The Structure and Function of the Chinese Copular

Home , Chew (surname), Chiu, Duan (surname), Interjection, Lai (surname), Shi (surname)

THE STRUCTURE AND FUNCTION OF THE CHINESE COPULAR

CONSTRUCTION

A DISSERTATION

SUBMITTED TO THE DEPARTMENT

OF EAST ASIAN LANGUAGES AND CULTURES

AND THE COMMITTEE ON GRADUATE STUDIES

OF STANFORD UNIVERSITY

IN PARTIAL FULFILLMENT OF THE REQUIREMENTS

FOR THE DEGREE OF

DOCTOR OF PHILOSOPHY

IN CHINESE

Fangqiong Zhan August 2012

This work is licensed under a Creative Commons Attribution- Noncommercial 3.0 United States License. http://creativecommons.org/licenses/by-nc/3.0/us/

This dissertation is online at: http://purl.stanford.edu/ps422bz6821

ii I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Chao Sun, Primary Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Elizabeth Traugott, Co-Adviser

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

Yoshiko Matsumoto

I certify that I have read this dissertation and that, in my opinion, it is fully adequate in scope and quality as a dissertation for the degree of Doctor of Philosophy.

John Wang

Approved for the Stanford University Committee on Graduate Studies. Patricia J. Gumport, Vice Provost Graduate Education

This signature page was generated electronically upon submission of this dissertation in electronic format. An original signed hard copy of the signature page is on file in University Archives.

iii

ABSTRACT

This thesis examines the structure and function of the Chinese copular construction within the framework of Construction Grammar (Goldberg 2006; Croft 2001; etc.) and

Constructionalization (Trousdale 2010; Traugott and Trousdale 2011; etc.). My analysis begins with the argument that shì is the systematic copula verb in Chinese. After identifying problems with previous accounts, I outline my own, original analysis of the syntax and semantics of the copular construction. I define the Chinese copula as an invariant non-inflectional verb that co-occurs with certain lexemes when they together form the predicate of a copular sentence. I propose that the copular construction is a form and meaning pairing: [(XPi) COP XPj] (XP=NP/VP/S)[SEMi copulative linking

SEMj] with [NP COP NP] as the prototype. The copular construction has two subschemas: specificational and predicational.

A cleft sentence is a special specificational copular sentence. The Chinese cleft construction is a form and meaning pairing: [NPi COP NOMj] (NOM=(ADV/TP/PP)

NP/VP/S de)[SEMi specificational+contrastive SEMj]. I suggest shì is consistently the copula verb in the cleft construction and signals the immediate post-copula element as contrastive focus. The cleft construction also has two subschemas: cleft-obj and cleft-sbj.

My constructional analysis improves on similar accounts of the cleft sentences in two ways. First, my analysis helps understand the grammatical status of shì and provides a schematic framework to understand the commonality and distinction between cleft sentences and copular sentences. Second, my analysis allows for a straightforward account of the relationship between the two subschemas of the cleft construction, and of

the relationship among variations of the cleft-sbj.

The thesis also examines the constructionalization processes of the copular construction and the cleft construction. I suggest that the Old Chinese (500 BCE- 200

CE) topic-comment construction, in which the demonstrative pronoun shì occurred at the subject position of the comment clause functioning as an anaphor, was reanalyzed as a subject-predicate construction via analogization to the construction of the Old Chinese verb wéi ‘to be.’ As the copular construction was entrenched and conventionalized in

Middle Chinese (200 CE -1000), it gave rise to the emergence of the cleft construction through host-class expansion, syntactic expansion (the nominalization was recruited into the predicate position of a copular sentence), and semantic-pragmatic expansion.

Together, my synchronic and diachronic analyses add up to a maximally explanatory account of the copula shì and the copular construction.

ACKNOWLEDGMENTS

My interest in the topic of grammaticalization and constructionalization was first inspired by a graduate seminar on the history of Chinese taught by Chaofen Sun during my first year of PhD study at Stanford University, and subsequently in a seminar on the theory of constructionalization taught by Elizabeth Traugott. More recently I have immersed myself in the theory of construction grammar and constructionalization, and have been amazed by the extent to which this framework can account for Chinese data both synchronically and diachronically. My dissertation process has basically entailed scrutinizing data from both classical and modern Chinese, consulting a broad range of scholarly works, and educating myself in the new critical framework and modes of analysis. None of this would have been possible without the guidance and support I have received from my graduate advisors, teachers, friends, and family.

First and foremost, I would like to thank my advisors, Chaofen Sun and Elizabeth

Traugott. Over the years, Chaofen has continuously stimulated my academic curiosity and supported my aspiration to search for new ideas and new methods. He has always been open and available for discussion, and his invaluable guidance and feedback have been the fuel that allowed me to develop this dissertation. I am also truly lucky and grateful to have had Elizabeth Traugott as my co-advisor; it was Elizabeth who introduced me to construction grammar and constructionalization, and saw me through the writing. She read every single draft of every chapter with meticulous care and her theoretical vision and analytical insight helped shape this project from its inception. I thank both Chaofen and Elizabeth for countless illuminating discussions and for their

unflagging patience, support, and encouragement. This dissertation could not have been written without them.

I also owe a great debt of gratitude to Yoshiko Matsumoto for serving on my committee, reading the chapters, and making time for discussions throughout the dissertation process. I thank her for raising insightful questions and offering informed, incisive comments and feedback.

I am also deeply grateful to John Wang, who has been serving as one of my references during my job hunting. He has always treated me with understanding and patience. I thank him for attending my oral defense, and for many wonderful and inspiring literature classes. I also thank Elizabeth Bernhardt for chairing my oral defense and Ban Wang for his constant advice throughout my graduate study.

I wholeheartedly thank Sandra A. Thompson and Hsiao-Jung (Sharon) Yu, my advisors at the University of California, Santa Barbara, who opened the door of functional linguistics to me and brought me into this fascinating world of linguistics to which I intend devoting the rest of my life. I also thank John Nathan, Ron Egan, and my other teachers at UCSB. They were tremendously helpful when I first came to United

States.

I am truly fortunate to have been surrounded by many great friends. I especially want to thank Zhang Yu. We came to Stanford from UCSB together, and we have shared so many laughs and memories. Without her, my graduate life would have been much lonelier and less colorful. I would also like to thank Xiaoman Miao, Jingxia Lin, Chenshu

Zhou, Tingting Zhao, Yanshuo Zhang, Jeff Knott, Melvin Su, Ming Chew Teo, Hisaaki

Wake, Judy Kroo, Yao Wu, Rui Wang, and many other friends in the department as well

vii

as Xiaofang Zhou, Hong Zeng, Huazhi Wang, Marina Chung, and Nina Lin who have helped me improve my language teaching.

I want to thank my dear friend Jerry Scots, who has been keeping me company from afar throughout my dissertation writing. I thank him for the extensive discussions and arguments that kept me awake, sane, and productive. I thank him for listening to my complaints and excitement, and offering support and encouragement. Without him, the writing process would have been a tedious chore.

Finally, I dedicate whatever I have achieved, with love and gratitude, to my parents

Huadong Zhan and Yajun Ma, and thank them from the bottom of my heart for their unconditional love, constant care and support throughout my life.

viii

TABLE OF CONTENTS

List of Tables ...... xii List of Figures...... xiii Abbreviations...... xiv 1 Introduction...... 1 1.1 The construction shì in Modern Chinese ...... 1 1.1.1 ‘Right, okay, or fine’...... 2 1.1.2 Demonstrative pronoun...... 3 1.1.3 Existential ...... 3 1.1.4 Copulative linking...... 4 1.1.5 “Focus marker” ...... 6 1.1.6 Bound morpheme...... 9 1.1.7 Some questions about shì...... 9 1.2 An overview of the literature on shì ...... 11 1.2.1 The grammatical category of shì...... 11 1.2.2 The origin of the copula shì ...... 16 1.2.3 The previous research on Chinese cleft sentences...... 19 1.2.4 The development of Chinese clefts...... 24 1.3 Data and methodology ...... 27 1.4 An outline of the structure of the thesis...... 29 2 Construction Grammar and Constructionalization ...... 32 2.1 Construction Grammar...... 32 2.1.1 Construction grammar as opposed to modular models...... 32 2.1.2 Three major constructional approaches ...... 35 2.1.3 Some relevant concepts of Construction Grammar ...... 40 2.1.3.1 Taxonomy and inheritance...... 40 2.1.3.2 Coercion...... 42 2.1.3.3 A usage-based model ...... 44 2.2 Constructionalization ...... 46

2.2.1 Two approaches to grammaticalization ...... 48 2.2.2 Motivation: analogy and ‘invited inference’...... 50 2.2.2.1 Analogy...... 50 2.2.2.2 Invited inference ...... 51 2.2.3 Mechanisms: reanalysis and analogization...... 52 2.2.3.1 Reanalysis ...... 53 2.2.3.2 Analogization...... 55 2.2.4 Constructional taxonomies...... 57 2.2.5 Constructionalization dimensions...... 59 2.2.6 Constructionalization and constructional changes...... 60 2.3 Summary...... 61 3 A copula analysis of shì in Chinese cleft sentences...... 62 3.1 Introduction...... 62 3.2 The syntactic concept of copula...... 66 3.3 The semantics of copula...... 74 3.4 The constructional framework ...... 78 3.5 The concept of cleft ...... 81 3.5.1 The cleft construction ...... 81 3.5.2 An adverb analysis of shì in the cleft sentence...... 87 3.5.3 My analysis on Chinese cleft sentences...... 95 3.6 Conclusion ...... 99 4 The constructionalization of shì: from a demonstrative pronoun to a copula...... 102 4.1 Introduction...... 102 4.2 Previous research on shì...... 106 4.3 The development of shì...... 111 4.3.1 A syntactic analysis of the classic copular sentence (CCS) in Old Chinese ...... 111 4.3.2 The constructionalization of the copula ...... 115 4.3.2.1 The semantic relatedness between the demonstrative pronoun and the copula ...... 115 4.3.2.2 The enabling context ...... 117 4.3.2.3 The mechanism of the constructionalization of shì...... 122

4.3.3 Statistical evidence for constructionalization of shì and further expansion...... 129 4.3.3.1 The increase of adverbs preceding shì ...... 130 4.3.3.2 The decrease of sentence final particles ...... 131 4.3.3.3 The decrease of the complex topic...... 132 4.3.3.4 The increase of [NP shì NP]...... 133 4.3.3.5 The competition between shì and wéi ...... 134 4.4 Typology and Conclusion...... 137 5 The constructionalization of the cleft construction...... 140 5.1 Introduction...... 140 5.2 The emergence of the cleft construction...... 142 5.2.1 Shì: the copula in early Middle Chinese (CE 200-CE 600) ...... 142 5.2.2 Nominalization [XP de]: in late Middle Chinese (CE 700-CE 1000)...... 150 5.2.3 The emergence of the cleft construction ...... 154 5.2.3.1 The emergence of [NP COP NOM] (NOM=XP de) ...... 154 5.2.3.2 The emergence of the cleft construction ...... 156 5.2.3.3 The emergence of the cleft-sbj ...... 159 5.3 Constructionalization ...... 166 5.3.1 Motivation: Analogy and pragmatic inferencing ...... 166 5.3.2 Mechanism: Analogization and Reanalysis ...... 169 5.3.3 Conventionalization: Frequency...... 172 5.3.4 Generality, productivity and compositionality...... 173 5.4 Constructionalization ...... 175 6 Conclusion ...... 177 6.1 Summary of the thesis...... 177 6.2 Thoughts on future study ...... 183 6.2.1 Shì as a bound morpheme...... 184 6.2.2 Relativization and nominalization...... 184 6.2.3 The development of contrastive focus...... 185 6.3 Summary...... 185 Bibliography ...... 186

LIST OF TABLES

Table 4.1 The adverb distribution of preceding shì in the string [(XP) shì XP (PTCL)] 130

Table 4.2 The final particle distribution in the string [(XP) shì XP (PTCL)]...... 131

Table 4.3 The pre-copula complex phrase and simple NP distribution in the string

[(XP) shì XP (PTCL)]...... 132

Table 4.4 The NP/VP distribution of the second XP in the string

[(XP) shì XP (PTCL)]...... 134

Table 4.5 The competition between shì ‘to be’ and wéi ‘to be’ I...... 135

Table 4.6 The competition between shì ‘to be’ and wéi ‘to be’ II...... 136

Table 4.7 The distribution of wéi...suǒ...... 137

xii

LIST OF FIGURES

Figure 2.1 Model of the symbolic structure of a construction in Radical Construction Grammar...... 39

Figure 2.2 An example of the taxonomic hierarchy of construction ...... 41

Figure 2.3 Constructional changes related to constructionalization ...... 60

Figure 3.1 The constructional schematic taxonomy of the prototypical Chinese copular construction ...... 80

Figure 3.2 The constructional schematic taxonomy of the Chinese cleft construction .....87

Figure 3.3 A comparison of the schematic copular construction and the cleft construction ...... 100

Figure 5.1 Model of the development of the cleft-obj and cleft-sbj ...... 170

Figure 5.2 The development of the constructional schematic taxonomy of the prototypical Chinese copular construction...... 171

Figure 6.1 The constructional schematic taxonomy of the prototypical Chinese copular construction ...... 178

Figure 6.2 A comparison of the schematic copular construction and the cleft construction ...... 179

xiii

ABBREVIATIONS

ASSOC Associative

AFF Affirmative

AP Adjectival phrase

ASP Aspect marker

BA BA construction

CCS Classical copular sentence

CRS Current related state

CL Classifier

COP Copula

DUR Durative

EXP Experiential

FO Focus operator

FM Focus marker

NEG Negative

NOM Nominalization/Nominalizer

NP Noun phrase

PERF Perfective

PL1 First person Plural

PP Prepositional phrase

PTCL Particle

xiv

REL Relative clause/Relativizer

S Clause

SG1 First person singular

SG2 Second person singular

SG3 Third person singular

TP Time phrase/Topic phrase

VP Verbal phrase

Chapter 1

Introduction

This thesis examines the synchronic structure and diachronic development of the Chinese copula shì and related constructions within the framework of Construction Grammar

(Goldberg 2006; Croft 2001; etc.) and Constructionalization (Trousdale 2010; Traugott and Trousdale 2011; etc.). In the constructional model, linguistic patterns are viewed as constructions. A construction is a form and meaning pairing. Constructions are of any shape from complex sentences to inflectional affixes. “It’s constructions all the way down.” (Goldberg 2006:18) Constructions are not considered the epiphenomenal byproducts of a combination of componential meaning and highly general rules. Instead, aspects of form and meaning are encoded by the construction itself. For my study of the

Chinese copular construction, by targeting both the meaning and form, the constructional approach accounts for a large range of data. (See Chapter 2 for a comprehensive and detailed theoretical discussion on Construction Grammar and Constructionalization)

1.1 The construction shì in Modern Chinese

The word shì occurs in pànduàn jù, a type of sentence that literally means ‘judgment sentence’ in the Chinese linguistic literature, serving to convey speaker’s belief or judgment. The word shì is interesting in that it is extensively productive and appears to occur in a variety of syntactic-semantic contexts. The Modern Chinese Corpus from

Peking University (CCL 2009) shows the tokens of shì rank as the third most frequent

within the corpus next to de (the structural particle) and yī ‘a, one’. The functions of shì in Modern Chinese can be listed as: 1) ‘right, okay, or fine;’ 2) demonstrative pronoun; 3) existential; 4) copulative linking; 5) “focus marker;” 6) bound morpheme. In the following discussion, I outline the examples of the major types of shì, and the thesis basically addresses the functions 2)-5).

1.1.1 ‘Right, okay, or fine’

In (1), shì means ‘right, okay, or fine,’ which is equivalent to the stative verb hǎo ‘okay,’ or duì ‘right.’1 In (1a), it indicates affirmation to the previous statement that the other speaker uttered. In (1b), shì denotes agreement to the command or suggestion given by the other speaker in the dialogue. Shì in (1c) is the verbal predicate modified by the adverb jìu ‘just’ following the clausal subject nǐ bié kàn ‘you do not watch,’ and it is evaluative, indicating speaker’s judgment.

(1) a. 是啊，这话没错 shì a, zhè huà méi cuò right PTCL, these words not wrong ‘Right, these words (are) not wrong.’

b. 你别看就是了 nǐ bié kàn jiù shì le SG2 not watch just okay CRS ‘If you don't watch, it will be fine.’

1 I follow Chao (1968) and Li and Thompson (1981) who treat adjectives in Chinese as intransitive verbs, or stative verbs in that just like regular verbs they are subject to aspect inflections.

c. 是，我一定完成任务 shì , wǒ yídìng wánchéng rènwù yes, SG1 definitely complete mission ‘Yes. I’ll definitely complete the mission.’

1.1.2 Demonstrative pronoun

Shì functions as a demonstrative pronoun occurring in the idioms in (2). It has been argued by scholars (Wang 1937; Feng 1995; Shi and Li 2001; etc.) that the copula shì in

Modern Chinese evolved from the demonstrative pronoun in Old Chinese, and examples

(2) present some of the Modern Chinese idioms in which shì retains the trace of the classical demonstrative pronoun

(2) 他唯利是图 tā wéi lì shì tú SG3 only profit this attempt ‘He attempts only profit.’

1.1.3 Existential

Example (3a) is an instance of existential in which shì serves to indicate existence. Shì in this type of sentence is often compared with the existential verb yǒu ‘have’ in Modern

Chinese, as in (3b). The difference between the existential yǒu and shì has been extensively discussed in the Chinese linguistic literature. Jin (1995) suggests the existential yǒu indicates a number of entities that exist at a certain location, one of which is specified as the complement of the verb yǒu, and hence in (3b), in front of my house,

there are a number of entities, one of which is the river; whereas the existential shì expresses there is the only entity that exists at the location. By employing the existential shì, the speaker ignores other entities that may co-exist at the same location, and profiles the one that is realized as the complement of shì. Therefore, in (3a), for the speaker, the river is the only relevant entity that appears in front of my house.

(3) a. 我家门前是条河 wǒ jiā mén qián shì tiáo hé my house door in front of COP CL river ‘In front of my house, there is a river (excluding other things).’

b. 我家门前有条河 wǒ jiā mén qián yǒu tiáo hé my house door in front of have CL river ‘In front of my house, there is a river.’

1.1.4 Copulative linking

Examples in (4) are prototypical copular sentences, in which shì is the copulative linking verb indicating the nominal relationship between the subject and the complement of shì.

Copular sentences exhibit a variety of nominal meanings: in (4a), tā ‘she’ and wǒ mèimei

‘my sister’ are equational; in (4b), the restricted set měiguó zǒngtǒng ‘the president of the

US’ is specified by Àobāmǎ ‘Obama;’ in (4c), the subject Mali ‘Mary’ is characterized by the post-copula complement huáng tóufà ‘blond hair.’

(4) a. 她是我妹妹 tā shì wǒ mèimei SG3 COP my younger sister ‘She is my younger sister.’

b. 奥巴马是美国总统 àobāmǎ shì měiguó zǒngtǒng Obama COP US president ‘Obama is the president of the US.’

c. 玛丽是黄头发 mǎlì shì huáng tóufà Mary COP blond hair ‘Mary is a blond.’

Li and Thompson (1976, 1981: 15-20) argue that “subject” in Chinese is not exactly equivalent to “subject” in English; in Chinese “‘subject’ is not a structurally definable notion” (1981: 19) since it can always be a zero. It is also not exactly equivalent to topic, as, unlike topic, it must have “a direct semantic relationship with the verb as the one that performs the action or exists in the state named by the verb” (1981: 15). In this view, the clause has a “subject” (defined on semantic rather than grammatical grounds), but there is often a topic that precedes the “subject” as well. For example:

(5) 豆腐我吃了 dòufu wǒ chī le Tofu SG1 eat-PREF ‘The tofu, I have eaten.’

Example (5) is a topic-comment sentence to be in response to a question like ‘have you eaten tofu?’ In (5) dòufu ‘tofu’ has been introduced and is the topic about which ‘I have eaten’ is said; wǒ chīle ‘I have eaten’ is the comment clause with the subject wǒ ‘I’ and the predicate chīle ‘have eaten.’ Based on this, Li and Thompson developed the idea of dividing languages into two types: “subject-prominent” (e.g. English) and “topic- prominent” (e.g. Chinese). Both types have both “subject” and “topic,” but the prominence of “subject” vs. “topic” differs in the two types. Many scholars including

Comrie (1981), Li and Thompson (1976) suggest that Chinese is a “topic-prominent” language, that is, a language in which the structure of the clause takes the form of a topic, about which something is to be said, and a comment, which is what is said about the topic, rather than being a language with a subject-predicate structure like that of English.

I take the position that term “topic” is an information-structuring and pragmatic notion (Lambrecht 2004). In the topic-comment construction such as (5) or topic- resumption construction such as ‘as for tofu, I have eaten it,’ the topic takes the slot preceding the comment clause. However, in the copular sentences like (4), the pre-copula

NPs are the topic, and the post-copula predicate NPs encode informational focus (Xu

2002, section 3.5.1 in Chapter 3), which means the subject usually encodes old/given information as topic, whereas the post-copula predicate as a whole indicates new information that is the informational focus.

1.1.5 “Focus marker”

A focus marker is a particle typically precedes the focus constituent (Hartmann and

Zimmermann 2006). Because Chinese is a verb-medial (VO) language, informational

focus in Chinese typically falls on the post-verbal elements, and therefore informational focus is unmarked. The so-called focus marker in Chinese is in fact a particle that comes before an element and marks it as contrastive focus. Many scholars (Huang 1998; Teng

1976; Fang 1995; Dong 2004; etc.) claim that shì is a contrastive focus marker in the sentences like (6), which marks the immediate post-copula time phrases zuótiān

‘yesterday’ in (6a) and qùnián ‘last year’ in (6b&c) as contrastive focus. However, this thesis proposes that (6) are examples of the cleft construction, and it is the cleft construction as a whole that marks the immediate post-copula element as contrastive focus. Shì is a copula verb rather than a particle in sentences like (6), and it signals and indicates contrastive focus, but it is not a focus marker. (See Chapter 3 for further discussion)

(6) a. 蛋糕是昨天做的 dàngāo shì zuótiān zuò de cake COP yesterday make NOM ‘It was yesterday that the cake was made.’

b. 我是去年来美国的 wǒ shì qùnián lái měiguó de SG1 COP last year come US NOM ‘It was last year that I came to the US.’

c. 我是去年来美国 wǒ shì qùnián lái měiguó SG1 COP last year come US ‘It was last year that I came to the US.’

Examples in (6) present a different information structure from (4), in which the pre- copula NP is still the topic, but the primary informational content is placed in the immediate post-copula focal position to assert what is presupposed by the speaker and the hearer. When the phonetic stress is located on shì, it affirms the whole sentence as an assertion, which means ‘it is true that…’ e.g., ‘it is true that this cake was made yesterday’ in (6a), and ‘it is true that I came to the US last year’ in (6b&c). This thesis does not address this case but only considers the case in which shì indicates the immediate following element as contrastive focus.

Examples like (6) are treated as instances of shì…de construction (Chao 1968; Li and Thompson 1981) or the cleft construction (Paris 1977; Hashimoto 1969; Teng 1979; etc.). A further property of the cleft sentences is that they exhibit contrastive focus, an exhaustiveness and exclusiveness implicature (Kiss 1998). For example, in (6a), we understand that ‘yesterday’ is the day and the only day that the cake was made. These sentences are presuppositional: in (6a), it is presupposed that ‘the cake was made on a certain day,’ and it is an appropriate answer to the question dàngāo shì shénme shíhou zuò de? ‘cake COP when make NOM; when was the cake made?’

Some scholars (Teng 1979; Huang 1998; etc.) distinguish (6a) from (6b&c) in that

(6a) seemingly have the syntactic configuration [NP shì NOM] (NOM=(ADV/PP/TP2)

NP/VP/S de), whereas (6b&c) have the structure of [NP shì VP (PCTL)] (see the structure of the Chinese cleft in 1.2.3). They are different structures and accordingly the shì in each has a different syntactic status: it is the copula verb in (6a), but it is not so in

(6b&c). Huang (1998) explicitly claims that shì in (6b&c) has the status of an adverb. I argue that shì cannot be an adverb in (6b&c) because it does not modify the verb lái ‘to

2 Here, TP=time phrase

come’ or the VP lái měiguó ‘come to US’, instead, it is still a copula verb that indicates copulative linking and indicates contrastive focus. A detailed discussion on this is provided in Chapter 3.

1.1.6 Bound morpheme

In (7), shì is a bound morpheme forming part of the connectives through coalescence with the preceding lexemes. In (7a), it co-exists with kě, which was a modal adverb in classical Chinese and they together form a disyllabic connective kěshì ‘but’ introducing a clause contrastive to the previous one. The adverb hái ‘still’ in (7b) and shì together form a connective háishì ‘or,’ denoting an alternative question.

(7) a. 他虽然很穷，可是很有志气 tā suīrán hěn qióng , kěshì hěn yǒu zhìqì SG3 although very poor, but very have ambition ‘He was very poor, but he was very ambitious.’

b. 你要喝咖啡，还是喝茶？ nǐ yào hē kāfēi , háishì hē chá SG2 want drink coffee, or drink tea ‘Do you want to drink coffee or tea?’

1.1.7 Some questions about shì

1.1.1-1.1.6 exemplify the major uses of shì in Modern Chinese. This thesis aims at the functions of shì from 1.1.2-1.1.5. The functions in 1.1.1 and 1.1.6 are relevant but out of

the scope of the copular construction that I focus on in this thesis. Examples (2-4) and (6) bring up a number of unsolved questions that will be addressed in the thesis.

First, in the above examples, what is the status of shì? Is it a copula verb in all uses of (2-4) and (6)? How do we define the copula in Chinese? What is its syntactic function?

Does it share similar syntactic properties with the English copula ‘to be’ and its Indo-

European equivalents? Does the copula encode any meaning? If not, where do the meanings of the copular sentences come from?

Second, what is the syntactic structure of the cleft sentence? What is the meaning of a cleft sentence? What is its information structure? What is its relationship with regular copular sentences? Do they really have distinct syntactic structures? Does shì have different syntactic status in (6a), (6b) and (6c)? If not, why do they appear to be distinct in terms of form? What contributes to this distinction? What is more, do shì… de sentences such as (6a) share the same syntactic and semantic properties as those like

(6b)? If not, what are the differences between them?

Third, shì in (2) retains a trace of the classical demonstrative pronoun, which supports the argument that the copula shì in Modern Chinese evolved from the demonstrative pronoun in Old Chinese. However, it is not clear how and why the demonstrative pronoun changed into a copula verb and where the first occurrences of copula shì occurred. A few scholars (Pulleyblank 1995; Shi and Li 2001; etc.) claim that the change underwent a process of grammaticalization (Hopper and Traugott 2003), and many proposals have been made to account for the process of the change, but the questions remain unanswered. This thesis seeks to answer them.

Finally, after the copula shì emerged, when did it start to signal contrastive focus?

Why and how did it develop this particular function? Did it undergo a process of further grammaticalization? In relation to this, what were the motivations and mechanisms that enabled the cleft sentence to emerge?

Attempting to answer the above questions, this thesis examines the structure and function of the Chinese copular construction within the framework of Construction

Grammar and Constructionalization. Construction Grammar was developed with a view to provide full and explanatory accounts of broad generalizations as well as specialized linguistic patterns. Constructionalization is a construction grammar perspective of the development of constructions over time. Since the structure and use of the copula and its construction are highly general and schematic, they are well suited to treatment within a constructional approach.

1.2 An overview of the literature on shì

In this section, I outline the major studies in the literature of Chinese linguistics on the word shì and its construction. I discuss the studies on the grammatical category of shì and its origin as well as the studies on the cleft construction and its development.

1.2.1 The grammatical category of shì

There has been a great amount of research on the grammatical category of the word shì.

Ma (1898) takes on the research of shì and considers it to be an anaphoric pronoun that can refer to an NP or can be demonstrative. Wang (1937) first treats shì as xì cí, a linking verb, a copula verb, which has become a widely accepted term for shì in Modern

Chinese. Following Jespersen (1924), Wang suggests that similar to the English verb ‘to be,’ the copula shì is colorless, and that it does not encode any meaning by itself. He treats shì as a xū cí ‘an empty word,’ and any modifiers that precede shì modify the whole predicate of the sentence, which consists of the copula shì and the post-copula phrase. He also points out that although shì is meaningless, it has two derived semantic functions: one is to determine cause; the other is to affirm or deny a proposition. These functions are derived from the fact that the copula shì is also evaluative, implying pànduàn ‘judgment,’ and accordingly copular sentences indicate judgment or assertion. (Wang 1984 Vol1:

159-163) Wang considers only the shì that precedes a nominal to be the copula, and believes that the function of shì preceding an adjectival phrase is influenced by the grammar of the western languages. He suggests that the shì followed by an adjectival or verbal phrase is not the real copula, but is used to affirm mood. (Wang 1984 Vol2: 475)

Lü (1979) systematically treats shì as a verb regardless of what follows. Its major function is to link, and to indicate affirmation and judgment. Following Wang, he considers that the verb shì together with the following element forms the predicate of the sentence. (Lü 2002 Vol5: 371) However, some Chinese linguists including Ding (1980),

Zhu (1982), Huang and Liao (1991) regard shì as a transitive verb, followed by an object.

Zhu (1982) suggests that the object following the verb shì can be either nominal or verbal. When shì is followed by a nominal object, it is pronounced neutrally, and the semantic relation between the subject and the object is equality or member-class. If the subject is an NP of location, shì denotes existential. When shì precedes a verbal object, it usually encodes contrast when it is pronounced neutrally, but denotes affirmation when it is phonetically stressed. (Zhu 1999 Vol1: 120) Chao (1968) also treats the complement of

shì as its object thus making shì a transitive verb, although he mentions that the formal properties are sufficiently different to set it apart from the classificatory verbs. (Chao

1968:716)

(8) a. * 他是了学生 tā shì-le xuéshēng SG3 COP-PERF student

b. * 他没(有) 是学生 tā méi(yǒu) shì xuéshēng SG3 not-have COP xuéshēng

c. *他有没有是学生？ tā yǒuméiyǒu shì xuéshēng SG3 have-not-have COP student

d. *他是一是学生 tā shì yí shì xuéshēng SG3 COP-one-COP student

e. *他是三次学生 tā shì sāncì xuéshēng SG3 COP three times student

f. *他是一天学生 tā shì yītiān xuéshēng SG3 COP one day student

g. *他 [是学生的] tā [shì xuéshēng de] SG3 [COP student NOM]

h. *他 [是的] 学生 tā [shì de] xuéshēng SG3 [COP REL] student

(8) shows that unlike other verbs, shì does not occur with aspect markers, e.g. (8a) and accordingly cannot be negated by the aspectual negator méi(yǒu) as in (8b). Also, shì does not allow the yǒuméiyǒu shì (have-not-have V) yes-no questions, e.g. (8c).

Furthermore, unlike other verbs, shì cannot be reduplicated and does not take iterative and durative adverbials as complements, e.g. (8d-8f). Finally, shì cannot be nominalized or relativized, e.g. (8g&h). However, shì also shares some properties with other verbs. (9) shows like other verbs, shì can be negated by the neutral negator bù and allows the V- not-V yes-no question formation. Modern Chinese has a number of non-inflectional verbs which have similar syntactic properties as shì shown in (8) and (9) such as the existential verb zài ‘to be at a location,’ the perception verb rènwéi ‘to think,’ etc.

(9) a. 他不是学生 tā búshì xuéshēng SG3 not COP student ‘He is not a student.’

b. 他是不是学生？ tā shìbúshì xuéshēng SG3 COP-not-COP student ‘Is he a student?’

Li and Thompson (1981:147), taking a functional descriptive approach, treat shì as an intransitive verb and suggest a simple copular sentence typically contains a referential subject noun phrase linked to a non-referential noun phrase by the copula verb. The verb phrase of the sentence is composed of the copula and a non-referential noun phrase that is not an object of the copula verb; and hence the verb phrase of the copular sentence is intransitive.

As for the fact that shì occurs preceding both nominals and verbals (VP/AP/PP/S3), many linguists such as Huang (1998); Hu (1979), suggest that the one followed by nominals is the linking verb, the copula verb, while the one followed by verbals is an adverb denoting emphatic mood. I will discuss Huang’s (1998) idea in more detail in

Chapter 3.

All the accounts above treat shì as an independent and atomic linguistic form. A copular sentence is made up of components with their aspects of meaning and form mapped onto one another by general rules. Thus, these accounts do not consider that the copula and the schematic copular construction encode aspects of form and meaning by themselves. This thesis argues that shì is consistently the invariant copula verb in Modern

3 According to Li and Thompson (1981), prepositions are considered as co-verbs in Chinese. Both adjectival phrases (AP) and prepositional phrases (PP) are classified as verbal phrases. S here is the acronym of clause.

Chinese and the copular construction is a schematic form and meaning pairing that has syntactic and semantic properties specific to the construction. I will provide a detailed discussion on this in Chapter 3.

1.2.2 The origin of the copula shì

Wang (1937) claims the copula shì emerged at the end of the Western Han (202 BCE - 9

CE). Qiu (1979) using the examples such as 是是帚彗shì shì zhòuhuì ‘this COP comet; this is the comet’ appearing in the then newly excavated texts from the Han tombs of

Mawangdui in Changsha, argues that the copula shì emerged in pre-Qin around the

Warring period (476 BCE- 221 BCE). Although whether these examples can demonstrate the claim that copula was produced in pre-Qin is controversial, Qiu’s claim is supported by a majority of Chinese scholars including Guo (1988); Tang (1991); Feng (1993); etc.

Wang (1937) suggests that the copula shì was conventionalized and became standard around 500 CE. He provides three criteria to determine the copula: 1) it was modified by adverbs; 2) the decrease of sentence final particle yě that denoted the copulative semantics in Old Chinese (500 BCE - 200 CE); 3) it was negated by the neutral negator bù, along with the decrease the occurrences of the Old Chinese nominal negator fēi. This claim has been widely accepted and scholars generally agree that the copula shì frequently occurred in Middle Chinese (200 CE -1000).

Since then the origin of the copula shì has been a hot topic among scholars. Three major proposals have been brought up:

The most widely accepted one is that the copula shì evolved from the demonstrative pronoun in Old Chinese. This claim is proposed by Wang (1937) and supported by Feng

(1993); Pulleyblank (1995); Shi and Li (2001); etc. Wang (1937) argues that in Old

Chinese, the subject was often followed by the demonstrative pronoun shì functioning as an anaphor referring to the subject. Shì frequently appeared in the position linking the subject and the predicate and denoting the copulative meaning, which enabled the process of changing into a copula. Feng (1993) suggests the copula shì originated and developed from a demonstrative pronominal form from the topic-comment construction (Li and

Thompson 1981) in Old Chinese and hypothesizes that one would expect a pause between the topic and comment in the topic-comment construction in early Old Chinese.

Feng proposes that the demonstrative pronoun shì evolved into the copula shì because of the weakening of the function as the demonstrative pronoun and the lack of necessity for the pause. However, I suggest that there is a problem with postulating a pause in classical

Chinese when we have no spoken data unless there are characters such as 兮xī in the text indicating interjection. Moreover, the lack of necessity for the pause that weakened the emphatic function of the demonstrative pronoun did not necessarily change the demonstrative pronoun into a verb.

Shi and Li (2001), in light of the theory of grammaticalization, argue that copula shì evolving from the demonstrative pronoun underwent a process of analogy, modeling regular transitive verbs in Old Chinese. They claim, following Kiparsky (1992), that the holistic structural property of a language at a certain period brings about grammaticalization through a process of analogy. They suggest that Old Chinese had already developed standard SVO word order, and the frequent occurrence of [NP shì NP] is the morphosyntactic context in which shì was influenced by and fitted into the extant transitive verb pattern. However, it is obvious that the copula shì in Modern Chinese

differs from the standard transitive verbs in some syntactically significant ways; if it underwent the analogical process modeling SVO structure, then why didn’t it develop into a full-fledged transitive verb?

The second proposal is suggested by Hong (1958) who challenges Wang’s theory and proposes that the copula shì evolved from the adjective shì, which functions to affirm a proposition. Feng (1992) believes that the copula shì originated from the adjective shì in Old Chinese that meant ‘real’ and ‘actual.’ The affirmative meaning denoted in the adjective shì was semantically related to the copula and was the main motivation that enabled the change to occur. However, this proposal does not account for the syntactic change of shì, nor does it explain how the meaning of ‘real’ or ‘actual’ changed into copulative meanings.

The third proposal is by Yen (1986) who further develops Hong (1958)’s theory and suggests that the use of shì as a copula came from the function as an affirmative particle.

Because of the contrastive meanings of fēi ‘wrong’ and shì ‘right,’ speakers started to use shì as an affirmative particle in an affirmative sentence to be in direct contrast with a corresponding negative sentence where the nominal negator fēi appeared. According to

Yen, fēi was replaced by bú shì ‘NEG shì’ later on. However, there is no evidence to show that fēi was replaced by bú shì. Bù in Old Chinese was used to negate verbs, VP and predicates, as in bù zhī ‘not know,’ bú wéi ‘not do,’ etc. Therefore, bù occurring preceding shì provides evidence that shì was part of the predicate.

None of the three proposals above considers the semantic relatedness between the original item and the target outcome, nor the morphosyntactic contexts in which new meaning can be seen and enabled to occur. Moreover, none of them considers the change

of shì in terms of a construction as a form and meaning pairing. This thesis, however argues, from the perspective of constructionalization, that the constructionalization of shì is part of the development of a schematic copular construction. I argue that the copular construction emerged out of the topic-comment construction in Old Chinese where shì was a demonstrative pronoun occurring at the subject position of the comment clause functioning as an anaphor referring to the previous topic phrase (NP/VP/S). I propose the grammatical constructionalization of the copular construction is the result of reanalysis of the topic-comment construction along with a process of analogization, modeling after the construction of the Old Chinese verb wéi ‘to be.’ The change was gradual consisting of successive micro-changes. A detailed analysis on the process of the constructionalization will be provided in Chapter 4.

1.2.3 The previous research on Chinese cleft sentences

Quirk et al (1985) defines a cleft sentence as a complex sentence in which a simple sentence is expressed using a main clause and a subordinate clause. Traditionally, the

English cleft includes it-cleft, and pseudo-cleft (WH-/ALL/TH-cleft).

Many scholars (Fang 1995; Huang and Fawcett 1996; Teng 1979; Tang 1988; etc.) hold that Chinese shì…de sentences share similar semantic and structural properties with the English cleft sentences, and hence they are termed cleft sentences as well. These scholars accept that Chinese cleft sentences involve a presupposition and a contrastive focus. Examples (6a&b), repeated here as (10a&b) are cleft sentences in which the post- copula element encodes the contrastive focus:

(10) a. 蛋糕是昨天做的 dàngāo shì zuótiān zuò de cake COP yesterday make NOM ‘It was yesterday that the cake was made.’

b. 我是去年来美国的 wǒ shì qùnián lái měiguó de SG1 COP last year come US NOM ‘It was last year that I came to the US.’

Many scholars e.g. Teng 1979; Zhu 1997; Huang 1998, distinguish the cleft

(equivalent to English it-cleft. Hereafter, I use the term “cleft” to be equivalent to English it-cleft) such as (10b) from the so-called pseudo-cleft such as (10a), because they believe there is a semantic distinction between them. They maintain that the cleft such as (10b) is not reversible, whereas the pseudo-cleft indicates equation and therefore is reversible.

They believe that example (11) is the reversed version of (10a) and that (10a) and (11) have identical meaning and they are pseudo-cleft sentences:

(11) 昨天做的是蛋糕 zuótiān zuò de shì dàngāo yesterday make NOM COP cake ‘What was made yesterday is the cake.’

The term “pseudo-cleft” in the English linguistic literature is used to account for wh-/the one/all sentences that are structurally distinct from the it-cleft although they both

involve two clauses: a relative clause and a copular clause. The pseudo-cleft and it-cleft in English share similar semantic and pragmatic indications, including exhaustiveness and exclusiveness (Prince 1978; Higgins 1979; Quirk et al. 1985; etc.), and specificational member-class relationship (Patten 2010).

I suggest that examples like (11) can be considered as a pseudo-cleft in Modern

Chinese, and the post-copula element in (11) ‘the cake’ also encodes contrastiveness. In

Chapter 3, I argue that (10a) is not a pseudo-cleft, but a cleft like (10b), because it has the form [NP COP NOM] (NOM=(ADV/PP/TP) NP/VP/S de). All cleft sentences are specificational, not equational, and cannot be reversed. Within the constructional framework, I propose that the cleft construction is schematic and has the form [NP COP

NOM] and the meaning of specificational plus contrastive. The pseudo-cleft such as (11) does not have the form [NP COP NOM] and it is not the focus of this thesis. Details on the Chinese cleft construction are provided in Chapter 3.

As for the structure of the Chinese cleft, there has been an extensive debate between two proposals: complex predication analysis and simplex predication analysis. The proposal of complex predication analysis involves a dichotomy: one is held by

Hashimoto (1969); Tang (1983); Ross (1983); Tsao (1990); Lü (1979); Chao (1968); etc, the other is proposed by Li and Thompson (1981); Zhu (1985); etc. Hashimoto (1969) argues that the deep structure of (10b) is:

[NP [V VP ]vp PTCL]s

[wǒ [shì qùnián lái měiguó]vp de]s

[I [COP last year come to the US]vp PTCL]s

For Hashimoto, the predicate of (10b) is complex as it consists of a serial verb construction, and the sentence final de is a particle that is independent of the predicate.

Li and Thompson (1981) suggest the cleft is a special copular sentence in which a nominalization is used, in which a nominalization equals (ADV/TP/PP) NP/VP/S plus the nominalizer de. Structurally, it consists of a subject and the copula verb shì followed by a nominalization, schematized as:

[NP [V TP V NP NOM ]vp ]s

[wǒ [shì qùnián lái měiguó de ]vp ]s

[I [COP last year come to the US NOM]vp ]s

For Li and Thompson, the sentence final de is a nominalizer that is part of the predicate, whereas Hashimoto claimed de was a particle that is not part of the predicate.

Simplex predication analysis is supported by Teng (1979), Huang (1998), Zhang and Fang (2001), Zhu (1997), Choi (2006), etc. Their analysis suggests shì is not a copula verb, but a focus marker that is placed preceding the focused constituent like a phonetic spellout of the focus. Huang (1998) treats shì as a focus operator (FO) having the status of an adverb and not part of the predicate. According to him, (10b) can be schematized as:

[NP [ADV VP ]vp PTCL]s

[wǒ [shì qùnián lái měiguó]vp de]s

[I [FO last year come to the US]vp PTCL]s

Example (6c), repeated here as (12), also treated as a cleft sentence, is taken by

Huang (1998) as evidence to support the simplex predication analysis.

(12) 我是去年来美国 wǒ shì qùniān lái měiguó SG1 FO last year come US ‘It was last year that I came to the US.’

Huang believes example (12) is (10b) with the final particle de omitted. It is a simplex clause with wǒ ‘I’ as the subject followed by the VO predicate lái měiguó ‘come to the US’ that is modified by the adverbial focus operator marking the adverbial time phrase as the contrastive focus.

The debate between the above proposals essentially lies in the syntactic status of shì and de: Hashimoto (1969), Li and Thompson (1981); etc. suggest shì is a copula verb, whereas Huang (1998); etc. claim shì is an adverb in (10b) and (12). Hashimoto; Huang; etc, treat de as sentence final particle, whereas Li and Thompson treat de as a nominalizer.4 Many linguists (Teng 1979; Ross 1983; Chiu 1993; Hsieh 1998; etc.) agree that sentence final de in the cleft can be optionally omitted.

In this thesis, I argue that shì is consistently a copula verb in Modern Chinese and following Li and Thompson (1981) I suggest the cleft construction has the structure consisting of a subject and the copula verb shì followed by a nominalization marked by the nominalizer de. Taking Construction Grammar as the given framework, I suggest that

4 Some scholars such as Shi (1994), Hsieh (1998) claim that the final de in cleft sentences is a perfective aspectual marker equivalent to the perfective marker le.

the cleft construction is a form and meaning pairing, with the form [NP COP NOM] and the specificational meaning with contrastive focus. The cleft construction is a subschema of the more schematic copular construction. In addition, it indicates an assertion with a presupposition asserted by a contrastive focus indicated by the copula shì. The cleft construction has two subschemas: cleft-obj like (10a) and cleft-sbj such as (10b). For the cleft-obj, the subject of the sentence is semantically co-referential with the object of the nominalization, and the final nominalizer de is obligatory; whereas in the cleft-sbj, the subject of the sentence is semantically co-referential with the subject of the nominalization and the nominalizer de is optional. A detailed account of my analysis along with Huang’s (1998) argument and my counter-argument to his analysis will be presented in Chapter 3.

1.2.4 The development of the Chinese cleft

Compared to the abundant literature on the grammatical status of shì and the structure of the cleft, accounts of the development of the cleft are rarely found.

Rather than targeting the emergence of the cleft as the emergence of a construction, some scholars focus on the development of shì as a focus marker. Shi and Li (2001) suggest diachronically the focus marker shì as a particle emerged from the copula shì in

Middle Chinese around 500 CE. They argue that the copula shì was further grammaticalized into a focus marker when it frequently occurred preceding and marking the interrogative wh-words as focus. They assume that around the same period of time, shì as the marker of a wh-word expanded to mark other categorical elements, e.g. NP, VP.

This thesis argues that around 500 CE the copular construction was entrenched and

conventionalized, and the word shì started to function in marking the following wh-words as focus. However, there is no evidence around 500 CE showing shì expanded the focus marking function from wh-words to others such as NPs, VPs. I will show that the examples that Shi and Li find from 500 CE texts to support their claim are simply ordinary copular sentences encoding copulative linking meanings. Therefore, Shi and

Li’s claim that the copula shì further grammaticalized into a focus marker around 500 CE is problematic. I argue that it is the cleft construction as a whole that marks the immediate post-copula element as contrastive focus, and that only along with the emergence of the cleft construction did shì start signaling contrastive focus. This process took place around 900 CE.

Shen (2008:387) gives a synchronic analysis of the formation of the cleft.

Following the cognitive linguistic approach theorized by Fauconnier and Turner (2003),

Shen claims that the cleft is generated through analogy and compounding, particularly through a process that he calls “analogical blending.” The sentence “belongs to a sentence pattern with its own constructional meaning of ‘subjective identity’ that is an emergent meaning as a result of conceptual blending.” He focuses on the cleft such as

(13):

(13) 他是昨天出的医院（的） tā shì zuótiān chū de yīyuàn (de) SG3 COP yesterday leave REL hospital (NOM) ‘It was yesterday that he left the hospital.’

Shen suggests that example (13) is derived from the analogical process in (14):

(14) a. 这是昨天出的病人 zhè shì zuótiān chū de bìngrén this COP yesterday leave REL patient ‘This is the patient that left yesterday.’

b. 他是昨天出的病人 tā shì zuótiān chū de bìngrén SG3 COP yesterday leave REL patient ‘He is the patient that left yesterday.’

x. 这是昨天出的医院  y. (13) zhè shì zuótiān chū de yīyuàn this COP yesterday leave REL hospital ‘This is the hospital that (he) left yesterday.’

In (14), b. and x. derive from a. through a process of analogy5, and y., representing

(13), is generated through the blending of b. and x. Therefore, (13) is derived through the process of “analogical blending.” This results in a structure in which tā is the subject, shì is the copula verb, yīyuàn ‘the hospital’ is the noun predicate, and zuótiān chū de is a modifying clause (or relative clause) modifying yīyuàn. Shen points out that the motivation of “analogical blending” is that the speaker wants to convey some new meaning. As for the particular structure of (13), the motivation is that the speaker wants to express the subjective recognition. It is a way to show the speaker’s empathy with the hearer.

5 Shen does not explain how b. and x. derive through analogy. One reason could be that the demonstrative zhè in a. and the pronoun in b. share definiteness, and bìngrén in a. and yīyuàn in x. are both regular nouns.

Although the pragmatic implicature of conveying new meaning is relevant in enabling a new construction, Shen’s analysis is problematic because he only covers one subtype of the cleft in which the post-copula predicate is an NP modified by a relative clause, with the nominalizer de optional in an ad hoc fashion. This thesis provides a diachronic analysis on the formation and development of the cleft construction in the framework of constructionalization. I propose that the cleft construction [NPi COP

NOMj][SEMi specificational+contrastive SEMj] evolved out of the conventionalized prototype of the specificational copular construction [NPi shì NPj][SEMi specificational SEMj] and the constructionalization process involved host-class expansion, syntactic expansion (recruiting nominalization into the predicate position), and semantic and pragmatic expansion. The development again involved successive micro-steps and therefore was gradual. I suggest that the emergent cleft construction marked the immediate post-copula element as contrastive focus that is signaled by the copula shì. I will discuss the constructionalization process in detail in Chapter 5.

1.3 Data and methodology

This thesis examines the Chinese copular configuration from both a synchronic and a diachronic perspective. My analysis basically consists of theoretical discussions on the topics that I have mentioned above and synchronic and diachronic tests on them. In the part on the synchronic theoretical discussion, I rely largely on examples that are either constructed or taken from the literature. I have chosen to exemplify my discussion in this way because the best way to illustrate the theoretical notions or concepts is to give

prototypical examples. However, since the issues surrounding the data are often complex,

I keep examples brief and choose examples that highlight the relevant features without requiring unnecessary explication. In this part of the thesis, the focus of my discussion is on prototypical copular sentences and cleft sentences.

In the part on the examination of the synchronic distribution and historical development of the copular and cleft construction, I undertake a corpus-based study and make use of data from the searchable Internet version of the CCL Chinese Corpus6 created and managed by Peking University. CCL was built in 2009 and includes data both in Modern Chinese and Classical Chinese. I use these data both to elucidate my discussion and to provide qualitative and quantitative evidence of synchronic distribution and diachronic change. My synchronic analysis informs my diachronic investigation and, in turn, the diachronic evidence is used to support my synchronic account of the copular construction.

The CCL Modern Chinese corpus consists of selective texts from 10 contemporary literary categories including newspaper journals, historical biographies, movies and TV dramas, translation works, Internet articles, dramas, institutional articles and literature works. It covers a variety of literary genres, from casual spoken language in movies and

TV dramas to formal institutional articles. The corpus is primarily written language and the casual spoken language only takes a very small portion (259,506 tokens) about

0.0356% in the corpus. It includes 728,909,261 tokens in total and contains 3,291,508 tokens of shì, which ranks the third most frequent token within the corpus next to de

(11,523,375) and yī (4,140,344). The high frequency of shì shows the importance of the

6 http://ccl.pku.edu.cn:8080/ccl_corpus/index.jsp

usage of shì in Chinese and the significance of the research on how this particular occurrence is used and processed by Chinese speakers and writers.

The CCL Classical Chinese corpus contains a list of Chinese texts from the East

Zhou Dynasty (Spring and Autumn and Warring states periods) (around 500 BCE) to

Republic of China (1911 or so), covering 1059 texts including standard records of history issued by royal family, historical narrations and their commentaries, collective quotes from Masters of a Hundred Lineages, poems and prose, Buddhist and Daoist texts, drama, short stories and philosophy notes. It includes 417,234,865 tokens in total and more than

414,984 tokens of shì, of which approximately 0.18% (about 7,733 tokens) are found in

Old Chinese. As for the periodization of Chinese language, I follow Sun (1996): Old

Chinese (500 BCE-200 CE), Middle Chinese (200 CE-1000), Early Chinese (1001-1900), and Modern Chinese (1900-present).

1.4 An outline of the structure of the thesis

In the present chapter, I have sketched a brief overview of the issues that this thesis addresses, provided some introductory background material and summarized the main arguments that I propose.

In the following chapter, I lay out the theoretical models of Construction Grammar and Constructionalization, the frameworks that I make use of in my account and the theoretical assumptions on which the rest of the thesis is based.

My analysis begins in Chapter 3 with the concepts of “copula” and “cleft.” I propose that shì is the systematic invariant copula verb in Modern Chinese. According to

Construction Grammar, a copular construction is a form and meaning pair that entails a proposition with the semantics of specificational or predicational. I suggest that the

Chinese cleft construction denotes the specificational plus contrastive meaning. The cleft construction, indicating a contrastive focus, can be schematized as [NPi COP

NOMj][SEMi specificational+contrastive SEMj].

In Chapter 4, I examine the historical development of the copular construction. I show that from the perspective of Constructionalization, the topic-comment structure with shì occurring at the initial position of the comment emerged in Old Chinese, and can be said to have been reanalyzed as a subject-predicate construction via analogization to the structure and meaning of the Old Chinese verb wéi ‘to be’. The chapter focuses on the two conditions in which constructionalization tends to take place: 1) the semantic relatedness between the original construction and the target outcome; 2) the morphosyntactic contexts in which the change was enabled. I argue that the process of demonstrative pronoun shì changing into a copula presents its functional change from discourse anaphoric function to syntactic linking function, which also gives rise to the decrease of the instances of the classical copular sentences.7

Chapter 5 comprises a historical investigation into the cleft construction. I make use of historical evidence to provide an explanation for the idiosyncratic characteristics of the cleft. As the copular construction was entrenched and conventionalized in early Middle

Chinese, the cleft construction emerged through host-class expansion (recruiting nominalization at the predicate position), and semantic and pragmatic expansion, where it became more general/schematic and more productive and less compositional.

7 It is generally believed that copular sentences in Old Chinese did not contain any copula verbs and the copulative indication was encoded in the sentence final declarative particle yě.

A summary of the thesis is given in chapter 6 along with my final conclusions and thoughts for future studies.

Chapter 2

Construction Grammar and Constructionalization

In this chapter, I introduce the principles and concepts that are specific to a constructional theory of grammar and a constructional approach to language change. The purpose of this chapter is to outline some of the basic assumptions that underlie this thesis and to explain the theoretical framework that I make use of in the analysis of the Chinese copular constructions. This chapter provides historical background on how the theory of construction grammar was developed and an outline of three constructional approaches to language. I explain the principles behind the different approaches, and draw on a number of insights and theoretical assumptions that are shared by most construction grammar models. I am also concerned with the nature of changes in constructions and present the framework of constructionalization, a construction grammar perspective of the development on constructions over time.

2.1 Construction Grammar

2.1.1 Construction grammar as opposed to modular models

Construction grammar was first developed to account for idiomatic cases in the speaker’s knowledge of a grammar of their language that intrinsically went beyond the capacity of generative grammar. It grew as an alternative to the modular (also called componential)

model of linguistic patterns proposed by theories of generative grammar from the 1960s to at least the 1980s. In a modular model, each component describes one dimension of the properties of a sentence and each type of linguistic knowledge. In other words, the general principle of a modular model is that “each component governs linguistic properties of a single type: sound, word structure, syntax, meaning, use.” (Croft and

Cruse 2004: 226). The components are intended to be highly general rules that apply to all structures of the relevant type, e.g. the rules of the syntactic component apply to all sentences and sentence types, and the same applies to rules for other components. One component is mapped onto another by general linking rules. For this model, lexical items are the only idiosyncratic and item-specific mappings between the components, and there are no idiosyncratic properties of grammatical structures larger than a single word.

Phrases and sentences are governed by the general rules of the syntactic component, semantic and phonological components, plus the highly general linking rules.

Constructions do not have theoretical status in this model; they are purely byproducts of componential meaning and general rules of the grammar that are usually conceived of as constituents or phrases.

For example, (1) is a prototypical copular sentence in Modern Chinese. In a modular model, the Chinese copular construction is a string of item-specific components.

Example (1) consists of three components: NP1 wǒ ‘I,’ the copula shì and NP2 xuéshēng

‘student, ’ and they are put together in the string by the general linking rule which in this case is that the copula must be followed by a predicate.

(1) 我是学生 wǒ shì xuéshēng SG1 COP student ‘I am a student.’

The modular model of grammar is consistent with the generative theory that syntax can be studied independently from meaning and other aspects of function, such as pragmatics and discourse function. Within generative grammar, constructions are simply syntactic configurations (strings); the fact that constructions may encode non- componential meaning is either not recognized or is considered to be outside of the scope of the “core” phenomena. (Patten 2010: 27) Therefore, within this model, the meanings of the three components in (1) are considered individually and mapped onto each other by general rules, but the meaning of them together as a whole is not considered.

However, the syntactic properties of idioms raise a great problem for the modular model of grammar. “Idioms are, by definition, grammatical units larger than a word which are idiosyncratic in some respect.” (Croft and Cruse 2004: 230) Idioms involve the feature of conventionality, i.e. “their meaning or use cannot be predicted, or at least entirely predicted, on the basis of a knowledge of the independent conventions that determine the use of their constituents when they appear in isolation from one another.”

(Nunberg, Sag and Wasow 1994: 492) The fact that some aspects of an idiom cannot be predicted by the general rules of the syntactic and semantic components and their linking rules poses a problem for the modular model. The linguists who proposed the original construction grammar, e.g. Fillmore, Kay and O’Connor (1988), aimed at the problem of idioms, and their analysis became the basis for a new grammatical framework.

Construction grammar treats all aspects of language as the proper objects of linguistic study, and there is neither “core” nor “periphery” knowledge. In this framework, constructions are symbolic form-meaning pairings, just like lexical items that encode idiosyncratic grammatical and semantic properties. The form-meaning mapping is represented as internal to the construction, as opposed to modular models, in which general linking rules map separate components of linguistic knowledge onto one another.

Not only syntax, all aspects of linguistic meaning, including semantics, pragmatics, and discourse function are required for a full account of grammatical knowledge. The model of construction grammar therefore assumes construction specific properties that are irrelevant to the general mapping patterns.

For Croft (2001) and Goldberg (2006), constructions are not only those with idiosyncratic properties, but also include the compositional strings that occur with sufficient frequency. Therefore, in the construction grammar model, (1) is an example of the copular construction, which has the form [NP COP NP] and the meaning that ‘a teacher’ is a property to characterize the subject ‘he.’ Both the form and meaning are internal and specific to the construction.

2.1.2 Three major constructional approaches

There are a number of different approaches that have been proposed with a constructional perspective on language. Overviews of these approaches can also be founded in Croft and

Cruse (2004); Fried and Östman (2004); Langacker (2005); Goldberg (2006); Traugott and Trousdale (2011); etc. The following discussion on constructional approaches is based on Traugott and Trousdale (2011).

Goldberg (Forthcoming) identifies four principles shared by all constructional approaches to language, and one principle (e) shared by most such approaches.

a) The construction is the basic unit of grammar. A construction is a conventional

form and meaning pairing. (Lakoff 1987; Fillmore, Kay and O’Connor 1988;

Goldberg 1995, 2006; etc.)

b) Both form and meaning are internal to construction with semantic structure

directly mapped on to surface syntactic structure, without derivations. (Goldberg

2002; Culicover and Jackendoff 2005)

c) Inheritance hierarchies are the forms that connect and associate constructions into

a network of language. (Langacker 1987; Hudson 1990, 2007; etc.)

d) Cross-linguistic (and dialectal) variation can be accounted for either by “domain-

general cognitive process” (Goldberg forthcoming) or by variety-specific

constructions (Croft 2001; Haspelmath 2008; etc.).

e) Language structure is shaped by language use (Barlow and Kemmer 2000; Bybee

2010; etc.)

Traugott and Trousdale (2011) distinguish three constructional models by referring to them as: Berkeley Construction Grammar, Cognitive Construction grammar, and

Radical Construction Grammar. All the three approaches share the first four principles above. Berkeley Construction Grammar theorized by Fillmore and his colleagues, initially focused on idiosyncratic idioms and expressions by suggesting, “the same analytic tools account for both most basic structures and these ‘special patterns’”

(Fillmore, forthcoming). This approach closely resembles certain formalist theories, in particular Head driven Phrase Structure Grammar, from which a sign-based theory (Sag

2012) is derived. The model takes on syntactic relations and inheritance hierarchies as well; the argument structure is syntactic as well as semantic.

Cognitive Construction Grammar as developed by Goldberg, originally focused on argument structure constructions, such as the English ditransitive constructions, e.g. ‘I gave/sent him a book’, and the way-construction, e.g. ‘He elbowed his way through the crowd’. The argument structure for Goldberg (1995:72) is construed as semantic but linked to syntax. In her 1995 model, she focuses on the patterns that are not predictable from their component parts and defines constructions as pairings of form and meaning where some aspect of the from, or some aspect of the meaning, is not derivable from either the combination of component parts, or from other pre-existing constructions.

(Goldberg 1995:4) In her 2006 book, Goldberg expands the scope of construction to the compositional strings that “are stored as constructions even if they are fully predictable, as long as they occur with sufficient frequency.” (Goldberg 2006:5) That is, constructions are of any shape, from complex clauses to lexical items, to inflectional affixes; “the network of constructions captures our grammatical knowledge of language in toto, i.e. it’s constructions all the way down” (Goldberg 2006: 18, italics and emphasis original).

This model of construction grammar is a usage-based framework, which claims that language is shaped by use, all grammatical knowledge is learned inductively from the input and constructions are “learned pairings of form with semantic or discourse function.” (Goldberg 2006:5)

According to Croft’s Radical Construction Grammar, “a construction is an entrenched routine ‘unit,’ that is generally used in the speech community ‘conventional,’ and involves a pairing of form and meaning ‘symbolic.’” (Croft 2005:274). This constructional model accounts for typological variation in a construction grammar framework. It also adopts the usage-based approach and takes a thoroughly non- reductionist approach to constructions by rejecting autonomous syntactic relations between elements in a construction.

It takes constructions as the basic or primitive elements of syntactic representation and categories are defined in terms of the constructions they occur in. For example, the components of the intransitive construction are defined as intransitive subject and intransitive verb (Vi), whose categories are defined as those words or phrases that occur in the relevant role in the particular construction. Vi is a category in the intransitive construction in English, not in UG. It also “highlights the taxonomic nature of constructional knowledge, the inheritance relationship between more general and more specific constructions, and the importance of language use in determining aspects of language structure.” (Traugott and Trousdale, 2011) Croft (2001) proposes that the form and conventional meaning of each construction are related by symbolic links that are internal to the construction. Thus, each complex construction contains units of form- meaning pairings. In addition to the internal link connecting individual elements to their conventional meanings, there is an additional symbolic link that relates the entirety of the construction’s form to the construction’s conventional meaning. Consequently, even a complex construction is itself a symbolic unit, a linguistic sign. Croft (2001: 18) represents the symbolic structure of a construction given in Figure 2.1.

Figure 2.1 Model of the symbolic structure of a construction in Radical Construction

Grammar (Croft 2001: 18; Croft and Cruse 2004:258)

As shown in Figure 2.1, the construction’s formal characteristics are made up of syntactic, morphological and phonological properties, while its conventional meaning comprises semantic, pragmatic and discourse-functional properties. Radical Construction

Grammar was originally conceived with language change in mind and its principles can be usefully integrated into the theory of constructionaliztion.

This thesis adopts the principles that are shared by the above three constructional accounts, and is most compatible with the views of Cognitive Construction Grammar and

Radical Construction Grammar (I will discuss the formalism in the next section). I treat the Chinese copular construction as a form and meaning pairing with both semantics and syntactic structure internal to the construction. For illustration of the construction, I adopt

Croft’s model of the symbolic structure of a construction in Radical Construction

Grammar as in Figure 2.1, as well as Booij’s (2010) schematization. Booij’s illustration of construction develops out of Croft’s. Instead of using a box, he illustrates a construction in a formula in which form and meaning are presented in brackets connected

by a two-direction arrow. The schematic formula for a construction is: [FORM]

[SEM (MEANING)]. In this thesis, I use this formula to illustrate Chinese copular construction as well as its subschemas.

2.1.3 Some relevant concepts of Construction Grammar

In this section, I introduce the relevant concepts and formalisms that are crucial to the thesis.

2.1.3.1 Taxonomy and inheritance

Construction grammar is an inventory of constructions including words, morphemes, morphological structures, syntactic constructions, etc., but they are not merely an unstructured list in construction grammar. Rather, constructions form a structured inventory of a speaker’s knowledge of the conventions of their language (Langacker

1987, 63-76). Croft (2001) refers to the structured inventory in terms of a taxonomic network of constructions with each construction constituting a node. Any construction with unique, idiosyncratic morphological, syntactic, lexical, semantic, pragmatic, or discourse-functional properties must be represented as an independent node in the constructional network in order to capture a speaker’s knowledge of their language. For example, the substantive idiom [sbj kick the bucket] must be represented as an independent node because it is semantically idiosyncratic. The more schematic but verb- specific construction [sbj KICK obj] must also be represented as an independent node in order to specify its argument linking pattern. Finally, the wholly schematic construction

[sbj Vt obj] is represented as an independent node because this is how construction

grammar represents the transitive clause (Croft 2001:25). Figure 2.2 shows an example of the taxonomic hierarchy of clause types.

schema Sbj Vt Obj

Sbj KICK Obj Sbj other Vt Obj subschema

micro‐construction Sbj kick Obj Sbj kick the bucket

Figure 2.2: An example of the taxonomic hierarchy of construction

As it shows in Figure 2.2, the taxonomic network is hierarchical, in which some constructions are more schematic or general than others and lower level less schematic constructions inherit attributes from higher-level more schematic constructions. Both

Cognitive Construction Grammar and Radical Construction Grammar suggest that the attributes of a dominating higher-level construction are inherited by the lower-level construction, but they also maintain that conflict between constructions from different schematic levels is permitted. When the information specified in inheriting constructions has a conflict with the information specified in more schematic constructions, “the more specific construction ‘wins out’ and inheritance is limited to only non-conflicting information.” (Patten 2010:32) That is to say, partial generalizations are recognized and allowed in constructional inheritance. That is: each constructional category contains some

constructions that are “better” (or more motivated) members, and they are the prototype of the constructional category; and it also contains non-prototypical members which inherit less attributes from higher-level constructions and extend from the prototype. For example, in Figure 2.2, a token of the micro-construction such as ‘he kicked the rabbit’ inherits more attributes of the subschema ‘sbj KICK obj’ and the schema ‘sbj Vt obj,’ than the token of the micro-construction ‘he kicked the bucket.’ Therefore ‘he kicked the rabbit’ is a more motivated member, the prototype of the construction; whereas ‘he kicked the bucket’ is an extension from the prototype, and it is a non-prototypical member. “As prototypical and non-prototypical instances coexist, the speaker forms an inductive generalization (or abstraction) that stipulates only those characteristics that are shared by all of its members.” (Patten 2010:33)

This thesis places the Chinese copular construction in the higher-level schematic node in a constructional taxonomy and I argue that constructions including the cleft construction are subschemas that inherit the attributes of the copular construction. I argue the copular construction contains the prototypical instances that inherits the most information and attributes from the schematic construction, as well as non-prototypical ones that override inheritance from the overarching construction. (See Chapter 3 section

3.4 for details)

2.1.3.2 Coercion

As is shown in the previous subsection, a constructional category contains prototypical members as well as non-prototypical ones. However, constructions as symbolic form- meaning pairs may be extended to accommodate non-prototypical items. That is to say,

although non-prototypical members of a constructional category are not as good and motivated as the prototypical members, they must still inherit some information and interpretation from the schematic higher-level construction. Michaelis (2003: 263) adopts the term ‘coercion’ for “the enriched interpretations that result from this procedure,” and restricts it to enrichment of lexical elements by grammatical ones. In Construction

Grammar, inheritance together with the symbolic nature of the construction account for coercion to take place. If conflicts occur between the construction’s entrenched meaning and the meaning typically associated with a designated lexical item, then the constructional requirements coerce the lexical item to conform to them. In this case, the construction as a whole overcomes the designated lexical meaning of the word. Goldberg

(1995: 158) identifies the conflicts in examples of the English caused-motion construction between the meaning of the verb and the information designated by the construction. She notes that in examples like ‘Joe kicked the dog into the bathroom’, motion is coded by the verb ‘kick’ and the preposition ‘into.’ However, in examples such as ‘Sam squeezed the rubber ball inside the jar’ and ‘Sam urged Bill outside of the house,’ “neither the verbs ‘squeeze’ or ‘urge’ nor the prepositions ‘inside’ or ‘outside’ independently code motion” (Goldberg 1995: 158), rather, the caused-motion construction coerces the “locative term into a directional reading” (Goldberg 1995: 159).

In this case, the construction’s conventional meaning overrides the meanings associated with its components when they are used independently outside of the construction.

Goldberg (1995: 159) comments, “In order for coercion to be possible, there needs to be a relationship between the inherent meaning of the lexical items and the coerced interpretation.” For example, the relationship between the locative prepositional meaning

and the constructional meaning of direction in the caused-motion construction is straightforward; the location given by the preposition is interpreted as the “endpoint of a path to that location” (Goldberg 1995: 159).

This thesis adopts Goldberg (1995)’s idea of coercion and argues that throughout its history the Chinese copular construction is a form and meaning pair that indicates a non- transitory state or situation. The construction contains prototypical members as in [NPi

COP NPj][SEMi copulative linking SEMj] like (1) and non-prototypical members such as those involving a verbal phrase or clause appearing in the subject or predicate position, as in (2) in which the verbal phrase zhǐ pǎo-le yī quān ‘run-PERF one lap’ appears in the predicate position. However, even if the verbal phrase or the clause is aspectually inflected, the copular sentence still denotes a non-transitory state or situation rather than a transitory temporal process. The constructional interpretation overrides what it designates for an aspectually inflected verbal phrase or clause. Coercion occurs at the schema level. (See Chapter 3 section 3.4 for details)

(2) 她的遗憾是只跑了一圈 tā de yíhàn shì zhǐ pǎo-le yī quān

SG3 ASSOC regret COP only run-PERF one lap

‘Her regret is that she has only run one lap.’

2.1.3.3 A usage-based model

As I have shown in 2.1.2, both Cognitive Construction Grammar and Radical

Construction Grammar share that language structure is shaped by language use. Two basic concepts within the usage-based model are the concepts of type frequency and

token frequency. A construction is high in type frequency if it is instantiated by many lexemes, i.e. if it occurs with many different lexical items, and low in type frequency if it only occurs with a few lexical items. Thus, the entrenchment of a more abstract (or more general) schema in the speaker’s inventory is a function of type frequency. On the other hand, a construction is high in token frequency if it is instantiated many times by the same lexical item, and low in token frequency if the construction, together with the lexical item, is infrequent in use. Entrenchment is a consequence of high frequency, either of types or of tokens.

Productivity is assumed to be a consequence of a construction’s high type frequency. The more general a construction is and the more lexical items that instantiate it, the more productive it is, and thus, the more likely it is that the construction attracts new items and that it spreads to other existing items that otherwise fulfill the relevant criteria for occurrence in that particular construction (Bybee 1985, 1995; Goldberg 1995).

Highly entrenched schematic constructions are productive since their type frequency is high. However, high token frequency of an instance does not mean high type frequency and does not contribute to the overall productivity of a schematic construction. Instead, it is only the token that is entrenched, as an independent unit in its own form, and not the higher-level schematic construction.

An important advantage to a usage-based theory of Construction Grammar is that it is able to intersect with and work alongside theories of language change. Changes to the tokens of the micro-construction of the constructional category may have consequences for the higher-level schematic construction in the taxonomy. This type of constructional change that leads to the creation and recruitment of the new constructions and the

reconfiguration of the existing ones proceeds upwards throughout the hierarchical taxonomy. This thesis argues the new occurrence of the cleft construction entered into the schematic specificational copular construction, and therefore gave rise to the increased schematicity of the super schematic copular construction.

A usage-based constructional approach therefore predicts that there are two different types of constructional change: one that is brought about by type frequency and the other that is dependent upon token frequency. Both the two changes involve the process of conventionalization and entrenchment of the schemas. From a constructional perspective of language change, these two different types of frequency are important contributors to the process of constructionalization. Barðdal (2008:176) suggests that low token frequency is associated with analogy and is an important factor for speakers when they extend lower-level constructions, whereas high type frequency is “an indicator of the highest level of schematicity each construction exists, and hence an indicator of the semantic scope of the construction and its productivity domain.” In the next section, I present the major concepts of the framework of constructionalization.

2.2 Constructionalization

Diachronic morphosyntactic change of language is widely accepted to involve grammaticalization. (Hopper and Traugott 2003; etc.) Traditional work on grammaticalization has generally focused on the development of atomic elements, i.e. grammatical markers, or lexical items changing into grammatical items. Recent research has shown that “grammaticalization may be considered as constructional emergence at an increasingly schematic level.” (Trousdale 2010:51) Traditionally, the term “construction”

has been used to refer to a string, a constituent that provides a context for a lexical item to become more grammatical. In Traugott and Trousdale’s (2011) framework of

Constructionalization, however, a construction is understood in the sense of Goldberg

(2006:5): it is a form-meaning pairing in which either the meaning of the whole is not strictly predictable from its component parts or the meaning is predictable from its parts, but which occurs with sufficient frequency for it to be stored as a pattern. (Trousdale

2010:51) Constructionalization is a construction grammar perspective of the development of constructions over time.

“Constructionalization is a process in which new (combinations of) signs are created through a sequence of formal and functional reanalyses. These new signs provide language users with new ways of encoding grammatical or lexical meaning. Minimally constructionalization involves reanalysis in terms of morphosyntactic form and semantic/pragmatic function; discourse and phonetic changes may also be implicated at various stages. Purely formal changes, or purely functional changes, are not constructionalizations. The relevant dimensions for constructionalization are generality, productivity and compositionality.” (Traugott and Trousdale 2011)

Form and meaning pairings subsume subcomponents, and each component can change independently, and therefore there may be many constructional changes on the way to constructionalization, but not all such changes result in constructionalization. “A constructionalization requires a new pairing of both meaning and form to occur, that is each one must be new in some way.” It is symbolized as formnew-meaningnew. (Traugott and Trousdale 2011) In the next subsections, I provide some relevant concepts that are based on Traugott and Trousdale (2011).

2.2.1 Two approaches to grammaticalization

Both grammaticalization and lexicalization are the extensively influential traditional terms that are prior to the development of the constructionalization theory. Many scholars including Traugott who currently focus on constructionalization come from this tradition.

I personally have the same experience. Some issues that I address in this thesis have been observed in the framework of grammaticalization, e.g. the development of copula from

Old Chinese (Pulleyblank 1995; Shi and Li 2001; etc.).

In the framework of Constructionalization, studies focus on the interface between grammaticalization and constructions (Croft 2001; Hilpert 2007; Traugott 2008;

Trousdale 2008; etc.), which is termed grammatical constructionalization. The framework also extends to consider aspects of lexicalization in the history of languages, which is termed lexical constructionalization. Trousdale (2008) shows how constructional approaches to language can account for both grammaticalization and lexicalization within a unified framework with the suggestion that in lexical constructionalization, constructions become less general, less productive, and less compositional, whereas in grammatical constructionalization, constructions become more general, schematic and more productive; yet they also become less compositional.

There have been two different ways in conceptualizing grammaticalization. One involves increase in dependency and reduction of various aspects of the original expression, suggested by Givón (1979); Haspelmath (2004); Heine, Claudi and

Hünnemeyer (1991); and Lehmann (1995); etc. Traugott and Trousdale (2011) call this approach grammaticalization-as-increased-reduction/dependency and abbreviate it as

GIRD. The concept of change in this tradition is primarily limited to morpho-syntactic

form, which does not include categories such as topic, focus and discourse markers, although Lehmann (1995) also refers to semantics, such as bleaching, as one of his

“parameters” of grammaticalization.

In the more recent research, scholars such as Himmelmann (2004) focus on grammaticalization as expansion of host class, semantic-pragmatic, syntactic contexts.

Traugott and Trousdale (2011) term this approach GE (the acronym of grammaticalization as expansion). This approach sees a way of reconciling reduction

(increase in dependency) and functional expansion and allows for reduction in form

(increase in dependency) along with the expansion in function of grammatical categories.

This thesis following Traugott and Trousdale (2011) adopts the GE view and treats grammatical constructionalization as the process of the development of “functional categories,” which involves pragmatic and semantic factors as well as morphosyntactic and phonological ones. The GE view not only explains traditional examples of grammaticalization that involving reduction and increasing dependency, e.g. be going to

‘be in motion in order to V’> be going to ‘future’> be gonna, it also includes less traditional examples such as discourse markers, marked focus construction, which involves expansion of its function range. Moreover, the GE view of constructionalization explains not only the types of changes that involve lexical sources, but also development of grammatical constructions and categories from non-lexical sources such as demonstratives, or to mark information structure. For example, this thesis argues that the

Chinese copula shì was constructionalized from the demonstrative pronoun when it occurred in the topic-comment construction functioning as an anaphor referring to the complex topic. The process of constructionalization involved host-class and syntactic

expansion, e.g. simple nouns occurred in the pre-copula position, semantic and pragmatic expansion, e.g. copulative linking meaning plus a new information structure. (see Chapter

4 section 4.3 for details)

2.2.2 Motivation: analogy and ‘invited inference’

2.2.2.1 Analogy

Fischer (2010) argues that analogy, which is based on both form and meaning, and which constitutes a fundamental cognitive principle, plays a primary role in language acquisition, and also in change. She argues analogies can be very concrete or quite abstract; an analogy may be based on ‘tokens’ (concrete items) as well as schemas or

‘types’ (abstract structures). By appealing to Anttila’s (2003) notion of an analogical grid,

Fischer suggests for analogy both iconic and indexical forces are important. In other words, analogy can operate on not only the paradigmatic (iconic) axis, but also the syntagmatic (indexical) axis. Fischer holds analogy is not only a formal mechanism of change, but also the cause that motivates change. For her, analogy encompasses analogical thinking and any analogical thinking can potentially bring about analogical change.

It is clear that analogical thinking along with categorization is one of the natural cognitive abilities that human entails, and it is an ability people bring to everyday activity. However, analogical thinking (cognitive matching), as an internal motivation for change, does not necessarily lead to analogical change, and therefore is not a cause to change. According to Traugott and Trousdale (2011), there is no causal relation but only

enabling relation between analogical thinking and analogical change. Analogical thinking may predict what might change, how it might change, but they can never cause it to change. Fischer uses the term “analogy” to cover both motivation and mechanism; by contrast, Traugott and Trousdale (2011) suggest the term “analogization” for the mechanism that brings about analogical change. (See 2.3.3.2 for details)

2.2.2.2 Invited inferencing

Semantic bleaching has been considered crucial in grammaticalization. (Lehmann 1985;

Heine and Reh 1984; etc.) Since the 1980s, many scholars have considered the effect that in the case of grammaticalization, semantic bleaching is accompanied by the coding

(semanticization) of pragmatic implicatures. (Traugott and Trousdale 2011) Sweetser

(1988:400) suggests that the lexical semantics of motion is lost along with the semanticization of the implicature of purposive motion into future in the development of future be going to out of motion with a purpose be going to. Traugott (1988:413) also suggests that the speaker adds the meaning of the target domain to the meaning of the word and that the emphasis is on increase “in the direction of explicit coding of relevance and informativeness that earlier was only covertly implied.” Therefore, Traugott and her colleagues hypothesize that most instances of change originate from “invited inferences” or “pragmatic implicatures” (see Grice 1989; Levinson 2000) that come to be semanticized, and both pragmatic and semantic changes precede syntactic/morphological change.

For example, Traugott (2007) discusses a range of examples, such as a lot of and a bit of, which underwent a development from partitive to degree modifier. As partitives, these constructions had meanings similar to a part of or a share of. Since a part of something suggests a quantity, these partitives were associated with quantifiers via invited inferencing or pragmatic implicature. For example, a bit of derives from ‘a bite out of’ and consequently implies a small piece or quantity. The quantity, along with a scalar meaning, as an implicature of the partitive was mapped onto entities and semanticized as the abstract semantics of quantifiers and degree modifiers.

This thesis argues the emergence of the cleft construction was essentially motivated by invited inferencing, e.g. the speakers’ communicative strategies in asserting a presupposition with a contrastive focus. The pragmatic inferencing led to semantic reanalysis (cleft specificational), and was combined with the newly emerged syntactic recruitment of the nominalization [XP de] motivated by analogical thinking modeling

Old Chinese nominalization [XP zhě]. (See Chapter 5 section 5.3 for details)

2.2.3 Mechanisms: reanalysis, analogization and subjectification

While ‘motivation’ has to do with the ‘why’ of change, ‘mechanism’ has to do with the

‘how’ of change. The main mechanisms for language change are usually considered to be reanalysis (the focus here is on difference from the original source8), and analogization or analogical extension (the focus here is on matching of the original source with some extant model). (Hopper and Traugott 2003)

8 The term “reanalysis” is not useful for cases of a child or second language learner has not yet learned a construction and interprets it in a different way from the speaker. Here, “re”-analysis has not occurred, only “different” analysis. (Traugott 2011)

Bybee (2003, 2006) treats frequency as mechanism. For constructionalization, frequency is key to Goldberg and Croft’s view of entrenchment, and entrenchment occurs after innovation when the new form is “being integrated and spread through the system”

(Leech, Hundt, Mair, Smith 2009:269). However, while repetition by members of a language community undoubtedly is a major factor in the fixing, freezing, and autonomizing associated with grammaticalization, frequency itself appears implausible as a mechanism for the onset of grammaticalization. (Traugott 2009) Moreover, mechanisms such as reanalysis and analogization may lead to individual innovations, but do not lead to change unless a community of speakers adopts and conventionalizes that innovation.

2.2.3.1 Reanalysis

Langacker (1977:58) defines reanalysis as “change in structure of an expression or class of expressions that does not involve any immediate or intrinsic modification of its surface manifestation.” There are two types of reanalysis: one is “resegmentation,” i.e. boundary loss, boundary creation, boundary shift; the other one is “syntactic/semantic reformulation,” i.e. re-bracketing, re-categorizing, and re-patterning. All examples of reanalysis involve changes in constituency (re-bracketing of elements in certain constructions), and reassignment of morphemes to different semantic-syntactic category labels. For example: be going to from [be+ main verb+ progressive aspect + to] to a future tense marker: [be going] [to V]> [[be going to] V]> [Auxiliary V].

Hopper and Traugott (2003:39) argue that reanalysis is the primary mechanism that leads to grammaticalization. Grammaticalization and reanalysis are distinct but intersect: grammaticalization cannot be realized without reanalysis, but it is not limited to reanalysis. Traugott and Trousdale (2011) mention one of the objections against the claim that reanalysis and grammaticalization are closely linked has been that reanalysis is abrupt but grammaticalization is gradual. Gradualness involves small steps in change, whereas in early generative work “reanalysis was associated not only with abrupt, discrete changes, but also large macro-parametric steps, saltations, or even ‘catastrophes’

(Lightfoot 1979, 1991).” However, “given current theories of micro-parameters, syntactic feature, reanalysis is not longer construed as saltation. It can be associated with gradualness, in the sense of micro-steps.” The changes are discrete, but step-wise, micro- step by micro-step, not “catastrophic saltations.” Constructionalization incorporates notions of both gradualness (diachronic) and gradience (the synchronic result of gradualness) (Traugott and Trousdale 2010).

This thesis argues that the emergence of the copular construction [(XPi) COP XPj

(PTCL)][(SEMi) copulative linking SEMj (Declarative)] is a process of grammatical constructionalization that involved reanalysis of the topic-comment construction [(XPi)

[shì XPj PTCL]][(Topici) [Anaphor SEMj Declarative]] through re-bracketing and re-categorizing, whereby the demonstrative pronoun shì that functioned as the anaphor referring to the topic phrase evolved into the copula shì. (See Chapter 4 section 4.3 for details)

2.2.3.2 Analogization

In the framework of constructionalization, analogy involves analogical thinking as one of the motivations and analogization (analogical change or analogical extension) as one of the mechanisms. As I have shown in 2.2.2.1, analogical thinking is one of the enabling factors that allow change, though it does not necessarily bring about change, whereas analogization works as mechanism that can give rise to new structures. At every stage, any language enables a set of options that speakers can take, and the options can be very unpredictable. Analogical thinking may predict what might have the potential to change; yet it may or may not further eventually enable analogical change. When analogical change occurs, analogization takes place. When analogization happens, it is simultaneously reanalysis. “All analogizations are instances of reanalysis, because each case of analogization involves a slight restructuring of what the speaker or hearer knows about a particular expression.” (Traugott and Trousdale 2011) Therefore, in the process of change, analogical thinking may precede analogization, and analogization occurs as reanalysis. All analogization (analogical changes) involve reanalysis, but not all reanalysis is motivated by analogical thinking. However, analogization and reanalysis as change mechanisms may lead to individual innovations, but do not lead to change unless they are frequently used and conventionalized by a community of speakers.

Analogization involves analogical match, which means the changing construction is matched to an extant exemplar through the process of analogization. Fischer (2010) claims any analogical match allows change. The analogical match can be very loose both with tokens on the concrete level and structures on the abstract level, as long as there is an analogical thinking that links two tokens or structures together. Therefore for Fischer,

in terms of change, the analogical model is very broad. For example, for be going to,

Fischer argues its changing into a future auxiliary takes other clausal patterns of this type, i.e. auxiliary-verb patterns such as the other future pattern I will go, as the iconic analogical model. Another example she gives is: in the process of the verbal adjunct i.e. undoubtedly, surprisingly, changing into pragmatic marker in English, the verbal adjunct appearing at the sentence initial position is analogized by the ellipted clausal phrases

(reduced modal clauses) that were used as a separate or independent phrase preceding the main proposition and with scope over this proposition (Fischer 2010:15).

I disagree with Fischer in that I believe analogical thinking does not cause change

(Fischer 2010:31), and pre-existing analogical model should not be very loose to allow change. In other words, the match that allows analogization should be constrained, i.e. the analogical model and the source construction should at least partially share the similar morphosyntactic environment and constraints. This is because analogical thinking is such a powerful mental action that can almost link everything in the world together. Traugott and Trousdale (2011) suggest in constructionalization, analogization involves a new structure with at least one subcomponent of meaning and one of form matched to another extant construction. For example, the existent English binominal quantifier a bit of/a lot of may have been taken as the analogical model for a shred of, because a shred of has at least one subcomponent of meaning, i.e. quantifier, and one of form, i.e. a … of, matched to a bit of/a lot of. One could argue that a shred of changing into a quantifier, just like a bit of/a lot of, underwent a process of reanalysis (rebracketing, boundary shift) motivated by semantic mismatch in speakers’ use, without analogization involved. However, from a usage-based perspective intending to avoid the suggestion that a construction or the

language itself is doing something, we can assume that a shred of should not have been developing by itself, but speakers might have well matched it to something, possibly a lot of/a bit of. This thesis argues the emergence of the copular construction [(XPi) COP XPj

(PTCL)][(SEMi) copulative linking SEMj (Declarative)] underwent a process of analogization modeling the structure of the full transitive verb wéi: [(XPi) wéi XPj

(PTCL)][SEMi BE SEMj (Declarative)] in Old Chinese. (See Chapter 4 section 4.3 for details)

2.2.4 Constructional taxonomies

In the constructional network, it is the relationships that exist between constructions that are crucial in any account of variation and change. Constructions display different degrees of schematicity; constructions lower in the taxonomy inherit properties form those higher in the taxonomy; and constructions intersect with each other at different levels in the taxonomy. This intersection of constructions is relevant not only to issues of mismatch in grammaticalization (Traugott 2007), but also to perceptions of synchronic gradience, which emerge as the result of gradual grammaticalizaiton. (Trousdale 2010)

Traugott (2007) proposes a hierarchy of levels of schematic constructions:

a. Macro-constructions: highly abstract, schematic constructions

b. Meso-constructions: fairly abstract constructions, subschemas that have similar

semantics and/or syntax

c. Micro-constructions: individual construction types

d. Constructs: instances of micro-constructions, tokens of actual use, the locus of

change

All four constructional levels are of importance in various stages of the grammaticalization process. (Trousdale 2010) The first three levels are abstract types, as distinct from actual utterances. The last one is data-points, or ‘tokens.’ Macro- constructions of a given taxonomy at the superordinate level are highly differentiated from other superordinate categories; by contrast, meso-constructions and micro- constructions at the subordinate level have high internal similarity. “The most salient level of categorization is normally referred to as the basic-level -- micro-level: it is at this level of the categorical taxonomy that ‘the largest amount of information about an item can be obtained with the least cognitive effort’ (Ungerer and Schmid 1996: 68).”

(Trousdale 2010:5)

Traugott and Trousdale (2011) modified the hierarchical taxonomy of schematic constructions proposed by Traugott (2007). In the new version, they retain the lower constructional levels of micro-construction and construct, and adopt the term of schema and subschema for the original macro- and meso- constructions. They suggest that macro- and meso-constructions are the mental representations of the linguistic notions of schemas, whereas schema is a more tangible concept and provides the notion of construction as a form and meaning pairing. In this thesis, I follow Traugott and

Trousdale (2011) and adopt the constructional taxonomy as:

a. Schemas: abstract, schematic constructions

b. Subschemas: less abstract, schematic constructions

c. Micro-constructions: individual construction types

d. Constructs: instances of micro-constructions, tokens of actual use, the locus of

change

2.2.5 Constructionalization dimensions

A constructionalization perspective sees both grammatical constructionalization and lexical constructionalization as directional with generality/schematicity, productivity and compositionality as the parameters. For grammatical constructionalization, it is as expansion on syntagmatic and paradigmatic dimensions (including host-class expansion) and reduction or increase dependence in the form. Therefore, the dimensions for grammatical constructionalization are increased generality, schematicity, productivity, and decrease in compositionality (Trousdale 2010; Traugott and Trousdale 2011).

Increase in generality involves semantic bleaching that indicates less restricted semantics and allows more collocation. According to Bybee and McClelland (2005: 391), collocational freedom leads to increase in the type frequency of a construction. Increase in productivity is associated with constructional expansion on different levels. For example: as for the binominal strings such as a lot of, a bit of and a shred of, increase in generality occurs on the level of micro-construction, e.g. a shred of was extended from collocates with partitive properties (a lot of, a bit of); it also occurs at the abstract, schematic level: the binominal construction became generalized and could be employed in a much larger set of discourse contexts including those which would be incompatible with their original meaning. Hoffmann (2005) makes the similar point by taking complex prepositions as examples. Increase in generality on the schematic level is linked with the increase of schematicity. Grammatical constructionalization is also characterized by decrease in compositionality. According to Traugott and Trousdale (2011), decrease in

compositionality is decrease in the transparency of the match between meaning and form and decrease in compositionality is gradient and gradual.

2.2.6 Constructionalization and constructional changes

As I have shown in the beginning of 2.2, constructionalization requires both form and meaning change, and a new form and meaning pairing to occur. Purely formal changes, or purely functional changes are not constructionalization. They are constructional changes. When the semantic or syntactic subcomponent of a construction changes independently, constructional change occurs.

Traugott (2012) convincingly argues that constructionalization is distinct from constructional changes in that constructional changes are language change including pre- constructionalizaiton changes, constructionalization, and post-constructionalization changes as in Figure 2.3.

pre‐ post‐ constructionalization constructionalization constructionalization

Figure 2.3: Constructional changes related to constructionalization

Other than those changes, constructional changes also involve systemic change, e.g. great vowel shift; phonology alone change, e.g. loss of rhoticity; syntax alone change, e.g. word order change; morphology alone change, e.g. clitic>inflection; semantic alone

change, or change in frequency use. However constructionalization is a subset of constructional changes in which morphosyntactic new form and meaning pairing is created through a sequence of small steps in which form and meaning are reanalyzed. All constructional changes including constructionalization are gradual processes involving many micro-steps.

Constructional changes also incorporate the traditional grammaticalization of GE and GIRD. Most instances of GE and GIRD occur in constructionalization and post- constructionalization. Some instances also occur in pre-constructionalization, e.g.

Diewald’s (2006) critical context and Heine’s (2002) bridging context.

2.3 Summary

In this chapter, I have focused on the major issues and principles that are crucial to the theories of construction grammar and made it clear that the framework I adopt in this thesis is consistent with the theories put forward by Croft (2001), and Goldberg (2006).

This chapter also focuses on the theoretical framework of constructionalization (Traugott and Trousdale 2011), a construction grammar perspective on the development of constructions over time.

In the following chapters, I focus on the structure and function of the Chinese copular construction within the framework of construction grammar and the constructionalization process of the copular construction and its subschemas coming into being in the history of Chinese.

Chapter 3

A Copula Analysis of Shì in The Chinese Cleft

Construction

3.1 Introduction

“Copula” is a Latin word meaning a connection, a link. It is generally used to refer to the uses of the English verb ‘to be’ and its equivalent in world languages. The copula in

Modern Chinese is shì, and typically occurs with predicative nominal, exemplified by xuéshēng in (1).

(1) 我是学生 wǒ shì xuéshēng SG1 COP student ‘I am a student.’

Example (1) is a typical copular sentence in Chinese with a subject followed by the copula shì and a nominal predicate. It has the form [NP COP NP] and the copulative linking meaning, specifically the post-copula predicate ‘a student’ is the property to characterize the subject ‘I.’ As for the information structure, (1) encodes predicate informational focus, which means the subject ‘I’ is the topic encoding given information,

and the post-copula nominal predicate xuéshēng ‘a student’ indicates new information that is the informational focus (see section 3.5 for details).

Non-nominals can also occur in either the subject position or the predicate position in a copular sentence. For example, in (2a), the verbal phrase kàn diànshì ‘watch TV’ appears in the predicate position, and a clause tā qù Shànghǎi ‘he goes to Shanghai’ in the subject position of (2b). Furthermore, as in any Chinese sentence, the subject can be a zero as in (2c).

(2) a. 我的爱好是看电视 wǒ de àihào shì kàn diànshì SG1 ASS hobby COP watch TV ‘My hobby is to watch TV.’

b. 他去上海是老板的决定 tā qù Shànghǎi shì láobǎn de juéding SG3 go Shanghai COP boss ASS decide ‘His going to Shanghai is boss’s decision.’

c. 是我的错 shì wǒ de cuò COP SG1 ASSOC fault ‘It is my fault.’

The examples in (1) and (2) suggest that the Chinese copular sentences appear to have the structure [(XP) COP XP], in which XP can be nominal or non-nominal. My search of copular sentences with copulative linking meaning in CCL Modern Chinese

Corpus demonstrates that 87.6% (4380 sentences out of 5000) of the copular sentences

are [NP COP NP], whereas occurrences such as those in (2) only take up 12.4%. The outcome of the survey confirms that [NP COP NP] is the prototypical structure of a

Chinese copular sentence.

Following Li and Thompson (1981), I consider example (3) to be a copular sentence consisting of a subject, the copula shì, and a nominalized predicate marked by the nominalizer de. It is also commonly known as shì…de construction or the cleft construction in Chinese linguistic literature (Li and Thompson 1981; Hashimoto 1969;

Teng 1979; Paris 1979; etc.). In a sentence like (3), the linguistic form immediately after shì i.e. zuótiān ‘yesterday,’ constitutes the key element of the contrastive focus.

(3) 她是昨天去上海的 tā shì zuótiān qù shànghǎi de SG3 COP yesterday go to Shanghai NOM ‘It was yesterday that she went to Shanghai.’

There has been an extensive debate among Chinese linguists on whether shì is a copula when it occurs in cleft sentences. Lü (1979); Zhu (1982); Chao (1968); Hashimoto

(1969); Li and Thompson (1981) among others, suggest that shì is simply a verb in

Modern Chinese. Thus, it functions as a copula verb in cleft sentences. However, other linguists such as Teng (1979); Zhu (1997); Shi (1994); Choi (2006), claim shì in cleft sentences is not so. The major reason for the debate is the co-existence of examples resembling (4) with those like (3). Comparing the two sentences in (3) and (4), we see that, in spite of the absence of de at the end of (4), they are essentially the same. Based on this observation, some linguists (Teng 1979; Huang 1998; etc.) believe examples such as

(3) and (4) share the simplex predication: [NP FM VP (PTCL)] and shì in cleft sentences is not a copula verb, but a focus marker. Huang (1998:213) claims that shì in cleft sentences is a focus operator having the status of an adverb.

(4) 她是昨天去上海 tā shì zuótiān qù shànghǎi SG3 COP yesterday go to Shanghai ‘It was yesterday that she went to Shanghai.’

This chapter, based on a cross-linguistic understanding of the concept of copula, argues for a systematic treatment of shì in Modern Chinese as a copula verb. I define the

Chinese copula shì as an invariant non-inflectional verb that co-occurs with certain lexemes when they together form a predicate. In light of the theory of Construction

Grammar, I treat the copular construction as a form and meaning pairing: [(XPi) COP

XPj][SEMi copulative linking SEMj] with the prototype form [NP COP NP]. The copulative linking involves two subschemas: specificational and predicational. Huang’s simplex predication treatment of the cleft construction and treating shì as an adverb are problematic as he fails to consider the fact that the cleft construction entails the specificational meaning rather than a transitory event or process that the simplex predication may encode. I propose that shì is a systematic invariant copula verb in

Modern Chinese, and the cleft construction is complex predication that can be

9 schematized as [NPi COP NOMj] (NOM=(ADV/TP /PP) VP/S/NP de)[SEMi specificational+contrastive SEMj]. “Complex” here is understood as a structure that

9 TP= time phrase

involves a subordinate nominalization (NOM). Semantically, the cleft construction is specificational with contrastive, which involves a restricted non-referential set encoded by the nominalization entering into a class-membership relation with a referential subject and the immediate post-copula element encoding contrastive focus.

The chapter is structured as follows: section 3.2 explains the syntactic concept of

“copula;” section 3.3 discusses the semantics of the Chinese copula; in section 3.4, I provide a constructional schematic taxonomy for the prototypical copular construction.

Section 3.5 discusses the concept of “cleft,” introduces the adverb analysis proposed by

Huang (1998) along with my counterarguments to it, and proposes my analysis on the cleft construction; section 3.6 is the conclusion.

3.2 The syntactic concept of “copula”

A common definition of “copula” found in dictionaries, as well as in the linguistic literature, is as follows:

(5) A copula is “a word that links a subject and a predicate.” (Narahara 2002: 16)

The terms “subject” and “predicate” need to be specified to fully understand the above definition of “copula.” In the generative literature, subject is the category that occupies the specifier position of IP ([Spec, IP]), and the predicate is the projection of a lexical category that assigns a theta-role to an argument. According to a typical definition in some functional descriptive linguistics, the subject of a sentence is the “phrase that has a ‘doing’ or ‘being’ relationship with the verb in that sentence.” (Li and Thompson 1981:

87) The predicate is the part of a sentence containing what is said about the subject. “The word ‘predicate’ is a functional term in opposition with the functional term ‘subject,’ … it commonly refers to the function of a verb phrase…. A predicate, however, is not necessarily a verb phrase.” (Li and Thompson 1981: 140) In terms of a copular sentence, traditionally, the term “predicate” has a broad sense, which consists of the copula verb and the post-copula phrase, and a narrow sense that excludes the copula and only indicates the post-copula phrase. The familiar terms such as nominal predicate, adjectival predicate, and verbal predicate all apply to the narrow sense of “predicate” (Kahn 1973,

Wang 1937). Jespersen (1924) introduced the term “predicative” to apply to predicates in this narrow sense to avoid ambiguity. However, most linguists find that once recognized, the ambiguity is harmless, and they keep using the term “predicate” for both the narrow and broad senses. In my discussion, following the tradition, I use the term “predicate” to signify either copula plus post-copula phrase or post-copula phrase itself.

The definition in (5) implies that in a copular sentence, the copula links up the subject and the predicate. However, the definition does not characterize the grammatical category of “copula,” and the grammatical relations among “subject,” “copula,” and

“predicate” are unclear. Radford (1997), applying the minimalist approach, modifies the definition in (5) and defines “copula” as:

(6) A verb used to link a subject with a non-verbal predicate.

The definition in (6) explicitly identifies “copula” as a verb and suggests that the predicate (the broad sense) of a copular sentence consists of both a copula verb and a

non-verbal predicate (the narrow sense). Radford further defines “verb” in certain languages as:

(7) A category of word that has the morphological property that can carry a range

of inflections including past tense.

To specify the category of the predicate as non-verbal in (6) amounts to saying that the function of a copula as a predicate maker is added to the elements that normally do not form predicates on their own. The following examples present the occurrences of the

Standard English copula.

(8) a. This is a cup. b. *This a cup

(9) a. The cups are full. b. *The cups full.

(10) a. The cup is on the desk. b. *The cup on the desk

(11) a. *He was break the cup. b. He broke the cup.

The above examples show that in Standard English, nominal, adjectival and prepositional phrases cannot function as predicates on their own but must be combined with a copula in the predicate position, as exemplified in (8)-(10). On the other hand,

verbal phrases that function as a predicate on their own are not compatible with a copula, as in (11).

Hengeveld (1992) considers the copula to be meaningless, “semantically empty,” and a mere carrier of inflectional features for predicate phrases. Stassen (1997:66) proposes the Dummy Hypothesis and argues that the idea underlying the hypothesis is that the copula is basically a “hat-rack” for categories of verbal morphology. This idea coincides with Lyons’s who claims that the principal function of the copula verb ‘to be’ in Russian, Greek and Latin is to serve as the locus in surface structure for the marking of tense, mood and aspect. (Lyons 1968: 322)

Radford’s definition in (6) and Hengeveld, Stassen and Lyons’ proposals unravel the syntactic functions of a copula:

a. A copula functions as a linker between subject and non-verbal predicate.

b. A copula functions as a syntactic “hat-rack” to attach tense and other verbal

inflectional features to a clause that contains a non-verbal predicate.

c. A copula functions as a predicate marker that is added to lexemes that do not form

predicates on their own.

Two interpretations can be inferred from the above discussion:

a. First, a predicate in which verbal inflectional categories are coded should never

contain a copula.

b. Second, all copulas should carry verbal inflectional categories.

In the following discussion, however, I will show that the Chinese copula is incompatible with both of these two interpretations.

In standard Modern Chinese, nominal phrases cannot function as predicate on their own and can only be combined with a copula in predicate position, as is exemplified in

(12). Other than nominal phrases, adjectival, prepositional and verbal phrases can function as a predicate on their own, as in (13).

(12) a. *我学生10 wǒ xuéshēng SG1 student

b. 我是学生 wǒ shì xuéshēng SG1 COP student ‘I am a student.’

(13) a. 她很漂亮 tā hěn piàoliàng SG3 very pretty ‘She (is) very pretty.’

b. 她在北京 tā zài Běijīng SG3 at Beijing ‘She (is) at Beijng.’

10 In colloquial spoken Chinese, (12a) can be acceptable with a prosodic pause between wǒ ‘I’ and xuéshēng ‘student.’ In fact, the copula can always be dropped in non-standard spoken Chinese as well as many languages. However, in standard or written Chinese, (12a) is ungrammatical.

c. 她打了小王 tā dǎ-le Xiǎowáng SG3 hit-PERF Xiaowang ‘She hit Xiaowang.’

Examples in (12) show that the copula is added to phrases that do not form a predicate on their own. In Chinese, the adjectives are treated as intransitive verbs (Li and

Thompson 1981: 141) and prepositions are considered as coverbs (Li and Thompson

1981: 356); both of them belong to the category of non-nominal and can form a predicate on their own.

According to Radford’s definition of “verb” in (7), the non-verbal predicates are those that do not have the morphological property of inflection, the non-inflecting category. In English and other European languages, nominal, adjectival and prepositional phrases are not subject to tense, modal, and aspect inflections, and they are referred to as the non-inflecting categories here. Since only the non-inflecting categories occur in the predicate position of a copular sentence, the copula’s function of linking subject and non- verbal predicate holds. However, in some languages, Chinese included, not only nominal predicates can occur in copular sentences, non-nominal predicates with coded verbal inflectional categories such as aspects can also be found following the copula. The following examples illustrate the difference:

(14) *What makes him happy is has been to Beijing.

(15) a. 我的爱好是看电视 wǒ de àihào shì kàn diànshì SG1 ASSOC hobby COP watch TV ‘My hobby is to watch TV.’

b. 她的遗憾是只跑了一圈 tā de yíhàn shì zhǐ pǎo-le yī quān SG3 ASSOC regret COP only run-PERF one lap ‘Her regret is that she has only run one lap.’

In (15a&b), the predicates of the copular sentences contain the copula and a verbal phrase. In (15b), the verbal predicate has a perfective marker attached to it. The fact that the inflected verbal category is also found in the predicate position of a copular sentence marks a significant distinction between the European and Chinese copular sentences.

In Standard English, the copula’s “hat-rack” function holds, as the copula verb ‘to be’ supplies tense and other verbal inflectional categories to the clauses that have nominal, adjectival and prepositional phrases as the predicates. However the Chinese copula never carries any verbal morphology; and therefore it is invariant. The following

English and Chinese examples demonstrate the difference:

(16) a. I am his student. b. I was his student last year.

(17) a. 我是他的学生 wǒ shì tā de xuéshēng SG1 COP SG3 ASSOC student ‘I am his student.’

b. 去年我是他的学生 qùnián wǒ shì tā de xuéshēng last year SG1 COP SG3 ASSOC student ‘I was his student last year.’

In (16a), the copula was expresses past tense, while in (17b), there is no past tense inflection attached to the copula shì. The sense of past in Chinese is coded only by the temporal phrase qùnián ‘last year’ in (17b). As for (16b), if the temporal phrase ‘last year’ is not specified, the clause is still past with the past tense marker in was. However, if qùnián ‘last year’ is not present in (17b), the sentence does not encode the meaning of the past.

(18) 他是 *了/*过/*着老师 tā shì *-le/*-guò /*-zhe lǎoshī SG3 COP PERF/EXP/IMP teacher

In Chinese a typical verb can be morphologically marked to denote different aspects, such as the verb pǎo ‘run’ in (15b). As an invariant non-inflectional copula verb, shì is not subject to any Chinese aspect markers in (18), be it perfective -le, experiential - guò, or imperfective -zhe.

The asymmetry between (14) and (15), as well as that between (16b) and (17b) suggests that the definition in (6) and its derived functions of “copula” are not cross- linguistically applicable. Therefore, I propose the following definition for the Chinese copula:

(19) Chinese copula shì is an invariant non-inflectional verb that co-occurs with certain lexemes whereby they together form the predicate of a copular sentence.

Semantically, it functions to signal either a predicational or a specificational meaning.

Based on the definition (19), in (17a), wǒ ‘I’ is the subject, the copula shì marks the noun phrase tā de xuéshēng ‘his student’ as a nominal predicate, and they together function as the predicate of the sentence. In (15a), the copula shì co-occurs with the verbal phrase kàn diànshì ‘watch TV,’ and they together form the predicate of the sentence. In the following, I will discuss the basic semantic functions of the Chinese copula shì.

3.3 The semantics of copula

As stated in the previous section, Hengeveld (1992) considers the copula to be meaningless, ‘semantically empty,’ a mere carrier of inflectional features for predicate phrases. Stassen (1997) proposes that the copula is a dummy, and does not contain any meaning. Pustet (2003:5) also points out that “a copula does not add any semantic content to the predicate phrase it is contained in.”

However, many linguists hold that although copula itself is a semantically null verb, copular sentences express a variety of copulative meanings. Blom and Daalder (1977) propose that copular sentences in the languages of the world fall into two semantic classes based on the relationship between the objective information encoded in the subject and predicate, which are often referred to as “predicational” and “specificational”

(Declerck, 1986: 2).

A predicational sentence predicates a non-referential property of the referential subject. “Predicational” has also been called “attributive” (Gundel 1977; Lyons 1968;

Halliday 1970), “characterizational” (Kuno and Wongkhomthong 1981; Quirk et al.

1985), “ascriptive” (Kahn 1973), etc. Higgins (1979: 214) suggests that predicational is

“being about” something. The following examples can be said to be predicational:

(20) a. Mary is a student. b. The cup is full.

The predicate in (20a), ‘a student’ signals the characteristic of Mary ‘being a student,’ and in (20b) ‘being full’ is a description of the cup.

The semantics of the specificational relationship is not as straightforward as that of the predicational relationship. The term “specificational” has been adopted in a broader sense for “identificational” (Kuno and Wongkhomthong 1981; Quirk et al. 1985),

“equative” (Halliday 1970b; Huddleston 1971; Kahn 1973), etc. Higgins (1979) argues that specificational sentences function like lists: the subject of a specificational sentence acts as the heading of the list and the post-copula elements serve as items on that list.

Higgins suggests that specificational sentences involve a “value-variable” relation. He notes, “the heading of a list provides a ‘variable’, thereby delimiting a certain domain, to which the items on the list conform as ‘values’ of that variable” (Higgins 1979: 155).

Higgins (1979: 214) argues that specificational sentences do not involve a predication relationship, since “The whole notion of being ‘about’ something is alien to a list”.

Likewise, Declerck (1986: 2) defines a specificational sentence as one whose semantic function is to specify a value for a variable.

Patten (2010), however, aiming at the definition of specificational meaning, argues that the “value-variable” relation is purely information-structural and is not the product of the semantic contribution of its components, consequently “the definition of specificational meaning as a “value-variable” relationship cannot help us identify what distinguishes specificational (identifying) copular sentences from predicational

(descriptive) copular sentences.” (Patten 2010:64) She proposes that specificational meaning is the product of a special type of nominal predication relation. Unlike predicational sentences, specificational sentences involve a restricted set (existentially presupposed or asserted) that enters into a class-membership relation with a referential expression. In other words, the crucial characteristic for creating specificational meaning is that a copular sentence denotes a universally quantified restricted or existentially presupposed set, which is inherent to the semantics of definite noun phrases, and a referential member that specifying the set. Example (21) presents what Patten calls the canonical specificational sentence:

(21) The best student we have is Sally.

The subject in (21), the definite NP ‘the best student we have,’ denotes a restricted, quantified set, which is paired with a referential nominal predicate ‘Sally.’

(22) a. The one who stole the money is Bill. b. Mr. Obama is the president of the United States.

(23) Mr. DuPont is my father.

The two examples in (22) are both specificational, which specifies a referential member for a non-referential and restricted set. ‘Bill’ is a referential member of the restricted set ‘the one who stole the money;’ ‘Mr. Obama’ is the referential member of the restricted set ‘the president of the United States.’ However, (23) is equational, in which both of the subject and the predicate are semantically referential and encode a one to one class-membership predication relation. I treat equational as a sub-class of specificational relationship. The major difference between an equational and a specificational copular sentence is that in equational sentences the subject and the predicate are both semantically referential, whereas a specificational sentence specifies a referential member for a non-referential restricted set.

Similarly, Chinese copular sentences can also be classified into the two semantic copulative categories: specificational and predicational. Examples in (24) are all specificational. (24a&b) are specificational sentences, in which the subject and the nominal predicate form a class-member relationship. (24c) is equational with a referential subject and predicate.

(24) a. 我做的是这个 wǒ zuò de shì zhè -ge SG1 make NOM COP this CL ‘What I made is this one.’

b. 奥巴马是美国总统 àobāmǎ shì měiguó zǒngtǒng Obama COP US president ‘Obama is the president of the US.’

c. 那个人是我妈 nà ge rén shì wǒ mā that CL person COP my mother ‘That person is my mother.’

Examples in (25) can all be said to be predicational, and to predicate a property or characteristic of the subject.

(25) a. 她是黄头发 tā shì huáng tóufà SG3 COP yellow hair ‘She has yellow hair.’

b. 她是个学生 tā shì ge xuéshēng SG3 COP CL student ‘She is a student.’

3.4 The constructional framework

Above I have discussed the syntactic functions of a copula and the semantics of copular sentences. Linguists of different theoretical traditions in general recognize the two categories of copular sentences: specificational and predicational, and the question arises whether the two categories involve one or two different copulas. Halliday (1967); Quirk and Greenbaum (1973); Kahn (1973); etc. believe the dichotomy of the semantics of copular sentences comes from there being two ‘be’s, which have different syntactic

functions. In Montague grammar (Montague 1973; Dowty et al. 1981; Partee 1999), a distinction is made between the two ‘be’s: the ‘be’ of predication and the ‘be’ of specificity (equality, identity and specificity). They are distinguished from each other by the types of their arguments. The ‘be’ of the predicational takes two arguments of type

, a subject entity, and type , the returned truth value when offered its subject entity. However, the ‘be’ of the specificational takes two arguments of type , two entities. Stowell (1989:255) considers the predicative ‘be’ to be a raising verb11, and the specificity ‘be’ is “a two-place predicate conveying a relation of identity holding between two referential NPs.”

Within the framework of Construction grammar, Croft (2001:266) suggests: a copular construction

“[p]rofiles the assertion classifying the subject as belonging to the category of the predicate nominal, or possessing the property of the predicate nominal. However, the copula verb itself is of minimal semantic content, adding only a predicative function to a maximally schematic categorization of the referent of the subject argument. It can be argued that the profile of the whole clause is determined partly by the copula and partly by the predicate noun/adjective— categorization of the subject referent as being of the type profiled by the predicate noun, or ascription of the property profiled by the predicate adjective.”

I suggest that there is only one copula involved and the dichotomy of the semantics of copular sentences is captured by the two subschemas under the schema -- the copular

11 By verb raising it means verb movement to V, which has been posited for infinitival verbs in languages e.g. English, German and Dutch. The hypothesis is that the verb of an infinitival complement, if the complement is not extraposed is moved and adjoined to its governing verb, thereby creating a verb-cluster. Dutch Verb Raising creates the structure in (ib) (assuming the SOV d-structure in (i)a). (Evers 1975; Rutten 1991; etc.)

(i) a. dat Jan [VP [VP hard werken1] willen2] heeft that Jan hard work want-to has b. dat Jan [VP [VP hard t1] t2] heeft willen2 werken1 that Jan has wanted to work hard

construction. In this thesis, the Chinese copular construction is treated to be a form and meaning pairing, which has the form [(XP) COP XP] with [NP COP NP] as the prototype and denotes copulative linking meaning involving specificational or predicational. Even if non-nominals occur at the XP position of the copular construction, the construction’s conventional meaning overrides the meanings associated with verbal phrases or clauses that denote temporal transitory process, and therefore it still denotes a proposition with the meaning of specificational or predicational. In other words, the meaning of non- transitory states or situations that can be described or specified is coerced out from the copular construction even if verbal phrases or clauses that occur in the subject or predicate position are temporally and aspectually inflected. The prototypical copular construction can be schematized as: [NPi COP NPj][SEMi copulative linking SEMj].

I propose a constructional schematic taxonomy for the prototypical Chinese copular construction:

[NPi COP NPj]<‐‐> [SEMi copulative linking SEMj] Schema

[NPi COP NPj]<‐‐> [SEMi speciicational SEMj] [NPi COP NPj]<‐‐> [SEMi predicational SEMj] Subschema

[NPi COP NPj]<‐‐> [SEMi [NPi COP NPj]<‐‐> [SEMi [NPi COP NOMj]<‐‐> [SEMi equational SEMj] speciicational SEMj] speciicational+contrastive SEMj] Sub subschema

Figure 3.1: The constructional schematic taxonomy of the prototypical Chinese copular construction

In the constructional schematic taxonomy of the prototypical Chinese copular construction, [NPi COP NPj][SEMi copulative linking SEMj] is the abstract schema, under which [NP COP NP] with specificational and predicational meanings are two less abstract schemas. I categorize three sub subschemas under the schema of [NPi COP

NPj][SEMi specificational SEMj]: [NPi COP NPj][SEMi equational SEMj],

[NPi COP NPj][SEMi specificational SEMj], and the cleft construction [NPi COP

NOMj][SEMi specificational+contrastive SEMj]. In terms of information structure, both equational and specificational copular sentences encode informational focus in which the subject is the topic encoding given information, and the post-copula predicate as a whole indicates new information that is the informational focus; whereas the cleft sentences indicate contrastive focus encoded by the immediate post-copula element. In the next section, I will discuss the concept of cleft.

3.5 The concept of cleft

3.5.1 The cleft construction

The cleft construction is a subtype of the copular construction. A cleft sentence is a complex sentence in which a simple sentence is expressed using a main clause and a subordinate clause. (Quirk at el. 1985; Levinson 1983; etc.) Traditionally, the English cleft includes two sup-types: it-cleft and pseudo-cleft (all/what/the-cleft).

In English, a prototypical it-cleft sentence has the form [It COP NP RC12]13 as in

(26).

(26) It is John who grimaced.

In an English pseudo-cleft sentence, the subordinated clause is a free relative clause headed by what, all, or a relative clause headed by the one, as in (27).

(27) a. The one who grimaced is John. b. What John did is grimace. c. All John did is grimace.

The English cleft is often said to involve an element (the immediate post-copula element) encoding contrastive focus and the relative clause functioning as presupposition.

(Prince 1978, etc.) Lambrecht (2004:52) defines presupposition as the set of propositions lexicogrammatically evoked in a sentence that the speaker assumes the hearer already knows, or is ready to take for granted, at the time the sentence is uttered. Focus is the semantic component of a pragmatically structured proposition whereby the assertion differs from the presupposition. (Lambrecht 2004:213) Scholars have proposed many versions of subclasses of focus and the most common one is to classify focus into two subtypes according to whether or not the focused item is in contrast with other alternatives in a limited set. Many terminologies have been used to refer to the

12 RC is the abbreviation for relative clause. 13 The modern day English clefts allow a range of categories to occur as the complement of the copula ‘be’ (Patten 2010:222), i.e. the position of NP in [It COP NP RC]. Prepositional and adverbial phrases are common in the post-copula position, as in ‘It was just here that we met,’ ‘It is in December that she’s coming;’ plus the gerund form of verb phrase in ‘It is writing the paper that he did last night.’

noncontrastive type and the contrastive type focuses, such as rheme vs. kontrast (Vallduvi and Vilkuna 1998), information focus vs. identification focus (Kiss 1998), informational focus vs. operational focus (Roberts 1998) and informational focus vs. contrastive focus

(Xu 2002). This thesis adopts Xu’s terms of informational focus and contrastive focus to highlight their functions.

Contrastive focus not only asserts what is different from its presupposition; it is also associated with exhaustiveness and exclusiveness as proposed by Kiss (1998).

Accordingly, in (26), the post-copula NP ‘John’ is the focus complementing the presupposition ‘someone grimaced.’ Moreover, ‘John’ is exhaustively the all and only one exclusively that grimaced; therefore it is a contrastive focus.

Payne (1997: 280) suggests that cross-linguistically “a cleft construction is a type of predicate nominal consisting of a noun phrase (NPi) and a relative clause (RC) whose relativized NP is co-referrential with NPi;” and “clefts in many languages exhibit the pattern [NPi [COP headless RCi]].” Within Payne’s cross-linguistic formulation of cleft, the form that RC takes depends on what relativization strategies the language employs, i.e. it could be nominalization, a participial clause, or a more prototypical relative clause.

According to Li and Thompson (1981: 579), nominalization is the equivalent relativization strategy in Modern Chinese. They note that nominalization is a grammatical process to turn a VP/S/NP into a noun14. In Construction Grammar, the nominalization construction (NOM, in short) can be schematized as [VP/S/NP de][entity, situation, or state].

14 The copula shì as well as copular clauses cannot be nominalized, e.g. *[shì xuéshēng] de ‘[COP student] NOM.’

Therefore, the equivalent of a headless RC in Chinese is a nominalization, and the cleft construction can be schematized as [NPi COP NOMj][SEMi specificational

+contrastive SEMj], in which the post-copula NP is a nominalization marked by the nominalizer de.

Patten (2010: 103) suggests that English cleft sentences belong to a family of specificational copular sentences, in that they also involve the restrictive but non- referential set in common with definite noun phrases, which is given, or recovered, in the form of a relative clause. In other words, English cleft sentences are copular sentences in which the post-copula NP is identified as the referential member for the restricted and non-referential NP (in the form of relative clause). What is more, the restricted set encompasses not just objects, but also actions, and properties, and therefore the referential member encoded in the post-copula element can range over clauses, VPs as well as NPs. (Patten 2010:261)

I have suggested the Chinese cleft construction also belongs to the family of specificational copular sentences, because it also involves a restricted, non-referential set denoted by the post-copula nominalization specified by a referential member encoded by the subject NP. In terms of information structure, just like in English, the Chinese cleft construction also encodes contrastive focus. The element immediately following the copula is the contrastive focus asserting an idea that is presupposed. For example, in

(28a), the contrastive focus is the adverbial time phrase zuótiān ‘yesterday,’ and the presupposition in the context is ‘she went to Shanghai at some time.’ Similarly, in (28b), the contrastive focus is encoded in the element right following shì, e.g. zài jiā ‘at home.’

If a verb phrase directly follows the copula in a cleft sentence, as in (29), the contrastive

focus can be the verb qù ‘go,’ the complement of the verb Shànghǎi, or the verbal phrase as a whole qù Shànghǎi ‘go to Shanghai,’ depending on the context.

(28) a. 她是昨天去上海的 tā shì zuótiān qù Shànghǎi de SG3 COP yesterday go Shanghai NOM ‘It was yesterday that she went to Shanghai.’

b. 这个是在家做的 zhè-ge shì zài jiā zuò de this CL COP at home make NOM ‘It was at home that this was made.’

(29) 她是去上海的 tā shì qù Shànghǎi de SG3 COP go Shanghai NOM ‘It was Shanghai that she went.’

I suggest examples (28a&b) and (29) are cleft sentences in Modern Chinese, as all the examples have the form [NP COP NOM] (NOM= (ADV/TP/PP) VP/S/NP de) with specificational+contrastive meaning.

However, some scholars e.g. Teng 1979; Zhu 1997; Huang 1998, distinguish (28b) from (28a) and (29), because they believe that the Chinese cleft is not reversible (Huang

1998：211), whereas (28b) indicates equation and therefore the subject and the predicate can be reversed ((30) is the reversed version of (28b)). They believe both (28b) and (30) share the identical equational meaning and they are Chinese pseudo-cleft sentences.

Here, reversibility is taken by these scholars as a criterion to distinguish pseudo- cleft from cleft. However, it is always a problem if it is theoretically correct to use linguistic reversibility to describe the linguistic aspects of analysis and of synthesis as it may put unrealistic constraints on linguistic descriptions.

(30) 在家做的是这个 zài jiā zuò de shì zhè-ge at home make NOM COP this-CL ‘What was made at home is this one.’

I suggest all cleft and pseudo-cleft sentences are not equational but specificational, and cannot be reversed. (28b) is distinct from (30) which is a pseudo-cleft in Modern

Chinese in that they have different forms and indicate different meaning. (28b) has the form [NP COP NOM] and the post-copula PP zàijiā ‘at home’ is the contrastive focus.

However, (30) has the form [NOM COP NP] and the post-copula NP zhè-ge ‘this one’ is in focus. Although the Chinese pseudo-cleft also includes exhaustiveness and exclusiveness (Prince 1978; Higgins 1979; Quirk et al. 1985; etc.), e.g. zhè-ge ‘this one’ indicates exhaustiveness and exclusiveness in (30), and specificational member-class relationship (Patten 2010), it has the form [NOM COP NP], which differs from the cleft

[NP COP NOM]. The Chinese pseudo-cleft is not the focus of this thesis.

I suggest that (28b) is a cleft rather than a pseudo-cleft because it has the form [NP

COP NOM], entails the specificational meaning and encodes contrastive focus. Both

(28a&b) are examples of the cleft construction [NPi COP NOMj][SEMi specificational+contrastive SEMj], however, they are also different in terms of semantic

referentiality and the optionality of the nominalizer de. In (28a), the subject of the sentence tā is co-referential with the implicit subject of the nominalization, and the final nominalizer de is optional. By contrast, in (28b) the subject of the sentence zhè-ge is co- referential with the implicit object of the nominalization, and the nominalizer de is obligatory. Therefore, within the constructional framework, I propose that the schematic cleft construction has two subschemas: cleft-sbj as in (28a) and cleft-obj as in (28b). The cleft-sbj involves subject-subject co-referentiality, and optionality of the nominalizer de.

The cleft-obj involves subject-object co-referentiality which means the verb in the nominalization cannot be intransitive, moreover, the obligatory nominalizer. Therefore the constructional schematic taxonomy of the Chinese cleft construction can be elaborated as Figure 3.2:

[NPi COP NOMj]<‐‐> [SEMi speciicational Schema +contrastive NOMj]

[NPi COP NOMj]<‐‐> [SEMi cleft‐ [NPi COP NOMj]<‐‐> [SEMi cleft‐ Subschema obj SEMj] sbj SEMj]

Figure 3.2: The constructional schematic taxonomy of the Chinese cleft construction

3.5.2 An adverb analysis of shì in the so-called shì cleft sentence

Huang (1998) claims cleft sentences do not have the meaning of an ordinary copular sentence: “they neither indicate equation or inclusion, nor do they predicate property as ordinary copular sentences may do.” (Huang 1998: 213) Therefore, he argues that shì in

cleft sentences cannot be treated as a copula verb. He provides a syntactic analysis to argue that the cleft shì has the status of an adverb. I will discuss this argument in detail in the present section. According to Huang, (31) are cleft sentences with the final particle de omitted.

(31) a. 小李是昨天打了小王 Xiáolǐ shì zuótiān dǎ-le Xiǎowáng Xiaoli COP yesterday hit-PERF Xiaowang ‘It was yesterday that Xiaoli hit Xiaowang.’

b. 小李昨天是打了小王 Xiáolǐ zuótiān shì dǎ-le Xiǎowáng Xiaoli yesterday COP hit-PERF Xiaowang ‘It was hitting Xiaowang that Xiaoli did yesterday.’

c. 是小李昨天打了小王 shì Xiáolǐ zuótiān dǎ-le Xiǎowáng COP Xiaoli yesterday hit-PERF Xiaowang ‘It was Xiaowang that hit Xiaoli yesterday.’

Huang (1998: 213) claims that shì in cleft sentences has “the status of an adverb on a par with negation and modals.” He argues, in English cleft sentences such as (32), what immediately follows the copula is taken as the focus of the sentence ‘John,’ with the rest of the sentence ‘that hit Bill’ backgrounded, i.e. presupposition. There is a structural dependency between the focus and a position within the presupposed clause. The structure of (32) is shown in (33).

(32) It is John that hit Bill.

(33) It is Johni [that ti hit Bill]

Therefore, Huang says, there is a legitimate value-variable relationship between the focus and the gap in the presupposition. However, he claims the situation with the

Chinese cleft is quite different. In Modern Chinese, “a cleft sentence differs from a non- cleft only in the presence vs. absence of the focus indicator shì. There is no overt dislocation of the focus.” (Huang 1998:214) Example (34) is a non-cleft declarative version of (31):

(34) 小李昨天打了小王 Xiáolǐ zuótiān dǎ-le Xiǎowáng Xiaoli yesterday hit-PERF Xiaowang ‘Xiaoli hit Xiaowang yesterday.’

Huang suggests the simplest way of looking at Chinese cleft formation is to say that it inserts the marker shì directly in front of the constituent in focus, as exemplified in (31).

Chinese cleft sentences, “unlike their English counterparts, involve neither structural dependency, nor the value-variable relation between the focus and the presupposition.”

(Huang 1998: 205) He further claims that treating the cleft shì as a copula verb on a par with ordinary copula shì or the pseudo-cleft shì as in (28b) is both semantically implausible and syntactically problematic. Semantically, cleft sentences like (31) do not indicate equation, identification, inclusion or predicative property. Syntactically, one

important restriction on the cleft formation is that no post-verbal phrase may be clefted, e.g. Xiǎowáng in (31), therefore (35) is ungrammatical.

(35) *小李昨天打了是小王 Xiáolǐ zuótiān dǎ-le shì Xiǎowáng Xiaoli yesterday hit-PERF COP Xiaowang ‘It was Xiaowang whom Xiaowang hit yesterday.’

If the cleft shì is an adverb, since the position of an adverb, e.g. negation, modals, time adverbials, in Chinese is pre-verbal, “it is of course the case that shì can never occur post-verbally between a verb and its complement.” (Huang 1998: 215)

Huang also maintains that the cleft shì may enter into scope relations15 with negation and modals. In (36), following Huang, shì is glossed as FO (focus operator).

(36) a.小李是明天不去 Xiáolǐ shì míngtiān bú qù Xiaoli FO tomorrow NEG go ‘It is tomorrow that Xiaoli will not go.’

b. 小李不是明天去 Xiáolǐ bú shì míngtiān qù Xiaoli NEG FO tomorrow go ‘It is not tomorrow that Xiaoli will go.’

15 The General Scope Principle (Kroch 1974: 145) suggests: “if within a simplex sentence there are operators with the surface word order X Y Z…, then the operators are indexed in order of appearance, giving X1 Y1 Z1…, and a scope marker is established as follows: [X1’ Y1’ Z1’…] where X is a quantifier of type X’.”

c. 是小李可能明天去 shì Xiáolǐ kěnéng míngtiān qù FO Xiaoli possibly tomorrow go ‘It is Xiaoli who will possibly go tomorrow.’

d. 可能小李是明天去 kěnéng Xiáolǐ shì míngtiān qù possibly Xiaoli FO tomorrow go ‘Possibly it is tomorrow that Xiaoli will go.’

According to Huang, (36) shows the focus operator shì, the negation operator bù and the modal operator kěnéng enter into scope relations with each other in free order.

The fact that shì may enter into scope relations with negation and modals, also with other adverbs suggests shì is simply another such quantificational adverb which has the property of bearing scope. (Huang 1998: 216)

Finally, Huang suggests that although the cleft shì has the status of an adverb, it is closely related to the copula verb shì used in copular sentences including the pseudo-cleft shì. He points out that verbal phrases such as yòng lì ‘use force’ in (37a) can function adverbially in (37b):

(37) a. 她用了力了 tā yòng -le lì le SG3 use-PERF force CRS ‘She has used force.’

b. 她用力打小王 tā yòng lì dǎ Xiǎowáng SG3 use force hit Xiaowang ‘She hit Xiaowang with force (forcefully).’

Similarly, shì is a copula verb in copular sentences, but in cleft sentences, it functions as a focus operator and has the status of an adverb.

In the following I will demonstrate that Huang’s arguments on cleft shì being an adverb fail to stand up on further scrutiny. Instead, shì is systematically a copula verb in these sentences. As I have shown in section 3.5.1, sentences like (31) are examples of the cleft-sbj [NPi COP NOMj][SEMi cleft-sbj SEMj], in which the subject NP is co- referential with the subject of the nominalization, therefore it involves the semantic co- reference. Furthermore, it entails the specificational plus contrastive meaning with the referential subject specifying the non-referential but restricted set encoded by the post- copula nominalization, and hence it encodes a legitimate value-variable relation. In fact, as I will show in the next section, the examples in (31) are instances of the cleft-sbj with the optional sentence final nominalizer de.

First, Huang points out that no post-verbal phrase may be clefted, as in (35).

However, many linguists including Zhang and Fang (1996) point out that in Chinese, not only can adverbs not separate the verb and its complement in Chinese, no verb with an unrelated argument structure can occur in a verbal phrase between the verb and its complement as well, e.g. *dǎ chī Xiǎowáng ‘hit eat Xiaowang.’ Following the general constraint, as a verb, shì cannot occur between the verb and its object in (35).

Second, Huang claims cleft shì may enter into scope relations with negation, modals, and quantificational adverbs in free order (see 36), thus shì is simply another such adverb that has the property of bearing scope. However, examples in (39) show that, if we substitute shì in (36) with a regular verb such as zhīdào ‘to know,’ the grammaticality of the sentences holds.

(39) a.小李知道明天不去 Xiáolǐ know míngtiān bú qù Xiaoli know tomorrow NEG go ‘Xiaoli knows that (he) will not go tomorrow.’

b. 小李不知道明天去 Xiáolǐ bú know míngtiān qù Xiaoli NEG know tomorrow go ‘Xiaoli does not know that (he) will go tomorrow.’

c. 知道小李可能明天去 zhidao Xiáolǐ kěnéng míngtiān qù know Xiaoli possibly tomorrow go ‘(someone) knows Xiaoli will possibly go tomorrow.’

d. 可能小李知道明天去 kěnéng Xiáolǐ zhidao míngtiān qù possibly Xiaoli know tomorrow go ‘Possibly Xiaoli know that (he) will go tomorrow.’

Verbs are generally not considered as operators (unless they serve as auxiliaries) and do not enter scope relations with operators. The fact that shì shows up in one

sentence with negation or modal does not necessarily mean that it bears scope relations with them. It could still be a verb, like zhīdào ‘to know’ in (39).

Finally, the scope phenomenon of the cleft shì that is considered by Huang to be like an adverbial verbal phrase such as yòng lì ‘use force’ in yòng lì dǎ Xiǎowáng ‘hit him with force (forcefully)’ as in (37) is called into question by (40). Note that the sentence in (40) shows that the same verb yòng in the same position may actually be the verb of a sentence that has a serial-verb construction. It then follows that verbal phrase yòng lì in (37) is not used adverbially but may be the verb in a serial-verb construction.

Consequently, the evidence to support the claim that shì is a verb in copular sentences, but an adverb in the cleft is questionable.

(40) 她用了很大的力打小王 tā yòng -le hěn dà de lì dǎ Xiǎowáng SG3 use-PERF very heavy REL force hit Xiaowang ‘She has used heavy force to hit Xiaowang.’

In light of the data in (38-40), it has become clear that none of the arguments for treating shì as an adverbial focus operator shows that it is uniquely adverbial. It follows then that shì can be analyzed as being systematically a copula verb. Moreover, the adverb analysis of shì in the cleft implies the simplex structure [NP ADV VP (PTCL)] that mainly indicates a transitory temporal process rather than a non-transitory state or situation. However, as I have pointed out in 3.5.1, the cleft construction indicates specificational non-transitory states or situations. Accordingly the adverb analysis is not consistent with the semantic nature of the cleft. I will elaborate this point in 3.5.3.

3.5.3 My analysis of Chinese cleft sentences

The examples in (41) lay bare the function of shì as a non-inflectional invariant verb in connecting two non-transitory situations. In the sentences of (41a&b) a transitory process qù ‘go’ marked by an adverb dōu ‘all,’ can be negated by either negator, bù or méiyou.

However, preceding the copula shì, only one negator bù is possible.

(41) a. 我们没有都去 wǒmen méiyǒu dōu qù PL1 NEG all go ‘Not all of us have gone.’

b. 我们不都去 wǒmen bù dōu qù PL1 NEG all go ‘Not all of us will go/went.’

c. *我们没有是去 wǒmen méiyǒu shì qù PL1 NEG COP go

d. 我们不是去 (的) wǒmen bú shì qù (de) PL1 NEG COP go (NOM) ‘We are not those who will go/went.’

This is so because the Chinese méiyǒu is a negator of a process in terms of its temporal structure. However, the Chinese bù negates a non-transitory state. For example,

in (42), none of the adjectives, hǎo ‘good,’ cōngming ‘smart,’ or gāo ‘tall,’ indicating a non-transitory state can be negated by méiyǒu.

(42) a. 他不好，不聪明，不高 tā bù hǎo , bù cōngming, bù gāo SG3 NEG good, NEG smart, NEG tall ‘He is not good, not smart, or not tall.’

b. 他*没有聪明， *没有高 tā *méiyǒu cōngming, *méiyǒu gāo SG3 NEG smart, NEG tall

Such a pattern in negation confirms my hypothesis that the cleft construction inherits the syntactic and semantic properties from the schematic copular construction, and it denotes a specificational non-transitory situation or state rather than a transitory process.

The examples in (43) and (44) illustrate the cleft-sbj with the explicit nominalizer de and the cleft-sbj with the implicit16 nominalizer de respectively.

(43) 她是昨天到达北京的 tā shì zuótiān dàodá Běijīng de SG3 COP yesterday arrive Beijing NOM ‘It was yesterday that she arrived in Beijing.’ (CCL)

16 Since the nominalizer de in the cleft-sbj is optional, it can be at present or not in a cleft-sbj sentence. I treat de at the present as explicit, and de not at the present as implicit.

(44) 我是到北京学习来了 wǒ shì dào Běijīng xuéxí lái-le SG1 COP to Beijing study come-CRS ‘It was to Beijing that I have come to study.’ (CCL)

In example (43), the post-copula contrastive focus zuótiān ‘yesterday’ asserts the exhaustiveness and exclusiveness with respect to the presupposition ‘she arrived in

Beijing someday’ in the context. In (44), the contrastive focus, the PP dào Běijīng ‘to

Beijing,’ is asserted of the presupposition ‘I came somewhere to study.’ The form of (44) is just like (43) except that de is implicit. I hypothesize that (43) and (44) are cleft-sbj sentences with the final nominalizer de optional.

I found 2760 tokens of the cleft-sbj sentences in CCL Modern Chinese Corpus, of which 2122 (76.9%) share the form [PRO shì (ADV/TP/PP) VP de], and 638 (23.1%) are those in which the final nominalizer de is not present. This shows that the rate of the nominalizer de being implicit is far less than that of those with it. According to the frequency, [PRO shì (ADV/TP/PP) VP de] is the prototypical structure of the cleft-sbj.

The prototypical cleft-sbj involves the subject as the doer of the event encoded by the nominalization, e.g. (43)-(44).

A copular sentence in (45a) can be used grammatically only in a highly specific context, i.e., speaking to a waiter in a restaurant clarifying who has ordered what. It is inconceivable that mǐfàn ‘rice’ serves a predicational function, as it is not a property of human being. Therefore, it interprets only the specificational meaning, because a copula, cross-linguistically, is observed to represent either predicational or specificational function. It is, therefore, reasonable to consider (45a) to be a short form for a cleft-sbj

sentence in (45b), in which only mǐfàn ‘rice,’ the contrastive focus, is kept after dropping the entire presupposition, ‘something that I have ordered.’

(45) a. 我是米饭 wǒ shì mǐfàn SG1 COP rice ‘?? I am rice.’

b. 我是点米饭的 wǒ shì diǎn mǐfàn de SG1 COP order rice NOM ‘It is rice that I have ordered.’

The structurally ambiguous sentence in (46a) further illustrates this point. If we take the head noun of the post-copula RC representing the restricted set, i.e., ‘she is the daughter who was born last year,’ then the referential member tā ‘she’ in subject position is a member of the set of the daughter who was born. But then where does the second reading come from, as the seemingly common referential member of tā in (46a) can have a mother reading?

(46) a. 她是去年生的女儿 tā shì qùnián shēng de nǚ’er SG3 COP last year bear REL daughter ‘She is the daughter who was born last year.’ ‘It was last year that she gave birth to a daughter.’

b. 她是去年生的女儿的 tā shì qùnián shēng de nǚ’er de SG3 COP last year bear REL daughter NOM ‘It was last year that she gave birth to a daughter.’

The sentence in (46b) with the nominalizer de at present unambiguously gives the mother reading. Example (46b) with the form [PRO shì RC NP de] is obviously a cleft- sbj sentence, in which the subject tā ‘she’ represents the referential member of the non- referential, but restricted set of ‘the one who gave birth to a daughter last year’ encoded by the nominalization. The immediate post-copula contrastive focus qùnián ‘last year’ is asserted of the presupposition ‘she gave birth to a daughter some time.’ The sentence in

(46a) is structurally ambiguous only because the nominalizer de in (46b) is optional, thus appearing in more or less the same fashion as the one in (46a). This ambiguity has further substantiated my hypothesis that the Chinese cleft-sbj has a [NP COP NOM] form and the ambiguity is caused by the optionality of the nominalizer de. Therefore, shì must be a copula verb and similarly the examples Huang provides in (31) are simply cleft-sbj sentences with the implicit final nominalizer de.

3.6 Conclusion

This chapter has given discussions of the concepts of ‘copula’ and “cleft.” I argue that shì in Modern Chinese is systematically an invariant non-inflectional copula verb, not a particle, a focus marker, a focus operator or an adverb in the cleft construction. The copular construction can be schematized as [(XPi) COP XPj][SEMi copulative

linking SEMj], in which although the XP can be non-nominal, its prototypical form is

[NP COP NP]. I have proposed a constructional schematic taxonomy of the prototypical

Chinese copular construction (see Figure 3.1).

I argue that the cleft construction [NPi COP NOMj][SEMi specificational

+contrastive SEMj] is a subschema of the copular construction, which inherits the attributes of the schematic copular construction. Figure 3.3 summarizes the form and meaning properties of the copular and cleft construction, and provides a comparison between them.

• SY: [NP COP NP] • SY: [NP COP NOM] (NOM=(ADV/TP/PP) VP/S/NP de)

• SM: speciicational • SM:speciicational • PR: topic (pre‐COP NP); • PR: topic (pre‐COP NP); informational focus (post‐COP NP) contrastive focus (the immediate post‐COP element) • DI: contrastive

the copular construction the cleft construction

Figure 3.3: A comparison of the schematic copular construction and the cleft construction

(form in the upper box and meaning in lower box. SY: syntax PH: phonology SM: semantics PR: pragmatics DI: discourse)

100

The cleft construction indicates the specificational plus contrastive meaning, and the element immediately following shì is the contrastive focus. Under the schematic cleft construction, there are the cleft-sbj and cleft-obj. The cleft-obj involves subject-object co-referentiality and the obligatory presence of nominalizer; whereas the cleft-sbj involves subject-subject co-referentiality, and the optionality of the nominalizer de, which overrides the general optionality of de, as nominalization is a construction from which [NP COP NOM] inherits. The result of my search of cleft-sbj sentences in CCL

Modern Chinese Corpus reveals that in actual discourse less than a quarter of them have the implicit nominalizer de, and [PRO shì (ADV/TP/PP) VP de] is the prototype. This chapter also calls into question the validity of the adverbial treatment of shì in some cleft- sbj sentences, as none of the arguments given so far shows it is uniquely characteristically adverbial. The primary semantic function of the cleft-sbj is to indicate a non-transitory class-membership rather than a transitory process that the adverbial analysis may imply. Furthermore, I have demonstrated the extra mileage provided by the copula hypothesis as it can provide a natural explanation of the somewhat idiosyncratic sentences in (45) and (46).

101

Chapter 4

The Constructionalization of Shì in Chinese: from a

Demonstrative Pronoun to a Copula

4.1 Introduction

As indicated in Chapter 3, the copula shì occurs in a copular sentence, a special type of declarative sentence traditionally termed “pànduàn jù” in Chinese linguistic literature, which indicates definition, people’s belief or judgment. It is used to link the subject of a sentence with a predicate (see Chapter 3 section 3.2 for details). For example:

(1) [他是老师] [tā shì lǎoshī] [NP COP NP] 3SG COP teacher ‘He is a teacher.’

Example (1) has the syntactic structure [NP COP NP]: tā is the non-agent subject, lǎoshī is the nominal predicate, and the copula shì links them together. The prototypical function of the copula in Modern Chinese is to link two NPs. However it is generally known that copular sentences in Old Chinese (500 BCE - 200 CE) need not contain any copula verbs, as exemplified in (2) and (3):

102

(2) [仲尼日月也] [zhòngní rì yuè yě] [NP NP yě] zhongni sun moon PTCL ‘Zhongni (is) the sun and the moon.’ Lunyu (400 BCE)

(3) [政者正也17] [zhèng zhě zhèng yě] [NP VP yě] politics NOM upright PTCL ‘Governors (are) upright.’ Lunyu (400 BCE)

Examples (2) and (3) are typical copular sentences in Old Chinese. In this chapter, I call them classical copular sentences (CCS, hereafter). Example (2) has the syntactic structure [NP NP yě], zhòngní is the subject, rì yuè is the nominal predicate, yě is a sentence final particle. There is no copula verb connecting the subject and the nominal predicate in either (2) or (3).

Also found in Old Chinese texts are sentences of the topic-comment construction

(Li and Thompson 1977) [XPi [Sj]][Topici [COMMENTj]], such as examples (4) and (5). In (4) and (5), the comment clause appears to be a CCS. Scholars (Wang 1937;

Feng 1993; Shi and Li 2001; etc.) generally agree that shì in (4) and (5) is a demonstrative pronoun functioning as an anaphor referring to the preceding topic phrase.

17 Adjectives in Chinese share the properties with verbs, i.e. they take aspectual inflections. I treat adjectives as stative verbs. (See Chapter 3 section 3.2 for details)

103

Accordingly, example (4) has the syntactic structure [NP [shì NP yě]] with the demonstrative pronoun shì referring to the topic NP fù yú guì and there is no linking verb between the subject shì and the nominal predicate rén zhī suǒ yù. Example (5) has the syntactic structure [S [shì VP yě]].

(4) [富与贵 [是人之所欲也]] [fù yú guì [shì rén zhī suǒ yù yě]] [NP [shì NP yě]] wealth and nobility this people ASSOC thing want PTCL ‘Wealth and nobility, these (are) the things people want. Lunyu (400 BCE)

(5) [如弃德不让 [rú qì dé bú rang if abandon moral NEG yield [是废先君之举也]] [shì fèi xiān jūn zhī jǔ yě]] this abolish former emperor ASSOC behavior PTCL [S [shì VP yě]] ‘If (you) abandon the moral and don’t yield, it (is) abolishing the behavior of the former emperor.’ Zhuozhuan (500 BCE)

There are still some CCS occurrences in Modern Chinese that share the structure of

(2) (3) and the comment part of (4) and (5), but they are confined to the casual spoken language. For example:

104

(6) [今天星期五] [jīntiān xīngqī wǔ] [NP NP ] today Friday ‘Today is Friday.’

(7) [她长头发] [tā cháng tóufà] [NP NP ] 3SG long hair ‘She (has) long hair.’

Except in oral spoken language, the copular construction in standard Modern

Chinese requires the copula shì, linking the subject and the predicate as exemplified in

(1).

Regarding the formation and development of the copula shì, many scholars have proposed theories and hypotheses, but to date, this issue has not been adequately accounted for. This chapter, aims at the emergence and development of the copula shì and the copular construction, arguing that, in light of Traugott and Trousdale’s theory of constructionalization (2011), the emergence of the copular construction involved reanalysis of the topic-comment construction [(XP)18[shì XP yě]] (as in examples (4) and

(5)), whereby the demonstrative pronoun shì that functioned as the anaphor referring to the topic phrase evolved into the copula shì through a process of analogization modeling the structure of the full transitive verb wéi: [XP wéi XP] in Old Chinese. This chapter focuses on the two conditions in which the grammatical constructionalization took place:

18 In this chapter, XP=NP/VP/S. I will explain this more in section 4.3.

105

1) the semantic relatedness between the original construction and the target outcome; 2) the morphosyntactic contexts in which the change was enabled. Furthermore, I argue the demonstrative pronoun shì changed into a copula in the process of the emergence of the copular construction. This is a functional change from discourse anaphoric to syntactic linking function, along with the decrease of the instances of CCS, such as are exemplified in (2) and (3). Section 4.2 outlines the major previous studies on the development of shì.

Section 4.3 presents a detailed analysis on the constructionalization process of shì and the copular construction. Section 4.4 is a typological analysis regarding the issue and the conclusion.

4.2 Previous research on shì

The most frequently cited theory on the development of shì is proposed by Wang (1937).

He argues that copula shì developed from the demonstrative pronoun shì (as in examples

(4) and (5)) and it did not occur until late Western Han (206 BCE - 25 CE) and early

Eastern Han (25 CE - 220). Hong (1958) challenges Wang’s theory and proposes that the copula shì evolved from the affirmative response shì, as exemplified in (8), which functions to affirm a proposition.

(8) 曰：「是鲁孔丘與？」曰：「是也」 yuē shì lǔ kǒngqiū yú yuē shì yě say this Lu Kongqiu Q say right PTCL ‘(Is) this person Kongqiu from the state of Lu? Right.’ Lunyu (400 BCE)

106

Yen (1986) further develops Hong’s theory and suggests that the use of shì as a copula came from the function as an affirmative particle. Because of the contrastive meanings of fēi ‘wrong’ and shì ‘right,’ speakers started to use shì--an affirmative particle, in the sentential environment where the nominal negator fēi appeared. By Yen’s theory, the development of shì may be analyzed as an analogic change:

fēi ‘wrong’:: shì ‘right’ = fēi ‘negative’ :: shì ‘Y’

Y= ‘affirmative’

(9) 如以鬼非死人，則其信杜伯非也 rú yǐ guǐ fēi sǐ rén , zé qí xìn dùbó fēi yě if think ghost NEG dead people, then their belief DuBo wrong PTCL ‘If they think that ghosts are not dead people, then their belief in the story of Du Bo is wrong.’

如以鬼是死人，則其薄葬非也 rú yǐ guǐ shì sǐ rén , zé qí bó zàng fēi yě if think ghost AFF dead people, then their simple funeral wrong PTCL ‘If they think that ghosts are dead people, then their advocacy of simple funerals is wrong.’ Lunheng (100 CE)

As example (9) shows, both fēi and shì appeared in the same sentential environment with opposite meanings. This is evidence for the link between fēi as a negative particle and shì as an affirmative particle. According to Yen, the use of fēi was replaced by bú shì, where bú is a negative particle and shì is an affirmative particle, in a later era. Yen indicates that the use of bú shì instead of fēi was a crucial point for shì to be reanalyzed

107

as a copula. However, there is no evidence to show that bù shì evolved from or replaced fēi. Bù in Old Chinese was used to negate verbs, verbal phrases and predicates, as in bù zhī ‘not know,’ bù wéi ‘not do,’ etc. Therefore, bù occurring preceding shì gave the evidence that shì was part of the predicate. One of the earliest occurrences of bù shì is found in Zuozhuan (500 BCE), in which bù negated the nominal predicate shì guò ‘this fault’ in which shì is a demonstrative followed by a noun:

(10) 文王所以造周不是過也 wén wáng suǒ yì zào zhōu bù shì guò yě King Wen it use build Zhou NEG this fault PTCL ‘That King of Wen used it to build the state of Zhou (is) not a fault.’ Zuozhuan (500 BCE)

Feng (1993) also disagrees with Yen and argues that the evidence from the pair of sentences containing shì as the antonym of fēi is not robust enough to be the triggering experience for speakers. He found 10 sentences in Lunheng (100 CE), in which shì was used as a copula but only one of them can be considered as a pair with fēi. The reason is that in Old Chinese, fēi could only negate noun phrases, as in [fēi NP], while shì could occur in a number of environments, such as [shì N], [shì NP], [shì VP], [shì S], etc. Feng discovered that among 23 [ADV shì XP] sentences from Lunheng, again only one pair of fēi and shì was found. The asymmetry of the distribution of affirmative shì and negative fēi shows Yen’s hypothesis is untenable. Feng furthermore pointed out that Yen’s theory cannot explain the case that in the string [shì ADV XP], shì can only be interpreted as a demonstrative pronoun, see example (11), in which zì is the adverb following shì, but in the string [ADV shì XP], shì can be interpreted as a copula modified by an adverb, see

108

example (12). Therefore Feng concludes, “although it is possible that under the analogical change, the affirmative shì might have developed from antonym usage with fēi, in fact copula shì does not seem to have been developed from the affirmative particle shì.” (Feng 1993:282).

(11) 仲尼曰：是圣人仆也，是自埋于民 zhòng ní yuē : shì shengrén pú yě , shì zì mái yú mín Zhongni say this saints servant PTCL, this self bury in people ‘Zhongni said, this (is) the saint’s servant, he buries himself in people.’ Zhuangzi (300 BCE)

(12) 审是掌之罪也 shěn shì Zhǎng zhī zuì yě really is Zhang ASSOC fault PTCL ‘(It) really is Zhang’s fault.’ Lunheng (100 CE)

Feng also finds problems in Wang’s theory and the most serious problem, as he points out, is: if copular sentences such as CCS in examples (2)-(5) contained no copula, there would have been no place for shì to move into and function as a copula between the subject and the predicate in those sentences; as shì evolved into a copula, the demonstrative pronoun would have to have changed its position from the subject position to a position a copula usually occurs in. Feng proposes that there was an overt pause obligatorily occurring in CCS between the subject and predicate in Old Chinese [NP pause XP yě]. Similarly there was a pause obligatorily occurring in the topic-comment structure between the topic phrase and the comment sentence [XP pause [shì XP]]. This

109

is the reason that an interjectional particle and adverb could emerge between the subject and the predicate in a CCS or between the topic phrase and the comment sentence in a topic-comment structure. He argues that the demonstrative pronoun shì had changed its lexical category from a [+n] to a [+v] because of the weakening of the emphatic function of the demonstrative pronoun shì with the lack of necessity for the pause and because adverbs pushed the shì to merge with the pause and the anaphoric function of shì became opaque; and finally shì was reanalyzed as a copula and the pause disappeared.

However, Feng’s theory also has some serious difficulties. First, it is difficult to determine whether the prosodic pause existed in Old Chinese, as we do not possess audio records. Second, the lack of necessity for the pause that weakened the emphatic function of the demonstrative pronoun does not necessarily change the demonstrative pronoun into a verb. Furthermore, adverbs should not have the power to push shì to merge with the pause and to become a verb. It was speakers who produced, and hearers who interpreted.

Adverbs in Old Chinese systematically occurred in the position preceding a predicate, either a nominal predicate or a verbal predicate. The adverbs that occurred preceding and modifying shì are evidence showing that shì was no longer a demonstrative pronoun, but already part of the predicate, a copula.

Shi and Li (2001) argue that copula shì evolving from the demonstrative pronoun underwent a process of analogy, modeling regular transitive verbs in Old Chinese. They claimed, following Kiparsky (1992), that the holistic structural property of a language at a certain period brings about grammaticalization through a process of analogy. They suggested that Old Chinese had already developed standard SVO word order, and the frequent occurrence [NP shì NP] is the morphosyntactic context in which shì was

110

influenced by and fitting into the extant transitive verb pattern. However, it is obvious that the copula shì in Modern Chinese differs from the standard transitive verbs in some syntactically significant ways (see details in Chapter 1 section 1.2.1); if it underwent the analogical process modeling SVO structure, then why did it not develop into a full- fledged transitive verb? Furthermore, Shi and Li treat the structure [NP shì NP] as the morphosyntactic contexts in which the change is enabled. As will be discussed in 4.3.2, the earliest evidence showing that shì was already a copula is found in Mengzi (300

BCE), which was long before the frequent occurrences of the structure [NP shì NP]

(around 500 CE)19. Therefore, to suggest [NP shì NP] as the onset context for the grammaticalization of shì is problematic. Moreover, Li and Shi fail to identify the specific context that triggered this change and what the semantic relatedness between the demonstrative pronoun shì and the copula shì is remains a mystery.

4.3 The development of shì

In this section, I turn to my own analysis of the development of shì.

4.3.1 A syntactic analysis of the classical copular sentence (CCS) in Old Chinese

Examples (2) and (3) in section 4.1 are instances of CCS in Old Chinese appearing as

[NP NP yě] and [NP VP yě]. Example (13) is another instance of CCS this time with the structure [NP S yě].

19 I will show statistical evidence for this point in 4.3.3.

111

(13) [陳良楚产也] [chén liáng chǔ chǎn yě] [NP S yě] Chenliang Chu produce PTCL ‘Chengliang (is) the product of Chu (was born in Chu).’ Mengzi (300 BCE)

Therefore the structure of CCS can be summarized as [NP NP/VP/S yě]. If

XP=NP/VP/S, then the syntactic structure of CCS in Old Chinese can be schematized as

[NP XP yě]. The declarative sentence’s final particle yě was required in CCS in Old

Chinese and some linguists (e.g. Shi and Li 2001) consider it as the grammatical marker for CCS in the period. However, not all CCS in Old Chinese are marked by yě. Other final particles such as yǐ and ér in declarative, zāi, hū and yú in questions also occurred in this position, but yě is the most frequent one. It is more correct to say a declarative sentence final particle was required in this type of copular sentences in Old Chinese.

Hereafter I use PTCL to represent sentence final declarative particles in structure. The semantic relations in [NP XP PTCL] between NP and XP are copulative linking.

Therefore, the CCS construction in Old Chinese can be schematized as:

[NPi XPj PTCL][SEMi (copulative linking) SEMj Declarative]

The copulative linking meaning of CCS includes specificational, i.e. examples (2) and (13), and predicational, i.e. example (3). In (3), zhèng ‘being upright’ is a description of the zhèngzhě ‘governors.’ As I have shown in the constructional schematic taxonomy of the copular construction in Chapter 3, there are three subschemas under the schema of the specificational copular construction: equational, specificational and

112

specificational+contrastive. (see Chapter 3 section 3.4 for details) The specificational

CCS in Old Chinese has the meanings of equational or specificational. In (2), both of the subject Zhòngní and the nominal predicate rì yuè ‘the sun and the moon’ are semantically referential, and (2) is equational. In (13), Chéngliáng is the referential member of the non-referential but restricted set of chǔ chǎn ‘the product of Chu.’ Example (13) encodes a class-membership predication relation and it is specificational.

Now consider the topic-comment construction with a CCS as the comment part also found in Old Chinese. As has been shown in section 4.1, example (4) is a sentence of the topic-comment construction, the demonstrative pronoun shì functions as the anaphor referring to the topic phrase fù yú guì ‘wealth and nobility.’ There is no linking verb between the subject shì and the nominal predicate rén zhī suǒ yù ‘the things that people want,’ and the final particle yě marks the comment part of (4) a CCS. The structure of (4) is [NP [shì NP PTCL]], noticing the topic in (4) is a complex NP. Example (5) is also a sentence of topic-comment construction and the topic in (5) is a conditional clause; in the comment part the demonstrative pronoun shì functions as the anaphor referring to the topic part rú qì dé bú ràng ‘if (you) abandon the moral and don’t yield’ and there is no linking verb between the subject shì and the VP predicate fèi xiān jūn zhī jǔ ‘abolish the behavior of the former emperor.’ Again, the final particle yě marks the comment part of the sentence as a CCS. The structure of (5) is [S [shì VP PTCL]]. Examples (14)-(18) are additional examples of the topic-comment construction with a CCS occurring in its comment part.

113

(14) [吾無行而不與二三子者 [是丘也]] [wù wú xíng ér bù yú èr sān zǐ zhě [shì qiū yě]] [NP [shì NP yě]] SG1 NEG behavior CONN NEG tell 2PL NOM this Qiu PTCL ‘I don’t have things that I cannot tell you guys, this (is) me.’ Lunyu (400 BCE)

(15) [既欲其生又欲其死 [是惑也]] [jì yù qí shēng yòu yù qí sǐ [shì huò yě]] [S [shì VP yě]] not only want 3SG live but also want 3SG die this confuse PTCL ‘Not only to want him to live, but also to want him to die, this (is) confusing.’ Lunyu (400 BCE)

(16) [无父无君 [是禽兽也]] [wú fù wú jūn [shì qínshòu yě]] [VP [shì NP yě]] NEG father NEG lord this animal PTCL ‘Those who have no fathers and no lords, they (are) animals.’ Lunyu (400 BCE)

(17) 曰： [ [是鲁孔丘與]]？ yuē [ [shì Lú Kǒngqiū yú]] [(NP) [shì NP yú]] say this Lu Kongqiu Q(PTCL) ‘(Changju) said, ‘(Someone), (is) this Kongqiu from the state of Lu?’’ Lunyu (400 BCE)

114

(18) [然而不胜者 [rán ér bú shèng zhě this but NEG win NOM [是天时不如地利也]] [shì tiān shí bù rú dì lì yě]] this heaven time NEG compare to terrain advantage PTCL [NP [shì S yě]] ‘Those who did this but not won, this (is) that its heaven time could not compete with the terrain advantage.’ Mengzi (300 BCE)

Examples (4)-(5) and (14)-(18) show that the form of this particular topic-comment construction can be summarized as [(XP) [shì XP PTCL]]. In the following discussion, I will discuss the meaning of this construction, and argue that the earliest copula shì emerged from this construction, and it was the onset context for the copular construction to change into being.

4.3.2 The constructionalization of the copula

4.3.2.1 The semantic relatedness between the demonstrative pronoun and the copula

As has been mentioned above, the semantic relations between the subject and the predicate of CCS, see examples (2), (3) and (13), include specificational and predicational meaning. In the topic-comment construction [(XP) [shì XP PTCL], since the comment part of the structure is a CCS, the semantic relations between the subject, the demonstrative pronoun shì, and the predicate XP are specificational and predicational as well. For example, the CCS in (4) (14) and (16) are specificational, in (17) is equational, and those in (5), (15) and (18) are predicational.

115

As for the topic-comment construction, Li and Thompson (1981:95) suggest that the relationship between the topic and the comment is wide open. That is to say, as long as the comment expresses something about the topic in the perception of the speaker and the hearer, the sentence will be meaningful. The demonstrative pronoun shì frequently functions as the anaphor referring to the complex topic of the topic-comment construction; therefore, it entails the semantics of the topic. In this case, the demonstrative pronoun as the anaphor links the semantic meaning of the topic and the predicate part of the comment. It is the demonstrative pronoun that has the predicational or specificational relation with the predicate since they together form a CCS, however since the demonstrative pronoun refers to the topic, therefore the topic is related to the predicate of CCS in terms of the semantic relation of specificational and predicational.

Accordingly, the specific topic-comment sentences under discussion such as (4), (14) and

(16) are specificational, (17) is equational, and (5), (15) and (18) are predicational.

As I have shown in Chapter 3 section 3.3, in a copular sentence of Modern Chinese, the semantic relationship between the subject and the predicate is copulative linking include specificational and predicational meanings. From the above discussion, the categories of the semantic relations between the subject and the predicate in a copular structure of Modern Chinese appear to be parallel with the semantic relations between the topic and the predicate of the comment in the structure [(XP) [shì XP PTCL] in Old

Chinese. Therefore, I conclude that when the demonstrative pronoun shì frequently functioned as the anaphor referring to the topic, it also linked the semantic relations between the topic and the predicate of the comment. This particular topic-comment construction can be schematized as:

116

[(XPi) [shì XPj PTCL]][Topici [COMMENT(Anaphor SEMj Declarative)]]

However, the semantic relations between the topic XP and the predicate XP in the structure [(XP) [shì XP PTCL]] did not evidently show that the demonstrative pronoun shì had been assigned the syntactic status as a copula. But as I will show in the next sub- section, it gave the original semantic context for the demonstrative pronoun shì to be construed as the semantic meaning of a copula, and the topic-comment construction to convey the copulative linking meaning of a copular sentence. In fact, the syntactic change, the analogization process modeling the structure of the verb wéi ‘to be,’ eventually speakers construed shì as the semantic meaning of a copula and the syntactic property of a copula, which gave rise to the emergence of the copula shì.

4.3.2.2 The enabling context

In Old Chinese, shì as a demonstrative occurred in the specifier position of an NP, exemplified as shì guò ‘this fault’ in (10). Shì also occurred as a demonstrative pronoun, e.g. in example (19) shì appears in the object position of a VP yǒu shì:

(19) [国之有是多矣] [guó zhī yǒu shì duō yǐ] [NP VP yǐ] state ASSOC have this many PTCL ‘The cases that state having this (is) numerous.’ Zuozhuan (500 BCE)

117

At the same period, as has been shown in 3.1, it began to function as an anaphor occurring at the subject position of the comment part of a topic-comment construction which has the form [(XP) [shì XP PTCL]], see examples (14)-(18).

In the Old Chinese topic-comment construction, the demonstrative pronoun shì functioned as the anaphor frequently referring to a complex topic, e.g. a clause, a VP, or a complex NP. If the topic phrase was a simple NP (a pronoun or a proper noun), then the anaphor demonstrative pronoun shì normally did not occur; it then would be a CCS, e.g.

(2). Therefore, when the anaphoric demonstrative pronoun shì frequently occurred in the topic-comment construction [(XP) [shì XP PTCL]], it created the semantic relation between the topic phrase XP and the predicate of the comment XP, and gave rise to a robust semantic context for constructionalization, that is, the demonstrative pronoun shì changed into a copula verb. I hypothesize that the original and onset context that enabled the constructionalization is:

[(XPi) [shì XPj PTCL]][Topici [Anaphor SEMj Declarative]]

In the following discussion, I will give evidence to show that the earliest copula emerged from the string [(XP) [shì XP PTCL]], and that this is the original and morphosyntactic context that enabled the constructionalization of shì.

As has been mentioned in section 4.2, when adverbs occurred preceding and modifying shì, shì was no longer a demonstrative pronoun, but already part of the predicate, already a copula. I argue that the earliest [ADV+shì] combination emerged from the context of [(XP) [shì XP PTCL]]. Example (20) is found in Mengzi, in which the adverb jūn ‘totally’ occurred preceding shì:

118

(20) 公都子问曰：「 [钧是人也] gōng dōu zǐ wèn yuē : [jūn shì rén yě] [ADV shì NP yě] Gongdouzi ask said: totally COP people PTCL

[或为大人] [huò wéi dà rén] [NP wéi NP ] some be big people

[或为小人]，何也？」 [huò wéi xiǎo rén], hé yě [NP wéi NP ] some be small people, why PTCL ‘Gongdouzi asked, ‘(they) are totally people; some of them are good people, some of them are bad people; why is that?’’ Mengzi (300 BCE)

In example (20), the adverb jūn occurring preceding and modifying shì shows that shì was already part of the predicate, a copula. In (20), the subject is unspecified, the VP predicate consists of the copula shì that is modified by the adverb jūn, and the nominal predicate rén, and yě is the final declarative marker. It entails the copulative linking meaning, specifically predicational meaning as ‘being human’ is a characteristic of the unspecified subject ‘all of us.’

If the adverb jūn did not occur in the position, the first sentence of example (20) would have been (21), which had the structure [(XP) [shì XP PTCL]].

119

(21) 公都子问曰：「[ [是人也]] gōng dōu zǐ wèn yuē : [ [shì rén yě]] [(XP) [shì NP yě]] Gongdouzi ask said: these people PTCL,…’

Example (21) is a sentence of the topic-comment structure [(XP) [shì NP PTCL]], in which the topic is unspecified, shì is the demonstrative pronoun20, the subject of the comment sentence referring to the unspecified topic, NP rén ‘people’ is the nominal predicate of the comment sentence, and yě is the final declarative maker. It entails the predicational meaning between the unspecified topic and NP rén ‘people.’ Example (20) is the earliest occurrence in which [ADV+shì] is found in the Old Chinese text in CCL

Classical Chinese Corpus. I suggest that shì in (20) is the earliest copula we can find in

Old Chinese texts, and the earliest copula emerged from the string [(XP) [shì XP PTCL]].

Examples (22) and (23) are another pair of examples showing that the copula verb emerged from the context of [(XP) [shì XP PTCL]]. Example (22) is found in Mozi (BCE

400), in which the adverb bì ‘definitely’ occurring preceding and modifying shì shows that shì was already part of the predicate, a copula. In (22), the subject is bú xiào zǐ ‘an unworthy son,’ the VP predicate consists of the copula shì that is modified by the adverb bì, and the VP yuàn qí qīn ‘grudge his parents,’ and yǐ is the sentence final declarative maker. The semantics of (22) is predicational as the verbal predicate ‘to grudge his parents’ is a characteristic of the subject ‘an unworthy son.’ If the adverb bì did not occur

20 According to the context, shì in [[shì] [rén yě]] can only be a demonstrative pronoun, but not a demonstrative. It cannot be analyzed as [[shì rén] yě] ‘this person.’

120

in the position, example (22) would have been (23), which again had the structure [(XP)

[shì XP PTCL]].

(22) [不孝子必是怨其亲矣] [bú xiào zǐ bì shì yuàn qí qīn yǐ] [NP adv shì VP yǐ] not worthy son definitely COP grudge his parents PTCL ‘An unworthy son definitely is to grudge his parents.’ Mozi (400 BCE)

(23) [不孝子 [是怨其亲矣]] [bú xiào zǐ [shì yuàn qí qīn yǐ]] [NP [shì VP yǐ]] not worthy son this grudge his parents PTCL

Example (23) is a sentence of the topic-comment structure [(XP) [shì XP PTCL]], in which the topic is an NP bú xiào zǐ ‘not worthy son,’ shì is the demonstrative pronoun, the subject of the comment sentence, and VP yuàn qí qīn ‘grudge his parents’ is the verbal predicate of the comment sentence. (23) entails the predicational relation between the topic bú xiào zǐ ‘an unworthy son’ and VP yuàn qí qīn ‘grudge his parents.’

The examples above strongly support that [ADV+shì] emerged in the context of

[(XP) [shì NP PTCL]], which also gives evidence to suggest that the copula verb shì emerged from the same context. Therefore it is reasonable to claim that [(XP) [shì XP

PTCL]] is the onset mophosyntactic context enabling the emergence of the copular shì and the constructionalization of the copular construction that can be schematized as:

121

(24) [(XPi) [shì XPj PTCL]][Topici [Anaphor SEMj Declarative]]

[(XPi) COP XPj (PTCL)][SEMi copulative linking SEMj (Declarative)]

4.3.2.3 The mechanism of the constructionalization of shì

In the previous sub-section, I have suggested that the copula emerged from the context of

[(XP) [shì XP PTCL]], in this sub-section, I am going to answer the questions: why and how did the change come into being? I argue that the analogy and analogization are the motivation and mechanism of the constructionalization of shì.

As has been shown in Chapter 2, reanalysis refers to the replacement of old structures by new ones. Analogy, by contrast, refers to the attraction of extant forms to already existing constructions (Hopper and Traugott 2003:64). Analogy concerns pattern match with other members of a category and the focus is on similarity. Analogization

(analogical change) as one of the mechanisms occurs due to being influenced by and fitting into extant patterns. When analogization happens, it is simultaneously reanalysis.

Shi and Li (2001) argue that the copula shì evolving from the demonstrative pronoun underwent a process of analogy, modeling the standard SVO word order in Old

Chinese, and the frequent occurrence [NP shì NP] is the morphosyntactic context in which shì was influenced by and fitted into the extant transitive verb pattern. I agree that the copula shì evolving from the demonstrative pronoun underwent a process of analogy.

However, there is no evidence that frequent occurrence preceded reanalysis, but the

“critical” contexts (Diewald's 2002) for the change are attested in the data.

122

I suggest that the “critical” context for the change to be ready to occur is [(XP) [shì

XP PTCL]] in which the demonstrative pronoun shì began to change, as I have shown in

3.2.2. That is to say [NP shì NP] is the later stage of the change. I hypothesize that the demonstrative pronoun shì in [(XP) [shì XP PTCL]] started to change into the copula verb through the process of analogization, modeling the structure of the extant full transitive verb wéi. In Old Chinese, wéi was a verb with a variety of meanings including both ‘to do’ and ‘to be’ in English, and had the grammatical structure [(XP) wéi XP]. For example:

(25) [为政以德]，譬如北辰 [wéi zhèng yì dé] pìrú běi chén [V NP V NP] do politics use virtue like north star ‘Doing politics with virtues is like the north star.’ Lunyu (400 BCE)

(26) [民为贵]，君为轻 [mín wéi guì] jūn wéi qīng [NP V VP] people be precious, lords be unimportant ‘People are precious, and lords are less important.’ Mengzi (300 BCE)

(27) [克己复礼为仁] [kè jǐ fù lǐ wéi rén] [VP VP V NP] control self restore ritual be benevolence ‘Controlling self and restoring rituals are benevolence.’ Lunyu (400 BCE)

123

Examples (25)-(27) show the major usages of the verb wéi in Old Chinese. In example (25) the verb wéi has the semantic meanings of ‘to do.’ In (26) and (27), the verb wéi is the verb ‘to be.’ Example (26) has the structure [NP V VP] with the predicational meaning, and example (27) has [VP VP V NP], in which the paralleled two

VPs kè jǐ ‘control self’ and fù lǐ ‘restore rituals’make up the subject of the sentence, and the VP predicate consists of the verb wéi and NP rén ‘benevolence.’ It entails the predicational meaning: ‘controlling self and restoring rituals’ is the characteristics that describe the nominal predicate ‘benevolence.’ A number of similar examples are found in

Old Chinese texts:

(28) [孝弟也者为其仁之本與] [xiào tì yě zhě wéi qí rén zhī běn yú] [NP V NP yú] filial piety PTCL NOM be his benevolence ASSOC root PTCL ‘The behaviors of filial piety are the root of the benevolence.’ Lunyu (400 BCE)

(29) [中国于四海內则在東南为阳 ] [zhōngguó yú sì hǎi nèi zé zài dōngnán wéi yang] [S V NP ] central state at four seas in then at southeast be Yang ‘The central state in the world being in southeast is at the side of Yang.’ Shiji (100 BCE)

124

(30) [凡人之欲为善者为性惡也] [fán rén zhī yù wéi shàn zhě wéi xìng è yě] [NP V S yě] all people ASSOC want do good NOM be nature evil PTCL ‘The reason that all the people want to do good things is that their nature is evil.’ Xunzi (200 BCE)

(31) 曰：「[为仲由]」 yuē [wéi zhòngyóu] [V NP ] say be Zhongyou ‘Zilu said, ‘(I) am Zhongyou.’’ Lunyu (400 BCE)

The verb wéi in examples (28)-(31) is ‘to be.’ The structure of (28) is [NP wéi NP yú]] having the semantics of canonical specificational, the structure of (29) is [S wéi NP] with the predicational meaning, the structure of (30) is [NP wéi S PTCL] entailing predicational meaning, and (31) has the structure [wéi NP] with the subject not specified, which is equational. Consequently, the sentences of (26)-(31) can be structured as [(XP) wéi XP PTCL], which entails the linking meaning of specificational or predicational. The construction of wéi ‘to be’ can be schematized as:

[(XPi) wéi XPj (PTCL)][(SEMi) BE SEMj (Declarative)]

This schematized construction of wéi ‘to be’ appears similar to the onset context of the constructionalization of shì. These paralleled constructional contexts make it easy for the analogical process to occur:

125

[(XPi) [shì XPj PTCL]][(Topici) [Anaphor SEMj Declarative]]

[(XPi) wéi XPj (PTCL)][(SEMi) BE SEMj (Declarative)]

Now, let’s take a look at the following examples occurring in Lunyu and Mengzi:

(32) a. 长沮曰：「[夫执舆者为誰？]」 chángjū yuē : [fū zhí yú zhě wéi shéi ] [ NP V NP ] changju said: PTCL drive wagon NOM be who ‘Changju said, ‘who is the one who drives the wagon?’’

b. 子路曰：「[为孔丘 ]」 zǐlù yuē : [wéi kǒngqiū] [V NP ] zilu said: be Kongqiu ‘Zilu said, ‘It is Kongqiu.’

c. 曰：「[ [是鲁孔丘與]]？」 yuē : [ [shì lǔ kǒng qiū yú]] [(NP) [NP NP yú]] [ V NP yú] said: this Lu Kongqiu Q(PTCL) Changju said, ‘(Is) this Kongqiu from the state of Lu?’

126

d.曰：「是也．」 yuē : shì yě said: right PTCL ‘Zilu said, ‘Right.’’ Lunyu (400 BCE)

Example (32a) consists of the structure [NP wéi NP] with the equational meaning; in (32b), the subject does not occur, and the structure is [(NP) wéi NP] having the semantics of equation. The verb wéi in both (32a&b) is the verb ‘to be.’ (32c) has the structure [(NP) [shì NP yú]], in which the topic Kongqiu has been activated in the previous discourse, and therefore does not occur; the demonstrative pronoun shì refers to the unspecified topic and has the semantic relation of equation with the nominal predicate of the comment sentence Lǔ Kǒngqiū ‘Kongqiu from the state of Lu,’ and the final particle yú inherits question. The parallel form and meaning between (32b) and (32c) in the cohesive discourse is obvious, which gives the appropriate context for analogization to occur. I propose that the construction of (32b) is the analogical model for the demonstrative pronoun shì to develop into a copula verb: [(NPi) wéi NPj] [(SEMi) equational SEMj] was analogized to the emergence of [(NPi) shì NPj][(SEMi) equational SEMj].

Based on analogization, (32c) was analyzed as a topic-comment construction [(NPi)

[shì NPj yú]][(Topici) [Anaphor SEMj yú]] as it has been shown above, however, it could be reanalyzed (re-bracketing and re-categorizing) as [(NPi) shì NPj][(SEMi) equational SEMj] by means of analogization modeling [(NPi) wéi NPj][(SEMi) equational SEMj]. As for the meaning, shì was originally the anaphor linking the semantic relation of equation between the unspecified topic and the predicate of the

127

comment sentence; however, it could have been reanalyzed as the copula ‘to be’ with the meaning of anaphoric reference bleached but strengthened as equational between the unspecified subject and the nominal predicate.

Example (33a) from Mengzi was not ambiguous any more because of the adverb jūn

‘totally’ preceding and modifying shì, which shows that shì was no longer a demonstrative pronoun, but already a verb. Therefore, (33a) has the structure [shì NP

PTCL], and shì is the copula verb linking the predicational meaning between the unspecified subject and the nominal predicate.

(33) a.公都子问曰：「[钧是人也] gōng dōu zǐ wèn yuē : [jūn shì rén yě] [ADV COP NP yě] Gongdouzi ask said: totally COP people PTCL

b. [或为大人 ] [huò wéi dà rén] [NP V NP ] some be big people

c. [或为小人]，何也？」 [huò wéi xiǎo rén], hé yě [NP V NP ] some be small people, why PTCL ‘Gongdouzi asked, ‘(they) are people; some of them are good people, some of them are bad people; why is that?’’ Mengzi (300 BCE)

128

Based on the above discussion, the analogical process of the constructionalization can be schematized as: the topic-comment construction [(XPi) [shì XPj

PTCL]][(Topici) [Anaphor SEMj Declarative]] was reanalyzed as the copular construction [(XPi) COP XPj (PTCL)][(SEMi) copulative linking SEMj

(Declarative)], in which the demonstrative pronoun shì changed into the copula shì through the analogicalization process, modeling the construction [(XPi) wéi XPj

(PTCL)][(SEMi) BE SEMj (Declarative)]. As has been shown above, wéi as a full verb in Old Chinese had functions other than ‘to be’ including ‘to do,’ however, as wéi

‘to be’ was the analogical source, shì ‘to be’ was the only product from analogization.

This also leads to the formal properties of the copula verb in Modern Chinese that are sufficiently different from the classificatory verbs (see Chapter 1 section 1.2.1). The first occurrences of the copular construction had the structure [(XP) COP XP PTCL]. Since its emergence, the copular construction underwent host-class expansion, e.g. [NP COP NP] became more and more frequent (see 4.3.3), and syntactic and semantic expansions, e.g. the cleft construction (see Chapter 3 section 3.5 and Chapter 5).

4.3.3 Statistical evidence for constructionalization of shì and further expansion

In this sub-section, I give the statistical evidence showing constructionalization of the copula shì and further expansion. For the statistical counting, I have chosen four major books as the data: two Old Chinese books Lunyu (400 BCE) and Mengzi (300 BCE), one book at the transition between Old Chinese and Middle Chinese Shiji (100 BCE)21 and one Middle Chinese book Shishuoxinyu (500 CE). For each of the four books, I count the

21 As for Shiji, due to the time limitation, I only focus on the part of Liezhuan in Shiji.

129

total tokens of the string [(XP) shì XP (PTCL)], and the tokens of a number of variables

(see following discussion for details), which together give the evidence of the process of constructionalization.

4.3.3.1 The increase of adverbs preceding shì

In section 4.2 and 4.3.2.2, I pointed out that adverbs occurring preceding and modifying shì in the string [(XP) shì XP (PTCL)] show that shì was no longer a demonstrative pronoun, but already a copula, and that [(XP) [shì XP (PTCL)]] was the onset morphosyntactic context enabling the constructionalization of shì.

Along with the constructionalization of shì, the adverbs preceding and modifying shì became more and more frequent, which means shì as the copula verb with the specificational or predicational meaning of the copular construction became more and more entrenched. According to the data, the earliest adverb occurring in this environment is jūn ‘totally’ as it showed up twice as in jūn shì in Mengzi. The other frequent adverbs that occurred in the environment are bì, dìng ‘definitely,’ jiē, bìng ‘totally,’ zì ‘self,’ guǒ

‘as expected,’ etc.

without adv. with adv. Total

Lunyu (BCE 400) 13 (100%) 0 (0%) 13

Mengzi (BCE 300) 29 (93.5%) 2 (6.5%) 31

Shiji (part) (BCE 100) 27 (93%) 2 (7%) 29

Shishuoxinyu (CE 500) 66 (72%) 26 (28%) 92

Table 4.1: The adverb distribution of preceding shì in the string [(XP) shì XP (PTCL)]

130

Table 4.1 shows in Lunyu, there were no adverbs occurring in the position of preceding shì, however, in the Middle Chinese text Shishuoxinyu, 28% of the occurrences had an adverb preceding and modifying shì. The token frequency means shì as the copula verb became more and more entrenched.

4.3.3.2 The decrease of sentence final particles

As I have shown in 4.3.1, the final particle, e.g., yě, yǐ and ér in declarative, zāi , hū and yú in questions, was required in CCS, [NP XP PTCL], in Old Chinese. In the string [(XP) shì XP (PTCL)] along with the constructionalization of the copula shì, the declarative sentence final particle became less and less frequent.

with PTCL without PTCL Total

Lunyu (400 BCE) 13 (100%) 0 (0%) 13

Mengzi (300 BCE) 30 (97%) 1 (3%) 31

Shiji (part) (100 BCE) 26 (90%) 3 (10%) 29

Shishuoxinyu (500 CE) 5 (5%) 87 (95%) 92

Table 4.2: The final particle distribution in the string [(XP) shì XP (PTCL)]

Table 4.2 shows in Lunyu, 100% occurrences with the structure [(XP) shì XP

(PTCL)] had the declarative sentence final particle, however, in the Middle Chinese text shìshuoxinyu, 95% of the occurrences did not have the final particle. It shows loss of the final particle by 500 CE. Since the final declarative particle in a CCS determines the copulative reading, the decrease of the occurrences of final particle in the string [(XP) shì

131

XP (PTCL)] gives evidence that the newly developed shì functions as a copula. The token frequency shows the new construction [(XP) shì XP (PTCL)] started to be fixed and frozen, and more and more frequent.

4.3.3.3 The decrease of the complex topic

As has been mentioned above, the pre-copula XP in the string [(XP) shì XP (PTCL)] could be a complex NP, VP, S or optional. Along with the process of the constructionalization, more and more simple NPs (pronouns and proper nouns) occurred in the pre-copula position, and eventually the pre-copula complex phrases gave way to simple NPs, and simple NPs became the most frequent, which gives evidence of the host- class expansion and syntactic expansion in the constructionalization process of the copula.

complex pre-copula phrase simple NP Total

(including unspecified)

Lunyu (400 BCE) 13 (100%) 0 (0%) 13

Mengzi (300 BCE) 31 (100%) 0 (0%) 31

Shiji (part) (100 BCE) 24 (83%) 5 (17%) 29

Shishuoxinyu (500 CE) 53 (58%) 39 (42%) 92

Table 4.3: The pre-copula complex phrase and simple NP distribution in the string

[(XP) shì XP (PTCL)]

132

Table 4.3 shows in Lunyu and Mengzi, all the strings occurring at the pre-copula position of [(XP) shì XP (PTCL)] were complex, however, in Shìshuoxinyu, almost half had changed into simple NPs. They show shì used as the anaphor referring the complex topic became less frequent, while the function of linking verb indicating copulative linking meaning between two NPs became more frequent. These data present the type frequency and give evidence of the host-class expansion and syntactic expansion.

As for the simple NP, most of them were pronouns, such as cǐ ‘this,’ wǒ ‘I,’ bǐ

‘he/they,’ and proper nouns, such as names of people and places. One thing that has to be pointed out is that in more than 90% of the occurrences where simple NPs occurred in the pre-copula position of [(XP) shì XP (PTCL)], the post-copula XP was also an NP. This shows the structure [NP shì NP], in which shì was already a standard copula, as [NP COP

NP] is the prototypical structure of the copular construction in Modern Chinese (see

Chapter 3 section3.1 for details).

4.3.3.4 The increase of [NP COP NP]

As has been mentioned in 4.3.1, the predicate XP of the comment sentence in the topic- comment construction [(XP) [shì XP (PTCL)]] could be NP, VP or S in Old Chinese.

Along with the constructionalization, more and more NP occurred in the post-copula position. Shi and Li (2001) counted the distribution of NP and VP as the predicate:

133

NP as the predicate VP as the predicate

Zuozhuan (500 BCE) 22% 78%

Lunyu (400 BCE) 48% 52%

Xunzi (200 BCE) 78% 22%

Table 4.4: The NP/VP distribution of the second XP in the string [(XP) shì XP (PTCL)]

Table 4.4 shows in Zuozhuan, VP was the major predicate of the last XP in the comment [(XP) shì XP (PTCL)], however the token frequency shows NP became much more frequent than VP at this slot of the construction in Xunzi. As shown in 4.3.3.2, along with the constructionalization, the final declarative particle gradually decreased, more simple NPs occurred in the pre-copula position, and the structure [NP COP NP] became more and more frequent and fixed. As [NP COP NP] is the prototypical structure of the copular construction in Modern Chinese, the frequency of [NP COP NP] shows shì had frozen into a standard non-inflectional copula verb.

4.3.3.5 The competition between shì and wéi

Over time, speakers came to prefer the copula shì ‘to be’ which originally developed from wéi ‘to be’ through the process of analogization, and eventually wéi ‘to be’ gave way to shì ‘to be.’ Here I counted all the tokens of wéi, shì, and wéi ‘to be,’ shì ‘to be’ in the four books. As for how to determine the copula shì ‘to be,’ I use the adverbs and the simple NPs (pronouns and proper nouns) that occurred preceding shì as the criteria to determine shì ‘to be.’

134

Table 4.5 shows in Lunyu, 8.4% of verb wéi was used as the linking verb ‘to be’22, but none of the shì was used as ‘to be.’ However in Shishuoxinyu, 24% of the shì was used as a copula ‘to be,’ but the verb wéi which was used as ‘to be’ had reduced to 6.8%.

This shows shì ‘to be’ significantly outnumbered wéi ‘to be’ in Middle Chinese, and wéi

‘to be’ eventually gave way to shì ‘to be.’

‘to be’ wéi Total wéi ‘to be’ shì Total shì

Lunyu (400 BCE) 15 (8.4%) 179 0 (0%) 60

Mengzi (300 BCE) 27 (5.2%) 517 2(0.8%) 258

Shiji (part) (100 BCE) 376(5.1%) 7374 6 (0.6%) 999

Shishuoxinyu (500 CE) 45 (6.8%) 651 64 (24%) 265

Table 4.5: The competition between shì ‘to be’ and wéi ‘to be’ I.

Table 4.6, from the perspective of the proportion of shì ‘to be’ and wéi ‘to be’ in all

‘to be’s also shows in the competition between ‘to be’ wéi and ‘to be’ shì, wéi ‘to be’ eventually gave way to shì ‘to be.’

22 I distinguish wéi ‘to do’ and wéi ‘to be’ by looking at their meanings in the contexts.

135

‘to be’ wéi ‘to be’ shì Total ‘to be’

Lunyu (400 BCE) 15 (100%) 0 (0%) 15

Mengzi (300 BCE) 27 (93%) 2 (7%) 29

Shiji (part) (100 BCE) 376 (98%) 6 (2%) 382

Shishuoxinyu (500 CE) 45 (41%) 64 (59%) 109

Table 4.6: The competition between shì ‘to be’ and wéi ‘to be’ II.

Starting with Shiji, along with the process in which wéi ‘to be’ started to give way to shì ‘to be,’ wéi ‘to be’ also became specialized. One significant pattern started to occur frequently, that is wéi...suǒ... For example:

(34) 大月氏王已为胡所杀 dà yuè zhī wáng yǐ wéi hú suǒ shā Dayuezhi King already be barbarians whom kill King of Dayuezhi has been the one whom the barbarians killed. Shiji (100 BCE)

Example (34) shows the pattern of wéi...suǒ... 23, in which wéi can still be analyzed as ‘to be,’ linking two NPs ‘King of Dayuezhi’ and ‘the one whom Hu killed,’ and suǒ is a pronoun equivalent to English ‘the one.’

23 Other linguists, e.g. Tang 1988, consider the pattern to be one of the passive structures in Old and Middle Chinese.

136

wéi...suǒ... wéi ‘to be’ Total wéi

Lunyu (400 BCE) 0 (0%) 15 (8.4%) 179

Mengzi (300 BCE) 0 (0%) 27 (5.2%) 517

Shiji (part) (100 BCE) 52 (0.7%) 376 (5.1%) 7374

Shishuoxinyu (500 CE) 25 (3.8%) 45 (6.8%) 651

Table 4.7: The distribution of wéi...suǒ...

Table 7 shows the occurrences of wéi...suǒ... increased significantly from Shiji to

Shishuoxinyu, which means wéi ‘to be’ was confined to the specific pattern, and was left the other occurrences of this usage to shì ‘to be.’ This is another fact that shì ‘to be’ outnumbered wéi ‘to be,’ and eventually became the standard Chinese copula.

This subsection presents the evidence showing the process in which the demonstrative pronoun shì was gradually constructionalized into a copula.

4.4 Typology and Conclusion

Heine and Kuteva (2002) point out that a demonstrative or demonstrative pronoun changing into a copula is a common grammaticalization process cross-linguistically. For example, in Egyptian, pw ‘this’ was a proximal demonstrative, and it changed into a copula verb.

137

(35) Egyptian (Gardiner 1957: 103ff) Nwn pw jt ncrw Nun this father gods ‘The father of the gods is Nun.’

Other languages such as Hebrew, Wappo, Swahili, and Palestinian Arabic also have a copula that developed from demonstratives or demonstrative pronouns. In a number of pidgin and creole languages, demonstrative pronouns also appear to have given rise to copulas. Heine and Kuteva (2002) suggest that demonstratives in their pronominal uses may give rise to various copula functions, such as existential, identifying and qualifying functions. Diessel (1999) classifies demonstratives cross-linguistically into four categories: pronominal, adnominal, adverbial, and identificational. In Diessel’s terminology, the demonstrative pronoun described in this chapter is treated as an identificational demonstrative that functions as anaphor. He discusses demonstrative-to- copula path-of-evolution and maintains that the reanalysis of demonstratives as copulas originates from a topic-comment construction. Hengeveld (1992:250) observes that this evolution “goes hand in hand with a reinterpretation of the theme-clause construction as a subject-predicate construction.” This remark serves as a good characterization of the development of the copular construction that emerged with the reanalysis of the topic- comment construction as a subject-predicate construction.

This chapter gives a detailed account for the formation and development of the copula shì. I argue that the onset morphosyntactic context that enabled the constructionalization was the Old Chinese topic-comment construction [(XPi) [shì XPj

PTCL]][(Topici) [Anaphor SEMj Declarative]]. The demonstrative pronoun shì

138

functioned as an anaphor referring to the topic and the semantic/pragmatic relation was copulative linking, specifically specificational or predicational, between the topic and the predicate of the comment. Together the two elements enabled reanalysis as a construction linked by a copula verb ‘to be.’ The demonstrative pronoun shì was gradually constructionalized into a copula through the analogization process modeling the construction of the verb wéi ‘to be’ [(XPi) wéi XPj (PTCL)][(SEMi) BE SEMj

(Declarative)] and reanalysis (rebracketing), and the copular construction [(XPi) COP

XPj (PTCL)][(SEMi) copulative linking SEMj (Declarative)] came into being. There is evidence showing the process and further expansion: the adverbs preceding and modifying shì became more and more frequent; more and more simple NPs occurred in the pre-copula position showing the host-class expansion; more and more NP occurred in the post-copula position; the declarative final sentence particle yě became less and less frequent; and shì ‘to be’ which was originally patterned like wéi ‘to be’ through the process of analogization became more and more competitive, and eventually wéi ‘to be’ gave way to the copula shì ‘to be.’

Along with the constructionaliztion of the copula, the occurrence of the CCS in Old

Chinese, such as examples (2) that had the syntactic structure [NP XP PTCL], became less frequent. This also shows that the sentence typology of the CCS had shifted from emphasis on clause boundary: two adjacent phrases NP and XP along with a sentence final particle marking it as a copular sentence to emphasis on clause cohesion, with a central linking verb and two phrases NP and XP (eventually two NPs) succeeding and following it as in Modern Chinese [NP shì NP].

139

Chapter 5

The Constructionalization of the Cleft Construction

5.1 Introduction

Examples in (1) present two instances of the cleft construction in Modern Chinese. A cleft sentence in Chinese is a special copular sentence in which the copula shì links a subject NP and a nominal predicate, a nominalization NOM=[XP de] (XP=NP/VP/S) marked by the nominalizer de: [NP COP NOM]. The cleft construction has two subschemas: cleft-sbj and cleft-obj. The cleft-sbj involves subject-subject semantic co- referentiality and the optionality of the nominalizer de, e.g. in (1a) the subject of the sentence wǒ ‘I’ is co-referential with the implicit subject of the nominalization qùnián lái měiguó de, and the nominalizer de is optional; whereas a cleft-obj involves subject- object co-referentiality and the obligatory nominalizer, e.g. (1b).

(1) a. 我是去年来美国 (的) wǒ shì qùnián lái měiguó (de) SG1 COP last year come US (NOM) ‘It was last year that I came to the US.’

140

b. 这个是昨天做的 zhè-ge shì zuótiān zuò de this CL COP yesterday make NOM ‘It was yesterday that this was made.’

The cleft construction entails the specificational meaning plus the immediate post- copula element, e.g., qùnián ‘last year’ in (1a), signaling the contrastive focus indicating the presupposition implied in the rest of the sentence, e.g. I came to the US some time.

The two subschemas cleft-sbj and cleft-obj also differ in meaning in that they have different semantic co-referentiality. (See above, and Chapter 3 sections 3.5 for details)

The copula shì is generally treated to signal contrastive focus in cleft sentences. In the history of Chinese, it is widely accepted, e.g. Wang 1937; Feng 1993; Pulleyblank

1995; Shi and Li 2001, that shì evolved from a demonstrative pronoun to a copula verb in

Old Chinese. In the previous chapter, I argued that the copula shì underwent the process of constructionalization from a demonstrative pronoun in Old Chinese, and the copular construction was constructionalized from the topic-comment construction in Old Chinese when the demonstrative pronoun shì occurred at the subject position of the comment clause functioning as an anaphor referring to the topic. In some fixed expressions of

Modern Chinese as I showed in chapter 1, shì retains this anaphoric function from Old

Chinese.

Many scholars hold that after shì had changed into a copula, it changed into a focus marker through a process of further grammaticalization in early Middle Chinese around

500 CE (Shi and Li 2001; Dong 2004; etc.). This chapter, however, argues that it is not that shì was further grammaticalized into a focus marker, but the new cleft construction

141

emerged, in which shì remained a copula verb but expanded its syntactic function, i.e. signaling contrastive focus. This emergence took place in late Middle Chinese (700-1000

CE).

This chapter aims to address how and why the cleft construction came into being in the history of Chinese. In the light of the theory of constructionalization, I argue that the development of the cleft construction in Chinese involved reanalysis in terms of syntactic form and semantic-pragmatic function. Section 5.2 outlines the historical emergence of the cleft construction. Section 5.3 provides a detailed analysis of how and why the cleft construction was constructionalized. Section 5.5 is the conclusion.

5.2 The emergence of the cleft construction

In this section, I will give a detailed analysis of the historical emergence of the cleft construction. Three facets will be focused on in the analysis: shì being crystalized as a standard copula in early Middle Chinese, the occurrence of the nominalization [XP de] in late Middle Chinese, and the emergence of the combination of the copula and the nominalization, i.e. the sequence of [NP COP XP de] in late Middle Chinese.

5.2.1 Shì: the copula in early Middle Chinese (200 CE - 600 CE)

In the previous chapter I argued that the onset mophosyntactic context that enabled the change from a demonstrative pronoun to a copula was the topic-comment construction, in which the demonstrative pronoun shì occurred at the subject position of the comment clause functioning as an anaphor referring to the topic. This gave rise to the copulative

142

linking semantic relation between the topic and the predicate of the comment that made the two elements ready to be linked by a copula verb ‘to be.’ Through the analogization process modeling the construction of Old Chinese verb wéi ‘to be,’ and reanalysis, the demonstrative pronoun shì was gradually constructionalized into a copula, and the copular construction came into being. The trajectory of the development of the copular construction can be schematized as follows, where XP=NP/VP/S:

[(XPi) [shì XPj PTCL]][Topici [Anaphor SEMj Declarative]]

[(XPi) COP XPj (PTCL)][(SEMi) copulative linking SEMj (Declarative)]

[NPi COP NPj][SEMi copulative linking SEMj]

Example (2) is the earliest occurrence in which shì appear to have the status of a copula verb:

(2) 公都子问曰：「 [钧是人也] gōng dōu zǐ wèn yuē : [jūn shì rén yě] [ADV shì NP yě] Gongdouzi ask said: totally COP people PTCL

[或为大人] [huò wéi dà rén] [NP wéi NP ] some be big people

143

After the emergence of the copular construction, the occurrences of [NP COP NP] became more and more frequent, and the structure [NP COP NP] was conventionalized as the prototype for the copular construction. In the early Middle Chinese book,

Shishuoxinyu (500 CE), 73% of the copular sentences are [NP COP NP] (48 occurrences out of 66 attested copular sentences), as exemplified in (3):

(3) a. [此三人並是高才 ] [cǐ sān rén bìng shì gāo cái] [NP ADV COP NP ] this three people totally COP high talent ‘These three people are totally of great talent.’ Shishuoxinyu (500 CE)

b. [我是李府君亲] [wǒ shì lǐfǔjūn qīn] [NP COP NP ] SG1 COP Lifujun relative ‘I am Lifunjun’s relative.’ Shishuoxinyu (500 CE)

144

With the emergence of the copular construction, the copulative linking meanings of predicational and specificational were also crystalized. (3a) is a predicational copular sentence with the post-copula predicate providing a property ‘high talent’ for the subject

‘these three people;’ whereas (3b) is a specificational sentence, in which the non- referential but restricted set ‘Lifujun’s relative’ is specified by the referential member the subject ‘I.’ The examples in (3) encode predicate informational focus in terms of information structure, that is to say the subject usually encodes referential given information, generally a topic, and the post-copula predicate as a whole is the informational focus indicating non-referential new information.

Other than the occurrences of [NP COP NP], in Shishuoxinyu 27% (16 out of 66) of the copular sentences are [(XP) COP XP], as exemplified in (4a) [NP COP VP] and (4b)

[NP COP S]. These occurrences are not as frequent as [NP COP NP], but they co-occur with [NP COP NP] all the way through the history of Chinese, as found in Modern

Chinese in (4c) [S COP NP]. Semantically, (4a&b) are predicational, and (4c) is specificational.

(4) a. [此是有情痴] [cǐ shì yǒu qíng chī] [NP COP VP ] this COP have love devoted ‘This is having devoted love.’ Shishuoxinyu (500 CE)

145

b. [此是屋下架屋耳] [cǐ shì wū xià jià wū ěr] [NP COP S ] this COP house under build house just ‘This is just building a house under another house.’ Shishuoxinyu (500 CE)

c. [中国人能够了解篮球是我的梦想 ] [zhōngguórén nénggòu liǎojiě lánqiú shì wǒ de mèngxiǎng] [S COP NP ] Chinese people can understand basketball COP SG1 ASSOC dream ‘That Chinese people can understand basketball is my dream.’ Yao Ming, my dream my world (2004)

Shi and Li (2001) argue that shì in (4a&b) was already further grammaticalized into a focus marker (FM). It marks VP yǒu qíng chī ‘have love devoted’ in (4a) and NP wū xià ‘house under’ in (4b) as contrastive focus (exclusiveness and exhausiveness). Their evidence for this claim is: in Early Middle Chinese, shì started to occur frequently preceding the interrogative wh-words, such as shéi ‘who’, hé ‘what,’ as in (5).

(5) 是誰教汝？ shì-shéi jiāo rǔ FM-who teach you ‘Who taught you?’ Beiqishu (530 CE)

146

Following Heine and Reh (1984), Shi and Li argue that wh-words have embedded focus, and the focus is often marked. In Old Chinese, the focus was marked by OV word order. A standard verb phrase in Old Chinese had the word order of VO, e.g. fèi zhī ‘get rid of it,’ xiǎng zhī ‘enjoy it’ (Zuozhuan, 400 BCE); whereas a verb phrase consisting of a wh-word, e.g. shéi ‘who,’ hé ‘what,’ employed the word order of OV, e.g. shéi lì ‘whom crown,’ hé rú ‘what like’ (Zuozhuan, 400 BCE). They claim that when shì changed into a copula verb, due to its low transitivity, wh-words could only occur following shì. (Shi and Li 2001:45) Consequently, the focus of the wh-words was no longer marked by word order; instead, speakers chose shì to carry on this mission. According to Shi and Li

(2001:48), in early Middle Chinese around 500 CE, shì served as a focus marker for wh- words and by analogy it started to mark other categorical elements immediately following it as focus, e.g. NP in (4b), VP in (4a). Therefore, shì was further grammaticalized into a focus marker in early Middle Chinese and has been in use all the way through nowadays.

As I have discussed above, in early Middle Chinese, with the emergence of the copular construction, the copula shì, like all the verbs in Chinese as a VO language, served to introduce the informational focus indicated by the post-verbal phrase. Since the informational focus is unmarked in Chinese, the term “focus marker” used by Shi and Li should refer to mark contrastive focus. The copula shì evolving into the copula in Old

Chinese underwent a process of analogization, modeling after the construction of the Old

Chinese verb wéi ‘to be.’ In Old Chinese, although wh-words occurred preceding regular transitive verbs, they systematically occurred following wéi ‘to be,’ i.e. zǐ wéi shéi ‘you are whom (who are you)?’ (Lunyu, 400 BCE) Therefore, when copula shì came into being, it occurred naturally preceding wh-words. At the same time, based on the

147

frequency, shì started to mark the focus of wh-words, but it did not become to signal contrastive focus as it is in Modern Chinese cleft sentences. In other words, if Shi and

Li’s term of “focus marker” refers to mark contrastive focus, then their claim regarding the copula shì further grammaticalized into a focus marker around 500 CE is insufficient and problematic. I propose that sentences such as (4a&b) that do not indicate contrastive meaning in discourse are regular copular sentences, in which shì does not function to signal contrastive focus, but encodes copulative meanings and informational focus just as in (3) and (4c).

In early Middle Chinese, the high frequency of instances of shì preceding a wh- word, such as shéi ‘who’ and hé ‘what,’ indicates that shì was employed to mark the interrogative focus of the wh-words. However, this function of marking interrogative focus did not extend to other linguistic categories since there was no exclusiveness and exhaustiveness in sentences like (4a&b), instead, shì as an element coalesced with the wh-word, and became an integral part of the grammatical word, as a result of grammatical constructionalization: [shì wh-word] [FM INTERR] > [wh-word] 

[INTERR]. Example (6) is the evidence:

(6) 汝為是誰？ rǔ wéi shì-shéi SG2 be who ‘Who are you?’ Xianyujing (CE.500)

148

In (6), wéi is the main verb following the subject and preceding the predicate that is wh-word shì-shéi ‘who.’ Clearly, shì here does not have the status of a copula or a further grammaticalized focus marker, instead, it was integrated into the wh-word shéi and they together formed one unit ‘who’ occurring at the predicate position of the sentence.

Lien (2009) proposes a similar idea by looking at the early Southern Min dialect dated back to the 16th. He argues the development of wh-word shìmiē ‘what’ (later shénme, with phonetic modification), underwent a process of lexicalization (Lehmann

2004) in which the focus marker shì preceding the wh-word mie lost its independence as a copula, fused into the latter and with the boundary collapse, the two elements blended into one wh-word.

Since shì in early Middle Chinese marking wh-word focus changed into part of a fixed interrogative item, it is unlikely that at the same period of time, the function of marking contrastive focus expanded from wh-words to a large range of other categorical elements such as NPs and VPs. In other words, [shì wh-word] did not enter into an abstract constructional schematic level [shì XP][FM Focus(contrastive)]. I propose that in early Middle Chinese, shì only marked the focus of wh-words, e.g. (5), and the two elements were gradually coalesced into one wh-word unit, e.g. (6). Other than marking wh-words, shì retained the copulative linking function indicating informational focus, e.g. (3) and (4a&b), and there was no cleft construction in early Middle Chinese.

Only along with the emergence of the cleft construction did the copula shì expand to the function of signaling NPs or VPs that immediately followed it as contrastive focus. This process took place in late Middle Chinese.

149

5.2.2 Nominalization [XP de]: in late Middle Chinese (700 CE- 1000)

The earliest utterances of nominalization [XP de] (also termed de construction among

Chinese linguists) appeared in Tang Dynasty (618-907 CE) with very low frequency, and the most-widely cited examples are in (7):

(7) a. 定知帏帽底24，仪容似大哥 dìng zhī wéimào de, yíróng sì dàgē must know hat NOM, appearance like big brother ‘(You) must know the one who’s hatted; he looks like (your) big brother.’ Chaoyejianzai (700 CE)

b. 张底乃我辈一般人，此终是其坐处 zhāng de nǎi wǒbèi yībān rén , cǐ zhōng shì qí zuòchù Zhang NOM PTCL we normal person, this eventually COP his sit-place ‘The one who’s named Zhang is a normal person like us; this will eventually be his seat.’ Suitangjiahua (700 CE)

Ota (1958) suggests that wéimào de ‘hat NOM’ in (7a) is an abbreviation of dài wéimào de ‘wear hat NOM (the one who’s hatted);’ similarly, Zhāng de ‘Zhang NOM’ in

(7b) is a short form of xìng Zhāng de ‘surname Zhang NOM (the one who’s named

Zhang).’ However, there is no evidence of the corresponding long versions of the similar examples at the moment. The two examples are the earliest occurrences of

24底 is a pre-modern variant of 的.

150

nominalization [XP de] in the history of Chinese. My data from CCL Classical Chinese

Corpus confirm that the earliest nominalizations [XP de] had the structure [NP de].

The particle de that occurs at the final position of a nominalization is termed nominalizer. Along with the emergence of the nominalizer, other functions of de also emerged, including associative (or genitive): shuǐ de làng ‘water ASSOC wave (the waves in the water),’ attributive (or relative): xiū de xíng ‘practice REL behavior (the behavior that one practices)’25 (Dunhuangbianwen 900 CE).

In terms of the formation of de, there has been an extensive debate among linguists and no consensus has been reached. Lü (1945) suggests that all the functions of de developed from the Old Chinese nominalizer zhě; Wang (1958) argues, based on the phonological liaison, that the Old Chinese attributive particle zhī, e.g. zuì wǒ zhī yoú

‘convict SG1 REL reason’ (the reason that you convict me) (Zuozhuan 400 BCE), was the origin of de; Mei (1988) further develops Wang’s argument stating that the attributive de first came into being from its source zhī, then with the influence of zhě which generally appeared at the phrase final position, the nominalizer de emerged; Jiang (1999) proposes de originally evolved from the localizer dǐ, e.g. dāng dǐ jiāo fàn ‘pan bottom burned rice’ (the burned rice at the bottom of the pan) (Shishuoxinyu 500 CE) and gradually assimilated the functions of zhě and zhī; Cao (1999) develops Jiang’s claim and argues that the three major functions of de that appeared in late Middle Chinese have three different origins: the attributive de developed from zhī; the associative de came from the localizer dǐ; the nominalizer de had the source of zhě.

25 The commonality and distinction between relativization and nominalization have been a hot topic among Chinese linguists, as they are marked by the same linguistic element de, and denote attributive meaning. In this thesis, I treat [VP/S de headNP] as relativization, and [NP/VP/S de] as nominalization.

151

How exactly the different functions of de emerged is not the focus of this chapter.

However, although there are different accounts proposed for the distinct functions of de, it seems most scholars agree that the nominalizer de is, in one way or the other, related to the Old Chinese zhě. In Old Chinese, zhě was a nominalizer that normally occurred at the phrase final position serving to turn an XP (XP=NP/VP/S) into a nominalization with the structure [XP zhě], which is equivalent to English ‘the one/thing.’ [XP zhě] appeared in a variety of positions26 including the subject position, the object position of a VO phrase, as in (8), etc. In early Middle Chinese around 500 CE, along with shì being frozen into a standard copula, it occurred in the predicate position of a copular sentence with very low frequency (only three occurrences are found in Shishuoxinyu), as in (9).

(8) 知之者不如好之者 zhī zhī zhě bù rú hào zhī zhě know it NOM NEG compare like it NOM ‘Those who know it cannot compare to those who like it.’ Lunyu (400 BCE)

(9) 讓是殺我侍中者，不可宥！ Ràng shì shā wǒ shìzhōng zhě , bù kě yòu Rang COP kill my servant NOM, NEG can forgive ‘Rang is the one who killed my servant; he cannot be forgiven!’ Shishuoxinyu (500 CE)

26 The use of the nominalizer zhě in Old Chinese is complex. It normally occurred in [XP de], but it can be optional. For example: we can find shàng zhě and shàng in one piece of text both meaning those that have higher position in the court. The nominalization in Old Chinese is out of the scope of this thesis.

152

Example (9) appears to be a specificational copular sentence with the structure [NP

COP NOM]. The context of (9) is: Rang kills one of the king’s servants and another general. Some high-ranked official in the court wants to excuse him. The king becomes furious and says that Rang is the one who kills his servant, and he cannot be forgiven.

From the context, the subject Rang is the one and only one who killed the speaker’s servant. The copular sentence here is background information to support the conclusion that Rang cannot be forgiven. Pragmatically, Rang is the topic of the copular sentence that is in contrast with anyone else who did not kill the speaker’s servant. The contrastive topic in (9) suggests that the copular sentences with the form [NP COP XP zhě] in 500

CE already had the function to signal contrastive. It is not entirely clear at this point how contrastive focus that is associated with the Modern Chinese cleft construction was systematically signaled. (9) is not necessarily a construction like the Modern Chinese cleft construction.

Around 700 CE, de was gradually used instead of zhě in the nominalization [XP zhě], and its function as a nominalizer emerged. The earliest instances of nominalization

[XP de] appeared to have the form [NP de], as exemplified in (7). They are found occurring at the subject position, as in (7b), or the object position of VO, as in (7a), but no examples are found in a copular sentence at this period of time.

After speakers chose de over zhě and the nominalizer de obtained its status, the host-class of [XP de] expanded from [NP de] to [XP de] (XP=VP/S/NP). (10) is an example:

153

(10) 师曰，说取行不得底， shī yuē : shuō qǔ xíng bù dé de, teacher say: say obtain do NEG obtain NOM, 行取说不得底 xíng qǔ shuō bù dé de do obtain say NEG obtain NOM ‘Teacher says: ‘through saying (you) obtain the things (you) cannot obtain through doing, through doing (you) obtain the things (you) cannot obtain through saying.’ Yunzhoudongshanwubenchanshiyulu (850 CE)

In (10), both of the two nominalizations have the serial verbs plus de with the second (main) verb negated and first verb indicating instrument/method [V Vneg de]: xíng bù dé de ‘the things through doing (you) cannot obtain’ and shuō bù dé de ‘the things through saying (you) cannot obtain.’ The nominalization, just like English restrictive relative clauses (Patten 2010), provides the semantically restricted set that can be specified by a definite referential member. As I will show in the next sub-section, around

900 CE, [XP de] was recruited at the predicate position of copular sentences, and it was extended to occur in more syntactic contexts.

5.2.3 The emergence of the cleft construction

5.2.3.1 The emergence of [NP COP NOM] (NOM=XP de)

In late Middle Chinese, along with zhě being gradually replaced by the nominalizer de and the occurrences of nominalization [XP de] becoming much more frequent, the sequence of [NP COP XP de] emerged.

154

As is shown in 5.2.2, the first occurrences of [XP de] around 700 CE did not appear in a copular sentence. With the host-class expansion of the nominalization, it is found to occur at the post-copula predicate position in a copular sentence. One of the earliest examples of [NP COP XP de] is (11), found in a Tang Dynasty Buddhist text

Jizhoulinjihuizhaochanshiyulu ‘The collective words from Master Linji and Master

Huizhao from Zhenzhou’:

(11) 道流是尔目前用底 dàoliú shì ér mùqián yòng de Daoism COP SG2 currently practice NOM ‘Daoism is the thing you are currently practicing.’ Zhengzhoulinjihuizhaochanshiyulu (880 CE)

(11) is one of the early instances of the copular construction in which a nominalization consisting of a clause plus the nominalizer de [S de] ér mùqián yòng de

‘the thing you are currently practicing’ was recruited in the predicate position.

Semantically, similar to (9), (11) is specificational in that the post-copula nominalization conveys a restricted non-referential set ér mùqián yòng de ‘the thing you are currently practicing’ and the definite referential subject dàoliú ‘Daoism’ specifies the referent of it.

Like (9), pragmatically, since (11) is found in a Buddhist text, the topic of the sentence dàoliú ‘Daoism’ is clearly in contrast with the Buddhism that is advocated by the Masters.

This shows again that the copular construction with the form [NP COP XP de] in 880 CE expressed contrastive meaning, but it had not yet developed into a cleft sentence. If (11)

155

was already a cleft sentence or had the salient implicature of one, should be the contrastive focus signaling the presupposition ‘someone is practicing Daoism currently,’ and ‘you’ should involve exclusiveness and exhaustiveness. The context27 does not provide this indication and therefore (11) is not yet a cleft sentence.

Accordingly, although the nominalization was recruited into the predicate position of the copular construction and the new structure [NP COP NOM] was used, the particular specificational meaning with contrastive focus as in a cleft is not found in (11).

Since constructionalization requires a formnew-meaningnew pairing, consequently, the emergence of (11) was simply a form change, and did not undergo constructionalization, but it was a micro-step in the stage of pre-constructionalization of the cleft construction.

5.2.3.2 The emergence of the cleft construction

The structure of the post-copula nominalizations in (12III) is [S de] (here S= N V):

(12) I. 莫將浮賄施為 I. mò jiāng fúhuì shīwéi I. not take bribe behave

II. 非是菩薩行藏 II. fēi shì púsà xíngcáng II. NEG COP Buddha behavior

27 The context is: the Master says, ‘Daoism is the thing that you are practicing. Just like the Buddha, whenever you have doubts, do you seek help from outsiders? That’s right. If you cannot get help from outsiders, you cannot obtain it from inside either.’

156

III. 此是俗門作底 III. cǐ shì súmén zuò de III. this COP layman do NOM ‘Don’t execute the bribe behavior; (it) is not Buddha’s behavior; it is laymen who do this.’ Dunhuangbianwen (900 CE)

The copular sentence in (12III) has the form [NP COP NOM] with the nominalization [S de] at the predicate position. The context of (12) is: Vimalakīrti tells one of his disciples not to execute the behavior of bribe, because that is not Buddha’s behavior; it is laymen who laymen do it. Semantically, it is specificational with the nominalization [S de] ‘the things that laymen do’ indicating a non-referential set that is specified by the subject cǐ ‘this’ (the bribe behavior). As for the information structure,

(12III) indicate the presupposition ‘some people execute the bribe behavior’ asserted by the post-copula NP ‘laymen’, whereby the NP indicates contrastive focus indicated by the copula shì turning the presupposition into an assertion. It expresses the exclusiveness and exhaustiveness, as laymen are the only people who execute the bribe behavior in this context. Moreover, the focus súmén is clearly in contrast with púsà ‘Buddha’ in (12II) immediately preceding it.

The above analysis shows that the semantics of (12III) has the specificational meaning, like (3b), (9), (11), but it also indicates contrastive focus. Example (12III) has the form [NP COP NOM], and its emergence from the regular specification copular construction underwent the process of constructionalization:

157

[NPi COP NPj][SEMi specificational SEMj]

[NPi COP NOMj][SEMi specificational+contrastive SEMj]

The process of constructionalization involves gradualness, micro-steps. As it shows below, each step of the process was a constructional change, as it only involved either form or meaning change. However, it was both form and meaning change from the frequent copular construction to the cleft construction.

[NPi COP NPj][SEMi specificational SEMj]

[NPi COP [XP zhě]j][SEMi specificational SEMj]

[NPi COP [XP de]j][SEMi specificational SEMj]

[NPi COP [XP de]j][SEMi specificational + contrastive SEMj]

158

The constructionalization involves: a) host-class expansion and syntactic expansion, i.e. nominalization was developed at the predicate position of a copular sentence; and b) semantic-pragmatic expansion, the specificational meaning with contrastive focus and presupposition emerged. I hypothesize that it was starting in sentences like (12III) that the copula shì was expanded to signal contrastive focus. If we take a further look at the newly emerging cleft sentence, we can see the subject of (12III) cǐ ‘this’ is semantically co-referential with the implicit object of the nominalization. I suggest (12III) is one of the first attested cleft-obj occurrences in the attested record of the history of Chinese.

Accordingly, (12III) can also be schematized as [NPi COP [XP de]j][SEMi cleft-obj

SEMj].

From the above discussion, I hypothesize that the cleft-obj emerged out of the frequent copular construction [NP COP NP] around 900 CE, and was constructionalized through host-class, syntactic, and semantic-pragmatic expansion. With the constructionalization of the cleft-obj, the more schematic cleft construction occurred.

5.2.3.3 The emergence of the cleft-sbj

In the cleft-obj (12III), the subject of the sentence cǐ ‘this’ is semantically co-referential with the implicit object of the nominalization, the two cleft sentences of (13) show that the two subjects are co-referential with the subjects of the nominalizations.

159

(13) I. 天下人总是学得底 I. tiān xià rén zǒng shì xué dé de I. people under heaven always COP study obtain NOM

II. 某甲是悟得底 II. mǒujiǎ shì wù dé de II. SG1 COP enlighten obtain NOM ‘It is through study that people under heaven always obtain (the state of Chan); it is through enlightenment that I obtain (it).’ Chanlinsengbaozhuan (1100 CE)

Example (13) is found in Chanlinsengbaozhuan (1100 CE), a Northern Song (960

CE-1127) Chan Buddhism classic. The context is about a little monk: one day when the little monk is meditating, he suddenly feels enlightened; he instantly gets up and goes to see the abbot; then he says to the abbot that it is through study that people always obtain the state of Chan, but it is through enlightenment that he himself obtains it. Both the sentences in (13) have the structure of [NP COP NOM] with the nominalization [V V de], in which the first of the serial verbs indicates instrument/method. In comparison with (13) and (12III), both of the subjects of the two sentences, i.e. tiānxiàrén ‘people under heaven’ in (13I) and mǒujiǎ ‘I’ in (13II), are co-referential with the implicit subjects of the two nominalizations. Both sentences in (13) express the specificational meaning: the post- copula nominalizations form two restricted non-referential sets ‘those who obtain (Chan) through study’ and ‘those who obtain (Chan) through enlightenment’ that are respectively specified by the referential subjects ‘people under heaven’ and ‘I.’ The two sentences are obviously in contrast with each other. In (13I), the immediate post-copula verb xué

160

‘through study’ is the contrastive focus indicated by the copula shì asserting the presupposition ‘people under heaven obtain the state of Chan through some method.’ As opposed to ‘through study’ in (13I), (13II) has the contrastive focus wù ‘through enlightenment’ asserting the presupposition ‘I obtain the state of Chan through some method.’

I suggest that the subject-subject co-referentiality in (13) shows further semantic expansion since the cleft-obj was constructionalized. What’s more, in the following discussion, I will show the development of the cleft-sbj also involves syntactic expansion that is the optionality of the nominalizer de.

An interesting example occurred in Wudenghuiyuan (1252) as in (14).

Wudenghuiyuan is a collection of classics of Chan Buddhism edited by Puji from

Southern Song (1127-1279) who compiled three Chan classics from Northern Song (960

CE-1127) including Chanlinsengbaozhuan (1100) and two from Southern Song. In

Wudenghuiyuan, the story of the little monk being enlightened was rewritten and (14) is the new version of (13):

(14) I. 天下人总是参得底禅 I. tiānxiàrén zǒng shì cān dé de chán I. people under heaven always COP meditate obtain REL Chan

II. 某是悟得底 II. mǒu shì wù dé de II. SG1 COP enlighten obtain NOM ‘It is through meditation that people always obtain the state of Chan; it is through enlightenment that I obtain it.’ Wudenghuiyuan (1252)

161

Comparing (14) with (13), since both mǒujiǎ and mǒu are variants of the first person singular pronoun, (13II) and (14II) are almost exactly the same. As for (13I) and

(14I), they have the same subject tiānxiàrén ‘people under heaven’, same copula modified by the adverb zǒng ‘always;’ the differences show up in the post-copula elements: first, xué ‘through study’ in (13I) correspnds to cān ‘through meditation’ in

(14I); second, the nominalization xué dé de ‘study obtain NOM’ is replaced by a head NP with a relative clause cān dé de Chán ‘meditate obtain REL Chan’.

In 5.2.2, I have mentioned that three major functions of de emerged in late Middle

Chinese: the nominalizer de, the attributive (relative) de and the associative (genitive) de.

The de in (14I) is the attributive (relative) de introducing a relative clause for a head noun, which probably developed from the Old Chinese zhī (although no consensus has been reached on this point). It is distinct from the nominalizer de in (13I), which is related to the Old Chinese nominalizer zhě. Consequently, (14I) seems to be a copular sentence

[NP COP NP] that does not involve a nominalization at the predicate position.

However, (14I) appears to indicate exactly the same semantics and information structure as (13I). The immediate post-copula verb cān ‘through meditation’ encodes focus in contrast to wù ‘through enlightenment’ in (14II). Both (13I) and (14I) indicate the same presupposition ‘people obtain the state of Chan through some method.’ Based on the observation, I propose that (14I) also has the form [NP COP NOM] involving a nominalization [REL NP (de)] at the post-copula position with the sentence final nominalizer de being optional. Moreover, just like (13), the two sentences in (14) involve

162

the semantic subject-subject co-referentiality. Therefore, I hypothesize (14) provides a piece of evidence of syntactic expansion after the cleft-sbj came into being.

Example (15) presents two cleft-sbj sentences with the subject-subject co- referentiality. The post-copula nominalization consists of [VP de], the VPs fā chū lái-le

‘discharge come out-ASP’ and shōuliǎn xiàng lǐ ‘retain toward inside’ are the focuses in contrast with each other. In (15I), the post-copula VP is attached with the aspectual marker le and the nominalizer de is implicit, whereas in (15II), the focus VP and the nominalizer de is specified.

(15) I. 蓋仁是箇發出來了， I. Gài rén shì gè fā chū lái - le, Alas benevolence COP CL discharge come out-ASP, 便硬而強 biàn yìng ér qiáng then hard and strong

II. 義便是收斂向裏底， II. yì biàn shì shōuliǎn xiàng lǐ de, Righteousness then COP retrain toward inside NOM, 外面見之便是柔 wàimiàn jiàn zhī biàn shì róu outside see it then COP soft ‘Alas, the benevolence is the thing that discharges and comes out of (the body), and thus it is hard and strong; whereas the righteousness is the thing that retains toward inside of (the body), and thus is seen to be soft from outside.’ Zhuziyulei (1270)

163

(16I&II) are also cleft-sbj sentences in contrast to each other with co-referential subjects. The nominalizations have the structure of [ADV VP de], the post-copula focuses, the two VPs, are modified by the adverbs cháng ‘always’ and yǐ ‘already’ respectively. For (16II), the VP yǐ sàn le ‘already scatter ASP’ is attached with the aspecutal marker le, and the final nominalizer de is present.

(16) 神便是氣之伸， I. 此是常在底 shén biàn shì qì zhī shēn , I. cǐ shì cháng zài de god then COP air ASSOC stretch, I. this COP always exist NOM

鬼便是氣之屈， guǐ biàn shì qì zhī qū , ghost then COP air ASSOC crook, II. 便是已散了底 II. biàn shì yǐ sàn le de II. then COP already scatter-ASP NOM ‘Gods are just the stretch of air; this always exists. Ghosts are just the crook of air; (it) has already scattered.’ Zhuziyulei (1270)

The cleft-sbj emerged out of the cleft-obj around 1100 through semantic and pragmatic expansion, and further developed with syntactic expansion. The process also involves micro-steps as follows:

164

[NPi COP [XP de]j][SEMi cleft-obj SEMj]

[NPi COP [XP de]j][SEMi cleft-sbj SEMj]

[NPi COP [XP (de)]j][SEMi cleft-sbj SEMj]

Each step here was a constructional change, however it was constructionalization with both form and meaning change from the cleft-obj to the developed cleft-sbj. When the cleft-sbj emerged, it together with the cleft-obj as two subschemas entered into the more schematic abstract cleft construction that parallels to equational and specificational copular constructions in the taxonomy of the prototypical copular construction. (See the taxonomy in Chapter 3 Figure 3.2)

This section has given a systematic analysis of the historical emergence of the cleft construction. I have argued the cleft-obj was constructionalized in late Middle Chinese out of the specificational copular construction [NP COP NP]. The cleft-obj was constructionalized through the syntactic and host-class expansion where the nominalization was recruited to the post-copula position, and the copula started to indicate contrastive focus. It developed the specificational meaning with contrastive focus out of regular specificational meaning. After the cleft-obj was constructionalized, it underwent further semantic-pragmatic expansion, e.g. subject-subject co-referentiality, thus the cleft-sbj emerged, and it further developed with syntactic expansion (the

165

optionality of the nominalizer). cleft-obj and cleft-sbj became two subschemas of the cleft construction. In the next section, I will focus on how and why the new constructions emerged using the framework of constructionalization.

5.3 Constructionalization

In this section, I consider the motivation and mechanism of the process of the constructionalization of the cleft construction [NPi COP NOMj][SEMi specificational+contrastive SEMj].

5.3.1 Motivation: Analogy and pragmatic inferencing

In 5.2.1, I have shown that in early Middle Chinese, shì was already a standard copula with the frequent form [NP COP NP], and the encoded meanings of predicational and specificational were also crystalized. In late Middle Chinese, the nominalization construction [XP de] occurred, whose encoded meaning, expressing a non-referential but restricted set, partially coincided with that of the specificational copular sentences. The semantic relatedness enabled the nominalization to be taken as part of the specificational copular sentences.

Along with [XP de] becoming more and more frequent, it was recruited at the predicate position of a copular sentence; thus, the sequence of [NP COP XP de] occurred.

I suggest that the recruitment of the nominalization into the predicate position of the copular construction of [NP COP NP] was motivated by their semantic relatedness and

166

the analogical thinking modeling the extant exemplar: the early Middle Chinese [NP COP

XP zhě]. Some early utterances of [NP COP XP de] expressed the regular specificational meaning following [NP COP XP zhě]: the nominalization denoted a non-referential restricted set, informational focus with new information, while the subject indicates the referential member, usually as the pragmatic topic to provide given information. They were propositions that did not indicate contrastive- presupposition.

However, around the same period of time in some other utterances, speakers started expressing contrast and presupposition to the string and the focus of [NP COP XP de] shifted-- the immediate post-copula element started to encode contrastive focus. The sentence was an assertion and provided contrast with a certain alternative in the previous or upcoming discourse. The emergent pragmatic inferencing of contrast and presupposition gave rise to reanalysis of the semantics of [NP COP XP de] and the bracketing of information. It still had specificational meaning, but the immediate post- copula element started to indicate contrastive focus and the rest of the sentence encoded the corresponding presupposition. The meaning of specificational+contrastive emerged.

The newly semanticized meaning provided the context for the copula shì to expand the function of signaling contrastive focus. Accordingly, the cleft-obj came into being.

After the cleft-obj emerged, and the specificational+contrastive meaning was crystalized, speakers started to express agentive meaning using this focusing structure, which gave rise to the semantic reanalysis as subject-subject co-referentiality and later on the syntactic expansion on the optionality of the nominalizer de.

167

Therefore, the steering factors that motivated the cleft-obj and cleft-sbj to emerge is essentially the speakers’ communicative strategies in asserting a presupposition with a contrastive focus and to express agentive meaning. The pragmatic inferencing led to semantic reanalysis, and combined with the newly emerged syntactic reconfiguration motivated by analogy. This observation is consistent with what is held by functional linguists working within a grammaticalization framework who concentrate on semantic- pragmatic forces, and with the emphasis on language use and language processing, they see change as gradual, non-discrete, and guided mainly by external or contextual forces.

Hopper and Traugott (2003:73) state what motivates language change are “speaker-hearer interactions and communicative strategies.”

5.3.2 Mechanism: Analogization and Reanalysis

I argue that the cleft-obj coming into being underwent syntactic reconfiguration through analogization. As has been pointed out in Chapter 2 section 2.3.3.2, analogical thinking is one of the enabling factors that allow change, though it does not necessarily bring about change, whereas analogization is a mechanism that gives rise to new structures. All analogizations involve reanalysis and “are instances of reanalysis, because each case of analogization involves a slight restructuring of what the speaker or hearer knows about a particular expression.” (Traugott and Trousdale 2011)

The recruitment of the nominalization [XP de], in which the non-referential restricted set in common with definite noun phrases is given, into the predicate position of a frequent copular sentence [NP COP NP] underwent analogization modeling the early

168

Middle Chinese string [NP COP XP zhě]. The process of syntactic reconfiguration is

[NP COP NP] [NP COP XP de], which later led to further reanalysis of semantic and syntactic reconfiguration including subject-subject co-referentiality and the optionality of the nominalizer de.

In the following discussion, I first sketch the constructionalization process of the change in terms of Croft (2001)’s Radical Construction Grammar model of construction, and then I illustrate the change in the frame of constructional schemas (a hierarchy) to show how the reanalysis took place. The constructionalization of the cleft construction is schematized as follows:

169

frequent copular construction  the cleft-obj  the cleft-sbj

(500 CE－) constructionalization (900 CE－) constructionalization(1100－)

• SY: [NP COP NOM] • SY: [NP COP NP] • SY: [NP COP NOM] NOM=XP (de) NOM=XP de • SE: speciicational; • SE: speciicational sbj co‐referential • PR: subject NP‐‐ • SE: speciicational; obj co‐referential • PR: contrastive focus topic; post COP NP‐‐ (immediate post‐COP informational focus • PR: contrastive focus element) (immediate post‐COP element) • DI: contrastive • DI: contrastive

Figure 5.1: Model of the development of the cleft-obj and cleft-sbj (form in the upper box and meaning in the lower box. SY: syntax PH: phonology SE: semantics PR: pragmatics

DI: discourse)

Figure 5.1 captures the process of constructionalization of the cleft-obj from the frequent copular construction and the constructionalization of the cleft-sbj from the cleft- obj. According to the development of constructional schematic taxonomy in Figure 5.2, change occurred in the level of instances of micro-constructions and tokens of use, e.g., cǐ shì súmén zuò de ‘this is what laymen do’ in (12), and later mǒujiǎ shì wù dé de ‘it is through enlightenment that I obtain (the state of Chan)’ in (13). The change is gradual

170

involving many micro-steps, from [NP COP NP] to [NP COP [S zhě]] to [NP COP [S de]], to [NP COP [VP de]], [NP COP [RC NP (de)]], [NP COP [ADV VP (de)]]. The constructionalization of the cleft-obj involved host-class expansion, syntactic expansion, semantic/pragmatic expansion, and the constructionalization of the cleft-sbj also involved semantic/pragmatic expansion and the syntactic expansion. When the cleft-obj and cleft- sbj occurred, the cleft construction emerged at a more abstract schema level, which became a subschema, paralleling to equational and regular specificational copular constructions in the constructional schematic taxonomy of the prototypical Chinese copular construction.

[NPi COP NPj]‐‐ [SEMi copulative linking SEMj] (500 C.E‐‐)

[NPi COP NPj]‐‐ [SEMi speciicational SEMj](500 C.E ‐‐) [NPi COP NPj]‐‐ [SEMi predicational SEMj](500 C.E‐‐)

[NPi COP NPj]‐‐ [SEMi equational [NPi COP NPj]‐‐ [SEMi SEMj](500 C.E‐‐) speciicational SEMj](500 C.E‐‐) [NPi COP NOMi]‐‐ [SEMi speciicational +contrastive SEMj] (900 C.E‐‐)

[NPi COP NOMj]‐‐ [SEMi cleft‐obj] SEMj (900 C.E‐‐) [NPi COP NOMj]‐‐ [SEMi cleft‐sbj SEMj] (1100‐‐)

Figure 5.2: the development of the constructional schematic taxonomy of the prototypical

Chinese copular construction

171

5.3.3 Conventionalization: Frequency

The mechanisms of change explain how one mental representation of a given expression can give rise to a different one. Bybee (2003) introduces frequency of use as one of the mechanisms: “new constructions come into being and spread by gradually increasing their frequency of use over time” (Bybee and McClelland 2005:387). Bybee (2003:602) states, “frequency is not just a result of grammaticalization, it is also a primary contributor to the process, an active force in instigating the changes that occur in grammaticalization. ” She distinguishes token frequency from type frequency suggesting token frequency is a mechanism that enables and brings about change at the first place and is also the outcome of change, whereas type frequency is the key to entrenchment or storage, which helps the outcome of change be frozen, fixed and conventionalized in a community.

Traugott and Trousdale (2011) point out that the enabling effect of token frequency is debatable. Their evidence is that according to the historical texts they have examined, several grammatical changes have started with very low frequency and sometimes continue to be used with low token frequency (also in Hoffmann 2005). My observation of the emergence of the Chinese cleft construction is consistent with Traugott and

Trousdale’s point here.

As I have shown in 5.2.2 and 5.2.3, the instances of nominalization [XP de] started with very low frequency in early Middle Chinese in the Tang Dynasty (618 CE -907).

According to Cao (1995) and Wu (1997), in Dunhuangbianwen (900 CE), in which the early instances of the cleft construction (such as example 12) are found, includes 12

172

tokens of the particle de and only 3 tokens of nominalization [XP de] in total including example (13) as the only one instance of [NP shì XP de] in the text. This shows when the pseudo-cleft construction emerged, in which a nominalization was recruited into the predicate position of a copular sentence, the structure of nominalization [XP de] itself was newly emerged and occurred with low frequency. What is more, the extant analogical exemplar that enabled the recruitment—the early Middle Chinese string [NP

COP XP zhě] was also sporadic. In Zhutangji (950 CE), 230 tokens of the particle de are found including 26 (11.3%) attested instances of nominalization [XP de] (Feng

2000:428), only one of them occurred in a pseudo-cleft sentence. After the cleft construction came into being, it still occurred with low frequency, and it remained low frequency for a long time even in Southern Song (1127-1279). In Zhuziyulei (1270), according to Zhu (1991) the token number of [NP COP XP de] is 606 (13.3% out of

4,560 tokens of particle de). From Zhuziyulei on, as [NP COP XP de] became more and more frequent, the cleft construction was entrenched, integrated and spread through the language system and was conventionalized in the language community. Therefore, frequency and repetition are undoubtedly a major factor in the fixing, freezing, and autonomizing associated with constructionalization; frequency itself appears implausible as a motivation for the onset of change.

5.3.4 Generality, productivity and compositionality

Traugott and Trousdale (2011) point out the relevant dimensions of constructionalization are generality/schematicity, productivity and compositionality.

173

Increase in generality is associated with generalization of meaning. With the emergence of the cleft-obj, the meaning of the nominalization as a whole was bleached. It no longer specifically denoted an entity, but indicated a presupposition asserted by a contrastive focus. The pragmatic implicature that enabled the constructionalization became part of the new meaning, which was more abstract, procedural and general than a nominalization on its own. When the cleft-sbj occurred, with the subject-subject co- referentiality, the nominal meaning of the nominalization was further bleached and generalized and it was possible for speakers to use the sentence final nominalizer de optionally.

Traugott and Trousdale treat increase in productivity as generalization of use. With the constructionalization of the pseudo-cleft and cleft construction, the copula shì functioned to mark the element immediately following it as focus. The increase in productivity is represented by the categorical expansion of the focus:

NPVVPADV, as exemplified in (12)-(16), which contributed to the micro-steps at the micro-construction level. What is more, as the cleft-sbj emerged, the nominal meaning of the nominalization became more bleached and generalized, and the verb of the nominalization started to entail temporal and aspectual events and situations. As we have seen in (15) and (16), both of the VPs in the nominalizations fā chū lái-le ‘discharge come out-ASP’ and yǐ sàn-le ‘already scatter-ASP’ were cliticized by the aspectual marker le. Accordingly, the nominalizations fā chū lái-le (de) ‘the thing that has come out’ and yǐ sàn le de ‘the thing that has scattered’ expressed situations involving aspectuality.

Increased generality and productivity led to increased schematicity. With the development of new micro-constructions, the schema came to have new construction-

174

types, and the cleft construction emerged, which became a sub schema in the in the constructional schematic taxonomy of the prototypical Chinese copular construction. The schematicity was increased.

The processes of the constructionalization of the cleft-obj and the cleft-sbj also involved decrease in compositionality. The contrastive focus indicated by the immedicate post-copula element within the nominalization gave rise to the decrease of its compositionality as a whole. Furthermore, its compositionality also decreased as the meaning of the nominalization became bleached and generalized, leading to the optionality of the nominalizer.

5.4 Conclusion

This chapter addresses the constructionalization processes of the cleft-obj and the cleft- sbj in the history of Chinese. I argued that the development of the cleft-obj involved the recruitment of the nominalization at the predicate position of the frequent copular construction through analogization, which led to pragmatic inferencing giving rise to semantic and syntactic reanalysis. Analogy and pragmatic inferencing were the enabling factors and analogization and reanalysis were the major mechanisms for the change.

After the cleft-obj came into being, the pragmatic inferencing of expressing agentive meaning enabled the semantic reanalysis of sbj co-referentiality, the cleft-sbj occurred.

After it occurred, it involved further syntactic expansion of the optionality of the nominalizer de. The cleft-obj and cleft-sbj became two subschemas under the more abstract schematic cleft construction.

175

Frequency and repetition were undoubtedly a major factor in the fixing, freezing, and autonomizing associated with constructionalization, however, frequency itself did not appear to be a motivation to enable change. I show Shi and Li’s claim that the copula shì was further grammaticalized into a contrastive focus marker around 500 CE is problematic. Instead, shì remained a systematic invariant copula verb since it was entrenched in early Middle Chinese, and it is with the emergence of the cleft construction that the copula expanded its syntactic function to signal contrastive focus.

176

Chapter 6

The Constructionalization of the Cleft Construction

This chapter provides a summary of the thesis and reviews issues for future research.

Section 6.1 is the summary is given and section 6.2 introduces possible directions for future study that related but not yet covered by this study.

6.1 Summary of the thesis

This thesis addressed the functions of shì in Modern Chinese and attempted to answer the following questions within the framework of Construction Grammar and

Constructionalization:

a. What is the copular construction?

b. What is the cleft construction?

c. How and why did the copular construction emerge?

d. How and why and did the cleft construction come into being?

This thesis started with the copula analysis of shì in Chinese cleft sentences. Based on a cross-linguistic understanding of the concepts of “copula” and “cleft,” I proposed that, in spite of its being a non-inflectional invariant predicate marker, the Modern

177

Chinese shì is a systematic copula verb. In Construction Grammar terms, the Chinese copular construction is a form and meaning pairing that has the form [(XP) COP XP] with [NP COP NP] as the prototype, and has the meaning of copulative linking, specifically predicational meaning and specificational meaning, i.e., a regular specificational sentence, an equational sentence, a cleft sentence.

I proposed a constructional schematic taxonomy for the prototypical Chinese copular construction [NPi COP NPj][SEMi copulative linking SEMj]:

[NPi COP NPj]<‐‐> [SEMi copulative linking SEMj]

[NPi COP NPj]<‐‐> [SEMi speciicational [NPi COP NPj]<‐‐> [SEMi predicational SEMj] SEMj]

[NPi COP NPj]<‐‐> [SEMi [NPi COP NPj]<‐‐> [SEMi [NPi COP NOMj]<‐‐ equational SEMj] speciicational SEMj] >[SEMi speciicational +contrastive SEMj]

[NPi COP NOMj]<‐‐ [NPi COP NOMj]<‐‐ >[SEMi cleft‐obj SEMj] >[SEMi cleft‐sbj SEMj]

Figure 6.1: The constructional schematic taxonomy of the prototypical Chinese copular construction

I suggested that the Chinese cleft sentences are special copular sentences and inherit the form and meaning properties from the schematic copular construction. The cleft construction has the form [NP COP NOM] (NOM= (ADV/TP/PP) VP/S de) and the specificational meaning with the contrastive focus that is encoded by the immediate post- shì element signaled by the copula shì.

178

Figure 6.2 presents a comparison of the form and meaning between the copular construction and the cleft construction:

• SY: [NP COP NOM]

• SY: [NP COP NP]

• SM: speciicational • SM: speciicational • PR: contrastive focus (the immediate • PR: informational focus (post‐COP post‐COP element) NP) • DI: contrastive

The copular construction The cleft construction

Figure 6.2: A comparison of the two constructions (form in the upper box and meaning in the lower box. SY: syntax PH: phonology SM: semantics PR: pragmatics DI: discourse)

I proposed that the cleft construction has two subschemas: cleft-sbj and cleft-obj.

The cleft-sbj involves the sbj co-referentiality (the subject of the sentence semantically co-referential with the subject of the nominalization) and the optionality of the nominalizer de; whereas a cleft-obj involves the obj co-referentiality (the subject of the sentence co-referential with the object of the nominalization) and the obligatory nominalizer.

Shì is generally treated to signal contrastive focus in the cleft construction. Some linguists including Huang (1998) claim that shì is not the copula verb in cleft sentences, but has the status of an adverb. I argued that the cleft sentences in these linguists’ view

179

are actually the examples of the cleft-sbj. Since the nominalizer de in the cleft-sbj is optional, there are two variants of the cleft-sbj: one with de at the end, the other without an explicit nominalizer. A search on CCL Modern Chinese Corpus shows the prototypical form of the cleft-sbj is [PRO COP (ADV/TP/PP) VP de] as only 23.1% (638 out of 2761 tokens) of the nominalizer de in the cleft-sbj is implicit in actual discourse.

The treatment of shì as an adverb fails to explain why shì can co-occur with an optional nominalizer de. With a systematic copula treatment of shì, the presence of de can be adequately explained as a nominalizer to preserve a specificational [NP COP NOM] structure.

In some Modern Chinese idioms, shì retains the classical trace of the demonstrative pronoun or demonstrative from Old Chinese, which supports the argument that the copula shì in Modern Chinese evolved from the demonstrative pronoun in Old Chinese. This thesis addressed how and why the demonstrative pronoun changed into a copula verb. I proposed that the emergence of the copular construction involved reanalysis of the topic- comment construction [(XPi) [shì XPj PTCL]][(Topici) [Anaphor SEMj

Declarative]], in which the demonstrative pronoun shì that occurred at the subject position of the comment clause (a classical copular clause) and functioned as the anaphor referring to the topic phrase evolved into the copula shì through a process of analogization modeling the structure of the full transitive verb wéi: [(XPi) wéi XPj

(PTCL)][(SEMi) BE SEMj (Declarative)] in Old Chinese.

The analysis focuses on the two conditions in which the grammatical constructionalization took place: 1) the semantic relatedness between the original construction and the target outcome: the function of anaphor linked the topic and the

180

predicate part of the comment clause, turning the whole topic-comment sentence into a copulative meaning; 2) the morphosyntactic contexts in which the change was enabled: the topic-comment construction [(XPi) [shì XPj PTCL]][(Topici) [Anaphor SEMj

Declarative]]. The constructionalization of the copular construction can be schematized as:

[(XPi) [shì XPj PTCL]][(Topici) [Anaphor SEMj Declarative]]

[(XPi) COP XPj (PTCL)][(SEMi) copulative linking SEMj (Declarative)]

Furthermore, I suggested the demonstrative pronoun shì changed into a copula in the copular construction. After the copula shì came into being, it competed with the Old

Chinese wéi ‘to be.’ Eventually shì won over and became the standard copula in Middle

Chinese. This is a functional change from discourse anaphoric to syntactic linking function, along with the decrease of the instances of classical copular sentence.

After the emergence of the copular construction [(XPi) COP XPj

(PTCL)][(SEMi) copulative linking SEMj (Declarative)], the occurrences [NPi COP

NPj][SEMi COP SEMj] became more and more frequent, and around 500 CE, the structure [NP COP NP] was conventionalized as the prototype for the copular construction and shì was already a standard copula verb. With the emergence of the copular construction, the copulative linking meanings of predicational and specificational were also crystalized.

181

The regular copular sentences encode predicate informational focus in terms of information structure, that is to say the subject usually encodes referential given information, generally a topic, and the post-copula predicate as a whole is the informational focus indicating non-referential new information.

I hypothesized that around 700 CE, the nominalization construction [XP de] emerged through analogization modeling the Old Chinese nominalization construction

[XP zhě]. Along with the syntactic and semantic expansion, [XP de] became more and more frequent, and was recruited at the predicate position of a copular sentence; thus, the sequence of [NP COP XP de] occurred. I suggested that the recruitment of the nominalization into the predicate position of the copular construction of [NP COP NP] was motivated by their semantic relatedness and the analogical thinking modeling the extant exemplar: the early Middle Chinese [NP COP XP zhě]. Early utterances of [NP

COP XP de] only expressed the regular specificational meaning: the nominalization denoted a non-referential restricted set, informational focus with new information, while the subject indicates the referential member, usually as the topic to provide given information. They were propositions that did not indicate contrastive-presupposition information structure.

However, around the same period of time in some other utterances, speakers started using [NP COP XP de] in contrastive context, and the contrastive focus emerged. The subject NP still indicated topic, but the immediate post-copula element started to encode contrastive focus signaled by the copula. The sentence was an assertion and provided contrast with a certain alternative in the previous or upcoming discourse. The emergent pragmatic inferencing of assertion gave rise to reanalysis of the semantics of [NP COP

182

NOM]. It still had the specificational meaning, on top of which the contrastive meaning was expressed. The newly semanticized meaning provided the context for the copula shì to expand the function of signaling contrastive focus. Accordingly, the cleft-obj came into being, in which the subject is semantically co-referential with the object of the nominalization.

After the cleft-obj emerged, and the meaning of specificational with contrastive was crystalized, speakers started to express agentive meaning using this focusing structure, which enabled the semantic reanalysis such as subject-subject co-referentiality and later on the syntactic expansion on the optionality of the nominalizer de.

This thesis analyzed the form and meaning of the copular construction and the cleft construction as well as how they developed in the history of Chinese. The analysis showed that it is in the cleft construction that the copula signals contrastive focus. Doing construction grammar avoids examining a linguistic element as an isolated atomic item, but analyzing it in the context of a construction. As we know, if a speaker intends to articulate a cleft sentence, but if s/he stops right after shì, the hearer will have no idea that the immediate following linguistic element will be a contrastive focus. Only if the speaker produces the whole sentence would the hearer possibly identify the contrastive meaning from the context. This shows it is the cleft construction as a whole that marks the contrastive focus, the copula only signals it, but does not mark it.

6.2 Thoughts on future study

This section reviews some possible issues for future study, including the development of shì as a bound morpheme, the historical development of nominalization and relativization

183

in the history of Chinese as well as the development of different approaches of contrastive focus.

6.2.1 Shì as a bound morpheme

In chapter 1, I showed that in Modern Chinese, shì is found as a bound form in a set of connectives, e.g. kěshì ‘but,’ jiùshì ‘even if,’ yàoshì ‘if,’ háishì ‘or.’ Previous research

(Chen 1993; Dong 2004; etc.) generally assumes that the development of the set underwent the process of lexicalization in the sense of univerbation (Lehmann 2002), with the syntactic strings [ADV/N/V COP] losing internal constituency and fusing into one unit over time. Did the copula shì coalescing with the preceding lexemes and changing into a bound morpheme undergo what Lehmann considers to be lexicalization, or a process of grammatical constructionalization? How and why did shì change into a bound morpheme? How does the change fit into the systemic change of Chinese over time?

6.2.2 Relativization and nominalization

In the discussion of how the cleft construction came into being, I briefly mentioned the development of the nominalization. In Old Chinese, zhě is the nominalizer that turns an

NP/VP/S into a nominalization, and zhī is a relative/attributive marker that introduces a relative clause preceding a head noun. However, zhī can also be found used as nominalizer in Old Chinese, and the nominalizer zhě in some circumstances can be optional. It is clear that both zhě and zhī were replaced by de since Middle Chinese, however there has been no consensus on how and why the changes occurred. How

184

exactly the nominalization and relativization emerge and develop in the history of

Chinese?

6.2.3 The development of contrastive focus

Because Chinese is a verb-medial (VO) language, informational focus in Chinese typically falls on the post-verbal elements. The information focus occurred along with the verbal construction in Old Chinese. This thesis addressed one type of contrastive focus: that indicated by the cleft construction. There are other ways to signal contrastive focus, e.g. lián construction, etc. What exactly are these approaches? How and why did they occur and develop? How did the cleft construction fit into the systemic change of contrastive focus?

6.3 Summary

This thesis investigated Chinese copular construction and its subschemas in the framework of Construction Grammar and Constructionalization. Future work remains to be done as how a particular change fits into the systemic change of Chinese overtime as well as whether the framework of this study can be extended to other languages and domains.

185

Bibliography

Anttila, Raimo. 2003. Analogy: the warp and woof of cognition. In Brian D. Joseph

and Richard D. Janda, eds., The Handbook of Historical Linguistics, 435-440.

Oxford: Blackwell

Barlow, Michael and Suzanne Kemmer, eds.. 2000. Usage Based Models of Grammar.

Stanford: CSLI Publications.

Benveniste, Emile. 1971 [1958]. Subjectivity in language. In Problems in General

Linguistics, 223-230. Trans. by Mary Elizabeth Meek. Coral Gables: University

of Miami Press.

Barðdal, Jóhanna. 2008. Productivity: Evidence from Case and Argument Structure in

Icelandic. Amsterdam: Benjamins.

Blom, Alied and Saskia Daalder. 1977. Syntaktische Theorie en Taalbeschrijving.

Muiderberg: Coutinho.

Biber, Douglas and Edward Finegan. 1988. Adverbial stance types in English.

Discourse Processes 11: 1-34.

Booji, Geert. 2010. Construction Morphology. Oxford: Oxford University Press.

Brems, Lieselotte. 2010. Size noun constructions as collocationally constrained

constructions: lexical and grammaticalized uses. English Language and

Linguistics 14: 83-109.

Brinton, Laurel J. and Elizabeth Closs Traugott. 2005. Lexicalization and Language

Change. Cambridge: Cambridge University Press.

186

Bybee, Joan, Revere Perkins and William Pagliuca. 1994. The evolution of grammar:

tense, aspect and modality in the languages of the world. Chicago: University of

Chicago Press.

Bybee, Joan L. 2001. Phonology and Language Use. Cambridge: Cambridge

University Press.

Bybee, Joan L. 2003. Mechanisms of change in grammaticization: the role of

frequency. In Brian D. Joseph and Richard D. Janda, eds., The Handbook of

Historical Linguistics, 602-623. Oxford: Blackwell.

Bybee, Joan and James L. McClelland. 2005. Alternatives to the combinatorial

paradigm of linguistic theory based on domain general principles of human

cognition. The Linguistic Review 22: 381-410.

Bybee, Joan L. 2006. From usage to grammar: the mind’s response to repetition.

Language 82: 711-733.

Bybee, Joan L. 2010. Language, Usage and Cognition. Cambridge: Cambridge

University Press.

Cao, Guanshun. 1986. Zutangji zhongde de, que, zhe. Zhongguo yuwen 3: 448-453.

Cao, Gangshun. 1995. Jindai hanyu zhuci. Beijing: Yuwen Press.

Chao, Yuen-ren. 1968. A Grammar of Spoken Chinese. Berkeley: University of

California Press.

Chen, Guanglei. 1993. Hanyu cifa lun. Shanghai: Xuelin chubanshe.

Chen, Jianming. 1986. Xiandai hanyu juxinglun. Beijing: Yuwen chubanshe.

Cheng, Lisa Lai-Shen. 2008. Deconstructing the shì…de construction. The

Linguistics Review 25: 235–266.

187

Choi, Kwok Tim. 2006. Formation of a Chinese cleft sentence. Paper presented at

The 22nd North West Linguistics Conference (NWLC 22).

Chomsky, Noam. 1981. Lectures on Government and Binding. Dordrecht: Foris.

Chomsky, Noam. 1995. The Minimalist Program. Cambridge: MIT Press.

Chu, Cheng-hsi. 1970. The Structure of shi and you in Mandarin Chinese. PhD

dissertation. The University of Texas at Austin.

Chiu, Bonnie Hui-chun. 1993. The Inflectional Structure of Chinese. PhD

dissertation, University of California in Los Angeles.

Comrie, Bernard. 1981. Language Universals and Linguistic Typology. Chicago:

University of Chicago Press.

Comrie, Bernard. 1998. Rethinking the typology of relative clauses. Language Design

1: 59-86.

Croft, William. 2001. Radical Construction Grammar: Syntactic Theory in

Typological Perspective. Oxford: Oxford University Press.

Croft, William. 2005. Logical and typological arguments for radical construction

grammar. In Jan-Ola Östman and Mirjam Fried, eds., Construction Grammars:

Cognitive Grounding and Theoretical Extensions, 273-314. Amsterdam:

Benjamins.

Croft, William and D. Alan Cruse. 2004. Cognitive Linguistics. Cambridge:

Cambridge University Press.

Culicover, Peter W. and Ray Jackendoff. 2005. Simpler Syntax. Oxford: Oxford

University Press.

188

Declerck, Renaat. 1986. Studies on Copular Sentences, Clefts and Pseudo-clefts.

Brussels: Leuven University Press.

Detges, Ulrich and Richard Waltereit. 2011. Moi, je ne sais pas vs. je ne sais pas,

moi. French tonic pronouns in the left vs. right periphery. Abstract presented at

IPra, Manchester, UK, July.

Diewald, Gabriele. 2006. Context types in grammaticalization as constructions.

Constructions SV: 1-9.

Dikken, Marcel den. 2005. Specificational copular sentences and pseudoclefts: a case

study. In Martin Everaert and Henk van Riemsdijk, eds., The Blackwell

Companion to Syntax. Oxford: Blackwell.

Dissel, Holger. 1999. Demonstratives: Form, function and grammaticalization.

Philadelphia: Benjamins.

Ding, Shusheng. 1980. Xiandaihanyu yufa jianghua. Beijing: Shangwu yinshuguan

Dong, Xiufang. 2004. Shi de jinyibu yufahua: you xuci dao cinei chengfen. Dangdai

yuyanxue 6.1: 35-44

Dowty, David R., Robert E. Wall, and Stanley Peters. 1981. Introduction to

Montague semantics. Synthese language Library. Dordrecht: Reidel.

Duan, Desen. 1991. Fuci zhuanhua wei lianci xianshuo. Guhanyu yanjiu 1: 47-51.

Engelbretson, Robert. 2007. Stance-taking in discourse. In Robert Englebretson, ed.,

Stance-Taking in Discourse: Subjectivity, Evaluation, Interaction, 1-12.

Amsterdam: Benjamins.

Evers, Arnold. 1975. The Transformational Cycle in Dutch and German. PhD

dissertation. Utrecht, distributed by IULC.

189

Fang, Mei. 1995. Hanyu duibijiaodian de jufa biaoxian shouduan. Zhongguo yuwen 4:

281-282.

Feng, Chuntian. 1992. Cong Wang Chong Lunheng kan guanxici ‘shi’ de wenti. In

Cheng Xiangqing, ed., Lianghan hanyu yanjiu. Jinan: Shangdong jiaoyu

chubanshe.

Feng, Chuntian. 2000. Jindai hanyu yufa yanjiu. Jinan: Shandong Education Press.

Feng, Shengli. 1993. The copula in classical Chinese declarative judgment sentences.

Journal of Chinese Linguistics 21: 277-311.

Fischer, Olga. 2010. An analogical approach to grammaticalization. In Katerina

Stathi, Elke Gehweiler, and Ekkehard König, eds., Grammaticalization: Current

Views and Issues, 181-220. Amsterdam: Benjamins.

Fillmore, Charles J., Paul Kay and Mary Kay O’Connor. 1988. Regularity and

idiomaticity in grammatical constructions: the case of let alone. Language 64:

501-38.

Fillmore, Charles J. Forthcoming. Berkeley Construction Grammar. In Hoffmann and

Trousdale, eds., Handbook of Construction Grammar. Oxford University Press.

Francis, Elaine J. and Laura A. Michaelis, eds.. 2003. Mismatch: form-function

incongruity and the architecture of grammar. Stanford: CSLI Publications, 259-

310.

Fried, Mirjam and Jan-Ola Östman, eds.. 2004. Construction Grammar in a Cross-

Language Perspective. Amsterdam: Benjamins.

Gao Mingkai. 1986. Hanyu yufalun. Beijing: Shangwu yinshuguan.

Givón, Talmy. 1979. On Understanding Grammar. New York: Academic Press.

190

Goldberg, Adele E. 1995. Constructions: A Construction Grammar Approach to

Argument Structure. Chicago: University of Chicago Press.

Goldberg, Adele E. 2002. Surface generalizations: an alternative to alternations.

Cognitive Linguistics 13: 327-56.

Goldberg, Adele E. 2006. Constructions at Work: The Nature of Generalization in

Language. Oxford: Oxford University Press

Goldberg, Adele E. Forthcoming. Constructionist approaches. In Hoffmann and

Trousdale, eds., Handbook of Construction Grammar. Oxford University Press.

Grice, H. Paul. 1969. Vacuous names. In Donald Davidson and Jaakko Hintikka,

eds., Words and objections, 118-145. Dordrecht: Reidel.

Gundel. Janette K. 1977. Where do cleft sentences come from? Language 53: 543-

559.

Guo, Xiliang. 1988. Guanyu xici ‘shi’ chansheng shidai he laiyuan lunzheng de jidian

renshi. Hanyu shilunji. Beijing: Shuangwuyinshuguan.

Harries-Delisle, H. 1978. Contrastive Emphasis and Cleft Sentences. In Joseph.

Greenberg ed., Universals of Human Language 4: Syntax. Stanford: Stanford

University Press.

Halliday, M.A.K. 1967. Notes on Transitivity and Theme in English. Parts 1 and 2.

Journal of Linguistics 8: 3-58.

Hartmann, Katharina and Malte Zimmermann. 2006. Morphological Focus Marking

in Gùrùntùm (West Chadic). Ms.. Humboldt University, Berlin.

Harris, Alice and Lyle Campbell. 1995. Historical Syntax in Cross-Linguistic

Perspective. Cambridge: Cambridge University Press.

191

Hashimoto, Anne Yue. 1969. The Verb ‘to be’ in Modern Chinese. Foundations of

Language Supplementary Series 9.4: 72–111.

Haspelmath, Martin. 2004. On directionality in language change with particular

reference to grammaticalization. In Olga Fischer, Muriel Norde, and Harry

Peridon, eds., Up and Down the Cline – the Nature of Grammaticalization, 17-

44. Amsterdam: Benjamins.

Heine, Bernd. 2002. On the role of context in grammaticalization. In Ilse Wischer and

Gabriele Diewald, eds., New Reflections on Grammaticalization, 83–101.

Heine, Bernd, Ulrike Claudi, and Friederike Hünnemeyer. 1991. Grammaticalization:

A Conceptual Framework. Chicago: University of Chicago Press.

Heine, Bernd and Tania Kuteva. 2002. World Lexicon of Grammaticalization. New

York: Cambridge University Press.

Heine, Bernd and Mechthild Reh. 1984. Grammaticalization and Reanalysis in

African Languages. Hamburg: Buske.

Hengeveld, Rob. 1992. Dynamic biogeography. New York: Cambridge University

Press.

Hengeveld, Kees. 1997. Non-verbal Predication. Berlin: Mouton de Gruyter.

Himmelmann, Nikolaus. 2004. Lexicalization and grammaticalization: opposite or

orthogonal? In Walter Bisang, Nikolaus P. Himmelmann, and Bjorn Wiemer,

eds., What Makes Grammaticalization? A Look From its Fringes and

Components, 21-42. Berlin: Mouton de Gruyter.

Higgins, Roger Francis. 1979. The Pseudo-cleft Construction in English. New York:

Garland.

192

Hilpert, Martin. 2008. Germanic Future Constructions: A Usage-based Approach to

Grammaticalization. Amsterdam: Benjamins.

Hoffmann, Sebastian. 2005. Grammaticalization and English Complex Prepositions:

A Corpus-based Study. London: Routledge.

Hong, Cheng. 1958. Lun Nanbeichao yiqian de xici. Yuyan yanjiu 2: 1-22.

Hopper, Paul. J. and Elizabeth Closs Traugott. 2003. Grammaticalization (revised

edn.). Cambridge: Cambridge University Press.

Hsieh, Chia-ling. 1998. Focusing Construction in Mandarin Chinese: Cleft and

Pseudocleft Sentences. Master Thesis. National Tsing Hua University.

Huang, Borong and Xudong Liao. 1991. Xiandai Hanyu. Beijing: Gaodeng jiaoyu

chubanshe.

Huang, C.-T. James. 1998. Logical Relations in Chinese and the Theory of Grammar.

New York and London: Garland Publishing Inc.

Huang, C.-T. James, Y.-H. Audrey Li, and Yafei Li. 2009. The Syntax of Chinese.

Cambridge: Cambridge University Press,

Huang, G. W. and Fawcett, R. P. 1996. A functional approach to two “focusing”

constructions in English and Chinese. Language Sciences 18: 179-194.

Huddleston, Rodney. 1971. The Sentence in Written English. Cambridge: Cambridge

University Press.

Hudson, Richard. 1990. English Word Grammar. Oxford: Blackwell.

Hudson, Richard 1997. The rise of auxiliary do: verb-non-raising or category

strengthening? Transactions of the Philological Society 95: 41-72.

193

Hudson, Richard. 2007. Language Networks: The New Word Grammar. Oxford:

Oxford University Press.

Hu Yushu. 1979. Xiandai hanyu. Shanghai: Shanghai jiaoyu chubanshe.

Jespersen, Otto. 1924. The Philosophy of Grammar. New York: Holt, Rinehart and

Winston

Jiang, Lansheng. 1990. Yiwen fuci ‘ke’ tanyuan. Guhanyu yanjiu 3: 44-50.

Jiang, Lansheng. 1999. Chusuoci de linggeyongfa he jiegouzhuci de de youlai.

Zhongguo Yuwen 2: 83-94.

Jin Lixin. 1995. “posi. you N” he “posi. shi N”, Yuyan jiaoxue he yanjiu 3: 82-84.

Kahn, Charles H. 1973. The Verb ‘be’ in Ancient Greek. In John W. M. Verhaar, ed.,

The Verb 'Be' and its synonyms Vol. 6. Dordrecht: Reidel.

Kay, Paul and Charles J. Fillmore. 1999. Grammatical constructions and linguistic

generalizations: the What’s X doing Y? construction. Language 75.1: 1–33.

Kiparsky, Paul. 1992. Structural case. Unpublished Ms. Berlin: Institute for

Advanced Study.

Kiss, Katalin É. 1998. Identificational focus versus information focus. Language 71:

245-273.

Kroch, Anthony. 1974. The Semantics of Scope in English. PhD dissertation. MIT.

Kuno, Susumo and Preeya Wongkhomthong. 1981. Characterizational and

identificational sentences in Thai. Studies in Language 5: 65-109.

Lambrecht, Knud. 1994. Information structure and sentence form. Cambridge:

Cambridge University Press.

194

Langacker, Ronald W. 1977. Syntactic reanalysis. In Charles N. Li, ed., Mechanisms

of Syntactic Change, 57-139. Austin: University of Texas Press.

Langacker, Ronald W. 1987. Foundations of Cognitive Grammar, Vol. I: Theoretical

Perspectives. Stanford: Stanford University Press.

Langacker, Ronald W. 2005. Construction grammars: cognitive, radical, and less so.

In Ruiz de Mendoza Ibañez, Francisco J., and M. Sandra Peña Cervel, eds.,

Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction, 101-

159. Berlin: Mouton de Gruyter.

Lakoff, George. 1987. Women, Fire and Dangerous Things. What Categories Reveal

about the Mind. Chicago: University of Chicago Press.

LaPolla, Randy. 2009. Chinese as a Topic-Comment (Not Topic-Prominent and Not

SVO) Language. In Janet Xing, ed., Studies of Chinese Linguistics: Functional

Approaches, 9-22. Hong Kong: Hong Kong University Press.

Leech, Geoffrey, Marianne Hundt, Christian Mair, and Nicholas Smith. 2009. Change

in Contemporary English: A Grammatical Study. Cambridge: Cambridge

University Press.

Lehmann, Christian. 1985. Grammaticalization: synchronic variation and diachronic

change, Lingua e Stile 20: 303-318.

Lehmann, Christian. 1995. Thoughts on Grammaticalization. Munich: Lincom

Europa (2nd, rev. ed. of Thoughts on Grammaticalization: A Programmatic

Sketch, 1982).

195

Lehmann, Christian. 2002. New reflections on grammaticalization and lexicalization.

In Ilse Wischer and Gabriele Diewald, eds., New Reflections on

Grammaticalization 1-18. Amsterdam: Benjamins.

Lehmann, Christian. 2004. Theory and method in grammaticalization. Zeitschrift für

Germanistische Linguistik 32: 152-187.

Lehmann, Christian. 2008. Information structure and grammaticalization. In Elena

Seoane and María José López-Couso, eds., in collaboration with Teresa Fanego.

Theoretical and Empirical Issues in Grammaticalization, 207- 229. Amsterdam:

Benjamins.

Levinson, Stephen C. 1983. Pragmatics. Cambridge: Cambridge University Press.

Li, Charles N. and Sandra A. Thompson. 1976. Subject and Topic: A new typology.

In Charles N. Li, ed., Subject and Topic. 458-489. New York: Academic Press.

Li, Charles N. and Sandra A. Thompson. 1981. Mandarin Chinese: A Functional

Reference Grammar. Berkeley: University of California Press.

Lightfoot, David. 1979. Principles of Diachronic Syntax. Cambridge: Cambridge

University Press.

Lightfoot, David. 1991. How to Set Parameters: Arguments from Language Change.

Cambridge: MIT Press.

Lü Shuxiang. 1979. Hanyu Yufa Fenxi Wenti. Beijing: Shangwu yinshuguan.

Lü Shuxiang. 1980. Xiandai Hanyu Babaici. Beijing: Shangwu yinshuguan.

Lü, Shuxiang. 1984. Lun di, de zhi bian ji di zi de youlai. In Shuxiang Lü, ed., Hanyu

yufa lunwenji. Beijing: Shangwu Yinshuguan.

Lü Shuxiang. 2002. Lü Shuxiang quanji. Shenyang: Liaoning jiaoyu chubanshe.

196

Lyons, John. 1968. Introduction to Theoretical Linguistics. Cambridge: Cambridge

University Press.

Lyons, John. 1982. Deixis and subjectivity: Loquor, ergo sum? In Robert J. Jarvella

and Wolfgang Klein, eds., Speech, Place, and Action: Studies in Deixis and

Related Topics, 101-124. New York: Wiley.

Ma Jianzhong. 1898. Mashi Wentong. Beijing: Shangwu yinshuguan.

Matsumoto, Yoshiko. 1997. Noun-modifying Constructions in Japanese: A Frame-

semantic Approach. Amsterdam: Benjamins.

Matthews, Stephen and Virginia Yip. 1994. Cantonese: A comprehensive Grammar.

London, New York: Routledge.

Mei, Zulin. 1984. Cong yuyanshi kan jiben yuanzaju binbaide xiezuo shiqi.

Yuyanxue conglun 13: 111-153.

Mei, Zulin. 1988. Ciwei di, de de laiyuan. Shiyusuo jikan 59.1: 141-172.

Montague, Richard. 1973. The proper treatment of quantification in ordinary English.

In K.J.J. Hintikka, J.M.E. Moravcsik and P. Suppes, eds., Approaches to Natural

Language, 221-242. Dordrecht: Reidel.

Michaelis, Laura A. 2003. Headless constructions and coercion by construction. In

Elaine J. Francis and Laura A. Michaelis, eds., Mismatch: Form-function

incongruity and the architecture of grammar, 259-310. Stanford CA: CSLI

Publications,.

Narahara, Tomiko. 2002. The Japanese Copula: Forms and Functions. New York:

Palgrave Macmillan.

197

Nunberg, Geoffrey, Thomas Wasow and Ivan Sag. 1994. Idioms. Language 70: 491-

538.

Nuyts, Jan. 2001. Subjectivity as an evidential dimension in epistemic modal

expressions. Journal of Pragmatics 33: 383-400.

Ota, Tatsuo. 1987. Zhongguo lishi wenfa. Beijing: Beijing University Press.

Partee Barbara H. 1999. Copula Inversion Puzzles in English and Russian. In

Katarzyna Dziwirek, et al eds., Annual Workshop on Formal Approaches to

Slavic Linguistics: 155-176. Ann Arbor: Michigan Slavic Publishers.

Patten, Amanda. 2010. Cleft Sentences, Construction Grammar and

Grammaticalization. PhD dissertation. University of Edinburgh.

Payne, Thomas E. 1997. Describing Morphosyntax. Cambridge: Cambridge

University Press.

Paris, Marie-Claude. 1979. Nominalization in Mandarin Chinese. Paris: Department

de Recherches Linguistiques, Université Paris VII.

Petré, Peter. 2012. General productivity: How become waxed and wax became a

copula. Cognitive Linguistics 23.1: 27-65.

Prince, Ellen F. 1978. A Comparison of wh-clefts and it-clefts in discourse. Language

54: 883-906.

Prince, Ellen. 1981. Toward a Taxonomy of Given-New Information, in Peter Cole,

ed., Radical Pragmatics, 223-255. New York: Academic Press.

Pulleyblank, Edwin G. 1995. Outline of Classical Chinese Grammar. Vancouver:

UBC Press.

198

Pustet, Regina. 2003. Copulas: Universals in the Categorization of the Lexicon. New

York: Oxford University Press.

Quirk, Randolph, Sidney Greenbaum, Geoffrey Leech and Jan Svartvik. 1985. A

Comprehensive Grammar of English Language. London and New York:

Longman.

Radford, Andrew. 1997. Syntax: A Minimalist Introduction. Cambridge: Cambridge

University Press.

Roberts, Craige. 1998. Focus, the flow of information and universal grammar. Syntax

and Semantics 29: 109-160.

Ross, Claudia. 1983. On the function of Mandarin DE. Journal of Chinese

Linguistics11.2: 214-246.

Rutten, Jean. 1991. Infinitival Complements and Auxiliaries. PhD dissertation.

University of Amsterdam.

Sag, Ivan A. 2012. Sign-Based construction grammar: an informal synopsis. To

appear in H.C. Boas and I.A. Sag, eds. Sign-Based Construction Grammar.

Stanford: CSLI Publications.

Schiffrin, Deborah. 1990. The management of a co-operative self during argument:

the role of opinions and stories. In Allen D. Grimshaw, ed., Conflict Talk:

Sociolinguistic Investigations of Arguments in Conversations, 241-259.

Cambridge: Cambridge University Press.

Stassen, Leon. 1997. Intransitive Predication. Oxford: Clarendon.

199

Stowell, Timothy. 1989. Subjects, Specifiers and X-bar Theory. In Mark Baltin and

Anthony Kroch, eds., Alternative Conceptions of Phrase Structure. Chicago: The

University of Chicago Press.

Shen Jiaxuan. 2008. Moving what? on emotional movement in ta shi qunian sheng

de haizi. Zhongguo Yuwen 326: 387-395.

Shi, Dingxu, 1994. The nature of Chinese emphatic sentences. Journal of East Asian

Linguistics 3: 81–100.

Shi, Yuzhi and Na Li 2001. Hanyu yufahua de lichen - xingtai jufa fazhan de dongyin

he jizhi. Beijing: Beijing University Press.

Sun, Chao Fen. 1996. 1996. Word-order Change and Grammaticalization in the

History of Chinese. Stanford: Stanford University Press.

Sun, Chao Fen. 2008. A study on subjectification and the BA construction. In Shen

and Feng, eds., Contemporary Linguistic Theories and Related Studies on

Chinese, 375-93. Beijing: Commercial Press.

Sweetser, Eve. 1988. Grammaticalization and semantic bleaching. in: Shelly

Axmaker et al., eds., General Session and Parasession on Grammaticalization,

389-405. Berkeley: Berkeley Linguistic Society.

Tang, Ting-chi. 1983. Focusing constructions in Chinese: cleft sentences and pseudo-

cleft sentences. In Ting-chi Tang, Robert L. Cheng and Ying-che Li, eds., Studies

in Chinese Syntax and Semantics, 127-226. Taipei: Student Book Co., Ltd.

Tang, Yuming. 1991. Zhuming zhongnian yuyanxuejia ziyuanji. Hefei:

Anhuijiaoyuchubanshe

200

Teng, Shou-Hsin. 1979. Remarks on cleft sentences in Chinese. Journal of Chinese

Linguistics 7.1: 101-114.

Traugott, Elizabeth Closs. 1988. Pragmatic strengthening and grammaticalization. In:

Proceedings of the 14th Annual Meeting of the Berkeley Linguistics Society:

Parasession on Grammaticalization, 406-416. Berkeley: Berkeley Linguistics

Society.

Traugott, Elizabeth Closs. 1995. Subjectification in grammaticalization. In Dieter

Stein and Susan Wright, eds., Subjectivity and subjectivisation, 31-55.

Cambridge: Cambridge University Press.

Traugott, Elizabeth Closs and Richard B. Dasher. 2002. Regularity in Semantic

Change. Cambridge: Cambridge University Press.

Traugott, Elizabeth Closs. 2003. Constructions in grammaticalization. In Brian D.

Joseph and Richard D. Janda, eds., A Handbook of Historical Linguistics, 624-

647. Oxford: Blackwell.

Traugott, Elizabeth Closs. 2007. The concepts of constructional mismatch and type-

shifting. Cognitive Linguistics: 18: 523-557.

Traugott, Elizabeth Closs. 2008a. The grammaticalization of NP of NP constructions.

In Alexander Bergs and Gabriele Diewald, eds., Constructions and Language

Change, 21-43. Berlin: Mouton de Gruyter.

Traugott, Elizabeth Closs. 2008b. Grammaticalization, constructions and the

incremental development of language: suggestions from the development of

degree modifiers in English. In Regine Eckardt and Gerhard Jaeger, eds.,

201

Language evolution: cognitive and cultural factors, 219-250. Berlin: Mouton de

Gruyter.

Traugott, Elizabeth Closs. 2010. Grammaticalization. In Silvia Luraghi and Vit

Bubenik, eds., A Companion to Historical Linguistics, 269-83. New York:

Continuum Press.

Traugott, Elizabeth Closs, 2011. Grammaticalization and mechanisms of change. In

Heiko Narrog and Bernd Heine, eds., The Oxford Handbook of

Grammaticalization, 19-30. New York: Oxford University Press.

Traugott, Elizabeth Closs. 2012. Toward a Coherent Account of Grammatical

Constructionalization. Draft for a volume on historical construction grammar

edited by Elena Smirnova, Mirjam Fried, Spike Gildea, and Lotte Sommerer.

Amsterdam: Benjamins.

Traugott, Elizabeth Closs and Graeme Trousdale. 2010. Gradience, gradualness and

grammaticalization: how do they intersect? In Elizabeth Closs Traugott and

Graeme Trousdale, eds., Gradience, gradualness and grammaticalization, 19-44.

Amsterdam: Benjamins.

Traugott, Elizabeth Closs and Trousdale, Graeme. 2011. Manuscript.

Constructionalization and Constructional Changes.

Trousdale, Graeme. 2008a. Constructions in grammaticalization and lexicalization:

evidence from the history of a composite predicate construction in English. In

Graeme Trousdale and Nikolas Gisborne, eds., Constructional Approaches to

English Grammar, 33-67. Berlin: Mouton de Gruyter.

202

Trousdale, Graeme. 2008b. Constructions in grammaticalization and lexicalization:

evidence from the history of a composite predicate construction in the history of

English. In Graeme Trousdale and Nikolas Gisborne, eds., Constructional

approaches to English grammar, 33-67. Berlin: Mouton de Gruyter,

Trousdale, Graeme. 2010. Issues in constructional approaches to grammaticalization

in English. In Katerina Stathi, Elke Gehweiler, and Ekkehard König, eds.,

Grammaticalization: Current Views and Issues, 51-72. Amsterdam: Benjamins.

Tsao, Feng-fu. 1990. Sentence and Clause Structure in Chinese: A Functional

Perspective. Taipei: Student Book Co., Ltd.

Ungerer, Friedrich and Hans-Jörg Schmid. 1996. An introduction to cognitive

linguistics. London: Longman.

Vallduví, Enric and Maria Vilkuna. 1998. On Rheme and Kontrast. In Peter Culicover

and Louise McNally, eds., The Limits of Syntax, 79–106. New York: Academic

Press.

Verhagen, Arie. 2005. Constructions of Intersubjectivity: Discourse, Syntax, and

Cognition. Oxford: Oxford University Press.

Wang, Li. 1937. Zhongguo Wenfa zhong de Xici. In Li Wang. 1958. Hanyushi

Lunwenji, 212-276. Beijing: Beijing Press

Wang, Li. 1980. Hanyu shigao. Beijing: Zhonghua shuju.

Wang, Li. 1984. Wang Li wenji. Jinan: Shandong jiaoyu chubanshe.

Wang, Li. 1989. Hanyu yufashi. Beijing: Shuangwu Yinshuguan.

Wiedermannn, H. 1986. The logic of being in Thomas Aguinas. In Knuuttial S. and

Hintikka J., eds., The logic of Being: Historical Studies. Dordrecht: Reidel.

203

Wischer, Ilse. 2000. Grammaticalization versus lexicalization: ‘methinks’ there is

some confusion. In Olga Fischer, Anette Rosenbach and Dieter Stein., eds.,

Pathways of change: grammaticalization in English, 355-370. Amsterdam:

Benjamins.

Wu, Fuxiang. 1997. Jindai hanyu gangyao. Changsha: Hunan Education Press.

Xu, Dan. 2006. Typological Change in Chinese Syntax. Oxford: Oxford University

Press.

Xu, Liejiong. 2002. Hanyu shi Huati Jiegouhua Yuyan ma? Zhongguoyuwen 5: 400-

410.

Xu, Liejiong. 2004. Manifestation of informational focus. Lingua 114: 277-299.

Xu, Liejiong and Haihua Pan. 2005. Jiaodian jiegou he yiyi de yanjiu. Beijing:

Waiyuyanjiu chubanshe.

Yen, Sian L. 1986. The origin of the copula shì in Chinese. Journal of Chinese

Linguistics 14.2: 227-241.

Zhang, Bojiang and Mei Fang. 2001. Hanyu gongneng yufa yanjiu. Nanchang:

Jiangxi jiaoyu chubanshe.

Zhu, Dexi. 1982. Yufa Jiangyi. Beijing: Shangwu yinshuguan.

Zhu, Dexi. 1985. Hanyu fangyan zhong de liangzhong fanfu wenju. Zhongguo yuwen

1: 62-80.

Zhu, Minche. 1991. Zhuziyulei jufa yanjiu. Wuhan: Changjiang Wenyi Press.

Zhu, Yao. 1997. The Focus-marking Function of Shi in Mandarin Chinese. PhD

dissertation. University of Minnesota.

204