, we will stay at home and read.’

b. ‘Because it rains, we stay at home and read.’

c. ‘Whenever it rains, we stay at home and read.’

75

There is no subordinating morpheme in (106) and the adjunct clause in the sentence can

be contextually understood as the antecedent of a conditional sentence, as in (106a), as an

adjunct clause of reason in (106b), and as an adjunct clause of time, as in (106c).

In summary, the relation between a main clause and a subordinating clause is

normally introduced by subordinating elements. Without a subordinating element, it is

the context that decides on their relation. A main clause can either precede or follow a

subordinating clause. Null pronouns are allowed in subordinating clauses only.

2.13 Island Effects

We observed that relativization obeys the complex NP constraint. In fact, this

operation is sensitive to other islands as well. For example, sensitivity to the Sentential

Subject Constraint is identified in (107b).

(107) a. [CP Lan đi theo ông già ấy] làm nhiều người rất ngạc-nhiên.

Lan go follow man old DEM make many person very surprise

‘That Lan followed the old man surprised many people a lot.’

b.??[ Ông già [CPLan đi theo ] làm nhiều người rất ngạc-nhiên ] ] vừa qua đời.

man old Lan go follow make many person very surprise just pass life

*‘The man that Lan followed surprised many people a lot just passed away.’

76

The object noun phrase ông già ‘old man’ in (107a) is embedded in a sentential subject

clause. The relativization of this noun phrase as shown in (107b) induces an island effect

as indicated by the question marks.

The violation of the Coordinate Structure Constraint, however, is strongly

ungrammatical as shown by the contrast in grammaticality between (108) and (109).

(108) Chúng tôi phải mua [NP sách giáo-khoa và văn-phòng-phẩm] .

we must buy book textbook and stationery

‘We must buy textbooks and stationery.’

(109) * Đây là sách giáo-khoa mà chúng tôi phải mua [NP và văn-phòng-phẩm ].

this be book textbook REL we must buy and stationery

*‘Here is the textbooks we must buy and stationery.’

There is a subtle difference with respect to the Adjunct Island Constraints. It is ill-

formed to relativize a noun phrase out of a because-clause as in (111).

(110) Tân về [CP vì họ sắp gặp cô gái ấy ].

Tan leave because they FUT meet girl DEM

‘Tan leaves because they will meet that girl.’

(111) *Đây là cô gái mà Tân về [CP vì họ sắp gặp ].

here be girl REL Tan leave because they FUT meet

*‘Here is the girl that Tan leaves because they will meet.’

77

It is, however, not as bad to relativize a noun phrase out of an if-clause, as in

(113).

(112) Tân sẽ ra về [CP nếu họ gặp cô gái ấy].

Tan FUT leave if they meet girl DEM

‘Tan will leave if they meet that girl.’

(113) ?Đây là [NP cô gái [ mà CP Tân sẽ ra về [CP nếu họ gặp]] ].

this be girl REL Tan FUT leave if they meet

‘Here is the girl that Tan will leave if they meet.’

Surprisingly, Vietnamese does not show any sensitivity to the Wh-Island

Constraint. A violation of this island does not result in ungrammaticality, as demonstrated

in (115).

(114) Tân muốn biết [CP ai sẽ đoạt tấm huy chương ấy].

Tan want know who FUT win CL medal DEM

‘Tan wonders who will win the medal.’

(115) Tấm huy chương mà [ Tân muốn biết [CP ai sẽ đoạt ] ] là huy chương vàng.

CL medal REL Tan want know who FUT win be medal gold

‘The medal that Tan wonders who will win is the gold medal.’

78

Next, let us look at topicalization. As we have observed in the beginning section,

in a pragmatically marked context, that is in topic-comment structures, a sentence such as

(116a) can be realized as (116b) with the object as topic being in the initial position

followed optionally by the topic marker thì.

(116) a. Tôi rất thích quyển sách này.

I very like CL book DEM

‘I like this book very much.’

b. Quyển sách này, (thì) tôi rất thích.

CL book DEM TOP I very like

‘This book, I like very much.’

c. Quyển sách này, [ tôi đã yêu-cầu [ cô Lan bảo sinh-viên đọc ] ].

CL book DEM I PST ask Ms Lan tell student read

‘This book, I asked Ms Lan to tell students to read.’

Note that the grammaticality of (116c) indicates that the long distance movement of the

noun phrase quyển sách này ‘this book’ to the topic position is available in Vietnamese.

In what follows, we will see that topicalization in Vietnamese obeys island constraints.

First, let us look at the Complex NP Constraint as shown in (117) and (118).

79

(117) a. Tôi đồng-ý với [ ý-kiến là [ sinh-viên nên đọc quyển sách này ]].

I agree with opinion COMP student should read CL book DEM

‘I agree with the opinion that students should read this book.’

b.?Quyển sách này, tôi đồng-ý với [ ý-kiến [ là sinh-viên nên đọc ] ].

CL book DEM I agree with opinion COMP student should read

* ‘This book, I agree with the opinion that students should read.’

Note that in (117b) the complex noun phrase is a noun-complement clause structure, and

the island effect induced by topicalization in (117b) is less serious than that in (118b),

where the complex noun phrase is modified by a relative clause. For the moment, I have

no explanation for this phenomenon.

(118) a. Tân đã bác-bỏ [ các nguyên-tắc [ giải-quyết vấn-đề ấy ] ].

Tan PST refute PLU principle solve problem DEM

‘Tan refuted the principles that solve that problem.’

b. *Vấn-đề ấy, Tân đã bác-bỏ [ các nguyên-tắc [ giải-quyết ]].

problem DEM Tan PST refute PLU principle solve

*‘That problem, Tan refuted the principles that solve.’

The Sentential Subject Constraint (SSC) and the Coordinate Structure Constraint

(CSC) islands also differ with respect to the severity incited by the topicalization of the

80

elements inside them. The violation of the SSC island is less unacceptable than the

violation of the CSC island, as illustrated below.

(119) ?? Ông già ấy, [ Lan đi theo ] làm nhiều người rất ngạc-nhiên.

man old DEM Lan go follow make many person very surprise

*‘That old man, that Lan followed surprised many people.

(120) * Quyển sách này, tôi đã cho Lan [ và chiếc sáo kia ].

CL book DEM I PST give Lan and CL flute DEM

‘This book, I gave Lan and that flute.’

Topicalization of elements in the Adjunct Island Constraint follows the same

pattern as relativization of elements in the Adjunct Island Constraint in that there is

dissimilarity between the because-clause and the if-clause. We have seen that

relativization of the if-clause is less critical than that of the because-clause. The same

distinction is found in the topicalization of the if-clause compared with the because-

clause.

(121) * Cô gái ấy, Tân về [ vì họ sẽ gặp ].

girl DEM Tan leave because they FUT meet

‘That girl, Tan leaves because they will meet.’

81

(122) ? Cô gái ấy, Tân sẽ về [ nếu họ gặp ].

girl DEM Tan FUT leave if they meet

‘That girl, Tan will leave if they meet.’

The sentence in (121) is ungrammatical, while that in (122) is not as bad.

Finally, topicalization is not sensitive to wh-islands, a behavior reminiscent of the

insensitivity of relativization in wh-island contexts, which we have observed.

(123) Quyển sách ấy, Tân muốn biết [ ai sẽ mua ].

CL book DEM Tan want know who FUT buy

*‘That book, Tan wonders who will buy.’

The data above show that topicalization patterns with relativization with respect

to the island effects they induce, especially in their lack of sensitivity to wh-islands.

82

Chapter 3

UNIVERSAL QUANTIFICATION 3.1 Introduction

We mentioned in the first chapter that a wh-phrase can occur in interrogative

contexts as well as non-interrogative contexts. In this chapter we are concerned with the

universal interpretation that a wh-phrase obtains in non-interrogative contexts. The

technical term ‘wh-universal interpretation’ in this chapter refers to this universal

interpretation. Some illustrative examples are given in (1).13

(1) a. Ai/ Người nào cũng vui-vẻ.

who/ person which CUNG happy

‘Everyone/Every person is happy.’

b. Ai/ Người nào vui-vẻ.

who/ person which happy

*‘Everyone/Every person is happy.’

‘Who is happy?’

13 The terms ‘wh-indefinite’, ‘wh-phrase’ and ‘wh-word’ are interchangeable in this chapter.

83

c. Cái gì / Cái nào cũng đẹp.

CL what/ CL which CUNG beautiful

‘Everything is beautiful.’

As shown in (1a), cũng can associate with the simplex wh-phrase ‘who’ or with

the complex one ‘person which’. Without the element cũng, the universal interpretation is

out: (1b) is best construed as a wh-question. The simplex non-human wh-phrase gì,

‘what’, and the complex non-human wh-phrase cái nào, ‘CL+ which’, can associate with

cũng and obtain a universal interpretation as well. It is, however, more common for the

simplex form to appear with a classifier, especially when it occurs in subject position as

in (1c).

If it is base-generated in object position, a wh-phrase must move to sentence-

initial position in order to associate with cũng. If it stays in-situ, it does not obtain a

universal interpretation and the sentence is interpreted as a question. This is illustrated by

the examples below.

(2) a. Ai / Người nào Tân cũng quen.

who / person which Tan CUNG know

‘Tan knows everyone/ every person.’

b. Cái gì / Cái nào Tân cũng thích.

CL what/ CL which Tan CUNG like

‘Tan likes everything.’

84

c. Tân cũng quen ai / người nào?

Tan CUNG know who/ person which

‘Who does Tan know too?’

d. Tân cũng thích cái gì / cái nào?

Tan CUNG like CL what/ CL which

‘What/Which thing does Tan like too?’

Note that cũng is translated as an additive particle ‘too’. The question in (2c) is felicitous

in a context where the speaker is informed that John knows X and Tan also knows X, but

is not clear who is X. Instead of asking about John, the speaker chooses to ask about Tan

for the identity of X. The question in (2d) needs a similar context. Namely, John likes X

and Tan likes X, and the speaker chooses to ask about Tan for the identity of X.

Except for ‘why’ other adjunct wh-phrases such as ‘when’, ‘where’, ‘how’, ‘how

much’, can associate with cũng as well, as shown in the examples below.

(3) a. Khi nào cô ấy cũng buồn.

time which she CUNG sad

‘She is always sad.’ ‘She is sad at all times.’

b. Ở đâu cô ấy cũng buồn.

in where she CUNG sad

‘She is sad wherever she is.’

85

c. Thế nào cô ấy cũng buồn.

how she CUNG sad

‘She is sad however it is.’

d. Bao nhiêu cô ấy cũng chê.

how much she CUNG refuse

‘She refuses however much (you give her).’

e. * Vì-sao cô ấy cũng buồn.

why she CUNG sad

‘She is sad for whatever reason.’

The wh-phrases meaning ‘where’, ‘how’, ‘how much’ occur sentence-finally in wh-

questions, but in this non-question context they must occur sentence-initially in order to

associate with cũng. The wh-phrase meaning ‘when’ encodes a future tense interpretation

when it occurs sentence-initially in wh-question, but in this context it does not.

In the data above a wh-phrase and cũng are clause-mates in the sense that they

both appear in the same clause, but in what follows, they do not. For example, in (4a), the

wh-phrase ‘what’ is the object of the verb ‘cook’ of the antecedent clause, and cũng

appears in the consequent clause. Note that the classifier is optional when ‘what’ appears

in object position.

86

(4) a. Anh nấu (cái) gì tôi cũng ăn pro.

you cook CL what I CUNG eat pro

‘I will eat what you cook.’ (If you cook X, I will eat X.)

b. pro gặp ai Tân cũng chào pro.

pro meet who Tan CUNG greet pro

‘Tan greets anyone Tan meets.’ (If Tan meets X, Tan will greet X.)

c. Khi nào pro gặp Tân Lan cũng mượn tiền pro.

time which pro meet Tan Lan CUNG borrow money pro

‘Lan borrows money from Tan whenever Lan meets Tan.’

The universal interpretation of wh-indefinites is cross-linguistically well attested,

and there are many theories that attempt to account for it. In this chapter we will look at

some well known analyses in the literature: The unselective binding theory of

Nishigauchi (1986, 1991), and Cheng (1991, 1995), and Shimoyama’s (2001) neo-

Hamblin semantics. In a nutshell, Nishigauchi (1986, 1991) and Cheng (1991, 1995) hold

that wh-indefinites are restricted free variables, not quantifiers, and their universal force

is derived from the operators that bind them. Shimoyama (2001), based on Hamblin’s

(1973) semantics, assumes that a wh-indefinite introduces sets of alternatives. The wh-

indefinite combines with other elements via pointwise functional application (see Rooth

1992), forming a constituent that serves as a restriction for the universal quantifier.

Overall, all of these analyses share one common feature, namely, wh-indefinites are not

87

quantificational inherently and their universal force derives from a separate universal

quantifier.

I argue that the neo-Hamblin theory provides the best analysis for the data in

Vietnamese. The chapter will be organized as follows. A literature review is given in

Section 3.2. The data in Vietnamese is provided in section 3.3. Section 3.4 discusses our

proposal. Section 3.5 introduces a language-typical construction of Vietnamese wh-

universals, followed by a proposed analysis in section 3.6. The conclusion is given in

section 3.7.

3.2 Previous Accounts

In what follows, we will review the accounts of wh-universals in two well-studied

languages: Japanese and Chinese.

3.2.1 Nishigauchi (1986, 1991)

In Japanese wh-phrases such as dare, ‘who’, and nani, ‘what’, are interpreted as

universal quantifiers when they occur with the quantificational element mo. Following

Heim’s (1982) analysis of indefinite NPs as variables, Nishigauchi (1986, 1991) holds

that wh-indefinites in Japanese are not quantificational, but are predicates with open

variables. Hence, a wh-indefinite such as dare, ‘who’, is represented as person (x). Thus

the universal interpretation of dare mo in (5a) is derived from mo, a universal quantifier.

The syntactic configuration of this expression is much like that of its English counterpart

every person as shown in (5b, c).

88

(5) a. Dare mo ga nani-ka o tabe-te-iru.

who MO NOM what KA ACC eating-be

‘Everyone is eating something.’14 (Nishigauchi 1991, p. 198)

b. PP c. DP

PP P D NP

NP Q ga every person

dare mo

This account nicely captures cases where the wh-indefinite and its associate, mo

are adjacent. However, unlike English quantificational expressions with the quantifier

every, where the quantifier and its NP restriction are always in a local relationship, a

Japanese wh-indefinite can be associated with mo non-locally as shown in (6a).

(6) a. Dare ga ki-te mo, boku-wa aw-a-nai.

who NOM come MO I TOP meet-not

‘For all x, if x comes, I would not meet x.’ (Nishigauchi 1991, p. 204)

14 In Nishigauchi’s (1991) dare-mo is glossed as ‘someone’, nani-ka as ‘something’. I revise the gloss as in (1b) for ease of reference. NOM=nominative case, ACC= accusative case.

89

b. CP

Spec C’

Wh x IP C

… x… mo

In (6a) mo is not attached directly to the wh-indefinite, but to a clause in which the wh-

indefinite dare is embedded. Nishigauchi assumes that the semantic properties of this wh-

indefinite are the same as in local cases. Namely, it is a variable bound by mo and obtains

its universal interpretation from mo. This binding, however, is subject to some

configuration. Working within the Government and Binding framework, Nishigauchi

assumes that in the example above, the Q-element mo is the head C and in order for mo to

bind the wh-indefinite, the former must govern the latter. Nishigauchi proposes a

movement account, according to which, the wh-indefinite moves covertly to Spec, CP

where it forms a Spec-head relation with mo and is governed by mo, as illustrated in (6b).

The movement analysis accounts for the apparent non-local relation between the

wh-indefinite and its binder by establishing a local relation at LF. Nishigauchi also

extends this analysis to cases that involve violation of Subjacency such as (7a).

(7) a. [[[ Dare ga kai-ta] hon]-o yon-de-mo] boku-wa manzoku-deki-nai.

who-N wrote book-ACC read MO I-TOP satisfied-can-not-be

‘For all x, y, x a person, y a book x wrote, I cannot be satisfied reading x.’

(Nishigauchi 1991, p. 213)

90

The wh-indefinite dare is embedded in a relative clause modifying the head noun

‘book’, but the quantifier mo is in the matrix clause. This is a Complex Noun Phrase

island. To account for the CNP island, Nishigauchi provides an LF pied-piping analysis,

according to which dare ‘who’ moves covertly from IP to Spec, CP of the relative clause.

Once the wh-indefinite occupies the operator position within the relative clause, the

feature [+wh] percolates to CP. This percolation has the effect of identifying CP as

[+wh], as illustrated in (8a). Nishigauchi assumes that a relative clause in Japanese is in

Spec, NP position. Following Selkirk’s (1982) hypothesis that when the head of some

projection is unmarked for some feature, the feature on Spec percolates up, Nishigauchi

proposes that the quantificational feature associated with the CP in Spec, NP gets

percolated up to the dominating NP hon ‘book’, which identifies the whole complex NP

as [+wh], as illustrated in (8b). This allows the complex NP to move and adjoin to the Q-

elements mo in the matrix Comp. Hence, there is no island violation.

(8) a. CP [+wh]

Spec C’

who x IP C

… x…

Covert movement within the relative clause and percolation of [+wh] feature to CP.

91

b. NP[+wh]

CP [+wh] N’

Percolation of [+wh] feature of the relative clause (CP) in Spec to the head of NP.

This analysis is syntactically and semantically problematic as pointed out in

Shimoyama (2001). First, the reason why a Complex NP is an island is that the Spec, CP

of the relative clause is occupied by an empty relative pronoun, which prevents wh-

phrases from cyclically undergoing wh-movement. If the Spec, CP of the relative clause

is available in Japanese then there is no need for the pied-piping mechanism. Second, as

indicated by Toyoshima (1996, quoted in Shimoyama 2001), the assumption that the

relative clause CP occupies Spec, NP requires justification given that relative clauses are

modifiers and are commonly taken to be adjoined to N’ (or NP in a DP hypothesis).

Semantically, Ohno (1989, quoted in Shimoyama 2001), demonstrates that there are

interpretations of the mo-construction that Nishigauchi’s (1990) proposal fails to capture.

For example, consider (9).

(9) Kono mise-de-wa [[dono kuni-de syuppansareta] hon]-mo utteiru.

this store-in-Top which country-in was published book-MO sell

a. ∀x, y [[ country (x) & book (y) & publish-in (x) (y) ] → sell (y) (this store)]

b. ∀x [country (x) →∃ y[book(y) & publish-in (x) (y) & sell (y) (this store)]]

92

(Shimoyama 2001, p. 29)

According to Nishigauchi’s LF pied-piping analysis, both the wh-indefinite and the head

noun of the relative clause receive the universal force from mo. Hence, the truth

conditions on this analysis would be as in (9a), where mo quantifies over country-book

pairs. Ohno (1989) indicates that (9) can be truthfully uttered if ‘each country is

represented with at least one publication’, as represented in (9b). Ohno proposes that on

this reading mo quantifies over the wh-indefinite ‘which country’, and (9b) arises from

the indefinite interpretation of the head of the relative clause ‘book’.

Additionally, Shimoyama argues that the movement analysis above does not

satisfactorily explain islands in Japanese.

Consider the contrast between CNP and wh-islands.

(10) [[ ø [Yamada-ga dare-ni nani-o okutta ka] sitteiru] syoonin]-mo

[[[Yamada-Nom who-Dat what-Acc sent Q] know] witness]-MO

damatteita.

was_silent

a. ‘The witness who knew what Yamada sent to whom was also silent.’

‘Even the witness who knew what Yamada sent to whom was silent.’

b. * ‘For every person x, the witness who knew what Yamada sent to x was silent.

c. ** ‘For every thing x, the witness who knew to whom Yamada sent x was silent.’

93

d. * ‘For every person x, for every thing y, the witness who knew whether Yamada

sent y to x was silent.’ (Shimoyama 2001, p.54)

(11) Wh-island

CP [+wh]

Spec C’

who x IP C

… x…

The wh-phrases in (10) are embedded in a wh-island. The absence of the interpretations

b, c, and d in the example above suggests that the wh-indefinites dare, and nani inside the

embedded question, a wh-island, cannot be associated with mo. By contrast, the

association of the wh-indefinite dare with mo is unproblematic when it is embedded in a

complex NP, as shown in (7). Another example appears in (12).

(12) [[[[Dono T.A.-ga ø osieta] gakusei]-ga ø syootaisita] sensei]-mo kita.

which T.A.-Nom taught student -Nom invited teacher -MO came

‘For every T.A. x, the teacher(s) that the student(s) that x had taught invited came.’

(Shimoyama 2001, p.57)

Shimoyama argues that whatever allows LF pied-piping in (12) must allow LF

pied-piping in (10) given that the condition for feature percolation in Nishigauchi’s

theory is satisfied in (10). The configuration of the wh-island in (10) is as in (11): The

94

wh-phrase moves covertly from IP to Spec, CP, as a result of which the [+wh] feature

percolates up to the embedded CP. This would allow the whole wh-island to pied-pipe to

form a Spec- head relation with mo, and the readings (10b, c, d) should be available.

Nishigauchi’s theory, therefore, does not predict the absence of these readings. In other

words, his theory does not predict differential island sensitivity.

3.2.2 Cheng (1991, 1995), Cheng and Huang (1996)

Following Nishigauchi (1986, 1991) and Heim (1982), Cheng (1991, 1995)

proposes that wh-indefinites in Chinese are indefinite NPs without any inherent

quantificational force and they obtain quantificational interpretations from neighboring

operators. Like wh-indefinites in Japanese, wh-indefinites in Chinese can be interpreted

as universally quantified expressions in association with their binders as illustrated in

(13): The wh-indefinite sheme obtains a universal interpretation in association with dou, a

universal quantifier and a binder.15 Note that in Cheng’s analysis, a Chinese wh-indefinite

is a negative polarity item, so it needs both a binder and a licensor. In universal contexts,

dou is the licensor and the binder. According to Cheng (1995), dou is an adverb adjoining

to either Asp, V or Asp’, V’.

15 In the context where dou is associated with a non-wh-indefinite, Cheng (1991, 1995) analyzes it as a distributor, quantifying over a plural NP as illustrated in (i).

(i) Tamen dou hen xihuan wo. they all very like I ‘They all like me.’ (Cheng 1995, p.198)

95

(13) Zhangsan shenme dou chi.

Zhangsan what all eat

‘Zhangsan eats everything.’ (Cheng 1995, p.202)

In the spirit of unselective binding, a binder unselectively binds any variable in its

c-commanding domain. Yet, dou is selective in that it licenses and binds only the closest

wh-indefinite on its left, and only one. This restriction, as illustrated in (14), is dubbed as

the leftness condition in Cheng (1991). The absence of the readings (14 ii) and (14 iii)

indicates that only the wh-indefinite shenme, ‘what’, being closer to dou than shei, ‘who’,

obtains the universal interpretation; the wh-indefinite shei, ‘who’, has to be construed as

an interrogative phrase.

(14) Shei shenme dou chi

who what all eat

i. ‘Who eats everything?’

ii. * ‘What does everyone eat?’

iii. * ‘Everyone eats everything.’ (Cheng 1995, p.203)

Unlike local wh-universals in Japanese, in Chinese a wh-indefinite and its licensor

do not form a quantificational phrase (QP) as in Nishigauchi’s analysis. Cheng

96

hypothesizes that the binder dou is an adverb, adjoining to an Asp’ to m-command the

wh-indefinite to license it. 16 This analysis accounts for the following data.

(15) a. * Shei gei Lisi dou xie-le xin

who to Lisi all write-ASP letter

‘Everyone wrote a letter to Lisi.’

b. Shei dou gei Lisi xie-le xin

who all to Lisi write-ASP letter

‘Everyone wrote a letter to Lisi.’ (Cheng 1995, p.220)

In the sentence (15a), the intervening PP, adjoined to V’ or VP, prevents dou from

adjoining to Asp’, so dou must adjoin to V’. This is not a configuration for m-

commanding (see diagram 16a).17 By contrast, in (15b), dou is not intervened by the PP,

so it can adjoin to Asp’ and be able to m-command the wh-indefinite (see diagram 16b).

The sentence is grammatical.

16 The term m-command in Cheng (1991, 1995) follows Chomsky (1986): A m-commands B iff neither dominates the other and the first maximal projection dominating A also dominates B. 17 Cheng (1991) assumes that in Chinese, Aspect is head of the sentence, not Inflection because Chinese is not inflectional.

97

(16) a. * AspP

NP Asp’

Shei Asp VP

V’

PP V’

P NP dou V’ gei Lisi V NP

xie-le xin

b. AspP

NP Asp’

Shei dou Asp’

Asp VP

V’

PP V’

P NP V NP gei Lisi xie-le xin

While dou in the examples above imposes a seemingly strict locality condition on

the variable it binds, there are data that show that a wh-indefinite and dou can be adjacent

or non-adjacent. Consider the sentence below, from Cheng (1995) where a wh-indefinite

and dou are not adjacent to each other.

98

(17) Shei chi shenme dou gen wo wuguan.

who eat what all to I irrelevant

Lit. ‘Whoever eats whatever is irrelevant to me.’

‘Whatever pairs of x and y such that x eats y are irrelevant to me.’

(Cheng 1995, p.222)

The sentence in (17) requires a new analysis for dou. We saw that in (14) dou licenses

and binds only one variable that it m-commands, which explains the absence of the

readings (14ii, iii). However, if we maintain this analysis, we cannot explain why (17)

can have a reading in which both wh-indefinites are universally interpreted. In fact,

Cheng (1995) proposes that (17) is similar to (18), where dou does not quantify over NPs,

but over propositions.

(18) [CP Hufei qu bu qu ] dou hao.

Hufei go not go all good

Lit. ‘Whether or not Hufei is going is good.’

‘Either Hufei is going or Hufei is not going and both options are fine.’

(Cheng 1995, p.223)

The sentential subject in (18) is a yes-no question. Following Hamblin (1973) and

Karttunen (1977), Cheng assumes that yes-no questions denote a set of propositions, and

dou quantifies over this set, just like when it quantifies over a plural NP. Cheng further

99

assumes that when the embedded question (the sentential subject) is quantified over by

dou, the interrogative interpretation disappears. Cheng (1995) does not explicitly provide

an analysis of (18), however.

Another construction in which a wh-phrase and dou are not adjacent is refered to

by Cheng and Huang (1996) as dou-conditionals. An example of dou-conditionals is in

(19).

(19) Ni jiao shei jin-lai, wo dou jian ta.

you ask who come in, I all see him/her

‘Whoever you ask to come in, I will see him/her (the person who you ask to come in).’

(Cheng and Huang 1996, p. 145)

Cheng and Huang (1996) propose that (19) involves elliptical embedded questions

because the antecedent clause can be fully spelled out with the question embedded under

the subordinating conjunction bulun ‘regardless of’. They assume that this conjunction

selects an interrogative proposition as its complement. In their analysis, the wh-

indefinites in (19) are not variables, but existential quantifiers. The antecedent in (19) is

an embedded question, over which dou quantifies, and thus contributes to the universal

quantification.

In summary, Cheng and Cheng & Huang propose two analyses for universals with

dou. In local cases, Cheng (1995) assumes that wh-indefinites are variables; they are

bound and given universal interpretations by dou, which is adjoined to the AspP or VP

100

domain and m-commands the wh-indefinites. Cheng and Huang (1996) analyze the non-

local universal cases of wh-indefinites and dou as conditional sentences and call them

dou-conditionals. The antecedent clause where wh-indefinites appear is analyzed as an

elliptical phrase containing an embedded question and wh-indefinites are treated as

existential quantifiers. Dou quantifies over the set of propositions denoted by the

question.

In order to account for non-local cases, Nishigauchi proposes LF movement,

while Cheng proposes two distinct denotations for wh-indefinites: They are restricted free

variables in local contexts, but existential quantifiers in non-local cases. Let us consider

how the neo-Hamblin semantics accounts for wh-indefinites in local and non-local cases.

3.2.3 Shimoyama (2001, 2006), Krazer and Shimoyama (2002)

Shimoyama (2001) and Krazer and Shimoyama (2002) (henceforth K&S 2002),

based on Hamblin semantics of wh-questions and Rooth’s (1992) focus interpretation

theory, propose that a wh-phrase denotes a set of individual alternatives. The Japanese

wh-phrase dare, ‘who’, for example, denotes the set of individuals {a, b, c...} in the

context. Other lexical items denote singleton sets of their standard denotations. For

example, the verb ‘slept’ introduces just one alternative, namely the property that is true

of an individual x in a world w if and only if x slept in w. The denotation of the Japanese

sentence dare-ga nemutta ‘who-Nom slept’ is the set {a slept, b slept, c slept…}, if a, b

and c... are the individuals available in the context. This set is obtained via applying

101

Hamblin Functional Application (20) in a pointwise manner, the computation of which is

illustrated in (21).18

(20) Hamblin Functional Application

If α is a branching node with daughters β and γ, and [[β]]w,g ⊆ Dσ and [[γ]]w,g⊆ D<στ>,

then [[α]]w,g = {a∈ Dτ :∃b ∃c [ b∈[[β]]w,g & c ∈ [[γ]]w,g & a = c (b)]}.

(21) a. [[ dare ]] w, g = { x: human (x) (w)}

b. [[ nemutta ]]w, g = { λxλw’. slept (x) (w’)}

c. [[dare-ga nemutta]] w, g = {p: [human (x) (w) & p = λw’. slept (x) (w’)]}

The set of propositions in (21c) results from combining (via Hamblin Functional

Application) the verb denoting a singleton set of functions of type <et> with each

individual in the set denoted by the wh-indefinite ‘who’. In other words, the individual

alternative set introduced by the wh-indefinite into the semantic computation expands via

pointwise Hamblin Functional Application, forming alternative sets of higher type, and

keeps expanding until captured by an operator that takes sets of alternatives as its

arguments. These operators are listed below.

18 Note that the denotation in (21c) is not what K&S 2002 propose. On their analysis, there is an existential quantifier as below. (i). [[dare-ga nemutta]] w, g = {p: ∃x [human (x) (w) & p = λw’. slept (x) (w’)]} Given that what we need is a set of proposions of the form: x is a person and x slept in w’, we think it is unnecessary to have an existential quantifier in the denotation.

102

(22) Sentential quantifiers

For [[ α ]]w,g ⊆ D <s,t> :

(i) [[ ∃α ]]w,g = {λw’. ∃p [p∈ [[ α ]]w,g & p(w’)=1]}

(ii) [[∀α]]w,g = {λw’. ∀p [p∈ [[ α ]]w,g → p(w’)=1]}

(iii) [[Neg α]]w,g = {λw’. ¬∃p [p∈ [[ α ]]w,g & p(w’)=1]}

(iv) [[ Qα]]w,g = [[ α]]w,g (Hamblin 1973)

(23) Generalized quantifiers

For [[ α ]]w,g ⊆ D <e> :

(i) [[ ∃α ]]w,g = {λPλw’. ∃a [a∈ [[ α ]]w,g & P (a) (w’)=1]}

(ii) [[∀α]]w,g = {λPλw’. ∀a[a∈ [[ α ]]w,g → P (a)(w’)=1]}

If the set in (21c) meets with an existential quantifier, the computation will be

stopped and the set will be existentially closed. In that case, we will have a singleton set

containing the proposition {that at least one person slept}, as in (24).

(24) [[ ∃α ]]w,g = {λw’. ∃p [p∈ [[ 21c ]]w,g & p(w’)=1]}

= {λw. ∃x [human (x) (w) & slept (x) (w)]}

= that at least one person slept

If the set in (21c) is captured by a Q(uestion)-operator, then by definition (22iv), the set

in (21c) is the denotation of the question.

103

In their system, there is no need to posit a local relation between the quantifier

and the wh-indefinite because the quantifier operates over the set formed by the wh-

indefinite. For example, both (25) and (26) obtain universal interpretations by the same

mechanism: The universal quantifier mo quantifies over the set of alternative individuals

denoted by dono gakusei, ‘which student’ in (25a), over the set of alternative mothers of

some student or other in (25b), and over the set of alternative children who read one book

or another in (26).

(25) a. Dono gakusei-mo odotta.

which student-MO danced

‘Every student danced.’ (Shimoyama 2006, p.139)

b. [Dono gakusei-no okaasan ] -mo odotta.

which student-Gen mother- MO danced

‘Every mother of some student or the other danced.’ (Shimoyama 2006, p.139)

(26) [[ dono hon-o yonda] kodomo]-mo yoku nemutta.

which book-ACC read child MO well slept

‘For every book x, the child who read x slept well.’ (K&S 2002, p. 2)

The wh-indefinite embedded in the complex NP combines with other elements to form a

set of alternatives A = {the child who read book a, the child who read book b, the child

who read book c …}. This set is captured by mo, a universal quantifier that operates over

104

sets of individual alternatives, and returns a singleton set of properties that all members

of the set A have. This set combines with the predicate ‘slept well’, yielding the set {the

child who read book a slept well, the child who read book b slept well, the child who read

book c slept well…}.

Their theory, therefore, accounts for the ‘local’ as well as ‘non-local’ cases

without the need for LF movement: mo is a universal quantifier, taking as its restrictor

any sister argument that introduces sets of alternatives. How does their theory account for

the island puzzle, mentioned in the section on Nishigauhi’s analysis?

The absence of the CNP island effect observed in (27a) falls out naturally with no

movement involved. As illustrated in the denotation (27b), if there are three students:

Taro, Ziro, Saburo in the context, then the set that mo quantifies over is {ι y [homework

assignment (y) & submit (y) (Taro)], ι y [homework assignment (y) & submit (y) (Ziro)],

ι y [homework assignment (y) & submit (y) (Saburo)]}. There is no need for covert

movement for mo to quantify over the embedded wh-indefinite properly.

(27) a. [[ Dono gakusei-ga teisyutusita ] syukudai ]- mo yuu-datta.

which student -Nom submitted homework - MO A-was

‘Every homework assignment that a student had handed in got an A.

b. ∀x [x∈ { ι y [homework assignment (y) & submit (y)(z)]: student (z)}

→ get an A (x)] (Shimoyama 2006, p.151)

105

What happens in wh-islands? Shimoyama (2001) proposes that the wh-island

effect is expected from the architecture of the interpretation system: The alternatives

introduced by the wh-indefinites expand until they meet the first operator of the relevant

kind, that is, one that takes a Hamblin denotation as its argument. In the case of wh-

islands, the first operator that the wh-indefinite meets is the Q operator ka. In wh-islands

the association of a higher particle like mo with the wh-indefinite is blocked by the lower

particle ka, while the complex NP case is free from this blocking because there is no Q

particle in the complex NP. This is schematized below.

(28) a. * [ …. [ …. wh-indefinites … ]-ka/mo ….]-ka/mo

b. [… […. wh-indefinite ….] CNP ]-ka/mo

In summary, the analysis of Shimoyama (2001) and K&S (2002) is the most

plausible of all the theories we reviewed for the following reasons: First, compared with

the Cheng and Cheng & Huang analyses, where a wh-indefinite is a variable in local

contexts, but is an existential in non-local contexts, their theory offers a uniform

denotation for wh-indefinites in local as well as non-local contexts: They consistently

denote sets of individual alternatives. They also provide a uniform mechanism for

universal interpretation in local and non-local contexts: The sets of alternatives of any

type are captured by the universal operators; hence, a universal interpretation obtains.

106

The island puzzle in Japanese, which is problematic on Nishigauchi’s analysis, is solved:

Wh-islands result because alternative sets are captured by the closer operator.

3.3 Wh-universals in Vietnamese

We have seen in the introduction that in Vietnamese a universal interpretation

arises when a wh-indefinite occurs along with the particle cũng, as shown by the

universal interpretation in (1a) and the lack of this reading in (1b), repeated below.

(1) a. Ai/ Người nào cũng vui-vẻ.

who/ person which CUNG happy

‘Everyone/Every person is happy.’

b. Ai/ Người nào vui-vẻ.

who/ person which happy

*‘Everyone/Every person is happy.’

‘Who is happy?’

With respect to its position in the sentence, the data we have so far show that cũng

precedes the verb. In fact, it appears in a position higher than the tense particle as in (29).

(29) a. Ai Tân cũng sẽ phải giúp.

who Tan CUNG FUT must help

‘Tan will have to help everyone.’

107

b. * Ai Tân sẽ cũng phải giúp.

who Tan FUT CUNG must help

‘Tan will have to help everyone.’

c. *Ai Tân sẽ phải cũng giúp.

who Tan FUT must CUNG help

‘Tan will have to help everyone.’

The sentences in (29b, c) show that cũng precedes neither the main verb, nor the modal

verb. It instead precedes a tense marker (29a).

Now, let us investigate the syntactic behavior of the particle cũng in universal

contexts. First, this particle requires the element it associates with to occur to its left, a

property dubbed as the leftness condition in the literature (see Cheng 1991). For example,

as we have observed in the introduction, an object base-generated wh-indefinite must

move to sentence-initial position to be associated with cũng as in (2a), repeated below.

Otherwise, it is construed as an interrogative-phrase and the sentence is a wh-question as

in (2c), repeated below.

(2) a. Ai / Người nào Tân cũng quen.

who / person which Tan CUNG know

‘Tan knows everyone/ every person.’

108

c. Tân cũng quen ai / người nào?

Tan CUNG know who/ person which

‘Who does Tan too know?’

Recall that cũng is an additive particle in (2c). This additive is associated with the

constituent to its left. Hence, the question in (2c) is felicitous in a context where there is

someone other than Tan who knows a person unknown to the speaker, and the speaker

chooses to ask about Tan for the identity of that person. It does not mean ‘who else does

Tan know?’Another example about this property is given in (30).

(30) Tân cũng đã giúp May.

Tan too PST help May

‘Tan too helped May.’

The sentence in (30) is felicitously uttered in a context where there is someone other Tan

who helped May. The speaker talks about that person, and then continues with Tan. In a

word, the leftness condition applies when the particle cũng is an additive particle as well

as when it occurs in universal contexts.

This leftness condition can account for the ambiguity of the sentence in (31).

109

(31) Tân ai cũng thích.

Tan who CUNG like

i. ‘As for Tan, he likes everyone.’

ii. ‘As for Tan, everyone likes him.’

The sentence in (31) is ambiguous: In the first reading ‘Tan’ is the subject. The

wh-phrase ai, ‘who’, moves from the object position to a position adjacent to cũng to

associate with it and obtain a universal reading. In the second reading, the wh-phrase ai,

‘who’, is the subject and is associated with cũng in its base-generated subject position.

Secondly, cũng is associated with one wh-indefinite at a time, and with the closest one.

(32) a. Cái gì ai cũng mua.

what who CUNG buy

i. ‘What is it that everyone buys?’

ii. *‘Who buys everything?’

iii.* ‘Everyone buys everything.’

b. Ai cái gì cũng mua.

who what CUNG buy

i. ‘Who bought everything?’

ii.* ‘What did everyone buy?’

iii.* ‘Everyone bought everything.’

110

The absence of the reading (32a ii) and (32b ii) indicates that only the wh-indefinite

closer to cũng, is associated with it. The absence of the reading in (32b ii) and (32b iii)

shows that only one wh-indefinite is associated with cũng.

Finally, as in Japanese and Chinese, a wh-indefinite can associate with cũng non-

locally, as in (33), where it appears that the wh-phrase meaning ‘who’ is in the antecedent

of a conditional and cũng is in the consequent. 19

(33) Ai xin pro/ anh ấy anh ấy cũng cho * ai / pro.

who beg pro/ he he CUNG give who/ pro

‘If X begs him (for money), he will give (money) to X.’

Note that the object of the verb ‘beg’ in the antecedent clause can be taken by an empty

pronoun or a pronoun, but the object of the verb ‘give’ in the consequent clause must be

an empty pronoun.

Another non-local example is given in (34). As previously noted, a classifier does

not obligatorily co-occur with the simplex wh-phrase ‘what’ when it is in object position.

(34) Anh nấu (cái) gì tôi cũng ăn * gì /pro.

you cook CL what I CUNG eat what/pro

‘If you cook X, I will eat X.’

19 The terms ‘local’ and ‘non-local’ refer to mono-clausal and bi-clausal contexts, namely when a wh-indefinite and its associate are in the same clause or in different clause respectively.

111

Like in (33), only the empty pronoun, not the wh-indefinite is allowed in the consequent

clause of (34).

In fact, the sentence in (34) is synonymous with (35a), a conditional sentence with

a subordinating element ‘if’ and a particle ‘then’. Note that the occurrence of cũng in

non-local contexts is not obligatory: (35b) means roughly the same as (35a).

(35) a. Nếu anh nấu gì thì tôi cũng ăn pro.

if you cook what then I CUNG eat pro

‘If you cook X then I will eat X.’

b. Nếu anh nấu gì thì tôi ăn pro.

if you cook what then I eat pro

‘If you cook X then I will eat X.’

The minimal difference between them is while (35a) conveys a concessive flavor,

meaning somewhat like ‘no matter what and how you cook, I don’t care’, (35b) does not.

It simply means ‘if you cook something then I will eat it, but if you do not cook, then I

will be OK without something to eat’.

112

3.4 Proposal

3.4.1 Cũng: Focus Sensitive Particle

We have seen that cũng is an additive particle, glossed as ‘too’, preceding tense

particles and obeying the leftness condition. In the following example, it is shown that

cũng is a focus-sensitive particle.

(36) a. Tân cũng giúp Hoàng.

Tan too help Hoang

‘Tan too helps/helped Hoang.’

Presupposition: Someone other than Tan helps/ helped Hoang.

b. Hoàng, Tân cũng giúp.

Hoang Tan too help

‘Tan too helps/helped Hoang.’

Presupposition 1: Tan helps/helped someone other than Hoang. (stress on Hoang)

Presupposition 2: Someone other than Tan helps/ helped Hoang. (stress on Tan)

The two sentences in (36a, b) share the same predicate-argument structure, namely, in

these sentences, Tan is the subject and Hoang is the object of the verb ‘help’. However,

while (36a) makes available only one presupposition, that is, there is someone other than

Tan that helps/helped Hoang, (36b) has two possible presuppositions depending on focus.

The first presupposition of (36b) obtains when ‘Hoang’ receives the main stress. The

second presupposition of (36b) arises when the main stress is on ‘Tan’. In this case,

113

‘Hoang’ is construed as a topic. With this interpretation, (36b) should be translated as ‘As

for Hoang, Tan too helps/helped him’. Since in a Vietnamese sentence, stressing an

element indicates that the content it introduces is new information, we can say that in the

former reading, ‘Tan’ is focus-marked (37a), while in the latter ‘Hoang’ is focus-marked

(37b).

(37) a. Hoàng [Tân ] F cũng giúp.

b. [ Hoàng ] F Tân cũng giúp.

On Rooth’s (1992) theory of focus, a focus-marked element is provided with two

semantic values: the ordinary semantic value and the focus semantic value. For example,

the denotation of (37a) with the ordinary semantic value of ‘Tan’ is as in (38a), and with

the focus semantic value as in (38b).

(38) a. λw. that Tan helped Hoang in w

b. {p: x is person & x helped Hoang in w}

= that May helped Hoang, that Nhi helped Hoang, that Tan helped Hoang…

The ordinary value of (37a), as in (38a), is a proposition, namely, a set of

possible worlds in which Tan helped Hoang. The focus value of the focus-marked ‘Tan’

denotes a set of individual alternatives: {x: x is a person}. The combination of this set

with the VP {λxλw. x helped Hoang} via pointwise functional application yields (38b).

114

In summary, cũng is a focus-sensitive particle and associates with the focus-

marked element on its left.

3.4.2 Cũng as a Universal Quantifier

Now let us consider how neo-Hamblin semantics works in local universal

contexts. Following the standard analysis of neo-Hamblin semantics (K &S 2002), I

assume that a Vietnamese wh-indefinite denotes a set of individual alternatives. Then

what is the syntactic and semantic role of the focus-sensitive particle cũng?

Consider the sentence in (39).

(39) Ai cũng đã giúp Tân.

who CUNG PST help Tan

‘Everyone helped Tan.’

I assume that cũng is head of a Focus Phrase that takes TP projection as its complement.

The surface word order of the sentence in (39) results from the movement of the subject

wh-indefinite ‘who’: First it moves from its base-generated position inside vP to Spec,

TP, then it moves from Spec, TP to Spec, FocP. This is shown in the diagram in (40),

where the striked-through elements indicate the traces.

115

(40)

FocP

ai Foc’ Foc TP cũng ai T’ T vP đã

ai giúp Tân.

While the surface word order in (39) matches quite well with the analysis, the

sentence in (41) imposes a challenge to our theory in that it is not the wh-indefinite, but

the subject NP ‘Tan’ that is adjacent to cũng.

(41) Ai Tân cũng đã giúp.

who Tan CUNG PST help

‘Tan helped everyone.’

Another way of expressing the propositional content in (41) is (42).

(42) Tân ai cũng đã giúp.

Tan who CUNG PST help

‘As for Tan, he helped everyone.’

116

To account for the surface word order of (42), we assume that the wh-indefinite ‘who’

moves from its object position to Spec, FocP, while the subject ‘Tan’ moves from Spec,

TP to Spec, Top. This is shown in (43).

(43) TopP Tân Top’ Top FocP

ai Foc’

Foc TP cũng Tân T’ T vP đã

Tân giúp ai. The surface word order of (41) results from scrambling the wh-indefinite ‘who’ from

Spec, FocP as shown in (44).

(44) Ai i [TopP Tân [FocP ti cũng đã giúp ]].

who Tan CUNG PST help

‘Tan helped everyone.’

117

For the universal interpretations of these sentences, I propose that at LF cũng

attaches to a constituent of sentence level, namely FocP, and functions as a universal

quantifier over sets of propositional alternatives, created by the wh-indefinites in

combination with other elements via Hamblin functional application. The LF of (44) is as

in (45). The denotation of cũng is given in (46).

(45) LF: [ cũng [FocP aii [Foc’1 [TP Tân đã [VP giúp ti]]]]].

(46) [[cũng ]]w,g = {λαλw’. ∀p [p ∈[[α ]]→ p(w’)=1]}, where [[ α]] w,g ⊆ Dst

The syntactic-semantic derivation of (44) with the LF in (45) is as follows.

(47) [[cũng FocP ]] w,g = {λw’.∀p [p ∈[[FocP ]] → p(w’)=1]}

cũng ∀ [[ FocP ]] w,g = {p: ∃x. [human (x) & helped (Tan) (x) (w’)}

ai i 1 [[TP]] w,g ={λw’. helped (Tan) (g(i)) (w’)} [[ai]] w,g ={x: human (x)(w’)}

[[ VP ]] w,g = {λyλw’. y helped g(i) in w’} Tân

giúp t i

If May, Nhi, and Gi are all people in the context, then (44) denotes the set {that Tan

helped May, that Tan helped Nhi, that Tan helped Gi}.

118

We have seen that neo-Hamblin semantics accounts nicely for the local universal

interpretations, namely when a wh-phrase and cũng are in mono-clausal contexts. We

know that Shimoyama’s (2001) theory aims to solve the so-called non-local cases without

the need of LF movement. Does her theory address the non-local case, that is, when a

wh-phrase and cũng are in bi-clausal contexts?

Actually, Shimoyama (2001) proposes that the universal quantifier mo in

Japanese is a cross-categorical universal quantifier, that is, it can operate over

constituents of different categories so long as they are its sisters. The denotation of this

cross-categorical universal quantifier is given in (48). In (49a, b), it attaches to clausal

constituents and quantifies over those clauses.

(48) [[ mo]] = λPλQ∀x [P(x) →Q(x)], where x ∈ D τ and P, Q ∈ D <τ,t>

(49) a. Taro-wa [dare-ga denwasi-te]-mo deru.

Taro-TOP who-NOM call- TE MO answer

‘No matter who calls, Taro answers.’

b. Taro-wa [Yoko-ga itu denwasi-te]-mo rusudatta

Taro-TOP Yoko-NOM when call-TE MO was-not-home

‘No matter when Yoko called, Taro was not home.’

Shimoyama (2001) assumes a situation-based analysis in which mo quantifies over the set

of situations denoted by the clause. For example, in (48a), it operates over the set of

119

situations in which someone calls. The sentence roughly means: ‘Every situation in

which someone calls is extendable to a situation in which Taro answers.’

I propose that Shimoyama’s (2001) situation-based semantics can account for

non-local cases such as the sentence in (50a).

(50) a. Anh nấu gì tôi cũng ăn pro.

you cook what I CUNG eat pro

‘If you cook X then I eat X.’

b. [IP ∀ [IP [CP 1 [IP 1 anh nấu gì ]] [IP 2 tôi cũng ăn pro]]]

The problem we have is how to combine the situation semantics in Shimoyama (2001)

with the possible world semantics in K&S (2002) for the task at hand.

I adopt Portner’s (1997) assumption that possible situations may be thought of as

spatiotemporal parts of possible worlds, giving rise to a natural part-of relation ‘≤’, where

the maximal possible situations are the worlds. I propose that cũng is not a quantifier in

(50a); instead, it introduces a covert universal quantifier with the following denotation.

(51) a. [[∀]]w,g = λpλq. ∀s [s ≤ w & p (s) = 1] → [∃s’[s ≤ s’. q(s’) = 1 ]]

b. For all situations s, such that s ≤ w, if p is true in s, then there exists an s’

such that s ≤ s’, q is true in s’.

120

This assumption is motivated by the following facts: First, cũng is not obligatory

as it is in non-local contexts. Secondly, while an adverb of frequency changes the

interpretation of the wh-indefinites in non-local contexts, it does not do so in local

contexts. For clarification, consider the following examples.

(52) Ai cũng hay dậy muộn.

who CUNG often get.up late

‘Everyone often got up late.’

The sentence in (52) is truthful in the context where ten people go camping for ten day

and ten of them get up late in seven days. It is not truthful if ten people go camping for

ten days, but seven of them get up late in ten days. This truth condition confirms that

cũng is a universal quantifier.

(53) Ai thức khuya pro cũng hay dậy muộn.

who stay.up late pro CUNG often get.up late

‘If X stays up late, then X often gets up late.’

However, cũng in (53) is not. The sentence is not truthfully uttered if ten people stay up

late for ten day and ten of them get up late in seven days. The truth condition for (53) is

that out of ten people who stay up late for ten days, seven of them get up late in ten days.

121

This truth-condition suggests that the adverb of frequency hay, ‘often’, plays the role of a

quantifier, not cũng.

Now, let us return to (50a). Structurally, it is a conditional sentence. Following

Iatridou (1991) I assume that the sentence-initial if-clause involves IP-adjunction. The

wh-indefinites in the antecedent are existentially closed by an operator over sets of

propositional alternatives. 20 The null pro in the consequent is construed as an e-type

pronoun.

Accordingly, the semantic composition of (50a) with the LF in (50b), where the

universal operator introduced by cũng is attached to the IP forming a tripartite

construction, is in (54).

(54) [[IP]] w,g = {λw.∀s [s ≤ w & [[CP1]] (s)] → [∃s’ s ≤ s’. [[IP2]] (s’) = 1 ]} ∀ IP CP1 IP2 [[CP1]] w,g = {that you cook at least one thing in s} [[IP2]] w,g = {that I eat in s’the thing that you cook in s} ∃ IP1 [[IP1]] w,g = {p: ∃x. you cooks x in s & s ≤ w } ↓ [[CP1]] w,g = [[∃ IP1]] w,g = {λs.∃p [p ∈[[IP1]] w,g & p (s) & s ≤ w]}

20 As will be discussed in Chapter 4, the wh-indefinite in the antecedent clause is licensed by a conditional operator.

122

[[ 50a ]] w,g = ‘For all s, s is part of w, if you cook at least one thing in s, there exists a

situation s’ such that s is part of s’, I eat in s’ the thing you cook in s.’

The situation-based analysis can account for cases where the appearance of an

adverb of frequency affects the quantificational reading. In (55a, b), the universal

interpretations do not obtain due to the occurrence of the adverb of frequency ‘often’.

(55) a. Ai có tiền thì cũng hay đi chơi.

who have money then CUNG often go play

i. ‘If people have money, they often go out enjoying themselves.’

ii. ‘(No matter who they are) if they have money they often go out enjoying

themselves.’

b. Ai có tiền thì hay đi chơi.

who have money then often go play

‘If people have money, they often go out enjoying themselves.’

As mentioned, the presence of cũng conveys a concessive flavor, which is indicated by

the second reading in (55a). Without it, as in (55b) this concessive connotation does not

arise. What is important to our current discussion is that disregarding the concessive

issue, the two sentences share the same truth conditions: They are true if and only if most

people who have money go out enjoying themselves. For example, they are true in a

context where there are five rich people, people with money, but only three of them go

123

out enjoying themselves. It does not mean ‘all people that have money often go out

enjoying themselves’, that is, in a context where there are five people with money and

they all go out enjoying themselves on and off, infrequently. On situation-based

semantics, the sentences in (55a, b) can be construed as ‘most situations s such that there

exists a person who has money in s is extendable to a situation s’ such that s’ is part of s,

and the person who has money in s goes out enjoying himself/herself in s’’. On this

analysis the adverb of frequency is translated as a quantifier over situations: MOST. Note

that just like non-local universals with cũng, the proposition alternative sets in the

antecedent of (55a, b) are existentially closed by a Hamblin operator.

In summary, cũng is a focus-sensitive particle and triggers the projection of a

Focus Phrase that requires a focus-marked element in Spec, Foc P. In local wh-universal

contexts cũng is a universal quantifier over sets of proposition alternatives. In non-local

contexts, it introduces a covert operator over situations.

3.5 A Special Case of Wh-universals

There is a special structure of wh-universals in Vietnamese. In this kind of wh-

universal a wh-indefinite needs to be associated with an element that only occurs in this

construction. Consider the example in (56).

(56) Ai nấy gặp Tân.

who NAY meet Tan

‘Everyone met Tan.’

124

The wh-indefinite ai ‘who’ obtains a universal interpretation when attached to

element nấy in (56). The constituency tests below in (57a, b, c) indicate that ai nấy is a

constituent consisting of the wh-phrase ai, ‘who’, and nấy.

First, inserting a sentential adverb in between the two elements results in

ungrammaticality as in (57a), and second, topicalization of the two elements together is

grammatical.

(57) a. *Ai rồi nấy ra-đi.

who then NAY leave

‘Eventually, everyone leaves/left.’

b. Rồi ai nấy ra-đi.

then who NAY leave

‘Eventually, everyone leaves/left.’

c. Tôi nghĩ [ ai nấy đang rất lo].

I think who NAY ASP very worry

‘I think everyone is very worried.’

d. Ai nấy [ tôi nghĩ [ đang rất lo ]].

who NAY I think ASP very worry

‘Everyone, I think is very worried.’

125

The element nấy only co-occurs with the simplex wh-phrase ai ‘who’, and never

appears with complex wh-phrases such as người nào, ‘which person’, or the non-human

wh-phrase gì, ‘what’.

(58) a. *Người nào nấy gặp Tân.

person which NAY meet Tan

‘Every person met Tan.’

b. * Gì nấy rất đẹp.

what NAY very beautiful

‘Everything is very beautiful.’

The wh-universal constituent ai nấy occurs only in subject position.

(59) a. *Tân thích ai nấy.

Tan like who NAY

‘Tan likes everyone.’

b. *Ai nấy Tân thích.

who NAY Tan like

‘Tan likes everyone.’

c. Ai nấy thích Tân.

who NAY like Tan

‘Everyone likes Tan.’

126

Neither fronting ai nấy from object position to sentence-initial position as in (59b), nor

keeping it in object position as in (59a), yields grammaticality.

While a non-human wh-phrase is not allowed to occur with nấy, there seems to be

a non-local association between a non-human wh-phrase and this element.

(60) Anh nấu gì (thì) tôi ăn nấy.

you cook what (then) I eat NAY

‘You cook x then I eat x.’

A piece of evidence supporting that (60) is bi-clausal and a conditional sentence

is that a conditional particle, glossed as ‘then’ and identical to the topic marker, can

optionally be inserted, separating the antecedent and the consequent. Therefore, sentence

(60) can receive the same analysis as the non-local wh-universals we have discussed in

which a covert quantifier over situations is responsible for the universal interpretation.

This hypothesis is similar to that of Bruening and Tran (2006), discussed below.

Bruening and Tran (2006) assume that the constituent ai nấy in (61a) is an

existential quantifier and the universal interpretation is derived from a covert quantifier

over situations, as represented in (61b).

(61) a. Ai nấy gặp Tân.

who NAY meet Tan

‘Everyone met Tan.’

127

b. ∀ s [∃x. person (x) in s ] →[ ∃s’ s ≤ s’. the person in s met Tan in s’]

In non-local cases such as (60), they assume that the non-human wh-indefinite gì

‘what’ and nấy form a constituent in the consequent. This constituent is an existential

quantifier. To account for the surface non-local relation, they propose an analysis based

on Nunes (2001)’s ‘Sideward Movement’. The constituent gì nấy moves from the

consequent clause to the antecedent clause before the two clauses are merged. Because

neither link in the resulting chain c-commands the other, both copies have to be

pronounced in some form gì in the antecedent and nấy in the consequent.

Adopting adopt Fox’s (1999, 2002) rule for interpreting traces, Kratzer’s (1989) and

Heim’s (1990) situation semantics for conditionals, Bruening and Tran (2006) interpret

(60), repeated below, as in (62):

(60) Anh nấu gì tôi ăn nấy.

you cook what I eat NAY

‘If you cook x, I eat x.’

(62) ∀s [∃x. x is a thing & you cook x in s ] [ ∃s’. s ≤s’ & I eat the thing x in s’]

Their analysis provides a unified denotation of wh-indefinites and a consistent

mechanism for local and non-local cases, but it leaves unanswered the question of the

denotations of the wh-indefinite and the morpheme nấy. However, this is just trivial from

the neo-Hamblin semantics perspective: The wh-indefinite that forms a constituent with

128

nấy denotes a set of individual alternatives, while nấy is an existential quantifier with a

denotation as in (63).

(63) For [[ α ]]w,g ⊆ D <e> :

[[nấy α ]]w,g = [[ ∃α ]]w,g = {λPλw’. ∃a [a∈ [[ α ]]w,g & P (a) (w’)=1]}

The existential quantifier nấy captures the individual set denoted by the wh-indefinite and

returns a singleton set: {λPλw’. ∃a [a∈ [[ α ]]w,g & P (a) (w’)=1]}

3.6 Conclusion

We have provided a uniform analysis for wh-universals in local and non-local

contexts. Basically, wh-indefinites uniformly denote sets of alternatives and universal

interpretations in both contexts obtain from neighboring operators. In local contexts, the

universal operator quantifies over sets of propositional alternatives; in non-local contexts,

the universal operator quantifies over situations. The reason why in local cases a wh-

indefinite must appear on the left of the focus-sensitive particle is that a Focus Phrase is

headed by the focus-sensitive particle that requires a focus phrase in its Spec. The focus-

sensitive particle is not needed in non-local cases because the quantificational

interpretation derives from a covert universal operator or an adverb of frequency.

The proposal that a Focus Phrase is headed by the focus-sensitive particle allows us to

account for the descriptive fact that cũng associates with only one constituent on its left,

as illustrated below.

129

(64) a. Cái gì ai cũng thích?

what who CUNG like

‘What does/did everyone like?

b. Ai cái gì cũng thích?

who what CUNG like

‘Who likes/liked everything?’

c. *Ai cũng thích cái gì?

who CUNG like what

‘What does/did everyone like?

d. Ai cũng thích cái gì thế?

who CUNG like what Q

‘What does/did everyone like?

In (64a) ai ‘who’ moves from Spec, TP to Spec, Foc P. What happens to the wh-phrase

cái gì ‘what’? If it stays in its base position as in (64c), then the sentence is

ungrammatical because of the LF intervention effect. Namely, the wh-phrase in (64c)

must move covertly to the CP domain for the sentence to be construed as interrogative,

but this movement is blocked by the operator introduced by cũng. If a question particle is

present, as in (64d), the sentence is grammatical because, as will be discussed in chapter

5, no movement is involved in this case. If it moves overtly, as in (64b), then the sentence

is grammatical. Now, the reason why only one wh-indefinite and the closest one is

130

universally construed is probably because cũng is the head of a Focus Phrase, hence only

what is next to it can be in Spec, FocP.

131

Chapter 4

EXISTENTIAL QUANTIFICATION

4.1 Introduction

In the previous chapter, based on neo-Hamblin semantics (Krazter and

Shimoyama 2002), we analyzed wh-indefinites in universal contexts as denoting a set of

individual alternatives, forming a Hamblin set of a higher type with other elements and

obtaining universal interpretations in association with universal operators. In this chapter,

we will look at the existential interpretations of wh-indefinites. Illustrative examples are

given in (1).

(1) a. Cô ấy không gặp ai.

she NEG meet who

‘She did not meet anyone.’

b. * Cô ấy gặp ai.

she meet who

‘She met someone.’

132

c. Cô ấy gặp ai đó.

she meet who DEM

‘She met someone.’

The wh-indefinite in (1a) is bare in that it is without any additional morphology, while

the wh-indefinite in (1c) is non-bare because it is attached with a demonstrative. While

the former requires a negative marker to obtain an existential reading, as shown by the

contrast between (1a) and (1b), the latter does not.

This chapter is concerned with the contexts where existential interpretations arise,

and the scopal behavior of wh-indefinites in these contexts. The chapter is organized as

follows. Section 2 is a short review of the analyses of wh-existential quantification.

Section 3 is concerned with bare wh-indefinites. Non-bare wh-indefinites are discussed in

section 4. Section 5 is the conclusion.

4.2 Literature Review

4.2.1 Cheng (1991)

Cheng (1991) analyzes Chinese wh-indefinites along the lines of Nishigauchi

(1986, 1991) who holds that wh-indefinites in Japanese are predicates with open

variables. Hence, a wh-indefinite like shei,‘who’, is represented as human (x), of which

the noun human denotes the restriction and the variable x is free. Its quantificational

interpretation, thus, has to come from a neighboring quantifier. Unlike Japanese wh-

phrases, however, a wh-phrase in Chinese is a polarity item, so it requires both a trigger

133

to license it as a polarity item (for example, the negative marker, or a question particle)

and a binder to determine its quantificational force, as exemplified below.

(2) a. Jialuo mai-le sheme ma.

Jialuo buy-ASP what Q

‘Did Jialuo buy anything?’

b. Jialuo mei-yo mai sheme.

Jiaoluo not-have buy what

‘Jiaoluo did not buy anything.’ (Cheng 1991, p. 124)

In the examples above, Cheng proposes that the instances of the wh-phrase sheme

are bound by the covert existential quantifier introduced by the rule of existential closure

(Heim 1982, Diesing 1992), but they have different triggers: The yes-no question

particle ma in (2a) and negative marker mei-you in (2b). She adopts Diesing’s (1992)

theory that there are two types of indefinites: Non-quantificational and quantificational.

The former undergoes existential closure, while the latter undergoes quantifier raising

(QR). However, as existential closure applies to the VP domain, and indefinite NPs in

Chinese are not quantificational, wh-phrases or indefinite NPs in subject position,

therefore, cannot be interpreted existentially. This accounts for the fact that in Chinese

indefinites are not grammatical as subjects as in (3a, b), while a sentence is grammatical

with definite NPs in subject position, as in (3c).

134

(3) a. *Shei xiang chi pingguo ma?

who want eat apple Q

‘Does anyone want to eat apples?’ (Cheng 1991, p.127, citing Huang 1982)

b. *Yi-ge-ren lai-le.

one-CL-person come-ASP

‘A person came.

c. Nei-ge-ren lai-le.

that-CL-person come-ASP

‘That person came.’ (Cheng 1995, p.128)

(3a) is ungrammatical because shei ‘who’ is in subject position and cannot be bound by

existential closure even though its licensor, the question particle ma, is present. Similarly,

(3b) is ungrammatical with the indefinite NP yi-ge-ren, ‘a person’, in subject position.

However, when you ‘have’ is present, the sentence is grammatical shown in (4).

(4) you yi-ge-ren lai-le.

have one-CL-person come-ASP

‘A person came/There came a person.’ (Cheng 1995, p.128)

Cheng (1991), following Huang (1988), assumes that you is a modal that takes an AspP

as its complement and is equivalent to an existential closure. Hence, the indefinite NP in

(4) receives the existential force from you.

135

In summary, a wh-phrase in Chinese needs both a trigger and a binder to obtain an

indefinite reading, namely as an existentially quantified expression. Since a wh-phrase is

a variable, and Existential Closure applies only in the VP domain, wh-indefinites are not

allowed in subject position.

4.2.2 Lin (1998, 2004)

Lin (1998) refers to Chinese wh-phrases as existential polarity wh-phrases

(henceforth EPWs). While Cheng assumes that Chinese wh-phrases are polarity items,

Lin argues that EPWs are different from negative polarity items (NPIs) in English. While

EPWs occur in typical affective contexts such as negative clauses, conditionals, and yes-

no questions, which license negative polarity item (NPIs) in English, there also exist

contexts that license Chinese EPWs, but not English NPIs, and conversely, contexts that

license English NPIs disallow the occurrence of EPWs. For example, while Chinese

EPWs are licensed in epistemic modality contexts, English NPIs are not allowed, as

shown in the contrast between the grammaticality of (5a) with the EPW shei and the

ungrammaticality of (5b), with the NPI any man in the same epistemic context.

(5) a. Keneng/ xiangbi shei you qifu ta le.

possibly/ most probably who again bully him ASP

‘Possibly/Most probably, someone bullied him again.’ (Lin 1998, p. 222)

b. *It is possible that any man was there. (Lin 1998, p. 228)

136

Conversely, English NPIs are licensed by adversative predicates like be surprised,

but EPWs are not.

(6) a. Mary is surprised that John eats any carrots.

b. * Wo hen jingya ta chi shenme shucai.

I very surprised he eat what vegetables

‘I am surprised that he eats any vegetables.’ (Lin 1998, p.229)

Similarly, wh-questions licenses NPIs in English, but in Chinese they do not

license EPWs.

(7) ‘Who wants anything to eat?’ (Lin 1998, p.228, citing Ladusaw 1980, p.186)

(8) Shei/shenme ren xihuan shenme?

who/what man like what

i.‘Who likes what?’

ii.* ‘What does somebody/anybody like?

iii.*‘Who likes something/anything?’ (Lin 1998, p. 234, citing Li 1992, p. 128)

Accordingly, Lin (1998) proposes that the licensing condition, or in his terms, the

Non-Entailment-of Existence Condition on EPWs, is as follows.

137

(9) The use of an EPW is felicitous iff the proposition in which the EPW appears

does not entail existence of a referent satisfying the description of the EPW.

(Lin 1998, p. 230)

Lin (2004) also argues against Cheng’s assumption that Chinese indefinites are

not grammatical in subject position. As shown in (10a, b), indefinites may sometimes be

subjects even without you.

(10) a. Turan yi-lang xiao keche zhuangshang wo-de jiche.

suddenly one-CL small car hit my motorcycle

‘Suddenly a small car hit my motorcycle.’ (Lin 2004, p. 269)

b. shi-bu-shi shei diao-le qian le?

be-not-be who lose-ASP money ASP

Is it the case that someone lost his money?’ (Lin 2004, p. 470)

In addition, Cheng’s hypothesis that in Chinese Existential Closure only applies to

VP is not well-grounded given the example below.21

(11) Ta haoxing mei/bu chi shenme (dongxi) de-yangzi.

he seem not eat what thing seem

a. ‘It seems that he did/does not eat anything.’

21 Interested readers may wonder why Existential closure does not apply at the higher VP headed by haoxing, ‘seem’, (8). In fact, it is not a verb as noted by Lin (2004): ‘Although haoxiang is translated as English seem, it is an adverb, not a verb’ (Lin 2004, p. 259)

138

b. ‘It seems that there is something such that he did/does not eat it.’

(Lin 2004, p. 460)

The wh-phrase shenme (dongxi) ‘what thing’ is ambiguous in its scope with

respect to the negative marker. Cheng’s analysis does not predict that shenme (dongxi)

can outscope the negation shown in the translation (8b). If Existential Closure only

applies inside the VP domain, then it is difficult to account for the intermediate scope of

shenme, where it takes scope over the negation.

A similar scopal ambiguity is observed in conditional contexts.

(12) Yaoshi ni bu xiang yaoqing shei de-hua, qing shixian rang wo zhidao.

if you not want invite who if please beforehand let me know

‘If you do not want to invite someone, please let me know in advance.’

(Lin 2004, p.460)

The sentence in (12) has a reading in which the wh-phrase takes a scope lower

than ‘if’ but higher than the negative operator: ‘If there is somebody such that you do not

want to invite him, please let me know in advance.’

In summary, Lin shows that wh-indefinites can take intermediate scope, for

example over negation. This is impossible to achieve via existential closure. To account

for this scopal behavior, Lin assumes that wh-indefinites in Chinese introduce choice

function variables that can be existentially closed in the scope of a licensor. In this view,

139

the intermediate scope of the wh-indefinites in (11) and (12) results from the existential

closure of the choice function variable in the intermediate scope of haoxing,‘seem’, and

yaoshi,‘if’, both of which are licensors. 22

Lin’s analysis of wh-indefinites as choice function variables, taking scope under

their licensors, nicely accounts for scopal ambiguity of wh-indefinites in Chinese, a

property shared by Vietnamese wh-indefinites in existential contexts. However, it is

impossible to extend Lin’s theory to Vietnamese wh-indefinites for the following reasons.

First, if wh-indefinites are choice function variables that need licensors and take scope

under the licensors, then what allows a wh-indefinite with additional morphology to

occur without licensors and takes any scope? Secondly, having a uniform denotation of

wh-indefinites in non-interrogative as well as interrogative contexts entails that wh-

questions in Vietnamese are insensitive to locality constraints. This is contrary to fact and

will be shown in chapter 5.

4.3 Bare Wh-indefinites

Now let us start our discussion of Vietnamese wh-indefinites. Wh-indefinites in

Vietnamese can be bare, namely, wh-indefinites are identical to wh-phrases, or non-bare,

namely, wh-indefinites are formed by adding some morphology to the wh-phrases.

Bare wh-indefinites in Vietnamese require licensing. A sentence with a bare wh-

indefinite as in (13) is ungrammatical.

22 A function f is a choice function (CH(f)) if it applies to any non-empty set and yields a member of that set (Reinhart 1997).

140

(13) * Cô ấy gặp ai.

she meet who

‘She met someone.’

Licensors are yes-no question operators (14a), conditional operators (14b),

epistemic operators (14c), an existential operator, meaning roughly something like ‘there

is/are’ or ‘have’ (14d), or negation operators (14e).

(14) a. Cô ấy có gặp ai không?

she Q meet who Q

‘Did/ Does she meet anyone?’

b. Nếu ai đến thì Anh Thơ sẽ rất vui.

if who arrive then Anh Tho FUT very happy

‘If someone arrives, AnhTho will be very happy.’

c. Hình như ai vừa gặp Tân.

seem who just meet Tan

‘It seems someone just met Tan.’

d. Có ai gặp Tân.

have who meet Tan

‘Someone met/ meets Tan.’

141

e. Tân không gặp ai.

Tan NEG meet who

‘Tan does/did not meet anyone.’

The wh-indefinites in the examples above take scope under their licensors. Thus

(14a) does not mean ‘there exists a person such that the speaker wants to know if she

met/meet that person’, where the existential closure of the wh-indefinite is wider than the

question operator. The wh-indefinite in (14b) takes scope under the conditional operator,

so the sentence means ‘if there exists a person such that that person arrives, then AnhTho

will be very happy’, not ‘there exists a person such that if that person arrives then

AnhTho will be very happy.’ Similarly, since the scope of the epistemic operator in (14c)

is wider than the wh-indefinite, and the sentence does not mean ‘there exists a person

such that it seems to the speaker that that person met Tan.’ Instead, it means it seems to

the speaker that there exists a person such that that person met Tan. Finally, the scope of

negation in (14d) is wider than that of the wh-indefinite, so the sentence does not mean

that ‘there exists a person such that Tan does/did meet that person.’

In sum, bare wh-indefinites in Vietnamese require licensing and take scope under

their licensor.

4.3.1 Licensing Condition

In the examples in (14), the licensor and the licensee are clause-mates. However,

as illustrated below, licensing does not require any clause-mate condition.

142

(15) Anh có tin là [cô ấy đã gặp ai ] hay không?

you Q believe COMP she ASP meet who or Q

‘Did/Do you believe that she met anyone/someone?’

In (15), the yes-no question operator có...không is in the matrix clause, and the wh-

indefinite is in the embedded clause.

Instead, a licensor must c-command a licensee to license it, as illustrated in the

examples below.

(16) [ Ai gặp anh ] có bất tiện không ?

who meet you Q inconvenient Q

‘Is it inconvenient that someone met you?’

The wh-indefinite is embedded in the subject clause, while the yes-no operator is in the

matrix clause. In this configuration the yes-no operator c-commands the wh-indefinite

and the sentence is grammatical. By contrast, when the licensor is embedded in the

subject clause and the wh-indefinite is in the matrix clause, as shown in (17), the licensor

does not c-command the licensee. The sentence is ungrammatical.

(17) *[ Anh ta không đến ] làm ai rất buồn.

he Neg arrive make who very sad

‘That he did not arrive makes someone very sad.’

143

4.3.2 Intermediate Scope

We have seen that a wh-indefinite and its licensor do not need to be clause-mates.

In what follows, we will see that when there is more than one potential licensor, a wh-

indefinite can take scope wider the lower licensor.

(18) Nếu anh không muốn mời ai thì báo cho tôi biết. IF > ∃ > NOT

if you NEG want invite who then report for I know IF > NOT > ∃

‘If you do not want to invite anyone/someone, let me know.’

There are two licensors in (18): The conditional operator ‘if’ and the negation operator

‘not’. The sentence in (18) is ambiguous: In the narrow reading, the sentence is truthful in

a context where there are ten people and if the hearer does not want to invite ten of them,

then the hearer is expected to notify the speaker so. In this reading, the wh-indefinite

takes scope lower than negation. In the intermediate reading, the sentence in (18) means,

‘if there is a person such that you do not want to invite that person, then let me know the

identity of that person’. The wh-indefinite takes scope wider than the negation in this

reading.

Similarly, the sentence in (19) is ambiguous between the narrow and the

intermediate scope reading. In the narrow scope reading, (19) asserts that ‘it seems that

she likes no one’. In the intermediate reading, (19) asserts that ‘it seems that there is a

person such that she does not like that person’.

144

(19) Hình như cô ấy không thích ai. SEEM > ∃ > NOT

seem she NEG like who SEEM > NOT > ∃

‘It seems she does/did not like someone.’

From the data above, we can come to the following characterization: bare wh-indefinites

in Vietnamese need licensors and normally take their scope under the scope of their

licensors. When other licensors higher in the construction are available, they can take the

scope in the immediate scope of these licensors. In other words, they can take scope in

the immediate scope of their potential licensors.

4.3.3 Analysis

It is obvious that Cheng’s (1991) analysis does not work for Vietnamese wh-

indefinites in existential contexts. The intermediate scope of a wh-indefinite is not

predicted by her theory since it confines existential closure to the VP. Lin’s (1998, 2004)

analysis provides a satisfactory account for the scopal behavior by analyzing wh-

indefinites as choice function variables that are existentially closed in the immediate

scope of a licensor. Hence, when there is more than one licensor, a wh-indefinite can take

scope in the immediate scope of the higher licensor and wider than the lower licensor.

However, Lin’s choice function analysis cannot be extended to wh-questions. As will be

shown in chapter 5, wh-questions are sensitive to island constraints and Beck’s

intervention effects. It is difficult for a choice function analysis to account for these

145

effects in wh-questions given that a choice function variable is standardly closed off in

the Spec CP domain in wh-questions without any sensitivity to islands.

In what follows, we will show that K&S 2002 accounts nicely for the data in

Vietnamese. In terms of neo-Hamblin semantics, a sentence such as (13), repeated below,

is not legitimate because the propositional alternative set lacks an operator that takes this

set as an argument and returns a singleton set, which is the denotation of a declarative

sentence. The scopal phenomena and the locality constraint can be explained on the basis

of the interaction between the alternative set denoted by the wh-indefinite and the

operator.

(13) * Cô ấy gặp ai.

she meet who

‘She met someone.’

4.3.3.1 Locality Condition and Narrow Scope

The narrow scope falls out easily by virtue of the interpretative mechanism in

neo-Hamblin semantics. Let us see how this mechanism works. K&S 2002 provide us

with the following two sentential operators that allow us to account for the existential

interpretations in sentences licensed by the existential operator, and the negation

operator.

146

(20) a. Sentential quantifiers (K&S 2002).

For [[ α ]]w,g ⊆ D <s,t> :

(i) [[ ∃α ]]w,g = {λw’. ∃p [p∈ [[ α ]]w,g & p(w’)=1]}

(iii) [[Neg α]]w,g = {λw’. ¬∃p [p∈ [[ α ]]w,g & p(w’)=1]}

For illustration, consider the examples below.

(21) a. Có ai mới gặp Tân.

have who just meet Tan

‘Someone just met Tan.’

b. LF: [Có [TP ai mới gặp Tân ] ]

The LF of (21a) is (21b); the propositional alternative set created by the wh-

indefinite and other elements in the construction, as in (22a), is existentially closed by the

Hamblin operator có (20i).23

23 The existential operator có can co-occur with lexical indefinites in subject position in existential sentences, where it quantifies over the set introduced by the NP, not sets of alternative individuals. We assume that có is ambiguous between a quantifier over sets of individuals and a quantifier over sets of alternative individuals. For example, the LF of (i) is in (ii), where the existential operator quantifies over the set of students.

(i) Có hai sinh-viên đang ngủ. have two student ASP sleep ‘There are two students sleeping.’

(ii) ∃x [student (x) & two (x) & sleep (x)]

147

(22) a. [[ TP ]] w,g = {p: y is a person & p = λw’. y met Tan in w’}

= that Nhi met Tan, that Lem met Tan, that Tho met Tan….

b. [[ ∃ TP]] w,g = {λw’. ∃p [p ∈[[TP]] w,g & p (w’) = 1]}

= that at least one person met Tan.

The sentence in (23a), with an LF as in (23b), follows a similar procedure as in

(24). The set of propositional alternatives in (24a) is captured by the negative operator,

yielding (24b).

(23) a. Tân không gặp ai.

Tan NEG meet who

‘Tan does/did not meet anyone.’

b. LF: [không [TP Tân gặp ai]]

(24) a. [[ TP ]] w,g = {p: y is a person & p= λw’. Tan met y in w’}

= that Tan met Lem, that Tan met May, that Tan met Nhi….

b. [[Neg TP ]]w,g = {λw’. ¬∃p [p∈ [[ TP ]]w,g & p (w’)=1]}

= that it is not true that Tan met a person

What happens when wh-indefinites are licensed by epistemic operators, the yes-

no question operator or a conditional operator? I assume that the denotation of the

epistemic operator is in (25) taken from K&S 2002. For the yes-no question operator I

148

provide the denotation as in (26). The conditional operator introduces a covert existential

quantifier in the antecedent clause as in (27).

(25) Possibility Modals: (K&S 2002)

For [[ α ]] w,g ⊆ D <st> : [[ kann α ]] w,g =

{λw’. ∃w’’ [w’’ is accessible from w’ & ∃p [p∈ [[ α]] w,g & p(w’’)= 1]]}

(26) Yes-no question operator

For [[ α ]]w,g ⊆ D <s,t> :

[[ Q y/n α ]]w,g = {{λw’. ∃p [p∈ [[ α ]]w,g & p(w’)=1]}, {λw’. ¬∃p [p∈ [[ α ]]w,g

& p (w’)=1]}}

(27) Sentential quantifiers (K&S 2002).

For [[ α ]]w,g ⊆ D <s,t> :

[[ ∃α ]]w,g = {λw’. ∃p [p∈ [[ α ]]w,g & p(w’)=1]}

The computations of sentences with these operators are illustrated below.

Consider the sentence in (28a), (29a) and (30a).

(28) a. Cô ấy có gặp ai không?

she Q meet who Q

‘Did/ Does she meet anyone?’

b. LF: [Q y/n [TP Cô ấy gặp ai ]]

c. [[ TP ]] w,g = {p: x is a person & p = λw’. she met x in w’}

149

d. [[Q y/n TP ]] w,g = {{λw’. ∃p [p∈ [[ TP]]w,g & p(w’)=1]},

{λw’. ¬∃p [p∈ [[ TP ]]w,g & p (w’)=1]}}

= that she met at least a person or that it is not true that she met at least a person

At LF the yes-no operator attaches to the sentence level (28b). It takes the alternative sets

of propositions (28c) and returns an alternative set in (28d), consisting of a two member

set of sets, which is the denotation of the yes-no question.

Next, consider (29a).

(29) a. Nếu ai đến thì Anh Thơ sẽ rất vui.

if who arrive then Anh Tho FUT very happy

‘If someone arrives, AnhTho will be very happy.’

b. LF: [Nếu [TP ai đến ]]

c. [[ TP]]w,g = {p: x is a person & p = λw’. x arrives in w’}

d. [[ ∃ TP]] w,g = {λw’.∃p [p ∈[[TP]] w,g & p (w’)]}

= that at least one person arrives

The LF of (29a) is (29b). The alternative set in the antecedent clause (29c) is existentially

closed, yielding (29d), which is the meaning of (29a). That is, ‘if at least one person

arrives, then Anhtho will be happy.’

Finally, the existential interpretation derived from an epistemic operator is

illustrated in (30).

150

(30) a. Hình như ai vừa gặp Tân.

seem who just meet Tan

‘It seems someone just met Tan.’

b. LF: [Hình như [TP ai vừa gặp Tân ]]

c. [[ TP]] w,g = {p: x is a person & p = λw’. x met Tan in w’}

d. [[ kann TP ]] w,g = {λw’. ∃w’’[w’’ is accessible from w’ & ∃p [p∈ [[ TP ]] w,g

& p(w’’) = 1]]}

= that there is some w’’ that is accessible from w’ where at least one person met

Tan in w’’

The computational derivations that result in existential interpretations account for

why wh-indefinites only take narrow scope with respect to their licensor: the licensor, or

the operator introduced by the licensors always takes as its argument a sentential

constituent that embeds the wh-indefinite. Therefore, they always take scope over the wh-

indefinites.

Now let us look at the locality constraint. It is well-known that quantifiers are

scopally constrained. They must take scope within the sentence where they occur. K&S

2002 propose that wh-indefinites do not fit into this picture. The only locality constraint

on wh-indefinites is that they must be captured by the first available operator. Hence,

they may not associate with a higher operator across a lower one:

(31) * [Op 1… [ Op 2…. wh-indefinite ….] ]

151

How does the locality constraint account for the contrast below?

(32) a. Hình như cô ấy muốn biết [ ai vừa mua gì ].

seem she want know who just buy what

i. ‘It seems she wants to know who just bought what.’

ii. *‘It seems she wants to know if someone just bought something.’

iii. ‘It seems she wants to know who just bought something.’

iv. ‘It seems she wants to know what someone just bought.’

b. LF: [CP SEEM … Q [CP … wh… ]]

The first reading in (32a) suggests that the locality constraint in (31) is operative:

As will be discussed in detail in chapter 5, the matrix verb ‘want know’ selects an

interrogative clause and thus licenses a question operator, the set propositional

alternatives denoted by an embedded clause is captured by the question operator, not the

higher licensor.24 Hence, it must be construed as a question. However, the availability of

the third and fourth readings indicates that it is the higher epistemic operator that renders

these indefinite interpretations of the wh-phrase available. This is confirmed by the fact

that when the epistemic operator is taken away, the indefinite readings of the embedded

wh-phrases disappear, as illustrated in (33). 24 For the embedded clause in (32a) to obtain the second reading (32aiii), the yes-no question must be employed as in (i). (i) Hình như cô ấy muốn biết [*(có) ai vừa mua gì hay *(không)]. seem she want know Q who just buy what or Q

‘It seems she wants to know if anyone bought anything.’

152

(33) Cô ấy muốn biết [ ai vừa mua gì ].

she want know who just buy what

i. ‘She wants to know who just bought what.’

ii.*‘She wants to know who just bought something.’

iv.*‘She wants to know what someone just bought.’

The example in (32) shows that licensing is not sensitive to wh-islands.

Similarly, in (34a), the indefinite reading of the wh-phrase embedded inside the Complex

NP Island suggests that it is captured by the operator outside the complex NP, namely the

epistemic operator.

(34) a. Hình như Nim mới ăn [CNP cái bánh mà [ Gi vừa mua cho ai ]].

seem Nim just eat CL cake REL Gi just buy for who

‘It seems Nim just ate eat the cake that Gi just bought for someone.’

b. LF: [CP SEEM … [CNP … wh… ]]

In sum, the data show that Vietnamese bare wh-indefinites do not obey the

locality constraint cast in K&S 2002 as shown in the diagram in (31). The next section

will look at an issue that patterns with what we have observed and analyzed.

153

4.3.3.2 Intermediate Scope: Not Always Available

The previous section illustrates that while narrow scope facts in Vietnamese can

be nicely captured by neo-Hamblin semantics, island facts in the language indicate that

the locality constraint in (31) is not obeyed, that is, licensing is not sensitive to islands.

More data given below confirm what we have observed and analyzed.

(35) a. Nếu anh không muốn mời ai thì báo cho tôi biết. IF > ∃ > NOT

if you NEG want invite who then report for I know IF > NOT > ∃

‘If you do not want to invite anyone/someone, let me know.’

b. Hình như cô ấy không thích ai. SEEM > ∃ > NOT

seem she NEG like who SEEM > NOT > ∃

‘It seems she does/did not like someone.’

The wh-indefinites in the sentences above display scopal ambiguity: They can

obtain both narrow and intermediate scope. Disregarding the narrow scope, the

intermediate scope reading of (35a) means ‘if there is a person such that you do not want

to invite that person, then let me know’. Similarly, the intermediate reading of (35b)

asserts that ‘it seems that there is a person such that she does not like that person’. From

the data above, we can come up with the following characterization: When other

licensors higher in the construction are available, wh-indefinites can take scope in the

immediate scope of these licensors. At an explanatory level, we can revise K&S 2002

system and say that in intermediate scope reading, the wh-phrase moves covertly out of

154

scope of the first licensor and forms an alternative set with other elements in its new

position. This alternative set is then captured by the higher operator, resulting an

intermediate reading. For illustration, consider how the intermediate and narrow scope

readings of (35b) are computed as in (36).

(36) Hình như [TP cô ấy không thích ai ].

SEEM she NEG like who

a. LF1: [ModP SEEM [TP2who i 1 [TP1 Tan NEG like t i ] ] ].

b. LF 2: SEEM [NegP NEG [TP Tan like who]]

The intermediate scope reading, represented as LF1 in (36a), obtains when the

wh-phrase ‘who’ moves out of the scope of the first licensor, NEG. The derivation of this

reading is in (37), based on the syntactic configuration in (38).

(37) a. [[ TP1 ]] w,g = { λw’. Tan did not like g (1) in w’}

b. [[ 1 TP1]] w,g = {λxλw’. Tan did not like x in w’}

c. [[ TP2 ]] w,g = {p: x is person & Tan did not like x in w’}

d. [[ ModP ]] w,g = {λw’. ∃w’’ [w’’ is accessible from w’ & ∃p [p∈ [[ TP2]] w,g &

p(w’’)= 1]]}

= that there is some w’’ that is accessible from w’ that Tan did not like a person in

w’’

155

(38) ModP

Seem TP 2

who i 1 TP 1

t i

As shown in (38), the wh-indefinite ai, ‘who’, moves covertly from the TP

domain (TP1) where the negation operator is located and forms a set of propositional

alternatives with the TP (TP2). This set is captured by the epistemic operator residing

over the TP domain.

In the narrow scope reading, the negation operator takes the set introduced by the

TP as shown in LF 2. The negation operator thus blocks the epistemic operator from

taking this set. The derivation of the narrow scope reading is in (39).

(39) a. [[ TP]] w,g = {p: x is a person & p = λw’. she likes x in w’}

b. [[ Neg TP ]] w,g = {λw’. ¬∃p [p ∈[[ TP ]] w,g & p(w’) = 1]}

= that it is not true that she likes a person

The covert movement of the wh-indefinite over the lower operator is confirmed

by the following data.

156

(40) Hình như Nim không ăn [cái bánh mà Gi mua cho ai].

seem Nim NEG eat CL cake REL Gi buy for who

‘It seems that Nim did not eat the cake that Gi bought for someone.’

(41) Nếu Nim không ăn [ cái bánh mà Gi mua cho ai] thì Gi rất vui.

if Nim NEG eat CL cake REL Gi buy for who then Gi very happy

‘If Nim did not eat the cake that Gi bought for someone, then Gi would be very

happy.’

(42) [ Licensor 2 [ Licensor 1 [CNP …wh…]]]

The configuration of the two sentences (40) and (41) is as in (42), where the wh-

indefinite is embedded in a CNP island, and the two licensors are in the matrix clause. If

LF movement is what happens in intermediate scope reading, then we expect that a wh-

indefinite in this configuration cannot obtain an intermediate scope. This is because

movement is restricted by islands. 25Therefore, the wh-indefinite cannot move over the

lower licensor, Op1, to be captured by the higher one, Op2. This prediction is borne out.

The sentence in (40) does not mean ‘it seems that there is a person such that Tan did not

eat the cake that Gi bought for that person’. Similarly, (41) does not mean ‘if there is a

person such that Tan did not eat the cake that Gi bought for that person, then Gi would be

very happy’.

However, wh-indefinites cannot always take intermediate scope. Consider the

examples in (43a, b), where a different negative marker is used. 25 LF movement is considered to be insensitive to islands by other authors (see Huang 1982). LF movement in Vietnamese is island sensitive (see chapter 5, section 5.2.2).

157

(43) a. Hình như Tân chẳng gặp ai. SEEM > NOT > ∃

seem Tan NEG meet who

‘It seems Tan did not meet anyone.’

b. Nếu anh chẳng muốn mời ai thì cho tôi biết. IF > NOT > ∃

if you NEG want invite who then let I know

‘I you do not want to invite anyone then let me know.’

The wh-indefinites in (43a, b) cannot scope out of the negation. This fact allows us to

hypothesize that the blocking mechanism is universal; hence the low scope of a wh-

indefinite with respect to its licensor is generally expected. The exceptional scope,

namely the intermediate scope, is available in cases where the operator allows the wh-

indefinite to move over it. 26 There is a non-movement alternative analysis: The

propositional alternative set introduced by the TP is allowed to expand over the sentential

negation operator, instead of being captured by it. This set is then captured by the higher

epistemic operator. However, this alternative analysis is not as elegant as the covert

movement analysis. First, it brings us the same result as the covert movement analysis at

the cost of violating the K&S 2002 constraint. Namely, the lower operator blocks the 26 Pragmatics plays a role as well. In the conditional sentence, if we replace the imperative in the consequent with a declarative clause, then the blocking effect is operative.

(i) Nếu anh không mời ai thì cô ấy sẽ rất buồn. if you NEG invite who then she FUT very sad * ∃ > NOT ‘If you do not invite anyone then she will be very sad.’ NOT > ∃

It is not clear to us why no intermediate scope is allowed in (i).

158

higher operator. Secondly, the covert movement of the wh-indefinite allows us to reach a

unified account when it comes to LF movement in wh-questions, as will be discussed in

the next chapter, section 5.2.3 on intervention effect, the negative marker chẳng blocks

LF movement as in (44), while the other không does not as shown in (45).

(44) Tân chẳng mời ai.

Tan NEG invite who

a.* ‘For which person x, Tan did not invite x?’

b. ‘There is not any person x such that Tan invited x.’

(45) Tân không mời ai.

Tan NEG invite who

a. ‘For which person x, Tan did not invite x?’

b. ‘There is not any person x such that Tan invited x.’

Briefly, in a wh-question without a wh-particle, a wh-phrase must move to Spec, CP. The

absence of the wh-question reading in (44) as opposed to its availability in (45) suggests

that this movement is blocked in the former, but is not blocked in the latter.

To summarize, wh-indefinites require licensors. A licensor does not need to be a

clause-mate with its licensee. However, it does need to c-command its licensee.

Existential interpretations of wh-indefinites derive from sentential operators and modal

operators. Bare wh-indefinites display properties well-known in the literature, namely, a

lower operator blocks a higher operator from taking the alternative sets. However, some

159

operators allow the wh-indefinite to move covertly over them and form a set with the

constituent that it moves out of, and this set gets captured by a higher operator. This

behavior results in intermediate scope.

4.4 Non-bare Wh-indefinites

Non-bare wh-indefinites are composed of a bare wh-indefinite and a

demonstrative meaning ‘that’. What is special about non-bare wh-indefinites is that they

do not need licensing, as shown in (46). This sentence is grammatical without any

licensor.

(46) Tân vừa gặp ai đó.

Tan just meet who DEM

‘Tan just met someone.’

When a non-bare wh-indefinite is c-commanded by a negation operator, as in (47)

it always takes scope over negation. Thus (47) unambiguously means ‘there exists a

person such that Tan does/did not meet that person’.

(47) Tân không gặp ai đó. ∃ > NOT

Tan NEG meet who DEM

‘Tan does/did not meet someone.’

160

Non-bare wh-indefinites also take scope over conditional operators, and yes no

question operators, as shown below.

(48) a. Nếu ai đó đến thì Thơ sẽ rất vui. ∃ > IF

if who DEM arrive then Tho FUT very happy

‘If someone arrives, Tho will be very happy.’

b. Tân có gặp ai đó không? ∃> Q

Tan Q meet who DEM Q

‘Did Tan meet someone?’

c. Nếu Tân không mời ai đó thì Thơ sẽ rất buồn. ∃>IF>NOT

if Tan NEG invite who DEM then Tho FUT very sad IF>∃>NOT

‘If Tan does not invite someone, then Tho will be very sad.’

The only wide scope reading in (48a) is ‘there is a person such that if that person

arrives then Tho will be very happy’. Similarly, the question in (48b) is about a person

such that the speaker wants to know whether Tan met that person. In (48c), the wh-

indefinite can take either the widest scope: ‘there is a person such that if Tan does not

invite that person, Tho will be very sad,’ or intermediate scope: ‘if there is a person such

that Tan does not invite that person, Tho will be very sad.’

Russian displays a similar property with respect to the contrast between bare and

non-bare wh-indefinites. Yanovich (2005) proposes that bare wh-indefinites in Russian

are Hamblin indefinites, namely, they introduce sets of individual alternatives. They need

161

licensing and take scope under their licensors. As shown in (49), without any licensor kto,

‘who’, is an interrogative phrase.

(49) Kto prišel?

who came

‘who came?’

When licensed by a yes-no question operator (50a) or a conditional operator

(50b), it is construed as a wh-indefinite and takes scope under its licensor. Thus, (50a)

does not mean, ‘there is a person such that the speaker wants to know whether that person

came.’ Similarly, (50b) does not mean, ‘there is a person such that if that person comes,

then the speaker wants to be notified.’

(50) a. Razve kto prišel?

is-it-true who came

‘Is it true that someone came?’

b. Esli kto pridet, zovi menja.

if who come call me

‘If anyone comes, call me.’

Non-bare wh-indefinites, by contrast, do not need licensing (51) and take scope

over negation (52).

162

(51) Kto-to vosel v komnatu.

who-TO came into the room

‘Someone came into the room.’

(52) Petja ne zametil kogo-to iz svoix odnoklassnic.

Petja NEG has-noticed who-TO of his girl-classmates

a. ‘There is a girl from his class that Petja did not notice.’

b. * ‘Petja did not notice any of the girls in his class.’

Non-bare wh-indefinites can also scope out of islands.

(53) Petja budet ščastliv esli kakaja-to devuška pridet.

Petja will be happy if which-TO girl will come.

a. ‘There is a property p such that Petja will be happy if a girl y such that p(y) =1

comes.’

b. ‘Petja will be happy if there is a girl who comes.’

The sentence in (53) allows two readings. In the first reading, the non-bare wh-indefinite

scopes out of the if-clause (53a). On the second reading, it stays inside the if-clause

(53b).

Yanovich argues that Russian wh-indefinites with to is like a certain NP, which is

always choice functional on Kratzer’s (1998) analysis. Yanovich thus analyzes to as a

choice function variable or to be exact a Generalized Choice Function as this element is

163

cross-categorical in that it can attach to AdjPs, AdvPs, and PPs. The denotation of the

element to is in (54).

(54) [[ to ]] = λp<τt,τ> . [f (p) ], where f is a Generalized Choice Function, and the

speaker knows f.

We saw that Vietnamese non-bare wh-indefinites can take scope out of the islands

in the data above, where if-islands are weak.27 Let us see if it can take scope out of a

strong island, namely CNP. Consider the example in (55).

(55) Tân không ăn [ cái bánh mà Thơ mua cho ai đó ].

Tan NEG eat CL cake REL Tho buy for who DEM

‘Tan did not eat the cake that Tho bought for someone.’

The sentence in (55) is two way ambiguous. In one reading, it means ‘Tan did not

eat the cake that Tho bought for a person,’ and the speaker may not know who that

person is. The other reading is ‘there is a person such that Tan did not eat the cake that

Tho bought for that person’. This reading indicates that the non-bare wh-indefinite can

scope out of a strong island.

27 As we saw in chapter 2, if- islands are more violable that other islands.

164

This scope behavior and the fact that non-bare wh-indefinites always take scope

over negation suggest that they are similar to the English lexical indefinite NPs of the

form a certain NP, and Russian non-bare wh-indefinites kto-to ‘who-TO’.

As shown in what follows, non-bare wh-indefinites can take wide, intermediate

and narrow scope.

(56) Tất cả sinh-viên phải đọc [tất cả các quyển sách [mà ai đó đã giới-thiệu]].

all student must read all PLU CL book REL who DEM ASP recommend

‘Every student must read every book that someone recommended it.’

a. ‘For every student x, there is a person y such that for every book z that y

recommended, x must read z.’

b. ‘There is a person y such that for every student x, every book z that that

person recommended, x must read z.’

c. ‘Every student x must read every book z that a person y recommended.’

From these observations I extend Yanovich’s (2005) analysis to account for the

non-bare wh-indefinite ai đó. That is, it can be analyzed compositionally with ai

providing an individual alternative set and the operator đó creating choice functions from

it. There are two well-known choice function approaches. Reinhart (1997) holds that a

choice function variable may be existentially closed at any CP level, as indicated by its

scopal behavior: it can take narrow, intermediate and wide scope. Kratzer (1998), on the

other hand, does not posit an existential quantifier to bind the choice function variable

165

and proposes that choice function variables are free variables, having their values fixed

by the context. Choice function variables on this analysis always take wide scope, but not

intermediate scope. While it is still controversial as to which is the best analysis, for the

task at hand, I think the scopal behavior of non-bare wh-indefinites in Vietnamese as

shown in (56) suggests that Reinhart’s (1997) approach would be best for Vietnamese.

The denotation of đó is given in (57). The semantic composition of a sentence

with a non-bare wh-indefinite as (58) is in (59).

(57) [[đó]] = λp<et,e>. [f (p) ], where f is a choice function

(58) [TP Tân [VP thích ai đó ] ].

Tan like who DEM

‘Tan likes someone.’

(59) [[ VP ]] w,g = {λxλw’. x likes f [[ai ]] in w’}

[[ TP ]] w,g = {λw’. Tan likes f [[ai ]] in w’}

= that Tan likes [f {a, b, c} in w’]

= that there is a way of choosing f from {a, b, c} such that Tan likes the

person f chooses.

Finally, Reinhart’s choice function approach can account for the following data.

166

(60) a. Tân không gặp ai đó. ∃ > NOT

Tan NEG meet who DEM

‘Tan did not meet someone.’

b. Thơ không biết Tân gặp ai đó. NOT > ∃

Tho NEG know Tan meet who DEM ∃ > NOT

‘Tho did not know that Tan met someone.’

The narrow scope of the negation operator in (60a) results from the fact that

choice function variable is existentially closed at the CP level above NEG. In (60b) the

choice function variable can be bound by an existential quantifier inserted at the

embedded CP level, resulting in scope below negation.

4.5 Conclusion

The neo-Hamblin semantics, adopted here, provides a satisfactory analysis for

wh-indefinites in existential contexts in Vietnamese. Vietnamese wh-indefinites require

c-commanding licensors and obey the locality constraint in K&S 2002, namely, the

alternative set must be captured by the lowest operator. However, a wh-indefinite can

move covertly over certain operators and be captured by a higher operator. This has the

effect of giving them intermediate scope.

Non-bare wh-indefinites are choice functions. They are not constrained by

locality and can take widest scope.

167

Finally, based on neo-Hamblin semantics, we are able to offer a uniform analysis

for wh-indefinites in existential as well universal contexts. As we will see in the next

chapter, neo-Hamblin semantics accounts nicely for wh-questions as well.

168

Chapter 5

A NEO-HAMBLIN ANALYSIS OF WH-QUESTIONS

5.1 Introduction

In the previous chapters, we have investigated wh-phrases in non-interrogative

contexts from a neo-Hamblin semantics perspective. This chapter is concerned with the

analysis of wh-questions on the basis of neo-Hamblin semantics. In fact, it is a revision of

an article on wh-questions in Vietnamese by Bruening and Tran (2006). The chapter is

organized as follows. Section 5.2 presents basic data that are taken from the article.

Section 5.3 reviews the analysis in the article. The reanalysis of wh-questions in neo-

Hamblin terms is given in section 5.4. Section 5.5 investigates wh-island effects in

Vietnamese and their implications for Kratzer and Shimoyama’s (2002) theory. Section

5.6 is concerned with wh-adjuncts, an issue not discussed in Bruening and Tran (2006),

followed by a report on an experiment on wh-adjuncts in section 5.7. The conclusion that

is given in section 5.8 is that neo-Hamblin semantics works perfectly for wh-questions

and provides a uniform account for wh-phrases in all contexts.

169

5.2 Data

Vietnamese is a wh-in-situ language: An argument wh-phrase must appear in the

position of a corresponding non-wh argument (1b); fronting as in English is not allowed

(1c).

(1) a. Tân vừa gặp May.

Tan just meet May.

‘Tan just met May.’

b. Tân vừa gặp ai (thế)?

Tan just meet who Q

‘Who did Tan just meet?’

c.*Ai Tân vừa gặp?

who Tan just meet

‘Who did Tan just meet?’

Like other wh-in-situ languages (Chinese, Japanese, Korean…) wh-questions in

Vietnamese may employ question particles. Unlike other languages that have been

described, the presence or absence of these elements affects grammaticality in island

sensitive configurations. Bruening and Tran’s (2006) account of Vietnamese wh-

questions analyzes a wh-phrase as being ambiguous between a weak indefinite and a

strong quantifier. This is a well-known view in the literature (see Pesetsky (1987), Cole

and Hermon (1994), Reinhart (1998), among others). As a quantifier a wh-phrase must

170

move covertly to its scope position (detected by island effects), and as a variable it is

bound by a binder (and is immune to island constraints). This analysis, while it does

account for the facts in Vietnamese wh-questions, requires that wh-phrases be lexically

ambiguous. In contrast, the Hamblin analysis that we adopt offers a unified account for

wh-phrases, where they always denote sets of individual alternatives in interrogative as

well as in non-interrogative contexts. Basically, on our theory, wh-questions result from

two mechanisms: either the propositional alternative sets introduced by the wh-phrases

are captured by a question operator (hence, no LF movement is needed) or the wh-phrase

moves to its scope position to signal the expansion to stop.

5.2.1 Argument Wh-phrases

Argument wh-phrases allow long distance readings despite the fact that there is no

overt wh-movement in a wh-question. With respect to selectional restrictions Vietnamese

behaves just like English in that it shows a correlation between scope interpretation and

selectional restrictions. For example, the verb biết ‘know’ can occur with either an

interrogative or a non-interrogative complement. The long-distance reading entails that

the verb selects a non-interrogative clause, and the local reading an interrogative clause:

(2) Tân biết ai đi New York.

Tan know who go New York

i. ‘Tan knows for which person x, x went to New York.’

ii. ‘For which person x, Tan knows that x went to New York?’

171

In the matrix question reading, (2i), the wh-phrase takes long distance scope.

The verb nói ‘say’ selects only non-interrogative clauses; embedded yes-no

questions with this verb are ill-formed, as shown in (3a). When a wh-phrase occurs in the

embedded clause, only the matrix wh-question reading is allowed.

(3) a.*Tân nói Thơ có gặp Lan hay không.

Tan say Tho Q meet Lan or Q

‘Tan said whether Tho met Lan.’

b. Tân nói Thơ đã gặp ai ?

Tan say Tho ASP meet who

i. ‘For which person x, Tan says Tho met x?’

ii. * ‘Tan says for which person x Tho met x’.

In contrast, the verb muốn biết ‘want to know’ selects only an interrogative

complement. Hence, no matrix wh-question reading is allowed, as shown in (4a): The

long distance scope is not available (see 4a ii). When the embedded clause is a

declarative, the sentence is bad as in (4b).

(4) a. Tân muốn biết Thơ đã gặp ai

Tan want know Tho ASP meet who

i. ‘Tan wants to know for which person x Tho met x.’

ii. * ‘For which person x, Tan wants to know Tho met x?’

172

b. *Tân muốn biết [ Thơ đã gặp Tiên.]

Tan want know Tho ASP meet Tien

‘Tan wants to know that Tan met Tien.’

Note that long-distance readings are available in (2) and (3): the wh-phrase can

take scope in the matrix clause even though it occurs in the embedded clause on the

surface.

Vietnamese offers another means for obtaining the matrix scope interpretation:

appending a particle to the sentence, as shown in (5).

(5) Tân biết ai đi New York thế?

Tan know who go New York PRT

i. ‘For which person x, Tan knew that x went to New York?’

ii. * ‘Tan knew for which person x, x went to New York.’

Note that this particle appears in the matrix clause only, as indicated by the lack of

ambiguity in (5). Because the particle is linearly sentence-final, it is not clear from the

surface word order whether the particle goes with the embedded clause or the matrix

clause. Along with the lack of ambiguity in (5), however, constituency tests place the

particle in the matrix clause. As shown by (6b), when the embedded clause is fronted, the

particle may not front with it:

173

(6) a. [ CP Ai đi New York ]i Tân biết ti.

who go New York Tan know

‘Tan knows for which person x, x went to New York.’

b.*[CP Ai đi New York thế ]i Tân biết ti.

who go New York PRT Tan know

‘Tan knows for which person x, x went to New York.’

Sentence (6a) has only the embedded reading, but matrix question readings are available

in sentences with fronted embedded clauses when they have both a sentence-final particle

and the particle mà as a focus marker of some sort, as shown in (7a). But sentence (7b) is

still ungrammatical even with mà.

(7) a. [CP Ai đi New York]i mà Tân biết ti. thế?

who go New York PRT Tan know PRT

‘For which person x, Tan knew x went to New York?’

b.*[CP Ai đi New York thế ]i mà Tân biết ti.

who go New York PRT PRT Tan know

‘Tan knew for which person x, x went to New York.’

The contrast shown above confirms that the particle thế only appears in matrix clauses.

174

5.2.2 Island Effects

The following data indicate that questions with argument wh-phrases in

Vietnamese obey island constraints. When a wh-phrase appears inside an island and the

question lacks a particle, it is ungrammatical, but with a particle the sentence is

grammatical.28 Consider the case of a complex NP, for instance:

(8) a. *Tân sẽ chụp hình [NPcon hổ [CP đã dọa ai ]]?

Tan FUT catch picture CL tiger ASP scare who

‘Tan will take a photo of the tiger that scared who?’

b. Tân vừa chụp hình [NPcon hổ [CP đã dọa ai ]] thế ?

Tan just catch picture CL tiger ASP scare who PRT

‘Tan took a photo of the tiger that scared who?’

The ungrammaticality of (8a) indicates that LF movement takes place.29 There must be

no corresponding movement in (8b) since the sentence is grammatical.

The same holds for the Sentential Subject Constraint:

28 The judgments on this section are from four native speakers of Vietnamese: One Northern dialect speaker and three Central dialect speakers. I am a speaker of the Central dialect. 29 There are other variants of the covert movement approach. One claims that the movement takes place in the overt syntax, but the tail of the chain is pronounced rather than the head (Bobaljik 1995, Groat &O’Neil 1996, Pesetsky 1998). Another alternative is that the wh-phrase itself never moves, but an operator associated with it does (Watanabe 1992, Aoun and Li 1993, Hagstrom 1998).

175

(9) a.* [CP Ai sẽ bỏ đi ] làm mọi người bối-rối?

who FUT leave make everyone embarrass

‘That who will leave will make everyone embarrassed?’

b. [CP Ai vừa bỏ đi ] làm mọi người bối-rối thế?

who ASP leave make everyone embarrass PRT

‘That who left made everyone embarrassed?’

We see the same thing with the Adjunct Island Constraint: 30

(10) a.* Tân sẽ thua cuộc [CP vì ai làm hư xe của anh ta]?

Tan FUT lose event because who make damage vehicle belong he

‘Tan will lose the race because who will damage his car?’

b. Tân thua cuộc [CP vì ai làm hư xe của anh ta] thế ?

Tan lose event because who make damage vehicle belong he PRT

‘Tan lost the race because who damaged his car?’

The (a) sentences are good only if they are echo questions. They are ungrammatical as

genuine requests for information. However, the (b) sentences are grammatical with the

question particle.

In summary, wh-questions with argument wh-phrases allow long distance

readings without overt movement and obey Subjacency when they occur without a 30 As will be shown shortly, questions without particles are irrealis, while questions with particles are realis. Hence, the contrasting pairs in this section differ in Mood as well.

176

particle. Can we take this Subjacency effect as a diagnostic of LF movement? It is clear

that in languages like English overt wh-movement observes Subjacency, but it is

controversial whether or not LF movement has to obey it. There is no consensus in the

literature on this issue. Huang (1982) proposes that LF movement in Chinese, and in

general, does not have to obey Subjacency. By contrast, Pesetsky (1987, 2000) holds that

LF movement does obey Subjacency and other restrictions that govern overt movement.

This position has been argued for by numerous others as well, such as Cole and Hermon

(1994) and Richards (2001). The data above indicate that LF movement in Vietnamese

does obey all restrictions on movement, including Subjacency. It thus adds to the

evidence accumulated by the authors above indicating that LF movement is no different

from overt movement in this respect. The alternative is, as noted in footnote 2, that overt

but invisible movement of a null operator associated with the wh-phrase is taking place

rather than LF movement of the wh-phrase. However, the only reason we can see to posit

this is to maintain that LF movement does not have to obey Subjacency. This position

would complicate, rather than simplify the theory. First, we have to stipulate a difference

between overt and covert movement that is conceptually unmotivated. Second, we would

have to posit invisible elements whose interpretation would be unclear, at best. Finally, a

null operator undergoing overt movement would be equivalent to the Q-particle

(basically, an existential quantifier), but then we would wonder why it has to move when

it can be base-generated in CP when there is a particle. The theory that says that covert

movement also obeys Subjacency has the virtue of simplicity and unity.

Next, let us look at the data on intervention effects.

177

5.2.3 Intervention Effects

Beck (1996), with data from German, claims that the LF movement of quantifiers,

including wh-words, is blocked by other quantificational expressions. A formulation of

this restriction is given in (11), with illustrative examples in (12).

(11) * […X i …[ Q or Neg …[ … t i LF …]]]

(12) a. ?? Wer hat niemanden wo angetroffen?

Who has nobody where met

‘Who didn’t meet anybody where?’

b. Wer hat wo niemandem angetroffen?

Who has where nobody met

‘Who didn’t meet anybody where?’

In (12a), when the in-situ wh-phrase wo is preceded and c-commanded by the negative

quantifier niemanden, the sentence is ungrammatical. By contrast, when wo occurs

structurally above the negative quantifier niemanden, as in (12b), the sentence is

grammatical. Beck (1996) offers a syntactic analysis, according to which the

ungrammaticality in (12a) is due to a constraint that does not allow the wh-phrase in

(12a) to move covertly to its scope position across the negative operator. The overt

movement of this wh-phrase, by contrast, is grammatical as shown in (12b).31

31 Beck (2006) reanalyzes the ungrammaticality of sentences such as (12a) on the basis of focus semantics. Basically, in the configuration [Q… [Op [φ …XP F … wh… ]]], where Op is a focus sensitive operator and Q is a question operator, when focus is evaluated at

178

Vietnamese shows a similar blocking effect in the following examples. As shown

in chapter 3, a wh-phrase followed by cũng obtains a universal reading, as in (13a). When

another wh-phrase occurs in object position, the sentence is ungrammatical as a question

in (13b). However, the sentence becomes grammatical if a wh-particle is added, as in

(13c).32

the level of phrase φ, focus semantics enters into ordinary semantics. Since focus operators use both ordinary values and focus values, focused elements, being of ordinary as well as focused values, can satisfy this requirement while wh-phrases, inherently focused, cannot. A wh-phrase in such a configuration will lead to ungrammaticality because the whole phrase is undefined. Can this analysis be extended to Vietnamese? The answer is no. I will return to this issue in section 5.4, where I conclude that the account of Beck 1996 is better than that of Beck 2006. 32 As mentioned in chapter 2, the universal quantifier illustrated in (13) is only one of the universal quantifiers that Vietnamese employs. A universally quantified expression can also be made up of a universal quantifier, such as mọi, tất cả, meaning ‘all’, and an NP. For example:

(i) tất cả các học-sinh all PL student

‘all the students’ (ii) mọi học-sinh

all student ‘ all students’ Blocking effects do not show up with universal quantifers of this type. For instance:

(iii) Tất cả các học- sinh đều làm gì? all PL student DEU do what ‘What do all the students do?’

As suggested by the English translation in (iii) this universal quantifier tất cả các học sinh conveys a referential group interpretation, referring to the set of students salient to both the speaker and hearer as a whole. There is no pair list reading in (iii). If this NP is actually a group-denoting noun and not a quantifier, it would not be expected to give rise to an intervention effect.

179

(13) a. Ai cũng thích bóng đá.

who CUNG like football

‘Everyone likes football.’

b.* Ai cũng thích cái gì?

who CUNG like what

‘What does everyone like?’

c. Ai cũng thích cái gì thế?

who CUNG like what PRT

‘What does everyone like?’

The same pattern can be found in cases where a wh-phrase is preceded by a

negative quantifier. In Vietnamese a wh-phrase obtains an NPI reading when it is in the

scope of negation, as in (14a). (14b) is ungrammatical as a wh-question without a

particle, as opposed to (14c) with one. 33

33 There is an alternative negative particle, không. Sentential negation formed by this particle does not cause intervention, while that formed by chẳng does, as shown below.

(i) Tân không mời ai. Tan NEG invite who a. ‘For which person x, Tan did not invite x?’ b. ‘There is not any person x such that Tan invited x.’

(ii) Tân chẳng mời ai. Tan NEG invite who a. * ‘For which person x, Tan did not invite x?’ b. ‘There is not any person x such that Tan invited x.’

(iii) Tân không mời ai cả. Tan NEG invite who all a. * ‘For which person x, Tan did not invite x?’ b. ‘There is not any person x such that Tan invited x.’

180

(14) a. Chẳng ai mời Tân.

NEG who invite Tan

‘No one invited Tan.’

b. Chẳng ai mời ai

NEG who invite who

i. ‘No one invited anyone.’

ii. * ‘Who did no one invite?’

c. Chẳng ai mời ai thế?

NEG who invite who PRT

‘Who did no one invite?’

These blocking effects are detected in languages like Japanese and Korean as

well. However, Tomioka (2007) argues that there are two types of blocking effects: the

German type illustrated above, and the Japanese/Korean type. The latter, according to

Tomioka, are not due to LF movement. For instance, blocking effects become much

In chapter 4 we observed the contrast in the possibility of scopal ambiguity: A wh-indefinite can take scope over the negative operator không, but not over the negative operator chẳng. We accounted for this contrast by proposing that the former, but not the latter allows the wh-indefinite to move over it at LF and associate with the higher operator. Hence, it obtains wider scope than the negation. We propose that a similar mechanism happens in wh-questions. That is, the negative operator không optionally allows covert wh-movement across it. When it is accompanied by a reinforcer of negation cả ‘all’, the blocking is in effect. Hence no question reading is available as in (iii).

181

weaker in embedded contexts in Japanese and Korean. If these effects were due to LF

movement, there would be no root-embedded contrast as appears in (15).

(15) a. ?*Daremo-ga nani-o yom-da-no (Jp)

everyone-nom what-acc read-past-Q

‘What did everyone read?

b. ?*Amuto mues-ul ilkci-anh-ass-ni (Kr)

anyone what-acc read-neg-past-Q

‘What did no one read?’

c. ?(?) Kimi-wa [CP daremo nani-o yom-ana-katta-to ] omotteiru (Jp)

you-top anyone what-acc read-neg-past-dec-comp think-Q

‘What do you think that no one read?’

d. ?(?) Ne-nun [CP amuto mues-ul ilkci-anh-ass-ta-ko ] sayngkakha-ni (Kr)

you-top anyone what-acc read-neg-past-dec-comp think-Q

‘What do you think that no one read?’

(Tomioka 2007, pp. 1571-1573)

The sentences in (15a, b) are ungrammatical, apparently for the same reason that similar

German examples are ungrammatical. But when the exact same sentences are embedded

as in (15c, d), their grammaticality improves substantially. Tomioka’s explanation for the

contrast in (15) is basically that in Korean and Japanese every sentence must have a topic.

The default topic is the subject. Quantifiers such as amuto and daremo are not suitable

182

topics, since topics, old or familiar information, are typically referential expressions.

Hence, (15a, b) are ungrammatical not because of LF blocking, but because they lack

topics or have unsuitable topics. In contrast, the matrix subjects in (15c, d) are pronouns,

which are perfect as topics. That is why there is a root-embedded contrast in Japanese and

Korean.

So, to argue that the blocking effect in Vietnamese is due to LF movement, we

need to show that they are not of the Japanese/Korean type. They are not: Vietnamese

behaves like German in not showing a root-embedded contrast:

(16) a. Tân biết chẳng ai gặp ai.

Tan know NEG who meet who

i. ‘Tan knows no one met anyone.’

ii. *‘Who does Tan know no one met?’

b. Tân biết chẳng ai gặp ai thế?

Tan know NEG who meet who PRT

‘Who does Tan know no one met?’

Example (16a) is grammatical with the interpretation where the two wh-phrases

are NPIs, as indicated by the translation. However, it is ungrammatical as a question,

even though the matrix subject is a proper name, which would be a perfect topic. In

contrast, (16b) with a particle is grammatical. As in island effects, the same analysis can

be employed here: LF movement must take place when there is no particle, but LF

183

movement cannot cross a negative quantifier. When there is a particle, there is no

movement, so it is immune to any constraints on movement such as Beck’s filter, as

illustrated in (17):

(17) LF

* [ CP __ NPI or QP … wh … ] √ [ CP OP i NPI or QP … wh i … ]

To sum up the results of this section, the data above provide strong support for the

hypothesis that both covert movement and non-movement are involved in Vietnamese

wh-questions. When the question particle is absent, wh-words may not appear in islands,

and they may not be c-commanded by other quantificational elements. Islands are

standardly taken to diagnose movement, and if Beck (1996) is correct, blocking effects

specifically diagnose LF movement. Hence, we conclude that wh-words must move to

the Spec, CP at LF when there is no question particle. The total absence of these effects

when a question particle appears indicates that no movement takes place when a particle

is present.

5.2.4 D-linking and Question Particles

The alternative patterns of wh-questions with respect to the presence and absence

of wh-particles in dealing with island constraints and LF blocking resemble the patterns

shown by D-linked and non-D-linked wh-phrases in English. That is, when a wh-phrase

184

is D-linked, which is when the answers are drawn from a presupposed set (Pesetsky

1987), it is immune to movement constraints. For example, consider (18).

(18) a. Who bought what?

b. *What did who buy?

c. Which book did which person buy?

The wh-phrase in-situ in (18b) is non-D-linked, and is subject to the Superiority Effect

(Chomsky, 1973). In contrast, the in-situ wh-phrase in (18c) is D-linked, and there is no

Superiority Effect. Is there such a correlation in Vietnamese? Namely, does this mean a

wh-phrase is D-linked when it co-occurs with a wh-particle?

In fact, there is no correlation between particles and D-linking in Vietnamese. A

wh-word with a particle can be non-D-linked, and a wh-word without a particle may be

D-linked. Consider the following situations.

Suppose there are three possible prizes for a contestant in a game show, and the

show’s host wonders which prize the winner will choose. In this situation the possible

answers come from an established set, namely the three prizes that are plainly visible to

the audience and the contestants. That is, this is a D-linked context. Nevertheless, the

question the host will ask is (19a), without a particle, not (19b) with a particle. (19b) is

only felicitous in a situation where the contestant has already made his choice and the

enquirer (for example, an audience member) wonders which he has chosen.

185

(19) a. Anh chọn cái nào?

you choose Cl which

‘Which do you choose?’

b. ?#Anh chọn cái nào thế?

you choose Cl which PRT

‘Which did you choose?’

If the presence of the particle were related to D-linking, then it is expected that (19b)

would be the preferred form in this context and that (19a) would be infelicitous. But the

fact is just the opposite.

The situation for (19) shows that D-linking does not require a particle. The

converse also holds: particles can be used in non-D-linked contexts, meaning that

particles, when they appear, do not induce D-linking.

To see this, consider a scenario where a pedestrian sees a policeman in the middle

of a noisy crowd. This person, thinking that something may have happened, comes over

and asks the policeman:34

(20) a. Chuyện gì xảy ra thế ?

story what happen PRT

‘What happened?’

34 There is a vocative particle as a polite form of address that would normally be used in (20), but we leave it out to avoid confusion with the question particle.

186

b. ?# Chuyện gì xảy ra?

story what happen

‘What happened?’

It is clear that there is no presupposed set of possible answers in this context (anything in

the world could have happened). Yet, the use of the particle is required. Its absence

makes the sentence unacceptable.

At this point, it seems safe to conclude that a particle is not required in D-linked

contexts and a particle does not require a D-linked context to be felicitous. Therefore,

question particles are completely independent of D-linking in Vietnamese.

As we observed in the chapter on Vietnamese syntax, the occurrence of sentential

final particles is to indicate modality in the language. Consideration of a broad range of

examples, including those above, suggests that the wh-particle in Vietnamese is

associated with realis mood. When it is used the speaker presupposes the existence of the

entity described by the wh-phrase and the realization of the event described by the

predicate, and believes that the hearer shares this knowledge. For example, in the case of

the passer-by and the policeman, in order to ask (20a) the speaker must have a strong

presupposition that something has happened and that the policeman knows what it is.

(20b) would be used in other contexts, for instance, in a chemistry class where a teacher

demonstrates an experiment and asks the students, before performing the act of mixing

two chemicals, what the outcome of this act will be. In this situation the teacher does not

187

presuppose that the students know what chemical reaction follows the act, nor has the act

yet occurred. The use of the particle would not be felicitous in this case.

Imagine another context where A saw Tan talking to a group of students, and the

day after A asks Tan who they were. Sentence (21a) is felicitous. (21b) is unacceptable in

this context. Suppose A wants to know about Tan’s plans, that is, which students Tan is

going to talk to, then (21b) is felicitous.

(21) a. Anh nói chuyện với sinh-viên nào thế?

you talk story with student which PRT

For which student x, you talked to x?

b. Anh nói chuyện với sinh-viên nào?

you talk story with student which

For which student x, you talk to x?

The entities described by the wh-phrase in (21a) are known to exist by both the speaker

and the hearer, and the predicate ‘talk’ has been realized in (21a). In contrast, the entities

described by the wh-phrase in (21b) may or may not be assumed to exist by the speaker.

In addition, the event has not been realized.

So far the contrast between questions with particles and questions without

particles can be described as a contrast between future and non-future contexts.

Consideration of further examples shows that this is not exactly the case, however, and

that the actual contrast is realis-irrealis. Consider the following, for example.

188

Imagine an interview where subjects are asked about the billionaire Bill Gates.

The interviewer wants to know who would have suffered most if Bill Gates had passed

away two years ago, and who will suffer most if he passes away next year. The questions

in Vietnamese would both lack particles:

(22) a. Ai vô-cùng đau-khổ nếu Bill Gates qua đời cách đây hai năm (*thế )?

who endless suffer if Bill Gates pass life ago two year PRT

‘Who would have suffered most if Bill Gates had passed away two years ago?’

b. Ai sẽ vô-cùng đau-khổ nếu sang năm Bill Gates qua đời (* thế )?

who FUT endless suffer if come year Bill Gates pass life PRT

‘Who will suffer most if the Bill Gates passes away next year?’

In (22a) the hypothetical event is in the past, and in (22b) it is in the future. Yet, the use

of the particle is ungrammatical in both questions. This confirms that there is no

correlation between the use of the particle and a future-non-future contrast. Rather, the

particle encodes realis mood: the events characterized by the predicates in the questions

have already happened. The counterfactual events in (22b) have not happened, whether

they are located in the past or in the future.

189

5.2.5 D-linked Wh-phrases.

Although there is no correlation between particles and D-linking, there is a role

for D-linking in Vietnamese. D-linked wh-phrases turn out to be grammatical in islands

even without wh-particles.

Suppose at a book fair, any reader can meet the author of any book that is on

display. A asks B about Tan. Sentence (23a) with a wh-phrase embedded in an island is

ungrammatical. Sentence (23b), with a ‘which’ phrase in the same island, is substantially

improved. Sentence (23c) with a particle is perfect, as expected. What is unexpected is

that (23b) is not ungrammatical. Note, however, that B’s answer to (23b) must be either

the whole sentence as in (23a) or the whole island as in (23b). (23c) could be a felicitous

answer only to (23c) with a particle, not to (23b).

(23)

A. a. *Tân định gặp [NP người [CP đã viết gì ]?

Tan intend meet human PST write what

‘Tan intends to meet the person who wrote what?’

b. Tân định gặp [NP người [CP đã viết quyển sách nào ]]?

Tan intend meet human PST write Cl book which

‘Tan intends to meet the person who wrote which book?’

c. Tân định gặp [NP người [CP đã viết quyển sách nào ]] thế?

Tan intend meet human PST write Cl book which PRT

‘Tan intends to meet the person who wrote which book?’

190

(24)

B. a. Anh ấy định gặp [NP người [CP đã viết quyển sách bày trên quầy Z]]

he intend meet person PST write CL book display on counter Z

‘He intends to meet the person who wrote the book displayed on counter Z.’

b. [NP Người [CP đã viết quyển sách bày trên quầy Z.]]

person PST write CL book display on counter Z

‘The person who wrote the book displayed on counter Z.’

b. Quyển sách bày trên quầy Z.

CL book display on counter Z

‘The book displayed on counter Z.’

In accounting for the absence of island constraints in Japanese, Nishigauchi

(1986) proposes that wh-movement moves the entire island that contains a wh-phrase to

the operator position of the clause where it takes scope (as well as the wh-phrase, which

itself moves within the island). Can this analysis be extended to Vietnamese? In fact,

Bruening and Tran (2006) assume a pied-piping analysis along the lines proposed by

Nishigauchi to account for Vietnamese D-linked questions. We will return to this issue in

the section on the previous analysis.

5.2.6 A Matrix- Embedded Contrast

We have observed that when there is a particle, there is no movement. Given that

a particle occurs only in matrix clause, we wonder whether movement must apply in

191

embedded clause, hence island effects are detected in this context. It is interesting that

that there is a contrast between matrix and embedded question with respect to island

violation. The data below show that island constraints are always absent from embedded

clauses and that embedded wh-questions never have an overt particle.

(25) a. [Tân sẽ mua [NP ngôi nhà [CP mà ai đã xây dựng ]]*(thế )]?

Tan FUT buy CL house REL who PAST build PART

‘Who will Tan buy the house that built?’

b. [Lan muốn biết [CP Tân sẽ mua ngôi nhà mà ai đã xây dựng (* thế )]]

Lan want know Tan FUT buy CL house REL who PAST build PART

‘Lan wants to know who Tan will buy the house that built.’

The question in (25a) requires the particle, as shown above, since the wh-phrase is inside

an island. However, when (25a) is embedded, as in (25b), the occurrence of the particle

makes the sentence ungrammatical. This is because ‘want to know’ requires a question

complement; but the particle, which occurs in the matrix clause, requires a matrix

question reading.

What is surprising is that (25b) is grammatical at all. As an embedded question, it

cannot have a particle. Since particles are necessary to circumvent the island constraints,

we would expect that wh-phrases would always be ungrammatical in islands in embedded

questions. The exact opposite is the case, however: They are always grammatical.

192

5.3 Previous Analysis

As far as we are aware, the article by Bruening and Tran (2006)

(henceforth B&T 2006) is the only analysis of wh-questions in Vietnamese within a

formal framework. According to this analysis, Vietnamese employs two basic approaches

to the issue of wh-in-situ: covert movement and unselective binding. Covert movement

takes place when the wh-question lacks a particle. The motivation for this movement is

scope interpretation. However, unlike Chinese on Huang’s (1982) analysis where covert

movement is insensitive to island constraints, covert movement in Vietnamese is subject

to both island effects and intervention effects, which provides additional evidence in

support of a position advocated by Pesetsky (1987, 2000), Richards (2001), Cole and

Hermon (1994), and Bruening (2001), who hold that covert movement obeys the same

constraints as overt movement.

What happens when there is no wh-particle? In the absence of a wh-particle, the

quesion is interpreted via unselective binding. Following Baker (1970), Pesetsky (1987),

Cole and Hermon (1994), Reinhart (1998), among others, we proposed that wh-phrases in

wh-questions with particles are not quantifiers, but variables unselectively bound by a c-

commanding Q-operator. Following a suggestion by Tomioka (p.c.), we assumed that this

Q-operator is licensed by a selecting head. This head can be a higher question selecting

verb like ‘wonder, ask’, or the particle thế. This explains the difference between matrix

and embedded questions in (25). The binder licensed by the particle in (25a) binds the

wh-phrase in the island, and the sentence is grammatical. The binder in (25b) is licensed

by the matrix verb, binding the wh-phrase in the island of the embedded clause. Inserting

193

the particle will result in ungrammaticality because the higher operator will have nothing

to bind, a violation of the ban on vacuous quantification.

In other words, the question particle is not itself the binder; the binder is always

null and needs to be licensed. The role of the overt question particle is to license it.

Hence, islands are violable when there is an overt question particle. Question embedding

verbs also licenses the binder, so embedded questions are insensitive to islands. The

analysis above can be summarized in (26):

(26) a. Wh-phrase as a quantifier, of type <<ett><tt>> :

[[ ai ]] = λP<e,tt>. λp. ∃x [ x is a human & p= P(x)]

→ Movement

b. Wh-phrase as a variable, of type <et>:

[[ ai ]] = λx. x is a human

→ Null operator licensing

Licensing is illustrated in (27).

(27) a. Embedded

V/VP

Vlicensor CP

∅i IP

wh i

194

b. Matrix

CP

CP PART licensor

∅i IP

wh i

With respect to D-linked wh-phrases, as previously mentioned, Nishigauchi

(1986) proposes that wh-movement moves the entire island that contains a wh-phrase to

the operator position of the clause where it takes scope (as well as the wh-phrase, which

itself moves within the island). The evidence that Nishigauchi uses to argue for the pied-

piping of islands is possible answerhood. Let us assume that the following principle

proposed by Pesetsky (1987) universally holds.

(28) Felicity Principle

A felicitous answer to a wh-question consists of a phrase structurally identical to

the wh-phrase whose index is immediately dominated by the Comp of the question

at LF.

Consider again the answers in (24) to the D-linked wh-question in (23). (24a, b)

are felicitous since they recapitulate the entire island, as would be required by the Felicity

Principle above if the whole island moves at LF to Spec-CP. An answer that corresponds

195

just to the wh-phrase embedded in the island as in (24c) is unacceptable, as would be

predicted by the Felicity Principle if the whole island has moved at LF.

This idea also explains the contrast with a single-word answer in the case of

islands with a particle, shown in (29).

(29) A. Tân thích nói chuyện với [NPông bác sĩ [CP[IP đã chữa-bệnh cho ai]] ] thế?

Tan like talk story with CL doctor PST cure give who PART

‘Who does Tan like to talk to the doctor who cures?’

B. Cho Mary.

give ‘Mary’

‘Mary’

Because the particle induces no movement; hence no pied-piping takes place, and a

single-word answer satisfies the felicity principle.

In summary, evidence obtained through the answers to D-linked questions that

satisfy the Felicity Principle and the absence of island effects in D-linked questions

strongly suggests that LF pied-piping is employed in D-linking contexts.

5.4 Neo-Hamblin Analysis

The previous analysis, as far as we are aware, can satisfactorily account for the

given data. However, it requires the lexical ambiguity hypothesis of the wh-phrase: Wh-

phrases vary being a variable of type <et> and a quantifier of type <ett,tt>. Given that

196

wh-indefinites as well as wh-phrases are identical in form, it is intuitively more plausible

to hold that they have the same denotation. Neo-Hamblin semantics offers the solution.

We have known that wh-indefinites or Hamblin pronouns denote sets of individual

alternatives, consistently of type <e>. Their quantificational interpretations are derived

from the operators that take them or the alternative constituents in which they are

embedded as arguments. If the operators are universal quantifiers, they obtain universal

interpretations and if the operators are existential interpretations, they are indefinites. On

this analysis, the terms ‘wh-phrase’ and ‘wh-indefinite’ are just labels; their nature as sets

of alternative individuals remains unchanged..

Before we move on with our neo-Hamblin analysis of wh-questions, let us look at

Beck’s (2006) reanalysis of intervention effects first. Why? Because in this new version,

Beck employs a neo-Hamblin analysis of wh-phrases, and proposes that the blocking

effect is not due to LF movement of the wh-phrase, but to the fact that it is in a

configuration that requires both ordinary semantic values and focus semantic values

while it possesses only the latter. The blocking effect is generally summarized as follows:

‘A wh-phrase may never have a focus-sensitive operator other than the Q operator as its

closest c-commanding potential binder’ (Beck 2006, p.46). Recall that in our analysis of

universals, cũng is a focus sensitive particle and when a wh-indefinite appears on its left,

a universal interpretation arises. We saw in section 5.2.4 that the wh-phrase in (13b),

repeated below as (30a), is located exactly in the configuration that is not allowed on

Beck’s (2006) analysis as shown in (30b).

197

(30) a.* Ai cũng thích cái gì?

who CUNG like CL what

‘What does everyone like?’

b. [Q [Op [φ … wh… ]]]

The focus sensitive operator cũng is closer to the wh-phrase than the

question operator Q; hence (30a) is ungrammatical as Beck’s (2006) analysis predicts.

Yet, Beck’s (2006) theory incorrectly rules out sentences like (31a) that share the same

configuration as (30a), contrary to fact.

(31) a. Tân cũng thích cái gì?

Tan CUNG like CL what

‘What does Tan also like?’

b. [Q [Op [φ … wh… ]]]

In addition, Beck’s (2006) analysis cannot explain why (13c), repeated below as

(32a) is grammatical even though it has the same configuration as (30). The minimal

difference between them is the presence of the wh-particle in (32a) and its absence in

(30a).

198

(32) a. Ai cũng thích cái gì thế?

who CUNG like CL what PRT

‘What does everyone like?’

b. [Q [Op [φ … wh… ]]]

In summary, Beck’s (2006) can account for the ungrammaticality of a wh-

question without a particle in a blocking configuration, but it cannot explain why the

presence of a particle brings about grammaticality given that a wh-phrase should be

consistently of focus value in the two configurations. We therefore hold that Beck’s

(1996) LF movement analysis can account for the contrast in grammaticality in

intervention contexts, but her reanalysis in the 2006 paper cannot.

Now, to see how neo-Hamblin analysis works in Vietnamese. Let us adopt the

following technical tools (the type-driven composition principles) cast by Kratzer and

Shimoyama (2002).35

Hamblin Functional Application

If α is a branching node with daughters β and γ, and [[β]]w,g ⊆ Dσ and [[γ]]w,g⊆ D<στ>,

then [[α]]w,g = {a∈ Dτ :∃b ∃c [ b∈[[β]]w,g & c∈[[γ]]w,g & a = c (b)]}.

35 The Japanese wh-particle ka, which is a Q-operator on K&S’s (2002) analysis, is considered as either semantically trivial or taking a set and returning a singleton set whose sole member is the question denotation in Groenendijk & Stokhof’s theory (see Shimoyama 2006).

199

Predicate Abstraction

If α is a branching node whose daughters are an index i and β, where [[β]]w,g ⊆ Dσ then

[[α]]w,g = {f: f and f ∈ D<eσ> &∀a [ f (a)∈[[β]]w,g [a/i] ]}.

Pronouns and traces

For any index i, [[ i]] w,g = {g (i)}.

Question operator

[[ Qα]]w,g = [[ α]]w,g where [[ α ]]w,g ⊆ D <s,t> (Hamblin 1973)

The semantic derivation of a wh-question on neo-Hamblin semantics is as

follows: The wh-phrase denoting a set of individual alternatives combines in a pointwise

manner with other elements in the structure to construct sets of alternatives until the

constituents so formed are bound by an operator that takes sets of alternatives as its

arguments and returns a set. As we have observed, this mechanism works nicely in

universal and existential contexts. For example, the universal operator introduced by

cũng quantifies over a set of propositions and returns a singleton set. The role of the Q-

operator in Vietnamese, we assume, is not to yield a singleton set from the sets of

alternative propositions, but to signal the pointwise functional application to stop. Recall

that we maintain B&T 2006 analysis in assuming that both non-movement and movement

apply in wh-questions. How does this work in neo-Hamblin semantics? Consider the

sentences in (33a) and (34a).

200

(33) a. Tân gặp ai ?

Tan meet who

‘Who did Tan meet?’

b. [ CP ai 1 [ 1 C [TP Tân gặp t 1] ]]

LF movement

(34) a. Tân gặp ai thế?

Tan meet who PART

‘Who did Tan meet?’

b. [CP Q [TP Tân gặp ai] ]

The two sentences in (33a) and (34a) are minimal different: (33a) lacks a wh-

particle, while (34a) is accompanied by a wh-particle. To account for the asymmetry with

respect to island sensitivity, B&T (2006) propose that when there is no particle

movement is employed, and when there is a particle, there is no movement. However, on

our revised analysis, movement is not triggered by scope interpretation, as it was in B&T

2006. Instead, the wh-phrase moves covertly to the maximal projection of the clause to

signal the propositional alternative set to stop expansion at the site where it moves to, as

illustrated in (35).36

36 The reason why wh-phrases move to Spec, CP could be (i) for feature checking in the MP framework, (ii) for scope interpretation, or (iii) to signal the pointwise function application to stop at the site where it moves to. Given the interpretation mechanism of neo-Hamblin semantics, (ii) is not necessary. As a language with poor morphology, Vietnamese is not likely to provide substantive support in favor of the feature checking theory. (iii) is the most plausible as far as I can tell.

201

(35) [CP whi [TP ti ]]

The propositional alternative set, formed via pointwise FA, in the TP domain, is

stopped by the wh-phrase that moves out of TP to Spec, CP.

As illustrated in the diagram in (35), the wh-phrase ‘who’ moves from its base-

generated object position in TP domain to the CP domain. Following Heim and Kratzer

(1998), we assume that this movement operation introduces an index (interpreted as a

lambda abstractor) right below the position where the moved element lands. Thus the

denotation of this index and C’ is: [[ 1 C’]] w,g {λx. λw’. met (Tan) (g (1)) (w’)}.

Combining this node with the denotation of the individual alternatives introduced by ai,

‘who’, we obtain [[CP ]] w,g = {p: human (x) & met (Tan) (x) (w’)}, the denotation of a

question.

(36) CP = [[ CP ]] w,g = {p: human (x) & met (Tan) (x) (w’)}

ai i 1 C’ C TP= [[TP]] w,g ={λw’.met (Tan) (g (1)) (w’)}

Tân gặp t i

The computation of (33a) is in (37).

(37) a. [[ai ]]w,g = { x: human (x)}

202

b. [[ gặp]] w,g = { λx. λy. λw’. met (y) (x) (w’)}

c. [[Tân]] w,g = Tân

d. [[ TP ]] w,g = { λw’. met (Tan) (g (1)) (w’)} PA

e. [[ 1 C’]] w,g = { λx. λw’. met (Tan) (g (1)) (w’)} HFA

f. [[ CP ]] w,g = {p: human (x) & met (Tan) (x) (w’)}

If May, Nhi, Gi, and Lem are all the people in the context, then (33a) denotes the set of

propositions: {that Tan met May, that Tan met Nhi, that Tan met Gi, that Tan met Lem}.

This is the denotation of a question in Hamblin’s (1973) theory.

What happens in (34a)? The LF of (34a) is (34b). As shown by the diagram in

(38), the covert Q-operator is licensed, in this case, by the wh-particle. There is no

movement needed because the propositional alternative set introduced by the TP is

captured by the null question operator residing in the CP domain.

(38) XP

CP thế

Q TP

Tân gặp ai

Wh-particle licensing the covert question operator Q

203

The semantic derivation of (34a) is as in (39).

(39) a. [[ai ]] w,g = { x: human (x)}

b. [[ gặp]] w,g = { λx. λy. λw’. met (y) (x) (w’)}

c. [[TP]] w,g = {λw’. Tân met [[ai ]] w,g in w’} HFA d. [[ TP ]] w,g = {p: human (x) & met (Tan) (x) (w’)} e. [[ Q TP ]] w,g = {that Tan met May, Tan met Gi, Tan met Lem…}

The denotation of the verb ‘meet’ is a singleton set whose sole member is its ordinary

denotation as in (39b). Hamblin Functional Application allows us to apply the function

λx λyλw. met (y) (x) (w) to each of members of the set of human individuals in a

pointwise manner and yield (39c), a set of propositions in (39d). The question operator

takes this set and returns the same set. If the individuals available are May, Gi, and Lem,

then, this set will be as in (39e).

How does neo-Hamblin analysis account for the data in B&T 2006 paper? We

have observed that a matrix question with a particle can evade island effects and

intervention effects while one without a particle cannot. We also witnessed that a wh-

phrase in embedded questions can take scope out of an island. All of these observations

can be explained as follows. (The bold letter Q with the index i refers to the scope of the

wh-phrase).

204

(40) a. Matrix [ CP Q i [TP [Island wh-phrase i ]] PART]

* [ CP Q i [TP [Island wh-phrase i ]] ]

b. Embedded [ CP [VP [CP Q i [Island wh-phrase i ]]]

Neo-Hamblin semantics nicely accounts for these facts. As mentioned, we

maintain B&T 2006 analysis in assuming that Vietnamese employs both movement and

non-movement mechanisms in wh-questions. In a configuration such as (41a), the

particle (thế/vậy) licenses a null question operator [Q] in the CP domain. The wh-phrase

in the island combines with other elements to form sets of alternatives up to the CP

domain. The null [Q] captures this set and yields the same set, namely a set of

propositions. A similar procedure happens in the embedded question configuration (41b):

The Hamblin functional application induced by the wh-phrase in the island creates sets of

alternatives that are captured by the covert [Q] licensed by the matrix verb.

(41) Non-movement

a. Matrix

XP

CP PART licensor

C’ 1

C[Q] … XP island

wh-phrase

205

b. Embedded

Vlicensor CP

C’

[Q] … XP island

wh-phrase

For illustration, consider wh-question with a configuration as in (34a), where the

wh-phrase is embedded in a CNP island. The CNP is computed on the principles below

(from Kratzer and Heim 1998).37

Predicate Abstraction

If α is a branching node whose daughters are a relative pronoun and β, then [[ α ]] =

λx ∈ D. [[ β ]] x.

Predicate Modification

If α is a branching node and {β, γ} the set of its daughters, then, for any assignment a, if

[[β]]a and [[γ ]]a are both functions of type <et>, then [[α ]]a = λx ∈ D. [[ β ]] a (x) =

[[γ ]]a (x) = 1.

37 In chapter 2, we proposed that mà is an operator that moves from IP to Spec, CP of the relative clause.

206

Traces and Pronoun Rule

If α is a pronoun or a trace, a is a variable assignment, and i ∈ dom (a), then [[αi]]a = a

(i).

In Vietnamese a noun phrase with a classifier tends to be construed as definite,

especially when followed by a modifier. Suppose that a covert iota operator is inserted in

the structure with a denotation like that of ‘the’ in English (K & H 1998):

[[ ι]] = λf : f ∈ D <et> & ∃! x [f(x) =1]. ιy [f(y) = 1], and that the relative marker mà

functions as a relative pronoun as in English.

The complex NP in (42a) with a configuration as in (42b) is derived as in (42c).

(42) a. [NP quyển sách [mà [ ai [ mới mua ]]

CL book REL who just bought

‘the book that who just bought’

b. CLP

CL NP

quyển N’ CP

sách mà i C’

C TP ai VP mới VP mua t i

207

c. [[ mua]] w,g = {λx. λy. λw’. bought (y) (x) (w’)}

[[ VP ]] w,g = {λy. y bought g(i) in w’}

[[ai ]] w,g = {x: human (x)}

[[ TP ]] w,g = {[[who]] w,g bought g(i) in w’} HFA

= {p: human (x) & x bought g(i) in w’}

[[ CP ]] w,g = {λx. [[who]] w,g bought x in w’} PA

= {P: y ∈De & P = λx. y bought x in w’}

[[ CP ]] w,g = {P: P = λx. y bought x | y ∈ {Gi, May, Nhi}}

[[ N’ ]] w,g = { λx. λw’. x is a book in w’} PM

[[ NP]] w,g = [[ CP]] w,g ( [[N’]] w,g) =

{λx. Gi bought x & x is a book, HFA

λx. May bought x & x is a book,

λx. Nhi bought x & x is a book}

[[ CLP ]] w,g = [[ ι NP ]] w,g = {the book that Gi bought, the book that May

bought, the book that Nhi bought}

The wh-question with a Complex NP island in (43) is semantically computed as

in (44).

208

(43) [CPTân [VPvừa mượn [NP quyển sách [mà [ ai [ mới mua ]]] ]]] thế ?

Tan just borrow CL book REL who just buy PART

‘Tan borrowed the book that who just bought?’

(44) a. [[VP]]w,g ={λz.λw. z borrowed [[ CLP ]] w,g }

b. [[CP]]w,g = {p: human (x) & Tan borrowed {the book that x bought}}

c. that Tan borrowed the book Nhi bought, that Tan borrowed the book May

bought, that Tan borrowed the book Gi bought…

In summary, Hamblin Functional Application works right through the island: The

set propositional alternatives keeps expanding up to the particle, where it is captured by

the covert operator licensed the particle.

Consider next the sentence in (45), where there is no particle and the wh-phrase

người nào ‘which person’ in the CNP is inherently D-linked. As shown in (46) the whole

CNP moves to the matrix CP domain. Hence, there is no island effect.

(45) [ CPTân [VP vừa mượn [NP quyển sách [mà [ người nào [mới mua ]]] ]]] ?

Tan just borrow CL book REL person which just buy

‘Tan borrowed the book that which person just bought?’

209

(46) LF pied-piping of the CNP to Spec, CP

CP

CNP i 1 C’

C [Q] TP

VP

Tan borrowed t i

The semantic derivation of (45) is in (47).

(47) a. [[ VP ]] w,g = {λx λw’. x borrowed g(1) in w’ }

b. [[ TP ]] w,g = {p: λw’. Tan borrowed g(1) in w’}

c. [[ 1 C’]] w,g = {λz λw’. Tan borrowed z in w’}

b. [[ CP]] w,g = [[ 1 C’]] ( [[ CNPi]])

= {λz. λw Tan borrowed z in w} ([[CNPi ]] w,g )

= {p: human (y) & Tan borrowed the book that y bought}

= that Tan borrowed the book Lem bought, that Tan borrowed the book May

bought, that Tan borrowed the book Gi bought…

The computational processes in the D-linked and non-Dlinked contexts end up

with the denotation of the question. Similar to B&T (2006), where the role of the wh-

210

particle is trivial: it serves as a licensor of a covert operator, in neo-Hamblin semantics,

the wh-particle also licenses a covert question operator that takes propositional

alternative sets. However, unlike B&T (2006), where wh-phrases are ambiguous between

weak indefinites and strong quantifiers, on neo-Hamblin semantics, the denotation of the

wh-phrase is uniform irrespective of whether a wh-particle is present or not.

5.5 Wh-islands

In chapter 3 we saw that one of the virtues of Shimoyama’s (2001) theory is its

ability to account for the island puzzle in Japanese. Wh-islands and CNP island contrast

with respect to the availability of a universal interpretation of a wh-phrase embedded in

those islands. The island puzzle falls out due to the interpretation mechanism: In the

configuration in (48a), the wh-phrase cannot associate with the higher Q given that a

lower Q is available to capture it. By contrast, the association between a wh-phrase and

the Q-operator in (48b) is not problematic. It is a long distance relation, but not one of

movement: the Q-operator takes as its argument the sister constituent, which is an

alternative set, formed by the pointwise functional application of the wh-phrase with

other elements in the structure. In (48b) there is no operator that stops Hamblin

Functional Application.

(48) a. *[Q2 [wh-island Q1 [ … wh … ]]]

211

b. [Q [CNP [ … wh … ]]]

Given that the interpretative mechanisms for wh-questions and wh-universals are

the same in Krazter &Shimoyama’s framework, we would expect that wh-islands are

inviolable in Vietnamese. In fact, wh-islands in Vietnamese present a tricky case

compared with other islands. I and most of the native speakers I have consulted with have

unclear and varying judgments on the acceptability of wh-island violations. Consider, for

example the sentence in (49).

(49) Anh muốn biết [ ai ăn cái gì ]

you want know who eat what

i. ‘For which person x, you want to know for what y, x ate y?’

ii. ‘For which thing y, you what to know for which person x, x ate y?’

iii. ? ‘You want to know for which person x, for which thing y, x ate y.’

(50) Anh ấy muốn biết [ ai ăn cái gì ]

he want know who ate what

i. ‘For which person x, he wants to know for which thing y, x ate y?’

ii. ‘For which thing y, he wants to know for which thing y, x ate y?’

iii. ‘He wants to know for which person x, for which thing y, x ate y.’

212

The embedded reading in (50) is easier to obtain than that in (49) for pragmatic

reasons, namely it is not pragmatically felicitous to assert the second person’s wishes,

knowledge. If (50) is uttered out of the blue, with a neutral intonation, the embedded

reading is the most prominent one. The availability of the first and second readings

indicates that wh-islands are violable. These readings are easily obtained when a heavy

stress is given on the wh-phrase that takes the matrix scope or a salient context is

provided. For example, in (49) the heavy stress is on ai ‘who’ for the first reading and on

cái gì ‘what’ for the second reading.

An appropriate context for sentence (49) could be one of the following. On

discovering that the pizza, the hotdog, and the fruitcake in the fridge may be poisoned, an

owner of a restaurant is very worried and wants to find out who ate one of these foods.

One of the cooks, understanding the owner’s worry, asks him, using (51).

(51) Anh muốn biết ai ăn cái gì?

you want know CL eat what

a. ‘For which x, you want to know for which y, x ate y?’

b. ‘For which y, you want to know for which x, x ate y?’

This is a D-linked context, with the set of eaters and the set of food being

presupposed. If we are correct, then wh-islands are violable under certain prosodic

conditions, like the use of heavy stress, and in D-linked contexts.

213

We have seen that a wh-particle can save an island constraint. This is true in wh-

islands as well. When a question particle is added, the double embedded reading is

canceled as illustrated in (52), leaving us only with the matrix wh-question.

(52) a. Anh muốn biết [ ai ăn cái gì ] vậy/thế?

you want know who eat what ] PART

i. ‘For which y, you want to know for which x, x ate y?’

ii. ‘For which x, you want to know for which y, x ate y?’

The data above indicate that wh-islands are like other islands in that they can be

circumvented in D-linked contexts and when they co-occur with wh-particles. In what

follows, however, it is shown that wh-islands are violable in non-D-linked contexts as

well. Consider a context where a suspect was investigated by the police. After the

investigation, the suspect was released. This is the dialogue between the suspect and his

friend.

(53) A: Cảnh-sát muốn biết cái gì vậy/thế

police want know CL what PART?

‘What did the police want to know?’

B: Cảnh sát muốn biết [ ai vừa giết ai đó ].

police want know who just kill who DEM

‘They wanted to know who just killed someone.’

214

A: Cảnh sát muốn biết [ ai vừa giết ai ]?

police want know who just kill who

‘Who did the police want to know killed?’

The data above suggest that wh-islands behave differently from other islands in that they

can be violated in D-linked as well as non-D-linked contexts. This fact is not predicted

under Kratzer and Shimoyama’s theory given that the wh-island configuration is exactly

(49a). The matrix question reading should not be available since the matrix Q-operator is

blocked by the embedded Q-operator. The violability of the wh-island indicates that

unlike the embedded overt question operator in Japanese, the embedded covert question

operators in Vietnamese allow a wh-phrase in their scope to move across them and to be

captured by a higher operator.

However, Vietnamese and Japanese may not differ after all. Kratzer (2004) shows

that Tokyo Japanese allows wh-islands to be violated. Her example is given below.

(54) [[ John-wa Mary-ga nani-o katta ka] kikimasita ka]?

John-Top Mary-Nom what-Acc bought Q asked Q

a. ‘Did John ask what Mary bought?’

b. ‘What did John ask whether Mary bought?’

The interpretation in (54a) is predicted by their theory. What is interesting is the

availability of the interpretation (54b), where the embedded wh-phrase takes matrix

215

scope. Kratzer (2004) proposes that to obtain this reading the embedded wh-phrase

moves to Spec, CP of the embedded clause, as in (55).

(55) [ […wh ... ka ] C’ ] CP...ka

Kratzer (2004) holds that this movement is local and therefore does not violate the wh-

island constraints and that local movement of this kind is licensed under certain prosodic

conditions. The effect of this movement is that the alternatives created by the moved wh-

phrase are captured by the higher Q, not the lower one.

Does Vietnamese employ a local movement like Japanese? Yes, it does,

optionally. Huang (1982) accounts for the absence of islands in Chinese by proposing

that Spec, CP is available in Chinese, which allows successive-cyclic movement. We

propose that in Vietnamese embedded questions do not require movement to Spec, CP,

presumably, because a covert question that is licensed by a question taking verb is always

available. This leaves Spec, CP as an escape hatch. Hence, embedded questions are not

islands at all. As schematized in (56), the violability of wh-islands in Vietnamese is

expected. In (56a), without a particle, the embedded wh-phrase moves to embedded Spec,

CP and the matrix Spec, CP. In (56b), where the matrix question operator is licensed by

the particle, the embedded wh-phrase moves to embedded Spec, CP, where it is captured

by the matrix question operator.

216

(56) a. [CP [CP [Q [TP …wh… …wh… ]]]

b. [CPQ [CP [Q [TP …wh… …wh… ]] PART]

In summary, we have seen that neo-Hamblin semantics account nicely for the data

in Vietnamese wh-questions. Neo-Hamblin semantics, however, does not deny covert

movement as illustrated by the robust evidence in Vietnamese and the-movement-under-

certain-conditions of Tokyo Japanese. We will look at wh-adjuncts in the next section

and see that the two mechanisms, movement and non-movement, are available in wh-

adjunct questions as well.

5.6 Adjunct Wh-phrases

There is a contrast between adjunct wh-phrases and argument wh-phrases in

Vietnamese. First, almost all the adjunct wh-phrases are morphologically complex. For

example, the wh-phrase meaning ‘when’ consists of a wh-element nào ‘which’ and a

noun (khi, lúc or hồi) meaning ‘time’. Secondly, adjunct wh-phrases’ behavior varies

considerably. As shown in (57), the phrase meaning ‘when’ varies in its position

according to tense: It occurs finally in the past tense, and initially in the future tense.

(57) a. Khi nào Tân về ?

time which Tan come.home

‘When will Tan come home?’

217

b. Tân về khi nào?

Tan come.home time which

‘When did Tan come home?’

There are two commonly used ‘how’ phrases in Vietnamese, each of which is a

combination of a preposition, a noun phrase, and nào ‘which’. Which ‘how’ phrase is

used is contingent on the properties of the predicate it quantifies over. For example, the

predicate cảm thấy ‘feel’ goes with the ‘how’ phrase như thế nào, and the predicate đi

‘go/travel’ with the ‘how’ phrase bằng cách nào.

(58) a. Anh cảm thấy như thế nào /* bằng cách nào?

you feel see like so which / by way which

‘How do you feel?’

b. Anh đi từ San Diego đến Newark bằng cách nào /* như thế nào ?

you go from San Diego to Newark by way which/ like so which

‘How do you travel from San Diego to Newark?’

The ‘how’ phrase always appears sentence-finally as shown by the ungrammaticality of

(58c, d).

218

c. ?? Bằng cách nào anh đi từ San Diego đến Newark ?

by way which you go from San Diego to Newark

‘How do you travel from San Diego to Newark?’

d. * Như thế nào anh cảm thấy ?

like so which you feel see

‘How do you feel?’

The wh-phrase meaning ‘why’ always appears in a fixed clause-initial position.

(59) a.* Thơ đi New York tại sao ?

Tho go New York why

‘For which reason x, Tho went to New York for x.’

b. Tại sao Thơ đi New York ?

why Tho go New York

‘For which reason x, Tho went to New York for x.’

It is worth noting that it may only take scope in the clause where it appears.

(60) a. Tân biết tại sao Thơ đi New York.

Tan know why Tho go New York

i. ‘Tan knows for which reason x, Tho went to New York for x.’

ii. * ‘For which reason x, Tan knows Tho went to New York for x?’

219

b. Tại sao Tân biết Thơ đi New York?

why Tan know Tho go New York

i. ‘For which reason x, Tan knows for x that Tho went to New York?’

ii. * ‘Tan knows for which reason x, Tho went to New York for x.’

iii. * ‘For which reason x, Tan knows that Tho went to NY for x?

No long distance reading is available in either (60a) or (60b).

There are two wh-phrases meaning ‘where’; one is simplex, the other complex.

They both normally occur combined with a co-verb (a preposition-like element derived

from a verb). The two most common co-verbs are ở ‘stay’ and đi ‘go’; the former co-

occurs with stative verbs such as ‘sleep, sit, live, stay’, and the latter with dynamic verbs,

denoting movement such as ‘go, run, drive’. Some illustrative examples are given in (61).

(61) a. Tàu đang chạy đi đâu?

ship ASP run go where

‘Where is the ship heading?’

b.Tàu đang đậu ở đâu?

ship ASP anchor stay where

‘Where is the ship anchored?’

While the adjunct wh-phrase ‘why’ does not show any long distance readings,

other adjunct wh-phrases can obtain distance readings as illustrated below.

220

5.6.1 Long Distance Readings

Long distance readings with argument wh-phrases are easy to detect. If a matrix

question reading obtains in association with a wh-phrase in an embedded clause, then we

have a long distance reading case. Adjunct wh-phrases do not necessarily display

explicitly their base positions. For instance, the wh-phrase meaning ‘where’ as shown

above always occurs sentence-finally, and so in a sentence with multiple clauses, it will

be structurally ambiguous. The availability of a long distance reading can be easily

detected if we choose a wh-phrase that associates with the embedded predicate, not with

the matrix predicate. For example, the wh-phrase đi đâu, ‘go where’, associates with the

embedded verb, which is a verb of motion meaning ‘drive’, but not with the matrix

stative verb ‘say’. Hence, in (62a) a long distance reading is the only reading. Note that if

the embedded verb is a stative verb like ‘buy’, then ở đâu, ‘stay where’, can be associated

with both the embedded and matrix verbs. This is indicated by the availability of the two

readings in (62b).

(62) a. Tân nói May đã lái xe đi đâu.

Tan say May PST drive car go where

i. ‘In which place x, did Tan say May drove to x?’

ii. * ‘In which place x, did Tan say in x May drove?’

b. Tân nói May đã mua xe ở đâu.

Tan say May PST buy car stay where

i. ‘In which place x, did Tan say Mary bought a car in x?’

221

ii. ‘In which place x, did Tan say in x Mary bought a car?’

The first reading in (62b) obtains when the wh-phrase ‘where’ attaches to the embedded

clause, and the second reading when it attaches to the matrix.

The wh-phrase ‘when’ displays a similar structural ambiguity. A long distance

reading of this wh-phrase is easily seen in (63a). As shown in (63b), when this wh-phrase

attaches to the embedded clause, we have the first reading, but when it attaches to the

matrix, we have the second reading.

(63) a. Tân nói khi nào Thơ đi New York .

Tan say time which Tho go New York

i. ‘At which time x, Tan said Tho will go to New York at x?’

ii. *‘At which time x, Tan said at x, Tho went to New York?’

b. Tân nói Thơ đi New York khi nào.

Tan say Tho go New York time which

i. ‘At which time x, Tan said that Tho went to New York at x?’

ii. ‘At which time x, Tan said at x that Tho went to New York?’

The initial position of the ‘when’-phrase in the embedded clause in (63a) leads to the

long distance reading only because it associates with the embedded verb. It cannot

associate with the matrix verb. To associate with the matrix verb it must occur sentence-

initially or sentence-finally, as illustrated in (64).

222

(64) Khi nào Tân nói Thơ đi New York.

time which Tan say Tho go New York

i. ‘At which time x, Tan will say at x that Tho went/ will go to New York?’

ii. *‘At which time x, Tan said that Tho will go to New York at x?’

The long distance reading of the ‘how’-phrase obtains in the context where the

embedded predicate is compatible with the ‘how’-phrase. For instance, the ‘how’-phrase

như thế nào ‘like so which’ is compatible with the verb ‘fear’ and a long distance

interpretation, as illustrated by the second reading in (65a), is available. By contrast, the

‘how’-phrase bằng cách nào ‘by way which’ is not compatible with the verb ‘fear’, so

there is no embedded question reading as indicated by the lack of the second reading in

(65b) and no long-distance reading is detected as shown by the absence of the third

reading in (65b). Note that while the ‘how’-phrase như thế nào ‘like so which’ cannot be

associated with the matrix verb in (65a) as indicated by the lack of the first reading in

(65a), the ‘how’-phrase bằng cách nào ‘by way which’ can be associated with the matrix

verb as shown by the availability of the first reading in (65b).

(65) a. Tân biết cô ấy sợ anh ta như thế nào.

Tan know she fear he like so which

i. *‘To which degree x, Tan knows to x, she fears him?’

ii. ‘To which degree x, Tan knows she fears him to x?’

iii. ‘Tan knows to which degree x, she fears him to x.’

223

b. Tân biết cô ấy sợ anh ta bằng cách nào.

Tan know she fear he by way which

i. ‘By which means x, Tan knows by x, she fears him?’

ii. *‘Tan knows by which means x, she fears him by x.’

iii. *‘By which means x, Tan knows, she fears him by x?’

The availability of long- distance readings with adjunct wh-phrases, albeit

complicated by idiosyncratic properties shown above, indicates that wh-questions with

adjunct wh-phrases in Vietnamese have one property of wh-movement, namely un-

boundedness. Moreover, the diagnostic tests below show that questions with adjunct wh-

phrases obey island constraints, and demonstrate the same asymmetry in island

sensitivity. When a wh-phrase appears inside an island and the question lacks a particle,

it is ungrammatical, but with a particle the sentence becomes less deviant and more

acceptable.

5.6.2 Subjacency

In what follows, we will have questions with either ‘when’ or ‘how’ inside an

island.38 For the ‘when’ phrase, we consider only past tense contexts, namely, the context

where it occurs sentence finally. For the ‘how’ phrase we devise a context where it is

compatible with the embedded verbs so that long distance readings can be available. 38 The judgments on wh-adjunct island effects in this section are given by three Vietnamese native speakers, and three Vietnamese heritage speakers. All of them are students at the University of California, San Diego.

224

Let us first consider the Complex Noun Phrase Constraint.

(66) *Tân mới gặp [ ông bác sĩ đã từng làm việc ở Newark khi nào ]?

Tan just meet CL doctor PST once work in Newark when

*‘Which time x, Tan met the doctor who once worked in Newark at x?’39

The sentence in (66) is ungrammatical to all of the native speakers I consulted with. All

but one consultant agreed that adding the wh-particle thế or vậy improves the

grammaticality the sentence as shown in (67). 40

(67) ? Tân mới gặp [ông bác sĩ đã từng làm việc ở Newark khi nào ] thế/vậy?

Tan just meet CL doctor PST once work in Newark when PRT

‘Which time x, Tan met the doctor who once worked in Newark at x?’

We have the same result with the Sentential Subject Island: Five out of six

speakers take it as ill-formed; only one of them considers it marginal.

39 Note that (66) is well-formed if the wh-phrase is interpreted as associated with the matrix verb ‘meet’. Under this interpretation, the wh-phrase ‘when’ is base generated as an adjunct of the matrix clause and is not inside the island. 40 The particles thế and vậy are dialectal variants: The latter belongs to the southern dialect and the former to the northern dialect. Out of the six students that we consult with, two of them speak the northern dialect, the rest are speakers of the southern dialect.

225

(68) *[Lan bỏ học khi nào ] làm cả nhà lo-lắng?

Lan drop out when make all family worry

‘That Lan dropped out when makes/made the whole family worried?’

The appearance of the wh-particle circumvents the island effect. All but one consultant

agree that (69) is much better than (68).

(69) ?[Lan bỏ học khi nào ] làm cả nhà lo-lắng thế/vậy?

Lan drop out when make all family worry PRT

‘That Lan dropped out when makes/made the whole family worried?’

The Coordinate Structure Constraint is obeyed as well, as shown by the

ungrammaticality of (70a). However, the degree of unacceptability reduces significantly

if the wh-adjunct phrase is a complex form, consisting of a noun meaning ‘day’ and

‘which’ as in (70c). Unlike the CNP and SSC island effects above, where the wh-particle

lessens the severity of the violation, the occurrence of a wh-particle does not much

improve the sentence’s grammaticality, as shown in (70b).

(70) a. *Họ đi học [ vào ngày thứ hai và khi nào ]?

they go study on day second and when

‘They went to school on Monday and when?’

226

b. ??Họ đi học [ vào ngày thứ hai và khi nào ] thế/vậy?

they go study on day second and when PRT

‘They went to school on Monday and when?’

c. ? Họ đi học [vào ngày thứ hai và ngày nào ]?

they go study on day second and day which

‘They went to school on Monday and which day?’

It is likely that the well-formedness of (70c) is due to the fact that the wh-phrase ngày

nào ‘day which’ is inherently D-linked: First, the morpheme meaning ‘which’ induces a

D-linked reading, and second the set of the weekdays is presupposed.

It is unexpected that the final particle does not seem to play any role in the

improvement of the Adjunct Island Constraint violation. For example, two of the

speakers surprisingly prefer sentence (71a) with no particle to sentence (71b) with the

particle. The rest of them consider both equally grammatical. 41

(71) a. ?? Họ từ chối [ vì Lan mời họ bằng cách nào] ?

they decline because Lan invite them how

‘They declined because Lan invited them how?’

b. ?? Họ từ chối [ vì Lan mời họ bằng cách nào thế/vậy ] ?

they decline because Lan invite them how PRT

‘They declined because Lan invited them how?’ 41 In the questions in (71), the wh-phrase meaning ‘how’ is to modify the adjunct clause, not the matrix.

227

Speakers’ judgment also varies considerably with respect to wh-island

constraints: Two of the speakers do not accept sentence (72). One speaker takes it as

slightly marginal and three regard it as well-formed.

(72) ? Họ muốn biết khi nào Lan gặp ai?

they want know when Lan meet who

‘They wonder when Lan will meet who?’

The sentence in (73) is ungrammatical according to two of the speakers. The rest of the

speakers take it as highly degraded.

(73) ??Họ muốn biết khi nào Lan gặp ai thế/vậy?

they want know when Lan meet who PRT

‘They wonder when Lan will meet who?’

The deviance of (73) is probably caused by a conflict between the realis mood encoded

by the particle, which has the effect of presupposing the propositional content of the

sentences, including the embedded clause as having happened, and the future tense

interpretation derived from the initial position of the ‘when’ phrase. If this is the case,

then we can conclude that (73) is not relevant to islands.

In summary, the data above illustrate that the final particle tends to circumvent the

island effects on adjunct wh-phrase questions. The judgments vary slightly among

228

speakers, however, probably because many subtle and delicate factors interfere, such as

the correlation between the wh-phrase’s position and the tense interpretation, and the

compatibility of the wh-phrase and the predicate it is intended to be associated with. It is

likely that the inconsistency in the judgments results from the fact that some adjunct wh-

phrases, by virtue of their morphological make-up, are inherently D-linked.

5.7 A Complex NP Island Experiment

It was shown in the previous section that the severity of the violation of

Subjacency varies among speakers and depends on what island is violated. The subtlety

and delicacy of the speakers’ judgment probably results from the reasons we have

discussed: The interaction between the wh-phrase’s morphological make-ups and the

predicates they quantify over, the inherent D-link of the wh-phrase, and so on. In this

section, we report an island experiment with two wh-phrases ‘how’ and ‘where’ and

provide an analysis of the experiment, arguing that it is D-linked contexts that increase

the ability of an adjunct wh-phrase to escape islands. 42

5.7.1 The Experiment

The experiment is designed as an elicited production task in which made-up

stories provide the context for an ambiguous wh-question, with the answers coming from

matrix or embedded position. Pictures are used to tell the story, and the potential answers 42 The experiment was done by Karl M. Chandler and Thuan Tran as part of the requirements for an Acquisition of Syntax course in 2004. The target population is Vietnamese speaking adults in Philadelphia, Pennsylvania. I thank Gaby Hermon for her support in this project.

229

are designed to be equal in salience. Once the story is completed, all the pictures for the

story are displayed simultaneously, and the experimenter asks the ambiguous question.

The answer reveals whether the wh-phrase is interpreted as a matrix or an embedded

question. For the test items, embedded wh-phrases occur within CNP islands. Thus, if a

person can give the embedded answer, this shows that the wh- phrase is not island

sensitive. Our experiment has six control items (ambiguous questions without island

violations) and six test items (ambiguous questions with potential island violations).

5.7.1.1 Control Items

The control items consist of stories followed by questions in complement clauses.

These questions have no structural restrictions on ambiguity and should reflect a chance

response rate on matrix and embedded responses. Three of the control items use the

question word ‘how,’ and three use ‘where’. The Vietnamese with an English gloss is

provided below, followed by an English translation. One of the test items goes as follows:

Girl in the Library

(English Version)

This girl is in the library. She really wants the yellow book that’s up high on the top

shelf. At first she couldn’t reach it, but then she got a tall ladder and used it to get to the

top shelf. See? She’s using the ladder to get the yellow book. After she got the book she

sat down to read it. She was so excited that she got the book that she wanted to tell

everyone. She picked up a microphone and announced that she got the yellow book.

230

Question: The girl announced that she reached the yellow book how?

Answers: With a microphone.

With a ladder.

(Vietnamese Version)

Cô gái này đang ở trong thư viện, cô ta muốn mượn một quyển sách màu vàng nằm rất

This girl ASP in library she want borrow one CL book color yellow lie very

cao trên giá sách. Lúc đầu, cô ta không lấy được, nhưng sau đó cô ta dùng một chiếc

high on shelf. At firstt she not take obtain but later she use one CL

thang và leo lên lấy quyển sách. Cô ta ngồi xuống đọc, nhưng một lát sau cô ta lại

ladder and climb up take CL book. She sit down read but one moment after she then

muốn cho mọi người biết là cô ta đã lấy được quyển sách.

want give all human know COMP she PAST take obtain CL book.

Cô ấy cầm micro và loan báo là cô ấy đã lấy quyển sách màu vàng.

She take microphone and announce COMP she ASP take Cl book color yellow

Question:

Cô gái ấy loan báo là đã lấy quyển sách màu vàng bằng cách nào?

Girl announce COMP pro PAST take Cl book color yellow by way which

231

5.7.1.2 Test items

Test items are designed to support a question with two possible answers; one of

them shows a violation of the CNP island. A test item begins with a story followed by a

question, as illustrated below (first the English translation, then Vietnamese with English

gloss). If Vietnamese speakers provide the embedded answer, this means the adjunct wh-

phrases are not sensitive to CNP islands in their grammars. One of the test items goes as

follows:

Frog Sees Big Bird Drawing

(English Version)

Big Bird is using a giant crayon to color a beautiful flower. Frog wants to see

what big bird is doing, but he is too far away. He is so far away that he cannot see the

picture big bird is drawing with his giant crayon. So, what does frog do? Look. He gets

some binoculars. He’s using binoculars to see what Big Bird is drawing with a giant

crayon. Frog can see that Big Bird is drawing a flower.

Question: Frog sees the picture that Big Bird is coloring how?

Answer: With binoculars.

With a giant crayon.

(Vietnamese Version)

BB đang dùng một cây viết màu lớn để tô một bông hoa.

BB ASP use one Cl write color big to color one Cl flower.

232

Frog muốn xem BB đang làm gì, nhưng nó ở quá xa.

Frog want see BB ASP do what, but he stay too far.

Nó ở quá xa đến độ nó không thể thấy bức hình mà BB đang tô

He stay too far come degree he not can see CL picture REL BB ASP color

bằng cây viết màu. Thế là Frog dùng ống nhòm.

by CL write color. Then Frog use tube watch

Nó đang dùng ống nhòm để xem BB đang tô màu cái gì bằng cây viết màu.

He ASP use tube watch so watch BB ASP color what by Cl write color

Frog thấy BB đang tô màu một bông hoa.

Frog see BB ASP color one Cl flower.

Question: Frog xem bức tranh mà BB đang vẽ bằng cách nào?

Frog see painting REL BB ASP draw by way which

The following test questions are asked after each story is told.

Test questions:

1. Frog xem bức tranh mà BB đang vẽ bằng cách nào?

Frog see painting REL BB ASP draw by way which

‘ Frog sees the picture that big bird is coloring how?

2. Julie ăn cái bánh sô cô la mà mẹ làm bằng cách nào?

Juile eat Cl cake chocolate REL mother make by way which

‘ Julie ate the chocolate cake that mother made how?’

233

3. Viên cảnh sát đã bắt tên trộm đã ăn cắp ví bằng cách nào?

Cl policeman ASP catch Cl steal ASP steal purse by way which

‘The policeman caught the thief that stole purses how?’

4. Bob kể cho mẹ nghe câu chuyện mà cô giáo đã kể ở đâu?

Bob tell give mother story REL teacher ASP tell stay where

‘ Bob told his mother about the story that his teacher played where?’

5. Người anh cho em gái gói kẹo mà mẹ mua ở đâu?

Cl brother give sister Cl candy REL mother buy stay where

‘The brother gave his sister the candy that Mom bought where?’

6. Big Bird đưa món thuốc đặc biệt cho cô gái bị té ở đâu?

Big Bird hand Cl medicine special give girl small suffer fall stay where

‘Big bird gave his special medicine to the girl that fell where?’

5.7.1.3 Results

Table 1 shows all of the results, including control items. The results presented

here indicate that ambiguity is possible. This result is predicted given that it is possible to

obtain long-distance readings with these wh-adjuncts. Because the wh-phrase meaning

‘how’ is morphologically D-linked, while the wh-phrase meaning ‘where’ used in the

experiment is not, it is expected that the two wh-adjuncts would not behave the same with

respect to island effects, namely, one would be more tolerant than the other in accepting

island violation.

234

Table 1

Stimulus Last heard item salience

Response Speaker 1

Response Speaker 2

Response Speaker 3

Response Speaker 4

Response Speaker 5

RELH1 M M M E M M CW2 E E E E E E RELH2 M M M M M M RELH3 M M M M M M CH1 M E E E M M CH3 E E M E M E RELW1 M E E E E M CW1 E E E E E E CH2 E E M E M E RELW2 M E E E E E RELW3 M M M E M M CW3 E E E E E E

Following are some notes on the abbreviations used in the table: RELH 1: Relative

Clause Island with wh-phrase ‘how’, RELW2: Relative Clause Island with wh-phrase

‘where’, CW1: Controlled items with ‘where’, CH1: Controlled items with ‘how’.

The cardinal numbers 1, 2… indicate the order in which the items are tested, and ‘Last

heard item salience’ refers to the question asked at the end of the story.

Table 2 shows the results which bear solely on potential island violations. This

format easily shows the idiosyncratic aspects of the results. It is surprising that for some

reason the wh-phrase ‘where’ is much less sensitive to islands than the phrase ‘how’.

235

Table 2

Stimulus Speaker 1

Speaker 2

Speaker 3

Speaker 4

Speaker 5

RELW1 E E E E M RELW2 E E E E E RELW3 M M E M M RELH1 M M E M M RELH2 M M M M M RELH3 M M M M M

Table 2 indicates that island sensitivity varies depending on what wh-phrase is in

the island. The ‘where’ islands are easier to be violated than ‘how’ islands: Out of three

relative islands with ‘where’, the first two relative islands, RELW 1 and RELW 2, are

violated by 4 speakers. Speaker 5 only shows violation with the second relative island.

The third island with ‘where’ is strictly obeyed, with only one violation (speaker 3). In

contrast, almost all speakers display their sensitivity to the relative islands with ‘how’:

There is only one instance of violation of RELH 1 (speaker 3).

From the results shown in table (2), it seems plausible to conclude that speakers

vary with respect to island sensitivity. For example, speaker 3 seems to be the most

tolerant of islands: Violating 4 out of 6 islands, while speaker 5 is the most sensitive to

islands: Only one out of 6 islands is violated. However, when a speaker replied with a

matrix answer, we are not sure if he can violate the islands or not. All we can be sure is

that the speaker did not violate the islands.

236

5.7.1.4 Analysis

We noted that the wh-phrase meaning ‘how’ which consists of a preposition, a

noun/pronoun and a wh-phrase meaning ‘which’, which suggests that this wh-phrase is

inherently D-linked. An island violation with this wh-phrase does not cause a serious

damage to the well-formedness of the sentence, as shown in (74), where this wh-phrase is

in a sentential subject island.

(74) ? [Lan mời bằng cách nào ] làm họ rất ngạc nhiên ?

Lan inivite by mean which make they very surprise

‘That Lan invited them how made them surprised?’

In contrast, the wh-phrase ‘where’, not inherently D-linked, is ungrammatical in a

sentential subject island (75), according to my judgment.

(75) *[ Tân mới gặp cô gái ở đâu ] làm họ rất ngạc nhiên?

Tan just meet girl stay where make they surprise

‘That Tan just met the girl where made them surprised?’

Therefore, it would be not very surprising if the D-linked wh-phrase ‘how’ shows no

sensitivity to islands. Yet, the result of the experiment illustrates that this phrase seems to

obey island constraints more strictly than the simplex wh-phrase meaning ‘where’ used in

the experiment, which is not D-linked inherently. The inherently D-linked wh-phrase

237

meaning ‘where’ consists of a noun meaning ‘place’ and ‘which’: chỗ nào ‘which place’.

The problem is that in the experiment, the inherently D-linked wh-phrase ‘where’ is not

used. However, there is reason to believe that the context in which the wh-phrase ‘where’

is used forces it to become D-linked: The answers are drawn from the presupposed set:

there are always two places made salient to the subjects of the experiment. In responding

to the questions, the subjects simply pick one out of the two places in each story as an

answer by pointing at the picture. This situation strongly suggests that a D-linked context

is activated.

5.7.2 Implications

The experiment shows that the adjunct wh- phrases ‘how’ and ‘where’ in

Vietnamese can be interpreted within CNP islands. It suggests that the insensitivity to

Adjunct Island Constraints of the wh-phrase ‘how’ in the section on adjunct wh-phrases

is not because Adjunct Island Constraints are weak islands, but because the wh-phrase

‘how’ is inherently D-linked. Regarding the wh-phrase ‘where’, we propose that it is

forced to be D-linked, permitting it to be interpreted within islands. The experiment also

indicates that forced D-linked wh-phrases are more insensitive to islands than inherently

D-linked wh-phrases.

The insensitivity of these wh-phrases in the experiment does not necessarily

refute the island constraints in Vietnamese. They are there in the language, but detecting

these constraints is strongly affected by pragmatic and prosodic contexts.

238

5.8 Conclusion

The neo-Hamblin analysis and the previous proposal by Bruening and Tran

(2006) are equally able to account for wh-questions in Vietnamese. Albeit different, the

two theories converge on important aspects. First, they provide a uniform

characterization of the Q-operator. That is, it is always null, as a binder in Bruening and

Tran (2006) and as a ‘stop signal’ that stops the propositional alternative set from

expanding in the neo-Hamblin analysis. Second, the role of the question particle is trivial

in wh-questions in the sense that it is not a ‘clause typing particle’ on Cheng’s (1991)

hypothesis (see Bruening 2007 for further discussion). Instead, question particles are just

is just a licensor of a null Q- operator and encodes realis modality in wh-questions.

Finally, both theories hold that Vietnamese wh-questions employ both movement at LF

and non-movement. LF movement is sensitive to constraints similar to those on overt

movement.

However, the neo-Hamblin analysis is more elegant than Bruening and Tran’s

(2006) analysis because it provides a uniformed denotation for wh-phrases as questions

and as indefinites. They are sets of individuals of type <e>. B&T (2006)’s analysis

assigns different denotation to the wh-phrases. They are of type <ett> as a wh-indefinite,

and of type <et>, and <ett,tt> as a wh-phrase. In addition, the neo-Hamblin analysis

offers a more simple semantic composition for wh-questions, that is, via Hamblin

functional application: Either the computation expands the constituent until it meets the

covert operator that captures it, or the wh-phrase moves covertly to Spec, CP, to signal

the expansion to stop at the site where it moves to. In contrast, the previous analysis

239

requires two distinct semantic computations: With unselective binding, the interpretation

is realized through an operator and variable binding relation; with LF movement, the wh-

phrase, which is a quantifier, must move to scope position.

240

Chapter 6

CONCLUSION

6.1 Summary

This thesis is mainly concerned with two questions: The first question is, ‘what is

a wh-phrase in Vietnamese?’ Our answer to this question is that a wh-phrase denotes a

set of individual alternatives. For instance, the denotation of a wh-phrase like ai, ‘who’, is

in (1).

(1) [[who]] = {x: x is person}

The denotation in (1) indicates that a wh-phrase is non-quantificational. Now, the

second question is, ‘what contributes to its quantificational interpretation?’ The answer to

this question is found in chapters 3, 4 and 5. In a nutshell, following the neo-Hamblin

theory of Kratzer and Shimoyama (2002), we hold that a wh-phrase combines with other

elements in the construction and forms an alternative set of higher type that is captured

241

by a neighboring operator. A wh-phrase varies its interpretations depending on which

operators capture the alternative set so formed. Specifically, in chapter 3 we propose that

the universal interpretation of a wh-phrase is derived from the focus-sensitive operator,

cũng. This operator functions as a universal quantifier operating over the propositional

alternative set, the denotation of which is in (2).

(2) [[cũng ]]w,g = {λαλw’. ∀p [p ∈[[α ]]→ p(w’)=1]}, where [[ α]] w,g ⊆ Dst

In bi-clausal contexts where a wh-phrase occurs in the antecedent clause of a conditional

sentence, and cũng is optionally present in the consequent, the universal interpretation

does not come from cũng, but from a covert universal quantifier over situations, the

denotation of which is in (3a), informally represented as in (3b).

(3) a. [[∀]]w,g = λpλq. ∀s [s ≤ w & p (s) = 1] → [∃s’[s ≤ s’. q(s’) = 1 ]]

b. For all situations s, such that s ≤ w, if p is true in s, then there exists an s’

such that s ≤ s’, q is true in s’.

The wh-phrase in the antecedent is captured by a covert existential quantifier over

individual alternative sets licensed by the conditional operator. Hence, it is existentially

construed.

The universal quantifier over situations is also responsible for the universal

interpretation of the constituent of the form ai nấy ‘who-NAY’, where nấy is an

242

existential quantifier over individual alternative sets, and that of wh-conditionals, a

construction similar to the Chinese bare conditional. In this construction, the wh-phrase

forms a constituent with nấy as an existential phrase, base-generated in the consequent

clause. This constituent, however, surfaces separately with the wh-phrase being in the

antecedent clause and nấy in the consequent clause. They are related via sideward

movement. As in bi-clausal contexts, in this structure, the universal interpretation is

derived from a quantifier over situations.

The account for the existential interpretation of wh-phrases is given in chapter 4.

Briefly, our analysis is that the existential interpretation of a wh-phrase is derived from a

covert existential quantifier licensed by epistemic operators, yes-no question operators,

negation operators and so on. For instance, the denotation of a yes-no question operator is

given in (4).

(4) For [[ α ]]w,g ⊆ D <s,t> :

[[ Q y/n α ]]w,g = {{λw’. ∃p [p∈ [[ α ]]w,g & p(w’)=1]}, {λw’. ¬∃p [p∈ [[ α ]]w,g

& p (w’)=1]}}

This mechanism entails that a wh-phrase always takes scope under its licensor.

Intermediate scope reading obtains when a higher licensor is available and the lower

licensor allows a wh-phrase to move out of its scope. There is no locality constraint

required. However, a licensor needs to c-command a licensee to license it. Existential

interpretations also obtain when a wh-phrase co-occurs with a demonstrative. In this case,

243

the demonstrative introduces a choice function over sets of individual alternatives, which

is existentially closed at any CP level in Reinhart’s (1996) spirit.

The interrogative interpretation of a wh-phrase is accounted for in chapter 5.

Basically, both movement and non-movement apply in wh-questions in Vietnamese.

When a covert question operator that requires the licensing of a question particle or a

question selecting verb is available, no covert movement is needed as illustrated in (5).

However, when a question operator is not available, a wh-phrase must move to Spec, CP.

On neo-Hamblin semantics, the function of the covert operator is trivial. We can propose

that as an operator it takes a propositional alternative set and returns the same set or we

can instead assume that a question operator functions as a scope marking operator,

namely to mark the scope of the question by stopping the propositional alternative sets

from expansion and obtain a question reading with a scope marked by the operator. If so,

then the covert movement is a way of scope marking just like the covert operator, as

illustrated in (6).

(5) [CP Op [TP…wh… ]]

(6) [CP […wh... ]]

The previous summary of chapters 3, 4 and 5 serves as the answer to the question of what

contributes to the quantification reading of the wh-phrase. The question that this chapter

is concerned with is, ‘what does Vietnamese quantification contribute to our

understanding of quantification in natural language?’ To answer this question, let us first

244

hold it true that: (a) a wh-phrase universally denotes a set of individual alternatives, and

(b) its quantificational interpretation derives from a neo-Hamblin operator, namely an

alternative set taking operator. The contribution that Vietnamese quantification provides

is to show how far it conforms to the above universal assumptions. Namely, it illustrates

what part of Vietnamese wh-quantification is universal and what part is language

specific. For instance, consider the use of the overt operator nấy, ‘NAY’. While it is

universal that a wh-phrase needs an operator to obtain an existential reading, the

requirement that the operator nấy, ‘NAY’ be attached to a human, simplex wh-phrase that

is base-generated in subject position is language-specific. Another language specific issue

is the covert movement of a wh-phrase in a wh-question without a question particle. This

is because on neo-Hamblin semantics, a question reading, which is a set of propositions,

naturally falls out as a result of combining a wh-phrase with other elements in the

structure via Hamblin functional application. Hence, there is no need for covert

movement in wh-in-situ languages.

6.2 Typological claims

We proposed previously that, cross-linguistically, a wh-phrase denotes a set of

individual alternatives and its quantificational interpretation derives from a neighboring

operator. We propose further that, typologically, languages differ with respect to which

type of operators is employed to assign a wh-phrase a quantificational interpretation. For

example, while wh-indefinites in Japanese and English (as in ‘somewhere’, where ‘some’

is analyzed as an operator operating over the wh-phrase ‘where’) use only overt

245

operators, Chinese employs covert operators. Vietnamese uses both covert and overt

operators in wh-indefinites.

Detailed typological claims of wh-quantification are provided in what follows.

6.2.1 Wh-indefinites

According to Haspelmath (1997), a wh-indefinite is built up of a wh-phrase and

an indefinite marker. For example, in Russian the wh-indefinite kto-to, ‘someone’,

consists of the wh-phrase kto, ‘who’, and the indefinite marker to. In his study, no

language is detected in which a wh-phrase is derived from a wh-indefinite, namely, a wh-

phrase is formed by attaching additional morphology to a wh-indefinite. Our analysis

nicely captures this generalization. That is, a clause where a wh-phrase occurs denotes a

set of propositions, the denotation of a wh-question. Without an existential operator, the

clause ends up being construed as a wh-question. Note that the semantic contribution of a

question operator is trivial. Its function is to stop the expansion of the alternative

propositional set at the site where it is located.

It is therefore universal that a wh-phrase needs to be operated over by an

existential operator to obtain an indefinite reading. Languages differ with respect to

whether a covert or an over operator is used. For example, Chinese uses covert operators,

but Japanese employs overt operators.

246

(7) Jialuo mei-you mai sheme. (Chinese)

Jiaoluo not-have buy what

‘Jiaoluo did not buy anything.’ (Cheng 1991, p. 124)

(8) Dare-ka-kara henna tegami-ga todoi-ta. (Japanese)

who-KA from strange letter-N arrived

‘A strange letter came from somebody.’ (Nishigauchi 1990)

The wh-phrase sheme ‘what’ in Chinese is construed as an indefinite phrase from a covert

operator licensed by the negative marker mei-you as shown in (7). The wh-phrase dare

‘who’ in Japanese obtains an indefinite reading from the overt operator ka, as shown in

(8).

Vietnamese employs both covert and overt operators. For example, the indefinite

reading of the wh-phrase ai ‘who’ in (9a) obtains from a covert operator licensed by the

negative marker, while the wh-phrase ai ‘who’ in (9b) takes on an indefinite reading from

the overt operator đó.

(9) a. Tân không gặp ai.

Tan NEG meet who

‘Tan does/did not meet anyone.’

b. Tân gặp ai đó.

Tan meet who DEM

‘Tan met someone.’

247

6.2.2 Wh-universals

It is universal that a wh-phrase needs to be operated over by an overt universal

operator quantifying over individual alternatives to obtain a universal interpretation, for

instance, the universal operator cũng in Vietnamese, dou in Chinese and mo in Japanese.

(10) Ai cũng quen Tân.

who CUNG know Tan

‘Everyone knows Tan.’

(11) Shei dou gei Lisi xie-le xin

who all to Lisi write-ASP letter

‘Everyone wrote a letter to Lisi.’ (Cheng 1995, p.220)

(12) Dono gakusei-mo odotta.

which student-MO danced

‘Every student danced.’ (Shimoyama 2006, p.139)

6.2.3 Wh-questions

With respect to wh-questions, our neo-Hamblin semantic analysis does not

require any syntactic movement cross-linguistically. Yet movement does take place in

matrix questions without particles in Vietnamese. However, there is never any evidence

of movement in languages like Chinese. The covert movement that takes place in

Vietnamese seems to be language-specific, and not driven by any semantic

considerations.

248

6.3 The Acquisition of Wh-questions

We know that wh-questions in Vietnamese employ both non-movement and

covert movement, with the former being insensitive to islands and the latter sensitive to

islands. Let us refer to this grammar as adult grammar to indicate that while this

grammar respects islands by generating sentences such as (13a), not (13b), it also permits

island violation as in (13c).

(13) a. Tân vừa chụp hình [NP con hổ nào]]?

Tan just catch picture CL tiger which

‘Which tiger Tan take a photo of? ’

b. *Tân vừa chụp hình [NPcon hổ [CP đã dọa ai ]] ?

Tan just catch picture CL tiger ASP scare who

‘Tan took a photo of the tiger that scared who?’

c. Tân vừa chụp hình [NPcon hổ [CP đã dọa ai ]] thế ?

Tan just catch picture CL tiger ASP scare who PRT

‘Tan took a photo of the tiger that scared who?’

This fact presents an interesting question regarding how Vietnamese children

acquire wh-questions. If children follow the Subset Condition proposed by Manzini and

Wexler (1987), namely, ‘given two languages, one of which is a subset of the other, if

both are compatible with the input data, the Subset Principle will state that the learning

249

function must pick the narrow one’, then what is the subset of the adult grammar, namely,

what is the grammar that generates either (13a), or (13c), but not both?

If the subset they pick is the grammar that generates sentences like (13c) freely,

then the question is, how do they unlearn that, namely, how do they know that a sentence

like (13b) is ungrammatical? It is certain that they do not find such sentences in the input.

Do they acquire this knowledge from negative evidence? As is thoroughly discussed in

the literature most of the corrections, if any, from parents, are related to function, not

form, it is unlikely that they learn that (13b) is ungrammatical from negative evidence.

If their subset grammar only permits sentences such as (13a), namely, their

grammar does not allow island violation, then how do they know that sentences like (13c)

are possible in the language? They can unlearn by observing in the corpora that sentences

like (13c) that violate islands are produced. In a word, the input data provide them with

evidence against their subset hypothesis so that they can revise it and come up with the

adult grammar, namely, the grammar that allow island violations when a particle is used

and does not if there is no particle.

In summary, the subset principle predicts that children acquiring Vietnamese will

start with the island-sensitive grammar. This prediction is probably correct in Chinese as

well. Children learning Chinese might start out respecting islands, and then get positive

evidence showing that island violating wh-in-situ questions do occur in Chinese corpora,

so they come up with Chinese adult grammar that allows island violation.

250

REFERENCES

Abusch, D.: 1994, ‘The Scope of Indefinites,’ Natural Language Semantics 2, 83-136.

Alonso-Ovalle, L., and Menéndez-Benito, P.: 2007, ‘Another Look at Indefinites in

Islands,’ ms., University of Massachusetts Boston and University of Massachusetts

Amherst.

Aoun, J., and Li, Y. A.: 1993, ‘Wh-Elements in Situ: Syntax or LF?’ Linguistic Inquiry

24: 199-238.

Beck, S.: 1996, Quantified Structures as Barriers for LF Movement,’ Natural Language

Semantics 4, 1-56.

Beck, S.: 2006, ‘Intervention Effects Follow From Focus Interpretation,’ Natural

Language Semantics 14: 1-56.

Bobaljik, Jonathan (1995), Morphosyntax: The Verbal Inflection, PhD Dissertation, MIT.

Bruening, B., and Tran, T.: 2006, ‘Wh-Questions in Vietnamese,’ Journal of East

Asian Linguistics 15 (4) 319-341.

Bruening, B., and Tran, T.: 2006, ‘Wh-Conditionals in Vietnamese and Chinese: Against

Unselective Binding,’ paper presented at the 32 rd Annual Meeting of the Berkeley

Linguistics Society.

251

Bruening, B.: 2007, ‘Wh-In-Situ Does Not Correlate with Wh-Indefinites or Question

Particles,’ Linguistic Inquiry 38, 139-166.

Buring, D. and Hartmann, K.: 2001, ‘The Syntax and Semantics of Focus-Sensitive

Particles in German,’ Natural Language and Linguistic Theory, 19 (2): 229-281.

Carlson, G., and Pellertier, F. G.:1995, The Generic Books, University of Chicago Press.

Chafe, W., L.:1975, Meaning and the Structure of Language, The University of Chicago

Press.

Cheng, Lisa L. S.: 1991, On the Typology of Wh-Questions, Ph.D. dissertation, MIT.

Cheng, Lisa L. S.: 1995, ‘On Dou-Quantification,’ Journal of East Asian Linguistics 4,

197-234.

Cheng, Lisa L.-S., and Huang, James C.-T.: 1996, ‘Two Types of Donkey

Sentences,’ Natural Language Semantics 4: 121– 163.

Chierchia, Gennaro: 1992, ‘Anaphora and Dynamic Binding,’ Linguistics and

Philosophy 15: 111-183.

Chierchia, Gennaro: 1998, ‘Reference to Kinds Across Languages,’ Natural Language

Semantics 6: 339 – 405.

Chierchia, Gennaro: 2000, ‘Chinese Conditionals and the Theory of Conditional,’

Journal of East Asian Linguistics 9: 1–54.

Chomksy, N.: 1976, ‘Conditions on Rules of Grammar,’ Linguistic Analysis 2: 303-351.

Chomsky, N.: 1995, The Minimalist Program, MIT.

Cohen, A., and Erteschik-Shir, N.: 2001, ‘Topic, Focus, and The Interpretation of Bare

Plurals,’ Natural Language Semantics 10, 125-165.

252

Cohen, A.: 2001, ‘On the Generic Use of Indefinite Singulars,’ Journal of Semantics 18,

183-209.

Cole, P. and Hermon, G.: 1994, ‘Is There LF Wh-Movement,’ Linguistic Inquiry 28, 239-

262.

Cole, P. and Hermon, G.: 1998, ‘The typology of Wh-movement: Wh-Questions in

Malay,’ Syntax 1, 221-258.

Collins, C.: 1997, Local Economy, MIT.

Comorovoski, I.: 1996, Interrogative Phrases and the Syntax-Semantics Interface,

Kluwer.

Dayal, V.: 1998, ‘Any as Inherently Modal,’ Linguistics and Philosophy 21, 433- 476.

Dayal, V.: 2000, ‘Scope Marking: Cross-Linguistic Variation In Indirect

Dependency,’ in Lutz, U., and Stechow, A.V. (eds), Wh-Scope Marking, 157-193.

Dekker, Paul: 1993, ‘Existential Disclosure,’ Linguistics and Philosophy 16: 561–587.

Diesing, M.: 1992, Indefinites, MIT.

Dingxu, S.: 1992, The Nature of Topic Comment Constructions and Topic Chains, Ph.D.

dissertation, University of Southern California.

Dimroth, C.: 2002, Topics, Assertions, and Additive Words: How L2 Learners Get From

Information Structure to Target Language Syntax, Linguistics 40-4: 891-923.

Duffield, N.: 2001, ‘On Certain Head-Final Effects in Vietnamese,’ WCCFL 20

Proceedings, 150-163.

Elbourne, P. : 2000, ‘E-Type Anaphora as NP-Deletion,’ Natural Language Semantics 9,

241-288.

253

Elbourne, P.: 2001, ‘When is Situation Semantics Allowed?’ paper presented at SALT

11, New York University.

Fintel, K. von: 1995, ‘A Minimal Theory of Adverbial Quantification,’ in Barbara Partee

and Hanas Kamp (eds.), Context Dependence in the Analysis of Linguistic Meaning:

Proceedings of the Workshops in Prague, 153-193.

Fintel, K. von and Gillies, A.S.: 2006, ‘Epistemic Modality for Dummies,’ ms. MIT and

University of Michigan.

Fox, Danny: 1999, ‘Reconstruction, Binding theory, and the Interpretation of Chains,’

Linguistic Inquiry 30: 157 - 196.

Fox, Danny: 2002, ‘Antecedent Contained Deletion and the Copy Theory of Movement,’

Linguistic Inquiry 33: 63- 96.

Fukui, N., and Takano, Y.: 2000, ‘Nominal Structure: An Extension of the Symmetry

Principle,’ in Svenonius, P. (ed.), The Derivation of VO and OV, 219-254.

Groat, E. & O’Neil J.: 1996, ‘Spell-Out at the Interface,’ in W. Abraham, S.D. Epstein,

H. Thráisson, and C-J. W. Zwart (eds.), Minimalist Ideas: Syntactic Studies in the

Minimalist Framework, 113-139.

Gundel, J. K.: ‘Shared Knowledge and Topicality,’ Journal of Pragmatics 9, 83-107.

Haiman, J.: 1978, Conditionals Are Topics,’ Language 54 (3), 564-589.

Haspelmath, M.: 1997, Indefinite Pronouns, Oxford.

Hasegawa, N.: 2005, The EPP Materialized First, Agree Later: Wh-Questions, Subjects

and Mo ‘also’-Phrases,’ Scientific Approach to Language 4, 33-80, Kanda University

of International Studies.

254

Hendriks, P.: 2004, Coherence Relations, Ellipsis and Contrastive Topics, Journal of

Semantics 21: 133-153.

Heim, Irene: 1982, The Semantics of Definite and Indefinite Noun Phrases, Ph.D.

dissertation, University of Massachusetts, Amherst.

Heim, Irene: 1990, ‘E-Type Pronouns and Donkey Anaphora,’ Linguistics and

Philosophy 13: 137–177.

Huang, C.T.J.: 1982, Logical Relations in Chinese and the Theory of Grammar, Ph.D.

dissertation, MIT.

Huang, C.T.J.: 1984, ‘On the Distribution and Reference of Empty Pronouns,’ Linguistic

Inquiry 15:4, 531-574.

Iatridou, S.: 1991, Topics in Conditionals, Ph.D Dissertation, MIT.

Ionin, T.: 2006, ‘This is Definitely Specific: Specificity and Definiteness in Article

Systems, Natural Language Semantics 14, 175 – 234.

Jayaseelan, K. A.: 2001, ‘Questions and Question-Word Incorporating Quantifiers in

Malayalam,’ Syntax 4, 63-93.

Kadmon, N., and Landman, F.: 1993, ‘Any,’ Linguistics and Philosophy 16, 353-422.

Karttunen, L.: 2002, ‘Syntax and Semantics of Questions,’ in Portner, P. and

Partee, B. H. (eds) Formal Semantics, Blackwell.

Katz, J. and Postal, P.: 1964, An Integrated Theory of Linguistic Description, MIT.

Keshet, E.: 2005, ‘Scalar Implicatures with Alternative Semantics, Ms., MIT.

Kiss, E.: 2007, ‘Topic and Focus: Two Structural Positions Associated with Logical

Functions in the Left Periphery of the Hungarian Sentence,’ in Féry, C., Fanselow,

255

and Krifka (eds.): The Notion of Information Structure, Interdisciplinary Studies on

Information Structure 6: 69-81.

Kitagawa, C.: 1982, ‘Topic Constructions in Japanese,’ Lingua 57, 175-214.

Kitagawa, Y.: 2005, ‘Prosody, Syntax and Pragmatics of Wh-questions in Japanese,’

English Linguistics 22.2: 302-346.

Kratzer, A.: 1989, ‘An Investigation of the Lumps of Thought,’ Linguistics

and Philosophy 12, 607- 653.

Kratzer, A., and Shimoyama, J.: 2002, ‘Indeterminate Pronouns: The View

from Japanese,’ paper presented at the 3rd Tokyo Conference on

Psycholinguistics.

Kratzer, A.: 2004, ‘Indefinites and the Operators They Depend on: From Japanese

to Salish,’ ms., University of Massachusetts Amherst.

Kratzer, A., and Heim, I.: 2000, Semantics in Generative Grammar, Blackwell.

Kuno, S.: 1995, ‘Negative Polarity Items in Japanese and English,’ Harvard

Working Paper in Linguistics 5, 165-197.

Lahiri, U.: 1998, ‘Focus and Negative Polarity in Hindi,’ Natural Language Semantics,

6, 57-123.

Lasnik,, H.: 1989, ‘On the Necessity of Binding Conditions,’ in Lasnik H. (ed), Essays on

Anaphora, Dordretcht: Kluwer Academic Publishers.

Lasusaw, B.: 1979, Polarity Sensitivity as Inherent Scope Relations,

Ph.D. dissertation, University of Texas at Austin.

Lee, C.: 2008, ‘Contrastive (Predicate) Topic, Intonation, and Scalar Meanings,’ in

256

Lee, C. at al (eds), Topic and Focus: Cross-linguistic Perspectives on Meaning

and Intonation, 151-175.

Lewis, D.: 1975, ‘Adverbs of Quantification,’ in Keenan E. (ed), Formal Semantics of

Natural Language, 3-15.

Li, Yen-Hui Audrey: 1992, ‘Indefinite Wh in Mandarin Chinese,’ Journal of East

Asian Linguistics 1: 125–155.

Li, Yen-Hui Audrey: 1996, ‘Definite and Indefinite Existential Constructions,’ Studies in

the Linguistic Sciences 26: 175–191.

Li, Yen-Hui Audrey: 1999, ‘Plurality in a Classifier Language,’ Journal of East Asian

Linguistics 8, 75 -99.

Lin, Jo-Wang: 1996, ‘Polarity Licensing and Wh-Phrase Quantification in Chinese,’

Ph.D. diss., University of Massachusetts, Amherst.

Lin, Jo-Wang: 1998, ‘Distributivity in Chinese and Its Implications,’ Natural Language

Semantics 6: 201–243.

Lin, Jo-wang: 2004, ‘Choice Functions and Scope of Existential Polarity Wh-phrases in

Mandarin Chinse,’ Linguistics and Philosophy 27, 451-491.

Link, G.: 2002, ‘The Logical Analysis of Plurals and Mass Terms: A Lattice-theoretical

Approach,’ in Portner, P. and Partee, B. H. (eds) Formal Semantics, Blackwell.

Longobardi, G.: 1993, ‘Reference and Proper Names: A Theory of N-Movement in

Syntax and Logical Form,’ Linguistic Inquiry 25: 609-665.

Manzini, M.R. and Wexler, K.: 1987, ‘Parameters, Binding, and Learnability,’ Linguistic

Inquiry 18: 413-444.

257

Matthewson, L.: 1999, ‘On the Interpretation of Wide Scope Indefinites,’ Natural

Language Semantics 7, 79-134.

Matthewson, L.: 2001, ‘Quantification and the Nature of Crosslinguistic Variation,’

Natural Language Semantics 9, 145-189.

Nishigauchi, T.: 1986, Quantification in Syntax, PhD dissertation, University of

Massachusetts, Amherst.

Nishigauchi, T.: 1990, Quantification in the Theory of Grammar, Kluwer, Dordrecht.

Nishigauchi, T.: 1991, ‘Construing Wh,’ in Huang, C.T.J, and May, R. (eds), Logical

Structure and Linguistic Structure, Kluwer, 197-231.

Noguchi, T.: 1997, ‘Two Types of Pronouns and Variable Binding,’ Language 73, 770-

797.

Nunes, Jairo: 2004, Linearization of Chains and Sideward Movement, MIT Press,

Cambridge, Mass.

Payne, T. E.: 1997, Describing Morphosyntax: A Guide for Field Linguists, Cambridge

University Press.

Pesetsky, D.: 1987, ‘Wh-in-Situ: Movement and Unselective Binding,’ in Reuland, E.,

and Meulen, A. (eds) The Representation of (In)definiteness, MIT.

Pesetsky, D.: 2000, Phrasal Movement and Its Kin, MIT.

Prince, E. F.: 1985, ‘Fancy Syntax and ‘Shared Knowledge’,’ Journal of Pragmatics 9,

65-81.

Prince, E. F., and Ward, G. L.: 1991, ‘On the Topicalization of Indefinite NPs,’ Journal

of Pragmatics 15.8, 167 -178.

258

Progovac, L.:1993, ‘Negation and Comp,’ Rivista di Linguistica 5.2, 329-347.

Portner, P., Yabushita, K.: 2001, ‘Specific Indefinites and the Information

Structure Theory of Topics,’ Journal of Semantics 18, 271-297.

Portner, P.: 2002, ‘Topicality and (Non-) Specificity in Mandarin, Journal of Semantics

19, 275 -287.

Ramchand, G. C.: 1997, ‘Questions, Polarity and Alternative Semantics,’ in K.Kusumoto

(ed.), NELS 27, 383-396.

Reinhart, T.: 1997, ‘Quantifier Scope: How Labor Is Divided Between QR and Choice

Functions,’ Linguistics and Philosophy 20, 335-397.

Richard, N.: 2001, Movement in Language, Oxford.

Rhimer, M.: 1998, ‘Quantification and Context,’ Linguistics and Philosophy 21, 95-115.

Rooth, M.: 1992, ‘A Theory of Focus Interpretation,’ Natural Language Semantics 1, 75-

116.

Rosén, V.: 1998, Topic and Empty Pronouns in Vietnamese, Ph.D. dissertation,

University of Bergen.

Rullmann, H.: 2003, Additive Particles and Polarity, Journal of Semantics 20: 329 -401.

Sag, I. A., and Fodor, J.D.: 1982, ‘Referential and Quantificational Indefinites,’

Linguistics and Philosophy 5, 355-389.

Schwarzschild, R.: 2004, ‘Singleton Indefinites,’ Journal of Semantics 19, 289-314.

Shimoyama, J: 2001, Wh-Constructions in Japanese, PhD dissertation, University of

Massachusetts, Amherst.

259

Shimoyama, J: 2006, ‘Indeterminate Phrase Quantification in Japanese,’ Natural

Language Semantics 14, 139-173.

Simpson, A.: 2005, ‘Classifiers and DP Structure in Southeast Asia,’ in Cinque, G. and

Kayne, R. (eds), The Oxford Handbook of Comparative Syntax, Oxford.

Song, J. J.: 2001, Linguistic Typology: Morphology and Syntax, Pearson Education.

Srivastav, V.: 1991, ‘The Syntax and Semantics of Correlatives,’ Natural Language and

Linguistic Theory 9, 637-686.

Stechow, A. V.: 2000, ‘Partial Movement, Scope Marking, and Transparent Logical

Form,’ in Lutz, U., and Stechow, A.V. (eds), Wh-Scope Marking, 447-478.

Stowell, T., and Beghelli, F.: 1997, Distributivity and Negation: The Synatx of Each and

Every,’ in Szabolsci, A. (ed.), Ways of Scope Taking, Kluwer.

Szabolsci, A.: 1994, ‘The Noun Phrase,’ in Kiefer, F. and Kiss, K. E (eds.), Syntax and

Semantics, Vol 27, The Syntactic Structure of Hungarian, Academic Press.

Tran, T.: 2005, ‘Wh-Conditionals in Vietnamese,’ paper presented

at Theoretical East Asian Linguistics, Harvard University.

Tran, T.: 2007, ‘Quantificational Strategies in Vietnamese,’ paper presented

at 17th Annual Meeting of the Southeast Asian Linguistics Society, University of

Maryland, College Park, Maryland.

Tran, T.: 2008, ‘Wh-Quantification: A Vietnamese Perspective,’ paper presented at

the 24th Northwest Linguistics Conference, University of Washington, Seattle.

Tsao, Feng-Fu: 1979, A Functional Study of Topic in Chinese: The First Step Towards

Discourse Analysis, Taipei, Taiwan.

260

Tomioka, S.: 2007, ‘Pragmatics of LF Intervention Effects: Japanese and Korean

Wh-Interrogatives,’ Journal of Pragmatics 39 (9), 1570-1590.

Umbach, C.: 2001, ‘Contrast and Contrastive Topic,’ paper presented at the ESSLLI2001

Workshop on Information Structure, Discourse Structure and Discourse Semantics.

Watanabe, A.: 1992, ‘Subjacency and S-Structure Movement of Wh-In-Situ,’ Journal of

East Asian Linguistics 1, 255-291.

Wilkinson, K.: 1996, ‘The Scope of Even,’ Natural Language Semantics 4, 193-215.

Yanovich, I.: 2005, ‘Choice Functional Series of Indefinite Pronouns and Hamblin

Semantics,’ in Georgala, E. and Howell (eds.), Proceedings of SALT XV,

Ithaca, NY: CLC Publications.