INFORMATION TO USERS

This manuscript has been reproduced from the microfilm master. UMI film s the text directly from the original or copy submitted. Thus, some thesis and dissertation copies are in typewriter face, while others may be from any type of computer printer.

The quality of this reproduction is dependent upon the quality of the copy submitted. Broken or indistinct print, colored or poor quality illustrations and photographs, print bleedthrough, substandard margins, and improper alignment can adversely afreet reproduction.

In the unlikely event that the author did not send UMI a complete manuscript and there are missing pages, these will be noted. Also, if unauthorized copyright material had to be removed, a note will indicate the deletion.

Oversize materials (e.g., maps, drawings, charts) are reproduced by sectioning the original, beginning at the upper left-hand comer and continuing from left to right in equal sections with small overlaps. Each original is also photographed in one exposure and is included in reduced form at the back of the book.

Photographs included in the original manuscript have been reproduced xerographically in this copy. Higher quality 6” x 9" black and white photographic prints are available for any photographs or illustrations appearing in this copy for an additional charge. Contact UMI directly to order.

UMI University Microfilms international A Bell & Howell Information C om pany 300 Nortfi Zeeb Road, Ann Arbor. Ml 48106-1346 USA 313/761-4700 800/521-0600

Order Number 9201707

Nouns, nominalization and denominalization in Classical Chinese: A study based onMencius and Zuozhuan

Liu, Cheng-Hui, Ph.D.

The Ohio State University, 1991

Copyright ©1092 by Liu, Cheng-Hui. Ali rights reserved.

UMI 300 N. Zeeb Rd. Ann Aibor, MI 48106

NOUNS, NOMINALIZATION AND DENOMINALIZATION

IN CLASSICAL CHINESE:

A STUDY BASED ON MENCIUS AND ZUOZHUAN

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy in the Graduate

School of The Ohio State University

By

Cheng-Hui Liu, B.A., M.A.

*****

The Ohio State University

1991

Dissertation Committee: Approved by

Frank Hsueh

Yan-Shuan Lao IAdviser James Tai Department of East Asian Languages and Literatures ACKNOWLEDGMENTS

I would like to express sincere appreciation to my adviser

Prof. Frank Hsueh for encouragement and insightful guidance through the years of my graduate study. Special thanks go to my committee members, Profs. James Tai and Y. S. Lao, whose valuable comments on the dissertation drafts allow me to improve my work. Thanks go to Debbie Knicely for typing the

Chinese characters for me. I am grateful for tremendous support from my parents and my friend Lee.

11 VITA

November 9, 1962 ...... Born--Kaohsiung, Taiwan

June, 1984 ...... B.A. National Tsing Hua University, Hsinchu, Taiwan

June, 1988 ...... M.A. The Ohio State University

Major Field: East Asian Languages and Literatures

Studies in Chinese Linguistics

111 LIST OF ABBRIVATIONS

ADJ = adjectival ADV = adverbial COMP = complementative DF = discourse functions 0 = objective P = predicator PRED = predicative PRED.a = PRED of descriptive sentences PRED.n = PRED of equational sentences PRED.V = PRED of narrative sentences SF = syntactic functions

BIHP Bulletin of the Institute of History and Philology, Academia Sinica BJDXXB Beijing Daxue Xuebao BMFEA Bulletin of the Museum of Far Eastern Antiquities BSOAS Bulletin of the School of Oriental and Aftican Studies CAAAL Computational Analyses of Asian & African Languages JGL Journal of Chinese Linguistics JCLTA Journal of Chinese Language Teachers Association LTS Language Teaching and Studies YYYJ Yuyan Yanjiu ZGYW Zhongguo Yuwen

IV LIST OF TECHNICAL TERMS

Agent Verbs: Used as in Clark & Clark (1979), which refer to denominal verbs whose parent nouns are in the agentive case.

Ambs : Short for ambivalents, a lexical category with both nominal and verbal capacities and naturally carries out the nominal and verbal functions (cf. Kennedy 1964).

Combinatorial Capacity: A term referred to by Lu (1942), Wang (1958), Kennedy (1964), Cikoski (1978), and Zhi (1985) as a factor for lower level phrasal constructions. For example, it is combinatorial capacity of verbs to take a nominal unit and form a [head-object] VP.

Decategorization: referring to a process which makes verbal items become less typical instances of the verb category so as to correlate with the nominal functions.

Denominalization: A term referring to zero conversion from nouns into adverbs or verbs. The conversion from nouns to verbs is better known as ’verbalization’.

Descriptive Sentences: One of the three basic sentence types classified by Liu (1958), the PRED of which is usually carried out by ambs.

Deverbal Entities: Verbs which are decategorized so as to correlate with the nominal functions. It is noteworthy that deverbal entities differ from typical nouns in the sense that they cannot naturally correlate with all nominal functions.

Discourse Functions : Functions classified from the pragmatic aspect, including reference, nucleus, and modification, each of which corresponds to certain syntactic functions. Reference corresponds to SUBJ/0; nucleus, to P/PRED; and modification, to ADJ/ADV/PRED.

Dominant Relation : The relation of a [head-object] VP, one of the two basic phrasal constructions of Classical Chinese which has to do with combinatorial capacity of verbs (cf. the term ’restrictive relation’). Equational Sentences: One of the three basic sentence types classified by Liu (1958), the PRED of which is usually carried out by nouns, and normally marked by a particle Ze-

Functional Categories: Classes of functions which include syntactic functions (SF) and discourse functions (DF).

Instrument Verbs: Used as in Clark & Clark (1979), which refer to denominal verbs whose parent nouns are in the instrumental case.

Lexical Categories: Classes of lexemes divided by virtue of corresponding functional categories.

Lexicalisation: a process through which contextually actuated denotations become fixed and are eventually listed in the inventory of the lexicon.

Location Verbs: Used as in Clark & Clark (1979), which refer to denominal verbs whose parent nouns are in the locative case.

Locatum Verbs: Used as in Clark & Clark (1979), which refer to denominal verbs whose parent nouns are in the objective case and denote objects to be placed in a location.

Marked: Adapted from Croft (1984), which refers to unnatural correlations between categories. An unnatural correlation is marked in that semantic drift is expected through the correlation (cf. the term 'unnatural correlation').

Narrative Sentences: One of the three basic sentence types classified by Liu (1958), the PRED of which is usually carried out by verbs.

Natural Correlation: A term from Croft (1984), which is concerned with the regular/unmarked correspondences between lexical categories and functional categories (cf. the term ’unmarked’).

Nominal Functions: The functions which naturally correlate with the lexical category of nouns, including reference (DF) and SUBJ/0 (SF).

Nominalization: A term used as in Hopper & Thompson (1984), which refers to decategorization of verbs into deverbal entities (cf. the term ’decategorization’).

VI Nouns: A lexical category primarily defined by the restrictive relation, the members of which naturally correlates with the nominal functions (cf. the term 'restrictive relation').

Parent (vs. Surface): A term used as in Clark & Clark (1979), which refers to the original state of a noun before zero- converted into a verb.

Profile: A term from Langacker (1987, 1988) which is used as the essential semantic criterion for identification of phrasal head, nominalization and verbalization.

Relative Positions: A term from Wang (1958) which refers to certain syntaxemes characteristic of certain functions (cf. the term ’syntaxeme’).

Relative Words: Nouns which occur in relative positions.

Restrictive Relation: The relation of a [modifier-head] NP, one of the two basic phrasal constructions of Classical Chinese which has to do with combinatorial capacity of nouns (cf. the term ’dominant relation’).

Result Verbs: Denominal verbs whose parent nouns are in the objective case and become resultative states implied in the derived verbs.

Surface (vs. Parent): A term used as in Clark & Clark (1979), which refers to the innovative state of a denominal verb.

Syntactic Functions: Functions classified in the syntactic aspect, which include SUBJ, PRED, P, 0, ADJ, ADV, and COMP.

Syntaxeme: A term from Mukhin (1985), which refers to basic syntactic unit/slot with a fixed grammatical function.

Target Verbs: The most readily computed denominal verbs in Classical Chinese, whose parent nouns are in the objective case.

Temporary Functions: A term from Wang (1920, 1940, 1958), used to label a syntactic function that the lexical category of a given lexeme is supposed not to correspond to.

Unmarked: Adapted from Croft (1984), which refers to natural correlations between categories. A natural correlation is unmarked in that no semantic drift is expected through this correlation (cf. the term ’natural correlation’).

Vll Unnatural Correlations: A term from Croft (1984), which is concerned with irregular/marked correspondences between lexical categories and functional categories (cf. the term ’marked’)»

Verbs: A lexical category primarily defined by the dominant relation, the members of which naturally correlates with the verbal functions (cf. the term ’dominant relation’).

Verbal Functions: The functions which naturally correlate with the lexical category of verbs, including nucleus (DF) and P/PRED (SF).

Zero Conversion: A conversion without morphological markers. In this study, this term is used to indicate conversion from nouns into verbs or adverbs.

Vlll LIST OF FIGURES

FIGURES PAGE

2.1. Croft Hypothesis ...... 15

2.2. Correlations between Syntactic and Lexocal C a t e g o r i e s ...... 27

2.3. Bi-Lexeme Correlations ...... 28

6.1. Markedness of Unnatural Correlations . . . 158

IX TABLE OF CONTENTS

ACKNOWLEDGMENTS ...... ii

VITA ...... iii

LIST OF ABBRIVATIONS...... iv

LIST OF TECHNICAL TERMS...... v

LIST OF FIGURES ...... ix

CHAPTER PAGE

I. INTRODUCTION ...... 1

1. Classical Chinese ...... 2 2. General Issues ...... 3 2.1. Tonal Derivation ...... 4 2.2. Zero C o n v ersion...... 5 2.3. Characters and Words ...... 5 2.4. Word Classes ...... 6

II. THEORETICAL ORIENTATION ...... 8

1. Preliminaries ..... 8 1.1. Lexical and Functional Categories . . 9 1.2. Prototype Theory ...... 12 1.2.1. Prototypicality in Nouns and Verbs 12 1.2.2. The Contrast of Marked vs. Unmarked T e r m s ...... 14 1.3. Semantic B a s i s ...... 16 1.3.1. Langacker’s Frame ...... 16 1.3.2. Implications ...... 19 1.4. Lexicalization ...... 21 2. Categorization ...... 22 2.1. Arguments against Categorization . . 23 2.2. Arguments for Categorization .... 24 2.3. Toward a Solution ...... 26 2.3.1. The Asymmetric Correlations .... 26 2.3.2. The Concept of ’Temporary’ .... 29 2.3.3. A Hypothesis ...... 31 3. Nouns as a Lexical Category ...... 33 3.1. Non-Verbs ...... 33 3.1.1. Adjectives or A m b s ...... 33 3.1.2. Nouns and A m b s ...... 3 5 3.2, Central and Peripheral Nouns , , , . 36 3.3, Nouns Converted into Adverbs and V e r b s ...... 38 4. Relevant Functional Categories , . . . 38

III, NOUNS AND THEIR SYNTACTIC FUNCTIONS . , , 40

1, Nominal Modifiers ...... 40 1.1. N u m e r a l s ...... 41 1.2. Demonstratives ...... 48 2, Nouns in Sentence T y p e s ...... 56 2.1, Sentence T y p e s ...... 56 2.2, Nouns in Equational Sentences , . , , 57 2.3, Nouns in Descriptive Sentences and Narrative Sentences ...... 58 3, Time and Place W o r d s ...... 59 3.1, Time and Place Words in Typical Nominal Functions ...... 59 3.2, Nouns in the Relative Positions , . , 62 3.2.1, Place Words in the Yu Phrases , , , 63 3.2.2, The Optionality of Y u ...... 66 3.2.3, Syntactic Positions of Relative Expressions of Time and Place , , . 73 3 ,'2 , 4 , A Discussion on Relative Positions 78

IV, NOMINALIZATION...... 80

1, Ambs in Nominal Functions ...... 83 1.1, Ambs as PRED of Equational Sentences 84 1.2, Ambs as S U B J / 0 ...... 84 1.2.1, Describing Ambs vs. Labelling Ambs 85 1.2.2, Conditions for De-ambs ...... 88 2, Verbs in Nominal Functions ...... 94 2.1, Verbs as PRED of Equational Sentences 94 2.2, Verbs as SUBJ/0 ...... 96 2.2.1, Verbs Which Do Not Serve as SUBJ/0 97 2.2.2, Deverbal Entities ...... 100 2.2.3, Relative Verbhood ...... 105 3, Verbs and the Sentence-Initial Position 110

V, DENOMINALIZATION ...... 115

1, Nouns as A D V ...... 119 2, Nouns as PRED in Non-Equational Sentences ...... 122 2.1. Nouns as PRED in Descriptive Sen t e n c e s...... 123 2.2, Nouns as P/PRED in Narrative Sen t e n c e s...... 124 2.2.1, Nouns as P R E D ...... 124 2.2.2. Nouns as P ...... 128 3, Categories of Denominal Verbs , , . . , 131 3.1, In the Agentive C a s e ...... 132 3.2. In the Instrumental Ca s e ...... 134 3.3. In the Locative C a s e ...... 136 3.4. In the Objective Ca s e ...... 139 3.4.1. Locatum Verbs ...... 139 3.4.2. Target Verbs ...... 140 3.4.3. Result Verbs ...... 142 4. Some Remarks on the Categories .... 142 5. The Pragmatic F a c t o r ...... 149 6. More Examples about Target Verbs . . . 152

VI. CONCLUSION...... 156

1. S u m m a r y ...... 156 1.1. The Full-Word C a t e g o r i e s ... 157 1.2. Correlations between Lexical and Functional Categories ...... 157 1.3. Unnatural Correlations and Semantic Evidence ...... 158 1.4. Context-Dependent Innovations .... 159 1.5. Significance of the Categorization . 160 2. Related I s s u e s ...... 161 2.1. Semantic Features of the SF C a t e g o r i e s ...... 162 2.2. Lexicon and Syntax ...... 164 2.3. The Pre-PRED P o s i t i o n ...... 165 2.4. Abstract N o u n s ...... 166 2.5. Denominal Verbs in Classical Chinese 167

LIST OF REFERENCES...... 169 CHAPTER I

INTRODUCTION

0. The purpose of the present study is to investigate how words with independent semantic contents, traditionally known as shici [ ] 'full words’, in Classical Chinese can be properly categorized. It is focused on two related issues,

1.e., the basic distinction of nouns and verbs and lexical conversion without morphological markers.

The study uses data from the texts of Zuozhuan [ ] and Mencius [ ] to analyze syntactic behaviors of full words in the language. It is assumed that lexical categories regularly correspond to functional categories. The lexical categories of nouns and verbs can be well-defined on the basis of some regular correlations. Lexical categories can go beyond the regular correlations and correspond to functional categories which regularly correspond to other categories.

Irregular correlations can serve as traces of lexical conversion.

It is important that irregular correlations always co­ occur with semantic drifts. On the basis of the consistent relations between irregular correlations and semantic drifts, this study contends that lexical conversion should exist in 2

the lexicon. Two basic processes of lexical conversion will

be discussed, viz. nominalization and denominalization.

Chapter II will provide the theoretical orientation on

compositional principles of the Chinese language and premises

from general knowledge of natural languages. It will explain

that the popular idea about temporary functions needs to be built on the notion of (zero) lexical conversion. Chapter III will demonstrate the basic roles typical nouns play in various

syntactic constructions. In Chapter IV, it will be displayed

that verbs can be nominalized and play some of the functions which are normally carried out by typical nouns. Chapter V

shows how nouns are converted into verbs and function as verbs do in various syntactic constructions. What kinds of semantic

drifts will occur with the lexical processes will also be discussed in Chapter IV and Chapter V.

1. CLASSICAL CHINESE. 'Classical Chinese’ has been used in various senses to refer to a language before Modern Chinese.

In its broad sense, it can be a label for the language of a

literary tradition wenvanwen [X'slX], which extended from the pre-Qin [ period through Tang-Song [ #5R] and down to the

Qing dynasty [ # ]. But in a narrow sense, it may mean the

language which was recorded mostly in the writings of the pre-

Qin intellectuals, particularly, those of the Warring States

[ ] period (475-221 B.C.). Norman (1988:83) thinks that it

refers to a written langauge from the end of the Spring and 3

Autumn [ ] period 1770-476 B.C) to the Han [91 ] (A.D. 24)

dynasty.

In the present study, ’Classical Chinese’ is used to

label the language of the texts of Zuozhuan and Mencius. It was probably recognized as a standard linguistic form in its own time.^

According to C h ’ien (1956), Mencius lived roughly from

390 to 310 B.C. The book of Mencius records his dialogs with people, and his comments on historical and political topics.

Zuozhuan is a chronical text covering the history around 240 years in the Spring and Autumn period.^ Yang (1982a) estimates

that it was written at some time between 403 and 386 B.C. by an unknown follower of Confucius. That is, the language used

in Zuozhuan and Mencius should be in circulation between the

5th and 3rd century B.C.

2. GENERAL ISSUES. In the following are some general issues about Classical Chinese which are closely related to this study. These issues are concerned with the grammatical nature of the language.

^ As far as this study is concerned, there is no serious grammatical discrepancy between Zuozhuan and Mencius.

2 Zuozhuan is an elucidation relative to events covered in Chunqiu.iing j (Spring and Autumn Annals). While the Annals collected records written by different historians, the elucidation was presumably completed by one auther. 4

2.1 TONAL DERIVATION. A number of important historical reconstruction works by mo d e m scholars support the hypothesis that there are word formation devices in Classical Chinese

(e.g., Karlgren 1933, 1949, 1956). Moreover, it is believed that tonal derivation is a device closely related to the derivation of nouns from verbs and vice versa (cf. Wang 1958,

Downer 1959, Chou 1961b, Mei 1990). Thus, despite the fact that there are no sufficient data to show word formation rules on a large scale, it is reasonable to assert the language’s sensibility to the ncun vs. verb distinction.

On the implication of tonal derivation to the categoriality of nouns and verbs, Zhu (1990:167) asserts:

Suppose that there is a language ... in which all the words in some syntactic positions denoting actions will in some other positions denote things with the same phonetic forms. At the same time, all the words in some positions denoting things will in some other positions denote actions with the same phenotic forms. Then, only under such circumstances can we say that this language makes no nouns/verbs opposition. (My trans.)

Due to the existence of tonal derivation (cf. Zhu’s B class), as well as the fact that there are unrelated words which are pure nominal or pure verbal (cf. Zhu’s C class) in Classical

Chinese, it is logical to say the language does hold the basic categorial distinction of nouns and verbs.^

^ Zhu’s B class and C class will be explained in the following section. 5

2.2. ZERO CONVERSION. Zero conversion is concerned with nominal and verbal variants derived from a single base form without overt marking. To realize zero conversion in a base form is to postulate one of the variants serves as root to the other variant (cf. Sanders 1988). The notion of zero conversion is useful for interpreting lexical conversion without overt morphological markers.

Zhu (1990) lists as examples nouns and verbs of Classical

Chinese in three different classes. Class A consists of noun/ verb pairs of exactly the same form without tonal derivation.

Class B consists of noun/verb pairs, one of which.is derived from the other by tonal derivation. Class C consists of nouns and verbs not in Class A and Class B.

Unlike pairs of Class B, pairs in Class A do not show overt evidence of derivation. But a derivational relation is still possible to postulate for the members of Class A, since a semantic relationship in between noun and verb in a pair is usually obvious. Moreover, parallel examples of many of the pairs in Class A can be found in Modern English (cf. Zhu

1990). This may suggest that such a zero lexical expansion phenomenon results from some universal principle of human languages.

2.3. CHARACTERS AND WORDS. Any linguistic study on historical Chinese documents has to cope with the discrepancy between the graphic system and the linguistic system. Despite 6 the fact that characters are used to stand for corresponding monosyllabic words, it is inevitable that two closely related words are occasionally represented under the same character, or one word is sometimes written by two characters.

The analysis on shi in Kennedy (1964) may serve as a good example for this potential problem in dealing with linguistic words on the data recorded by characters. Kennedy mistakenly treats the character [ $ ] as a single word (shi) in his study. As a result, he finds that the word shi cannot be analyzed to be a noun or a verb in Mencius since the nominal and the verbal uses of the word are statistically of the same ratio. However, as Chou (1961a) points out, the character stands for two lexical entries, a noun shi and a verb shi.

In the present study, the two terms ’character’ and

’word’ are used respectively to distinguish the ’graphic element’ from the ’lexical element’. On special occasions when a single character represents more than one word, the term

’ lexeme ’ will be used to put emphasis on the difference behind the form. Moreover, in order to avoid unnecessary controversy, those words which may cause confusion in analysis because of writing inconsistency will not be, as much as possible, chosen as examples to make an argument.

2.4. WORD CLASSES. Words in Classical Chinese are traditionally divided into two categories: the xuci [ J$|sl ]

’functional words’ and the shici ’full words’. The functional 7 words scarcely have independent denotations. Their role is to

indicate the syntactic relations between full words of a sentence (cf. Ma 1961, Yang 1920, Lu 1957, Dobson 1974).

Categorizing full words is one of the tasks of this study. As will be explained in Chapter II, there exist some problems in this task. On the distinction of nouns and verbs, although tonal derivation is traceable for some cases, the tonal evidence will not be actively used for our analyses in this study. One of the reasons is that dating tonal evidence remains an unsolved problem in previous studies on Classical

Chinese (cf. Zhou 1946, Downer 1959, Chou 1961b, Mei 1980)." CHAPTER II

THEORETICAL ORIENTATION

0. This chapter aims at setting the theoretical foundation for the present study of nominal expressions in Classical

Chinese. 'Nouns’ are to be categorized by a set of prototypes which show distinctive syntactic/semantic features from verbs.

This chapter will be divided into four parts. The first part is concerned with theoretical preliminaries. With respect to the nature of Classical Chinese, the adopted theories are function-oriented. The second part will review issues on full- word categorization and offer a tentative proposal on the verb/noun distinction. A key point in the third part is about the relation between ambs and abstract nouns in Classical

Chinese. Also in this part will be a detailed discussion on the noun category. Finally, functional words which are relevant to this study will be outlined in the last part of this chapter.

1. PRELIMINARIES. Classical Chinese is a language with a relatively limited marking system. For example, the derivational devices found in many other languages do not exist in Classical Chinese. But the lack of morphology by no

8 9 means implies that the language is immune to all the linguistic generalizations behind overt markers of other languages. As will be demonstrated in the following chapters, some morphosyntactically marked linguistic notions in other languages are also existent in Classical Chinese, though these notions are ’marked’ in some rather language-specific ways.

To be specific, the ’markedness’ in Classical Chinese is carried out by semantic variations. In order to display how the variations work in this language just as morphosyntactic markers work in other languages, a set of technical terms will be borrowed from studies in language typology as premises.

These premises are of explanatory importance for the discussions of the following chapters, and most of them underlie the typological, functional, and cognitive approaches.

1.1. LEXICAL AND FUNCTIONAL CATEGORIES. Two types of functional categories are regarded as essential to the grammar, viz. syntactic functions (SF) and discourse functions

(DF) .

The SF categories include subjective [ A # ] (SUBJ), predicative [ IR# ] (PRED), predicator [ ] (P), objective

[ ] (0), adjectival [ ] (ADJ), adverbial (ADV), and complementative [ ] (COMP). These categories are in a hierarchy as in the following : 10

S(ENTENCE) --- > (SUBJ) PRED

SUBJ ---> (ADJ) SUBJ

PRED --- > (ADV) PRED

PRED ---> PRED.a

PRED ---> (ADJ) PRED.n

PRED ---> PRED.v (COMP)

PRED.v --- > P (O)

O > (ADJ) 0

(SUBJ)-PRED is the highest level segmentation of a sentence, in which PRED is the nucleus and the only obligatory element.

ADV has a two-fold function; it modifies and locates PRED.

PRED stands for at least two types of predicatives in two types of sentences, the narrative sentences and the nominal sentences.

Some lower level constructions are referred to as outcomes of combinatorial capacity by Lu (1942), Wang (1959),

Kennedy (1968), Cikoski (1978), and Zhu (1985). P-0 indicates the combination of HEAD-OBJECT [ jîÈS ], verbs taking an internal argument to form a VP. ADJ-SUBJ, ADJ-0 and ADJ-PRED.n comes from the MODIFIER-HEAD [ fSJE ] construction, in which the head is usually a noun.^

^ It is under general compositional principles of Classical Chinese that a modifier of verbal expressions operates at a higher syntactic level. Therefore, it is compositionally infeasible to have such hierarchical rules as: P R E D > P 0 P > ADV P 11

In terms of the contrast of lexeme vs. syntaxeme, a VP with a verb and a nominal object means the combination of two lexemes, and P-0 means the combination of two syntaxemes.^

That is to say, any combination in terms of the SF categories is a combination of syntaxemes. Combinations of syntaxemes should be considered as purely functional, but combinations of lexemes are lexical. The notion of syntaxeme is important especially for ADV, which is usually a locative syntaxeme or a temporal syntaxeme.%

The SF categories have intrinsic properties which can be generalized into three DF categories: reference, nucleus, and modification (cf. Cikoski 1978, Croft 1984). SUBJ/0 falls into the scope of reference, PRED/P into that of nucleus, and

ADJ/ADV/COMP into that of modification. By virtue of Croft’s

(1984) natural correlation, PRED/P naturally correlate to nucleus. So do SUBJ/0 to reference and so do ADJ/ADV/COMP to modification.

^ The notion of syntaxeme comes from Mukhin (1985). As Mukhin puts it, 'a syntaxeme can be defined as an elementary syntactic unit (invariant), represented in the language by a system of its variants which may be expressed by both individual lexemes and syntactically indissoluble combinations of lexemes with auxiliary elements, for example, prepositions’ (p.404). Every syntaxeme can be labeled in terms of its semantic function, e.g. a locative syntaxeme, a temporal syntaxeme, and etc.

^ The notion of a locative or temporal syntaxeme is analogous to a term introduced by Wang (1958), viz. guanxi- wei [ ], which is crucially relevant to nouns. (More discussion on guanxi-wei will be in Chapter III.) 12

1.2. PROTOTYPE THEORY. The prototype theory adopted in this paper is from Rosch and her associates (e.g. Rosch 1973, 1977,

1978). Categorization in the prototype theory is a matter of

’centrality’. The central members, or prototypes, of different categories will define maximum categorial distinctiveness. In addition, the theory will be expanded to include markedness

in linguistic patterns, as suggested by Croft (1984).

1.2.1. PROTOTYPICALITY IN NOUNS AND VERBS. Hopper & Thompson

(1984) set up a basic distinction of nominal roots from verbal roots by virtue of their lexical contents. But semantics alone are insufficient for the roots to bring about categorial prototypicality. A root is unlikely to achieve the salience of its category in context-free situations. Only in the prototypical discourse functions will the roots fulfill their prototypicality.

The contrast of nouns and verbs is a matter of degree with respect to the prototype theory. The nominal or verbal prototypes define the maximum distinctiveness of nouns or verbs. But, on the other hand, when an instantiated entity

fails to achieve its categorial salience, it will be less distinctive from the other category (cf. Hopper & Thompson, p.708-9).

The categorial prototypes are overtly marked by morpho- syntactic trappings in many languages. The most prototypical

instantiation of an entity is attached by a full range of its 13 trappings. For example, when the English noun 'fox' achieves its salience in a sentence like 'Early in the chase the hounds started up AN OLD RED FOX, and we hunted him all morning', as exemplified by Hopper & Thompson (p.708), it is marked as either singular or plural, and it may take a determiner, an adjective or a demonstrative.

Moreover, de-categorization and category shift can be defined by the prototype theory. De-categorization indicates that an entity in certain discourse function becomes totally insulated from the categorial trappings and thus neutralized in terms of categoriality. Category shift indicates that a member of a specific category in a certain discourse function achieves the salience of some other category.

It should be noted that de-categorization and category shift are overtly marked in Modern English. For example, when a verb functions as a subject in a discourse, it will be de­ categorized and it will occur in a gerund or an infinite form.

When a noun functions as a predicator, it is converted into a verb. It will take an inflectional marker, -s or -ed, as a verb does. But, in Classical Chinese, no such signals of de­ categorization and category shift exist.

On the other hand, semantic variations can be observed in both languages. Since de-categorization and category shift are overtly marked in Modern English, the semantic evidence may be redundant. It is nevertheless the only evidence that can be relied on to recognize these grammatical processes in 14

Classical Chinese. As will be illustrated in the following chapters, semantic variations will be taken into account as essential markers for de-categorization and category shift.

1.2.2. THE CONTRAST OF MARKED VS. UNMARKED TERMS. Croft

(1984) interprets prototypes of syntactic categories by the markedness theory. A syntactic category is unmarked when it naturally correlates to semantic and pragmatic categories.*

Otherwise, it will undergo a category-changing morphosyntactic process. Croft tests his hypothesis on a set of Russian roots, and finds .that it correctly predicts the relationship of the roots with the semantic and pragmatic categories.

According to Croft, nouns naturally correlate to

'objects' (a semantic category) and 'reference' (a pragmatic category). Verbs naturally correlate to 'actions’ and

'predication', and adjectives to 'properties' and

'modification'.

Consequently, Croft's hypothesis predicts category- changing morphosyntax, as in Figure 2.1:

* Syntactic categories, in Croft's paper, mean nouns, verbs, and adjectives, which are identified as lexical categories here. The reason for this discrepancy is that morphosyntax makes syntactic categories and lexical categories the same phenomena in different aspects in Russian. For a language without morphology like Chinese, it becomes a goal to prove that syntactic and lexical categories are the same in different aspects. 15

Reference Modification Predication

Objects Unmarked Marked Marked

Properties Marked Unmarked Marked

Actions Marked Marked Unmarked

Figure 2.1.® Croft Hypothesis

In this figure, ’unmarked’ indicates a natural correlation, and ’marked’ indicates an unnatural correlation.

The hypothesis predicts that a category-changing morphosyntax will occur to unnatural correlations, as is true in the

Russian roots.

But, this hypothesis may not work in Classical Chinese exactly the way it does in Russian. In the first place, the markedness in Classical Chinese is primarily carried out by semantic variation, rather than morphosyntax. Moreover, the semantic category of ’properties’ and the pragmatic category of ’modification’ may be less explicit in Classical Chinese than they are in Russian.® (For the discussion about adjectives, see 3.1.)

Thus, we may conclude, cross-category correlations are a universal tendency, but the categories that actually are

® Figure 2.1 is a simplified version of the figure in Croft (1984:58).

® Croft also tests on an intermediate semantic category, viz. ’states’. As it turned out in the Russian data, ’states’ naturally correlates to nouns and adjectives. That is, in neither of these correlations does a morphosyntactic process occur. This may show that category boundaries are hot of equal distinctiveness. 16

correlated should vary from language to language. Moreover,

it is also a universal tendency that unnatural correlations are marked, though the markedness is often carried out in

language-specific ways.?

1.3. SEMANTIC BASIS. The semantic theory of the present

study is based on the frame described in Langacker (1987,

1988), which is used on semantic grounds to justify the distinction of nouns and verbs and the notion of category shift.

1.3.1. LANGACKER’S FRAME. According to Langacker (1988), polysemy should be treated as a natural phenomenon in human languages. In Langacker’s network model set up for polysemy, there are nodes within a lexeme, each of which represents a

frequently-used sense under the lexeme. Between the nodes is held either a specialization or an extension relationship.

Ideally speaking, there exists a prototype for every lexeme, although the prototype may not always be identifiable. (For the network, see Langacker 1988:50-3.)

? This conclusion is in accordance with a claim of Rosch (1977:3), which says that ’... the content of categories and the nature of prototypes of categories can be expected to vary across cultures but the principles of category formation and of the development of prototypes can be expected to be universal.’ 17

According to Langacker (1987), moreover, intrinsic

content itself is insufficient for the distinction of nouns

and verbs. The categorial difference rather depends on how

expressions are profiled, just as Langacker (p.68) states;

A nominal predication presupposes the interconnections among a set of conceived entities, and profiles the region thus established. By contrast, a relational predication presupposes a set of entities, and it PROFILES the INTERCONNECTIONS among these entities.

The entities are interconnected somehow in mental experience.

By predication, Langacker means a semantic structure of any size and complexity. Given a shared set of interconnected entities, a noun profiles a region of the entities, whereas a verb profiles the interconnections of the entities.

A region may be bounded or unbounded. A bounded region

indicates the sub-category of count nouns, while an unbounded one indicates the sub-category of mass nouns. Verbs are relational by nature, since they profile interconnections.

More importantly, verbs are relational predications which designate a process with time involved. In opposition, atemporal relational predications correspond to prepositions, adjectives, adverbs, infinitives, and participles.

As to profile, it has to do with the notion of base

(Langacker, p,56); 18

The BASE for a linguistic predication is its domain, i.e. the cognitive structures it presupposes; its PROFILE is a substructure of the base that is elevated to a distinctive level of prominence as the entity which the expression DESIGNATES.

On a common base, alternate profiles will bring about semantic distinctions. The choosing of profile explains the distinction between circle (n.) and round (adj.), or explode (v. ) and explosion (n.) (Langacker’s examples), i.e. categorial differences.

Profiling can further explain the procedure of nominalisation. In theory, a base may generate a set of profiles, depending on how the base is conceived. But in a specific language, only one or two of the possible profiles on a base will be conventionalized and lexicalized in the inventory of lexicon under a category.

The conventionalized domain, in turn, may serve as a base of another profile. That is how category shift happens.

Category shift from verbs to nouns, viz. nominalisation, is defined by Langacker (p.90-1) as follows:

... [nominalisation] takes the process designated by the verb stem as its base, and within this base it selects for profiling the higher-order region comprising the component states.*

* Category shift is not limited to the lexical level. Since the predication can be a linguistic expression of any size, and since profiling can operate on bases of different complexity, phrases and clauses are subject to category shift. 19

The higher order region may be bounded or unbounded. If the profiling designates an instance of the perfective aspect indicated by the verb, the region is bounded, and thus a count noun. When the profiling designates the imperfective aspect, the region is abstract, and hence an abstract noun. (For more about perfective vs. imperfective aspects, see Langacker

1987:78-84. )

In addition to category shift, profile also has to do with linguistic salience, since it functions to highlight a specific spot of the base. Thus profile explains how phrasal head can be conceptually recognized: In a phrase, 'that component which contributes its profile to the expression as a whole can be identified as the "head" ' (1988:75). As

Langacker exemplifies, in an adj + noun construction, the noun is the head because it is the region (the profile of the

'thing', noun) rather than the atemporal relation (i.e., adjective) that is designated to the phrase.

1.3.2. IMPLICATIONS. Langacker's frame may help explain the grammar of Classical Chinese in two aspects. First, this frame comprises conceptual components which are more analysable in a language with little morphology. For example, the meaning relation of the elements in a bi-lexeme construction may be easy to realize but difficult to describe, since the relation 20 is formally unmarked.® Now the modifier-head and the head- object relations can be defined by first recognizing the head, the profile which is contributed to the whole expression.

Secondly, this cognitive frame explains why polysemy is so prevailing and yet causes little confusion among native speakers. It also explains how category shift can be a frequent aspect of the language, implicitly marked, and yet never cause serious ambiguity. These perplexing phenomena thus acquire a possible explanation.

But, on the other hand, Langacker’s frame may raise a serious question about the necessity of lexical categorization in Classical Chinese. Given the assumption that category shift is conceptually easy to have access to and that polysemy is a norm for the lexicon, a lexeme can alternatively be visualized as consisting of related nodes with pure senses.

One of the senses will become prominent at a higher syntactic level, and only then the categoriality can be realized. If so, lexemes in context-free situations should indeed be category- free .

The strongest argument in favor of lexical categorization may be that it helps level off redundancies in the lexicon.

Given that conventionalized senses are retained in the memory system and that all possibly related senses are subject to be

® Wang (1955) admits that his effort of adopting the theory of relative ’ranks’ from Jespersen to describe semantic relations of immediate constituents should be given up. 21

innovated through certain operations, there will be as limited

information as possible in the inventory of lexicon.

From this point of view, profiling plays a critical role

of novel expressions. Again, in English, the operations of profiling for innovations can be recognized by derivational

or inflectional evidence. But in Classical Chinese, the operations are primarily recognized in semantic terms.

1.4. LEXICALIZATION. The present view about structure of the lexicon is similar to that described in Lehmann (1990). On the one side is the ideosyncratic sphere, consisting of an open- ended list of primitive conceptions under an infinite number of lexemes. On the other side, the lexicon contains systematic aspects which overlap with syntactic aspects. Meanwhile, lexicalisation will be regarded as a language-specific issue.

As illustrated by Lehmann, a grammatical concept may be lexicalized in the form of a primary category in one language but morpho-syntactically derived from a primary category in another language.

With Langacker’s network hypothesis, it will be hypothesized that the fewer the categories are involved in the lexicalisation, the more complex is the inner structure of lexemes. It seems that this can be confirmed by the lexical system of Classical Chinese. 22

2. CATEGORIZATION. Many authors have put much effort to

assert the universality of the noun/verb distinction in

natural languages (e.g., Robins 1952, Hopper & Thompson 1984,

Schachter 1985). Presumably, this distinction also holds in

Classical Chinese. However, its existence has been argued to

be marginal, mostly because of the isolating nature of the

language. With the isolating characteristic, the language has no overt marker to categorially contrast identical ’meaning’

like ’explore’ vs. ’exploration’ in English.

After Mashi Wentong [ ]» published in 1898, it was agreed that the traditional categorization of words into full vs. functional in Classical Chinese grasps the essential lexical differences.^® The lack of morphological marking is, however, still regarded as an obstacle for further subgrouping on full words, (see Gao 1953, 1954, Dobson 1959, Kennedy

1964). For example, Gao (1953) argues that morphology is the only reason for word categorization. Therefore, in a language with no morphology, there should be no word classes. Gao denies any validity of identifying lexical categories by means of syntactic categories, and he stresses that the syntactic categories exist in their own right. Wang (1959) represents another view on this issue. To Wang, word categorization is

fundamental and essential, and how to categorize is a matter of methodology. He suggests that both semantic and syntactic

^® The version of Mashi Wentong this paper makes use of is a proofread edition published in 1961 by Zhang Xishen. 23 signs should be taken into account.

In the following discussion, Sections 2.1 and 2.2 are summaries of arguments from a couple of representative studies on full words categorization, and a tentative resolution will be proposed in Section 2.3.

2.1. ARGUMENTS AGAINST CATEGORIZATION. In addition to Gao

(1953, 1954), Dobson (1959) and Kennedy (1964) are among those that have questioned the availability of full-word categories.

Dobson and Kennedy share the opinion that the notion of word classes is a matter of statistics, though they adopt different interpretations on the statistics.

Due to the fact that many words correspond to different syntactic categories, Dobson considers it empirically inadequate to establish full-word categories on the basis of inconsistent correspondences. Moreover, compared with assuming a word category of nouns and stating that nouns act as verbs, or nouns act as adjectives, as Dobson puts it, a grammatical category realized by distributions and environments has more stable features and therefore will be better defined.

Kennedy, on the other hand, investigates the frequency of occurrences for a set of most frequently used words in the text of Mencius. hoping that word categories can thus be generalized. But as he admits in the opening statement of the study on Mencius. statistics are unable to deal with a case like shi [ $ ], which is of equal chance to be a noun or a 24

verb.

2.2. ARGUMENTS FOR CATEGORIZATION. The challenge of full-

word categorization is how to cope with asymmetric

correlations between the syntactic categories and the

presupposed full-words categories, viz. nouns, verbs, and

adjectives.

Cikoski (1978) stands for the opinion that the syntactic categories directly correspond to lexical categories. The

strong point in this treatment is that the syntactic positions/slots, viz. subject and predicate, are regarded as units possessing intrinsic properties. Due to the intrinsic properties of the slots, only words with compatible properties are allowed to fill in. Those who hold this opinion have no choice but to ignore lexical overlapping brought up by the asymmetry, however. That this treatment results in overlapping

is because any word which fills in more than one slot will be categorized into more than one category. It is very unfortunate that many full words can be found in more than

one slot in Classical Chinese.

On the other hand, lexical overlapping is exactly what

Wang and Lu attempt to avoid. Wang (1958, 1959) and LU (1942)

come up with the idea of ’temporary’ [ ], although in

Chou (1961a) points out that Kennedy made a mistake in analyzing characters rather than words. The case of shi. for example, happens to be two words represented by the same character. 25 different senses, to sort out the possibility of overlapping.

Wang hypothesizes that full-word categories are implied in the major syntactic categories. Because the correspondences are far from straightforward, it is necessary to distinguish regular from temporary correspondences. While nouns regularly correspond to SUBJ/0, verbs regularly correspond to PRED/P.

Correspondences other than the regular ones are irregularities from words playing temporary functional roles [

Wang and Lu agree that combinatorial capacities (or phrasal principles) should also be attributed to categoriality. The most important phrasal patterns in

Classical Chinese are in bi-lexeme constructions with a dominant relation [ ] of HEAD-OBJECT or with a restrictive relation [ ] of MODIFIER-HEAD (also cf.

Gao 1955). The head in the dominant relation is regularly a verb, and the head in the restrictive relation is usually a noun.13 But, as can be expected, irregular combinations do occur in the language. Under such circumstances, the notion of 'temporary' is used to lift the irregularities.

The notion of 'temporary' is explanatorily powerful for the inconsistent correlations. But since a workable definition for 'temporary' has never been proposed by these scholars, it will eventually raise a question about the reality of

1^ Encountering the same irregularities, Lii (1954b) attributes them to temporary categories.

13 See Note 1. 26

’temporary’ vs. ’regular’.

2.3. TOWARD A SOLUTION. Regardless of pros and cons for

full-word categorization in the above-mentioned studies, a

general conclusion can be drawn from all the studies. That is,

it is implausible that we can reach absolute and clearcut criteria for lexical categories. Therefore, the attempt of the present proposal is to hypothesize a dynamic lexical system, in which full words are grouped into three primary categories, viz. nouns, verbs and ambs.^® These categories are defined by their own prototypes. The fuzzy boundary between the categories will be considered as a natural scenario resulting from less-prototypical instantiations.

2.3.1. THE ASYMMETRIC CORRELATIONS. In order to see how a dynamic system can deal with a problem that an absolute system cannot, the problem should be presented in the first place.

This section will provide a brief review of previous observations on the obfuscating correlations between lexicon and syntax.

Statistics alone may never be adequate for the definitions (cf. Kennedy 1964).

Instead of adjectives, this hypothesis recognizes the category of amb(ivalent)s proposed by Kennedy (1964). ’Ambs’ in this study are defined by the core members which function as PRED in a certain type of sentences. Three major sentence types will be discussed in 2.1 of Chapter III. 27

PRED/PSUBJ/0 DJ

Figure 2.2. Correlations between Syntactic and Lexical Categories

In Figure 2.2, SUBJ/0, PRED/P, and ADJ represent syntactic functions. N, V, and A represent the widely presumed full words categories. N stands for nouns, V for verbs, and A for adjectives.As shown in this figure, nouns and adjectives can carry out all the major syntactic functions. Verbs carry out all but ADJ. Therefore, it is difficult to decide on lexical characteristics with respect to surface syntactic correlations without further analyses.

The following figure displays another obfuscation, the odd relations between bi-lexeme combinations.The first two horizontal rows in Figure 2.3 list possible combinations of modifier-head. The third row shows the combinations of head-

In previous studies, the category of adjectives is usually constructed by virtue of semantics. In the studies which strictly rely on syntactic evidence like Chou (1961a), full words tend to be categorized into nouns and verbs only.

Figure 2.3 is a new design roughly modeled after two figures in Kennedy (1964:384). The capital letter A in Kennedy’s study represents the category of ambs, which is defined as a category with half-noun and half-verb characteristics (cf. the vocabulary charts on p.352 and p.360, and also the definition of ambs on p.371). But A in Figure 2.3 represents the category of adjectives. Moreover, the combinations of lexemes in parallel relation in Kennedy’s figures are omitted here. 28 object.

N-N = Nj’A N-A = Nd’A N-V = Nd’V

A-N = A j ’N A-A = Ad ’A A-V = Ad ’V

V-N = V.No V-A = V.Ao V-V = V.Vo

Figure 2.3. Bi-Lexeme Combinations

In this figure, the capital letters N, V, and A represent word categories nouns, verbs, and adjectives. The small letters j, d, and o represent ADJ, ADV, and 0 respectively, which are used to indicate the function a category plays in a specific combination. For example, Nj ’A means that in the combination of N followed by A, N functions as modifier of the head A.^®

There are two noteworthy points in Figure 2.3. In the first place, verbs have a tendency to hold a dominant relation when occupying the initial position of the cluster, as shown in the third row. That is, verbs tend not to act as modifiers.

Secondly, as shown in the first two rows, nouns and adjectives tend to get into a restrictive relation when occupying the initial position of the cluster.These points are in fact a paraphrase of Kennedy’s (1964:385) claim about basic patterns

This relational analysis is in conflict with the normal functions presupposed to nouns and adjectives. Conflicts of this kind will be ignored until ’adjectives’ are re-interpreted in 3.1.

The four combinations with ADV involved will be ignored in this discussion, because the present proposal considers that ADV operates at a higher syntactic level, (cf. Note 1.) 29 of the bi-lexeme combinations:

... it can, be said that apart from the coordinate relation there are basically only two: the first element may be descriptive of the second, as in N j ’N, or the first element may affect the second, as in V.No. This might be summarized by the statement that the first element may be either static or dynamic.

2.3.2. THE CONCEPT OF ’TEMPORARY’. Among the previous studies about full words categorization in Classical Chinese, the concept of ’temporary’ in ’temporary functions’ developed in Wang (1920, 1940, 1958) is the most appealing for a dynamic lexical system. But, the theory of temporary functions may need to be revised.

As pointed out in 2.2, the ’temporary’ concept itself needs an explanation. The markedness theory may be the best choice as a theoretical backup to ’temporary’: While words of a specific category are unmarked in regular functions, they are marked in temporary functions. Moreover, temporary functions and temporary categories should be treated as the same phenomenon in different grammatical aspects. When a word plays a temporary functional role at a higher syntactic level, it means that a category shift occurs in the lexeme. 30

The theory of temporary functions originates from Wang

(1920).2° Wang observes that a word is recognized to be in a temporary function by certain contextual factors, e.g., different word order or co-occurrence with specific functional words. The pronoun zhi [^] ’it, them’, for example, crucially indicates that the nouns immediately before it play the temporary function P.

Unfortunately, Wang did not pay attention to semantic drifts which co-occur with lexemes in temporary functions. It has been reported that there are two possible semantic drifts for verbs and adjectives functioning as 0, • or being nominalized by zhi [ 2. ] ’NOMINALIZER’ (in contrast with the pronoun zhi ) and g_i [ % ] ’NOMINALIZER’ (Lii 1942, Liu 1958).

One drift changes verbs and adjectives into an abstract noun.

The other drift changes verbs and adjectives into a concrete noun.Obviously, the marking by pronouns and nominalizers, and the semantic shifts are supporting evidence for the theory of temporary functions.

Only the first two chapters of Wang’s 1920 paper were published. But the most important paragraphs about temporary functions are quoted in Wang (1940). According to the quote, benxing [ ] ’basic property’ vs. bianxing [ ] ’derived property’ is the original idea about temporary functions.

One noteworthy point is that the second type of drift is more arbitrary and thus more lexicalized than the first type. The drift of the first type is more systematic and predictable, and thus more like a syntactic aspect than the second drift. (Details about these semantic drifts will be discussed in Chapter IV and Chapter V.) 31

2.3.3. A HYPOTHESIS. The present proposal rejects the idea that all members of the full-word categories are inherently multi-functional, mainly because a multi-functional treatment overlooks the systematic aspects in the lexicon. As shown in

2.3.1, full words will be difficult to sub-categorize if all the correlations are taken into account as being of equal categorial importance. However, if the correlations are separated into two groups; a regular/unmarked group and a temporary/marked group, then the first group will define characteristics for the three categories properly.

But a potential problem for identifying the three categories by natural correlations is circularity. It is cyclic to confirm a three-way distinction based on the correlations put forth on the assumption that three major categories exist in full words. That is why the correlations presented in Figure 2.2 and Figure 2.3 will not be adopted as the point of departure for a new hypothesis. These figures simply exemplify the complexity of full-word categorization.

The present hypothesis will start with a predominant phrasal rule, i.e., VP in the head-object construction, to identify a core category in the lexicon. In SF terms, this VP means P-0 in syntax. Then, since P-0 and PRED.v are in a hierarchical relation, these two are grouped together as core functions for the verb category. Therefore, verbs will be defined as 1) words that consistently serve as the head in the

HEAD-OBJECT construction and 2) words that consistently act 32

as P/PRED.V. Moreover, when verbs play atypical roles, some

formal marker or semantic drift will be expected as stated in

the following: a) when verbs act as a modifier in the

MODIFIER-HEAD construction, they will be peripheralized by nominalizers or b) when verbs function as reference/SUBJ/0,

their meanings will change.

The first two conditions positively feature the prototypes of verbs. Any verb, when taking a direct object or acting as PRED.v, achieves categorial salience. The latter two predict what will occur to a verb when de-categorized. In one case, as stated in a), it will be formally marked, and consequently its semantic drift is secondary. But as indicated in b), when it is not formally marked, the semantic drift provides the evidence for de-categorization. (For the content of semantic drifts, see Chapter IV.)

In Classical Chinese, the ’verb’ category has a large number of members among full words, because PRED.v is the only obligatory element of a narrative sentence, and the narrative sentence is the relatively unmarked sentence pattern in the language. The analysis of other categories of full words can thus be considered as an analysis of non-verbs. In the remaining part of this chapter, the categories of adjectives and nouns will be reconsidered. Moreover, major sentence patterns will be discussed in Chapter III. 33

3. NOUNS AS A LEXICAL CATEGORY. There is surely a lexical

’noun' category because among the non-verbs, nouns make up a group with distinctive features in opposition to verbs. But, however, nominal features are less prominent than verbal features demonstrated in the above section.

3.1. NON-VERBS. Non-verbs in full words are grouped into two major categories: nouns and amb(ivalent)s. Ambs exist in the fuzzy boundary between nouns and verbs, and closer to nouns. They are considered members of a category because of their function as PRED in descriptive sentences.

A number of previous studies claim that real abstract nouns are extremely rare in Classical Chinese and that they are usually derived from verbs and adjectives (cf. Liu 1958).

Wang (1940:128) asserts that when an adjective occurs after qi or zhi without the appearance of another noun, it will become an abstract noun. Since the concept of ’adjectives’ will be replaced by that of ’ambs’, abstract nouns will be considered as derived from verbs and ambs.

3.1.1. ADJECTIVES OR AMBS. Whether adjectives form a distinctive category in Classical Chinese or not needs reconsideration.^^ There are two points implied in former

There is a small set of words which are marked by suffixes like ran ] ’of such-and-such manner’ or ru [ io ] ’such-and-such-like’, and they are recognized as adjectives by many scholars (of. Wang 1958, Chou 1961a). Adjectives of this kind will be treated as belonging to a sub-category of 34 studies which deserve to be taken into account relative to this issue. The first point questions whether the function of modification is carried out by members of a specific category.

As a matter of fact, NPs in the modifier-head construction without being marked by the peripheralizer zhi are of a small proportion.23 Moreover, it is generally accepted that nouns can practice the function of modification without a genitive or possessive marker. As Liu (1958:34) points out, if two nouns in a cluster are not in parallel or dominant relation, they must be in the restrictive relation, in which the first noun modifies the second noun. In other words, in the small proportion of unmarked cases, nouns do have the function of modification.

The second point is about the semantic feature of adjectives. In general, adjectives are recognized in accordance with semantic features like ’qualities’ or

’properties’ (cf. Wang 1955, Liu 1958). These features are, however, vague and ambiguous, especially in Classical Chinese.

In comparison, ’conscious’ vs. ’consciousness’ in English is

ambs in Classical Chinese. This is mainly because they, like other ambs, usually serve as PRED in descriptive sentences. On the other hand, they differ from other ambs for not being used as nominal modifiers.

23 This may have a historical reason. As pointed out by Wang (1958), the parallel construction appeared earlier than the other two constructions. In Classical Chinese, although the modifier-head construction is mainly used to form NP, there are still a large number of NPs formed in parallel construction. At the same time, the modifier-head NPs are mostly marked by the peripheralizer zhi in the language. 35 in opposition to distinguish the discourse functions of

’modification’ vs. ’reference’. It is unclear whether this distinction really exists in the grammar of Classical Chinese, and yet we know for sure that it never holds in the lexicon.

Therefore, it will be empirically inappropriate to treat adjectives as a category.

On the other hand, the lexemes which are not pure nouns or verbs are categorized as ambs by Kennedy (1964). As Kennedy puts it, ambs are sometimes verb-like and sometimes noun-like, because ambs pass though all the tests he set up for nouns and verbs. It will be argued that ambs are much more.like nouns than verbs in the following discussion. In 3.1.2, the shared features of ambs and nouns will be presented. Even so, ’ambs’ should be a category in their own right. The proof is that ambs are the natural carriers of the function PRED.a.

3.1.2. NOUNS AND AMBS. As opposed to the conditions for verbs in 2.3.3, nouns will be defined by the following statements: Nouns are lexemes 1) that consistently function as the head of NPs in the modifier-head construction, 2) that consistently function as reference/SUBJ/0 in all major types of sentences, 3) that do not act as a head in the head-object construction without semantic drift, and 4) that function as

PRED.n in equational sentences.

The first two statements critically exclude ambs from nouns. When ambs serves as the head of NPs, it is required 36 that the head be marked by a nominalizer.^^ When a noun and an amb form a meaningful cluster with the noun preceding the amb, the cluster is likely to be in the SUBJ-PRED construction.^®

Moreover, when an amb functions as reference in a sentence, a specific referent is expected from the contexts. (For more details, see Chapter IV.)

The third statement happens to be a shared feature of nouns and ambs. To be specific, when ambs and nouns function as head in the head-object construction, they will undergo a grammatical process which will for the time being be labeled

’causativization’.

Accordingly, ambs will be defined as in the following statements: Ambs are lexemes 1) that tend to be marked by a nominalizer when functioning as head in the modifier-head construction, 2) that consistently undergo a semantic drift when serving as head in the head-object construction, 3) that naturally function as modifier in the modifier-head relation, and 4) that serve as PRED.a in descriptive sentences.

3.2. CENTRAL AND PERIPHERAL NOUNS. The core members of the noun category are by overwhelming majority concrete nouns,

i.e., nouns which refer to a concrete physical object. The

For nominalizers, see Chapter IV.

2® To be more specific, the PRED here means PRED.a for descriptive sentences. This point will be clarified in Chapter III under the discussion of sentence patterns. 37 prototypes of nouns make no distinction on person, case, or gender. Nouns are stored in a collective sense in the lexicon.

Thus, whether a concrete noun is singular or plural remains irrelevant until a numeral shows up to make a specification.

Proper nouns which label individual objects are among the most manipulable instances of nouns, because they need not to be further modified to achieve categorial salience. Common nouns which are modified by a numeral or a demonstrative should achieve similar salience to proper nouns. (For manipulability, see Hopper & Thompson 1984).

Moreover, there are nouns derived from full words of other categories. Some of them are marked by functional words, whereas some others are unmarked and yet can be recognized by their capacity to carry out the functions typical of nouns.

In both cases, semantic drifts are assumed.

Nouns which do not refer to a physical object may fall short of manipulability, and thus are peripheral cases of the noun category. A frequent case of peripheral nouns is that nouns serve as modifiers in the modifier-head construction.

Instantiations of this kind inevitably lose maniputablilty.

As to ’abstract nouns’, a small number of philosophical terms may serve as core members of this group. In most cases, abstract nouns are naturally instantiated from ambs or derived from verbs. As will be clarified in the following chapters, the status of ’abstract nouns’ in Classical Chinese is rather vague, and a summary about this issue will be provided in 2.4 38 of Chapter VI.

3.3. NOUNS CONVERTED INTO ADVERBS AND VERBS. In Classical

Chinese, proper nouns are scarcely converted into adverbs to function as ADV. When they are converted into verbs, they are most likely to be target verbs. (For target verbs, see 3.4.2 of Chapter V.) Common nouns may be converted into adverbs or denominal verbs. When they are converted into adverbs, an obvious semantic drift will co-occur. When common nouns are converted to carry out the verbal functions, they may shift to be transitive and function as P, or to be intransitive and function as PRED.v. Semantic drifts are expected from these shifts. (All about the semantic drifts will be discussed in

Chapter V.)

4. FUNCTIONAL WORDS. The concern in this section is the functional words which will be selected as test words for the traces of semantic variations.^® Pure functional words, such

^® For the sub-categorization for functional words, see Ma (1961), Yang (1920), Lü (1942), Chou (1961a). LÜ (1942) categorizes functional words into restrictives (adverbs), anaphoric words, connectives, and interjections. Anaphoric words include pronouns, demonstrative words, numerals and measure words. Connectives are of nominalizers, prepositions, and conjunctions. Restrictives consist of adverbs of time, place, degree, direction, concession, negation, etc. Chou (1961a) has a slightly different analysis. For example, there is a sub-category of descriptives covering adjectives and adverbs ending with some markers which has no equivalent in Lü’s analysis. The restrictives in Lü actually cover three subcategories in Chou, i.e. auxiliary predicatives, localizers, and adverbs. This discrepancy 39

as nominalizers, conjunctions, and adverbs, will receive

specific attention in the following chapters. Through the

interactions of the functional words and full words, the

semantic drifts and categorial shifts will be defined.

The negative bu [ ^ ] is an adverb of PRED.v. Therefore, when a PRED is modified by the PRED will be identified as a PRED.v.

Nominalizers are tentatively defined as 1) words which mark following ambs and verbs to function as reference/SUBJ/0, and 2) words which mark off VPs and sentential elements to function as reference/SUBJ/0. The nominalizers frequently used

in Classical Chinese are zhi [ 2. ], ai. [ % ] and zhe [ # ].

The conjunction used as a test word here is er [ jfij ] , whose primary function is to connect two PREDs or sentential units (cf. Hsueh 1990).

results from different criteria for the analyses. Moreover, Wang (1958) treats localizers in Chou’s and restrictives of time, place, and direction in Lü’s as nouns in ’relative positions’ which are under the cover term of guanxi yu [ ] ’relative words’. (More about relative words will be discussed in Chapter III.) These diverse analyses among scholars about functional words strongly suggest that functional concepts in the grammar are fuzzy to a certain degree. CHAPTER III

NOUNS AND THEIR SYNTACTIC FUNCTIONS

0. The purpose of this chapter is to survey the syntactic behaviors of nouns. Based on the previous discussions, two characteristics should justify the categoriality of nouns.

First, nouns are absolute heads in a [modifier-head] NP where the modifier is a numeral or an adjectival demonstrative.

Next, nouns are major carriers of the functions SUBJ/0 in all sentence types. Moreover, nouns are major carriers of PRED.n in equational sentences. Nouns are also expected to occur in certain syntexemes, identified by Wang (1958) as 'relative positions’.

This chapter will focus on three issues. The first issue is concerned with how typical nominal modifiers, viz. numerals and demonstratives, interplay with the head nouns. The second issue is about three major sentence types in Classical Chinese and how nouns perform in these sentence types. The third issue is about time and place words, and how they function in the

'relative positions’.

1. NOMINAL MODIFIERS. As mentioned in the previous chapter, typical nominal modifiers such as numerals and demonstratives

40 41 are markers that promote manipulability of concrete common nouns. The modification of numerals makes a noun counted and thus more specific. Consequently, the referent of the noun will not be understood in a collective sense. Adjectival demonstratives help determine ’proximity' of a referent to ’a point of reference’ (for ’proximacy’, see Lyons 1968). Since the modification of demonstratives pinpoints a designation boundary for the referent, it will certainly be more specific.

1.1. NUMERALS. In Classical Chinese, a common noun which denotes a palpable object is a label for a whole set of individuals. When context-free, this noun is realized in a collective sense. For example, ren by itself stands for the set of human beings. Whether ren is singular or plural may depend on contexts. This can be seen in the following sentences from Mencius ;

(la) Jin you ren ri rang qi lin zhi ji zhe. (3B/8)^

now-have-person-day-rob-his-neighbor-GENI-chicken- NOM ’Now there is a person who plunders his neighbor’s chicken everyday.’

^ The numbering system used hereafter is for examples from Mencius. It follows the numbering system in the Harvard- Yenching Concordance of Mencius. The English translations hereafter are made for better understanding of the analyses in the present study. The translations thus may differ from the published translations made for the purpose of studying literature. 42

(lb) Ren bu kan qi you, Yan-Zi bu gai qi le. (4B/29)

people-NEG-endure-DEMON-hardship,NAME-NEG-change- GENI-happiness ’People couldn’t endure that hardship, but Yan-Zi never changed his happiness.’

GENI means a genitive marker; NOM, a nominalizer; NEG, a negative; and DEMON, an adjectival demonstrative.

The English translation shows clearly that ren is interpreted as singular in (la) and plural in (lb), which is virtually determined by the contexts where ren occurs.

Ren in (la) refers to a thief who takes away his neighbor’s chickens. The thief’s image is compared to that of the addressee, an official, by the speaker. Therefore, ren is probably singular. Ren in (lb) is likely to be plural because, according to the context, ren should refer to contemporaries of Yan-Zi. This analysis can be supported by the fact that when unspecified, nouns are frequently used in a collective sense.

As to the numbers for j_i ’chicken’ and lin ’neighbor’ in

(la), they are rather vague. Although both are translated to be singular, it is equally good to have lin translated as plural. With reference to the following expression, vue rang yi ji [ H % — % ] ’to plunder one chicken every month’, ü may probably mean ’one chicken’. But it will still make sense to have ji in (la) to be plural. 43

The implicitness on number will become transparent

through the following parallelism of rou 'meat' and ma

'horse':

(2) Pao you fei rou, jiu you fei ma. (3B/9)

kitchen-have-fat-meat, stable-have-fat-horse 'There is fat meat in the kitchen and there are fat horses in the stable.'*

It is very likely that the native speakers of Classical

Chinese through the language perceived no difference at all on the count vs. mass distinction of 'horses' and 'meat' in a parallelism like (2). The countability of a common noun perhaps remains irrelevant unless a numeral occurs before it.

When counted by a numeral, the notions of 'count' and 'mass' will then matter to the grammar.

To count a noun like ren and ma which has a count equivalent in English, a numeral is placed directly before or after the noun:

(3a) Chen zhi wu ren yan. (2b/4)

Ë A-Wo servant-know-five-people-PAR 'I knew five of them.'

* In the free translation of Legge (1895, rep. 1960), pao is translated in a singular form, i.e., 'kitchen', and iiu is in a plural form, i.e., 'stables'. This may suggest that the number of the nouns in this specific context is subject to readers' judgement. It is even more likely that the number has no relevance for the readers. 44

(3b) Zhao-Zhan yi qi liang ma er ji qi xiong yu shu-fu. (Xuan 12)3

NAME-with-his-good-horses-two-aid-his-brother-and- uncle ’Zhao-Zhan aided his brother and uncle by giving them his two good horses.’

(3c) Xian Zi-Chan yi ma si pi. (Zhao 6) ÆfmwmggEç. see-NAME-with-horses-four-MEASURE ’[Qi-Ji] paid a respectful visit to Zi-Chan with four horses (as a gift).’

PAR means a particle and MEASURE means a measure word.

In (3a), ren is counted by a numeral wu ’five’ immediately before it. In (3b) the numeral er ’two’ occurs immediately after ma. But ma in (3c) is counted by the numeral si ’four’ plus a measure word pi.*

For those nouns which have a mass equivalent in English, they have to be counted by first adding a numeral to a related countable noun and then phrasing this NP into a genitive

3 The numbering system is used hereafter for examples from Zuozhuan. It follows the numbering system in the Harvard- Yenching Concordances to Ch’un-ch’iu. Kung-vang. Ku-lianiz and Tso-chuan. For readers who are not familiar with the Harvard- Yenching system, attention should be paid to the systematic differences between Mencius and Zuozhuan (cf. the uses of (3a) and (3b)). .

* It is noteworthy that measure words like ^i are rather rare in Classical Chinese. To be more specific, pi by itself functions more like a pure grammatical unit with little semantic content. Thus it is closer to classifiers developed in later times (cf. Tai & Wang 1990, Wang 1991). Most of the measure words nevertheless are close to nouns, because they designate a physical object by themselves. 45

structure marked by z h i . as shown in the following example:

(4) Wu yang zhi pi (5A/9)

f ive-sheep-GENI-leather 'The leather of five sheep’

In (4), the numeral wu 'five' is added to the noun van*

'sheep' as a measure unit. Then the whole NP is peripheralized to modify pi 'leather', a mass object.

When %i 'one' is added to the measure unit, the cluster can occur before the noun without the marking of zhi. as shown

in (5a). Meanwhile, as pointed out by Liu (1958:37), XJL tends to be omitted under such circumstances, as exemplified in

(5b) :

(5a) Yi dan shi, yi piao yin (4B/29)®

one-basket-food, one-gourd-drink 'a basket of food, one gourd of drink'

(5b) ... Dan shi dou geng xian yu se. (7B/11)

basket-food-vessel-soup-appear-PREP-looks '[... To transfer] a basket of food or a vessel of soup [to others] will make him show an irritated look.’

® Shi and yin are probably derived from verbs, and yet there is little evidence for date of the derivation. It can be observed, however, that these words are frequently used as nominal units in Classical Chinese. This may suggest that the nominal uses have been lexicalized at the stage in question. 46

Dan ’basket’, piao ’gourd’ and dou ’vessel’ are containers.

They are the typical measure words in Classical Chinese. Yang

in (4), on the other hand, is a lexical noun with an

independent denotation. Pi in (3c) is more like a classifier, although the category of classifiers is not fully developed in the language.

Moreover, numerals are used to indicate contrast or emphasis. Sometimes a numeral abruptly shows up in a discourse where contrast or emphasis is involved. In Mencius. for example, there is a story about a warlord of the Qi state. The warlord once made his servant change the sacrifice animal in a sacred ceremony from a cow to a sheep. His people thus thought that he was stingy. The warlord defended himself by arguing that it was really that he could not stand seeing a cow trembling. Thus, he asked the following rhetorical question :

(6) Qi-Guo sui bian xiao, wu he ai yi niu? Ji bu zen qi hu-su, ruo wu zui er jiu si di. Gu yi yang yi zhi ye. (lA/7)

NAME-although-small-small, I-QUES-care-one-cow? then-NEG-bear-its-trembling, like-NEG-guilt-CONN- approach-die-land. thus-with-sheep-change-it-PAR

’Although the Qi state is small in territory, why should I (its king) care for a single cow? I was just unable to bear seeing its trembling, like it is approaching the dead land without being guilty. Therefore, I changed it for a sheep.’ 47

QUES indicates a question word and CONN indicates a connective.

In the preceding context of (6), niu 'cow' and yang 'sheep' are referred to without being specified by numerals. Also as shown in (6), vang occurs without being modified by a numeral, although it is obviously singular. In yi niu 'one cow', the numeral yl 'one' was deliberatedly included by the warlord to belittle the value of a cow and to stress his kindness to save the cow's life.

In addition to common nouns, numerals are occasionally found to modify a proper noun:

(7) Yi Xue-Ju-Zhou, du ru Song-Wang he? (3B/6) -new, «aoaiw? one-NAME, only-treat-NAME-what® 'What can one Xue-Ju-Zhou alone do to the king of Song?’

Proper nouns are the only nouns which designate absolute singular. But when a numeral is added to a proper noun, its

inherent property will concede. As Liu (1958:20) notes, when a numeral is added to a proper noun, the noun at issue will change to a common noun. Thus, vi Xue-Ju-Zhou in (7) means

'one who behaves like Xue-Ju-Zhou'.

® Du is interpreted differently in Yang (1954:76) and Dobson (1974:232). For the former study, ^ is a question marker, but for the latter, du is an agential distributive, indicating 'the agent alone, to the exclusion of all others.' 48

1.2. DEMONSTRATIVES. To be specific, demonstratives

discussed in this section are adjectival demonstratives. It

deserves noting that most of the adjectival demonstratives

have a homophonie counterpart of anaphoric pronouns. One major

function of the demonstratives is to indicate ’proximity’ of

the head noun to a point of reference. Point of reference is

concerned with a situation that the speaker takes into account. The situation usually co-incides with the situation

that the speaker is in. But there are cases where the pointed

situation differs from the speaker’s current situation.

Pertaining to proximity of a point of reference, the demonstratives can be classified into two groups: one group

indicates proximity to the point and the other group indicates non-proximity to the point.

Since the situation that the speaker takes into account usually co-incides with the situation that the speaker is in,

the demonstratives of proximity usually reflect spacial or

temporal closeness to the speaker. Those of non-proximity usually reflect spacial or temporal distance from the speaker.

But, however, it is not required that the point of reference

should be identical with the current situation of the speaker.

For example:

(8a) Zai ci yue ye. (Zhao 17)

in-DEMON-month-PAR ’It will be/happen in this month.’ 49

(8b) Qi de bao bi min zai? (5A/3)

how-can,-torture-DEMON-people-PAR ’How could [Xiang] torture those people?’

(8c) Yi ta ma fan. (Xuan 12)

by means of-DEMON-horse-return ’[Zhao-Zhan] returned by another horse.’

Ci in (8a) and M in (8b) are the two most typical demonstratives which hold the contrast in proximate vs. non- promimate. In (8a), ci ’this’ indicates temporal closeness to the situation that the speaker is in, which is ’now’. In (8b), bi ’those’" indicates both temporal and spacial distances from the present situation.

The point of reference in (8c), on the other hand, has nothing to do with the present situation. (3b) and (8c) are in fact events in sequence. To combine (3b) and (8c) will make the designation of ta ’another’ in (8c) very clear. Because

Zhao-Zhan had given his two good horses to his brother and uncle (as read in (3b)), he was forced to escape from the dangerous spot by a horse other than the two he gave away.

Here, the point of reference is the two horses and the demonstrative ta specifies a horse other than these two horses.

In addition to ci, shi [ ]* M [ ^ ] and M 1 are used to indicate a proximate relation (of. Yang 1920, Liu

1958, Wang 1958). Wang further speculates that these 50 variations result from dialectal differences. In Zuozhuan. there is a case of shi which displays how the speaker plots proximity for a referent to a situation which is independent of his present situation:

(9) Jin-ren yu shi bi yu Yao, Yao you er ling yan. Qi nan ling, Xia-Hou-Gao zhi mu ye; qi bei ling, Wen- Wang zhi suo bi feng yu ye. Bi si shi jian, yu shou er gu yan. (Xi 32)

NAME people-defend-army-certainly-PREP-NAME, NAME- have-two-hills-PAR. its-south-hill, NAME-GENI-tomb- PAR; its-north-hill, NAME-NOM-NOM-avoid-wind-rain- PAR. certainly-die-DEMON-locality, I-collect-your- bones-PAR ’The Jin people will definitely block your troops at Yao. There are two hills at Yao. The south hill was where the tomb of Xia-Hou-Gao was located; and the north hill was where King Wen took shelter from winds and rains. [You] will for sure die in between these hills. I will collect your bones [there].’

This quote in (9) is what Jian-Shu [ ] told his son, who was a member of the troops planning to attack a neighboring state. Jian-Shu predicted that his son would die at Yao. He plotted a situation about the battlefield at Yao in advance, and he used shi to specify a referent .1 ian ’middle’, as in shi jian ’this spot [in between the hills]’. In this occasion, shi points proximity to a situation that the speaker Jian-Shu was concerned about, rather than the situation he was in.

In addition to Yang ( 1920) and Dobson (1974) consider fu as a non-proximate marker, or in Dobson’s term, a far- demonstrative. Examples of fu are as follows: 51

(10a) Fu qu yiu zai, ru qi xing hu! (Xi 24)

DEMON-sleeve-still-exist, you-perhaps-exil-PAR ’That sleeve is still there, you should perhaps exile yourself!’ (10b) Qing dong ren zhi neng yu fu er san you-si yan zhe, wu yu zhi xian. (Wen 13)

request-east-person-NOM-can-with-DEMON-two-three- off icials-speak-NOM , I-with-him-precede ’ I would request a person coining from the eastern state who can speak with those officials (who can speak the language those officials do) to precede with me .’

The above examples are adopted from Yang (1954:46). Fu indicates non-proximity to the situation where the speaker is located. In (10a), fu points to a sleeve which has to do with an event long before the present situation. Presumably, the speaker distanced himself from the situation which the sleeve belonged to, and therefore, used fu. In (10b), fu points to officials far from the present location.

Unfortunately, there is a problem with this non-proximity explanation on fu. Yang (1954:45) lists under the proximity explanation a couple of sentences with fju» two of which are from Zuozhuan:

(11a) Wei fu ren zhi li, bu ji ci. (Xi 30)

• not-DEMON-person-NOM-effort, NEG-reach-this ’Without this guy’s effort, [I] would never arrive at this point.’ 52

(11b) Fu er ren zhe, Lu-Guo she-ji zhi chen ye. (Cheng 16)

DEMON-two-persons-NOM, NAME-society-GENI-minister- PAR ’These two people are pillars of the Lu state.’

The English translations for fu ’this/these’ in (11) follow

Yang’s interpretation about the demonstrative fu.’ But this interpretation contradicts the general explanation on fu. Yang may have a good reason, though. In both cases, the head noun of fu refers to a previous event, a situation of concern to the speaker.®

To cope with the conflict, there are two options. The first is to admit two different fu, one for proximity, the other for non-proximity. The other choice is to neutralize the deictic implication, and to reinterpret the function of this word in question.

There are two interesting examples which indirectly support the second option, as in (12):

’ Based on the following argument on the function of fu. it will make little difference whether fu is translated as this/these or that/those.

® It is not absolutely infeasible to provide a non- proximate explanation for these cases. But it would do little good to the issue, because the definition about ’proximity to the concerned situation’ itself is tricky. Before a more operable definition becomes available, however, this one in question will be adopted. 53

(12a) Wang zhi fu miao hu? (lA/6)

king-know-DEMON-sprouts-PAR ’Does Your Majesty know about those sprouts?’

(12b) Gong sou fu ao yan. (Xuan 2)

duke-whistle-DEMON-mastiff-PAR ’The duke whistled for the mastiff.’

Dobson (1974:273) puts (12a) under the entry of particles of accentuation. It is undeniable that quite a few functional words can occasionally be interpreted as particles with implicit senses, conventionally noted by an expression like

*xuci wu yi’ [ ] (empty words have no meaning). Dobson re-interprets some of these occasions with ’particles of accentuation’, which means a particle ’precedes the accented word’ (p.65). This explanation is good, and yet, it further diverses the functions of the particle fu. More importantly, the explanation of accentuation may probably be secondary to the grammar on account of the fact that most of the other particles with an accentuation explanation are words of distinct grammatical functions.®

From the present point of view on functions of demonstratives, fu in (12a) is not just a particle that marks the accented word, it is also a particle that promotes

® Dobson (1974:65) exemplifies other markers of accentuation like ye [ •& ] and yi. [ ^ ]. It may be true for certain occasional uses of these particles. However, it is for sure that marker of accentuation never explains the primary purpose of these functional words. 54 manipulability of the head noun. Thus, fu in (12a) makes the

head noun miao ’sprouts’ more like a typical participant in

the discourse. As can be expected, the noun in the following context remains to be a focus of attention:

(13a) Qi ba yue zhi jian han, ze miao gao yi. (lA/6)

seven-eight-month-POSS-middle-dry, CONN-sprouts- wither-PAR ’In July and August, when there is a draft, the sprouts will wither.’

(13b) Tian yiu-ran zuo yun, pei-ran xia yu , ze miao xing zhi yi. (lA/6) imrm, rnimaza, sky-spontaneously-arise-clouds, abundantly-fall­ rains , then-sprouts-grow-them-PAR ’The clouds arise spontaneously in the sky, and rains fall abundantly, then the sprouts become grown up.’

The utterances in (13) report the subsequent events of (12a).

Since in (12a) the referent miao is specified by fu, miao in these events are more specific than any unspecified noun in the discourse.

Fu in (12b) is legitimate for an explanation similar to that in (12a). Fu here specifies a mastiff which remains a

focus in the following contexts:

(14a) Ming bo er sha zhi. (Xuan 2)

NAME-fight-CONN-kill-it ’Ming (Ti-Mi-Ming) fought and killed it.’ 55

(14b) Dun yue: qi ren yong-quan, sui meng he wei! (Xuan 2 ) fSB: NAME-say: eliminate-person's-dog, even if-brave- QUES-do 'Dun (Zhao-Dun) said, "[You’ve] got rid of other person’s dog; even if this was brave, it was of no use."’

Because of the occurrence of fu in (12b) before ^ ’mastiff’, it becomes clear that the mastiff is a focus of the subsequent events. As a result, zhi. the pronoun, in (14a) and ouan ’dog’ in (14b) should refer to the same mastiff that the duke whistled for.

To summarize, the function of numerals and demonstratives are to specify referents of common nouns. Through the specification, a common noun will become a typical and manipulable participant in discourse (cf. 6 and 12-14).

Moreover, although details about the demonstratives still need to be investigated, esp. deictic nature of the demonstratives, there is a good reason to agree with the former studies that the opposition of proximity and non-proximity exists in the language, represented by the typical contrast of ^ and ci.

Yet how and to what extent the distinction holds is open to

This interpretation of (14b) is a comment from Prof. Hsueh. Here, qi ren yong-quan is understood as a VP with the main verb ai ’to eliminate’. Yong-quan is analyzed as a nominal unit, meaning ’(employed) dog’, and ren vong-ouan means ’other person’s dog’. In a conventional reading, oi ren vong ouan is understood as two parallel VPs with main verbs ai and vong ’to use’ and this reading can be translated as ’to discard a person and to make use of a dog’. 56 question.

2. NOUNS IN SENTENCE TYPES. There are three major sentence types in Classical Chinese (Liu 1958). In all the sentence types, nouns are major carriers of the functions reference/

SUBJ/0. In addition, nouns are typical carriers of PRED in one of the sentence types, viz. PRED.n. in equational sentences.

2.1. SENTENCE TYPES. The three sentence types are equational, descriptive, and narrative sentences. The equational sentences have a nominal PRED, which will hereafter be represented as PRED.n. The descriptive sentences have an adjectival PRED, hereafter, P R E D . a . T h e narrative sentences have a verbal PRED, hereafter, PRED.v. In the most unmarked situations, nouns, ambs, and verbs hold a natural correlations with one sentence type respectively, which are as exemplified in (15):

(15a) Zeng-zi, shi ye. (4B/31)

NAME-teacher-PAR ’Zeng-zi is a teacher.’

(15b) Qi xian. (5A/6)

NAME-wise ’Qi is wise.’

As argued in Chapter II, adjectives are reanalyzed as ambs in this study. Therefore, ambs will be the major carriers of PRED.a. 57

(15c) Hu-li shi zhi. (3A/5)

Fox-eat-it ’Foxes ate it (the body).’

(15a) is a standard equational sentence and shi ’teacher’ is a noun as PRED.n. Xian ’wise’ in (15b) is an amb in a descriptive sentence as PRED.a. In (15c), shi ’to eat’ is a verb taking a direct object zhi ’it’ to function as PRED.v in the narrative sentence.

2.2. NOUNS IN EQUATIONAL SENTENCES. Nouns are the nuclei of nominal predicates:

(16a) Teng, xiao guo ye. (lB/15)

m , NAME-small-country-PAR ’Teng is a small country.’

(16b) Ren, ren xin ye. (6A/11) iz, A'D-Êo humaneness-human being-heart-PAR ’Humaneness is the human heart.’

The sentences in (16) are formed in the standard formation of equational sentences. Since there is no copula verb in

Classical Chinese, nominal predicates are juxtaposed with

There are many issues related to sentence types, esp. issues about unnatural/marked correlations between lexical categories and sentence patterns. For further investigation of these issues, Liu (1958:133-52) should make a good point of departure. 58

SUBJ.^^ Then the two NPs are followed by a final particle ye. as in NPl-NP2-ye. In (16a), xiao guo ’small country’ is an

NP with the head noun guo and the whole NP functions as PRED.n marked by ye. In (16b), ren xin ’human heart’ is an NP with the noun xin as the head to carry out the function of PRED.n.

The pragmatic function of PRED.n is referential predication. The first NP and the second NP are linked by labeling: the second NP is a label of the first one (cf.

Cikoski 1978). ’Human heart’, for example, labels for the philosophical term ’humaneness’, and ’small country’ labels for Teng.

2.3. NOUNS IN DESCRIPTIVE SENTENCES AND NARRATIVE SENTENCES.

In descriptive and narrative sentences, nouns normally function as SUBJ/0 or heads of SUBJ/0. For example:

(17a) Meng-Zi xian Liang-Hui-Wang. (lA/1)

NAME-visit-NAME ’Mencius visited the King Hui of Liang.’

The copula shi ’to be’ in Mandarin is an anaphoric pronoun, a demonstrative, or an emphasis marker in Classical Chinese. About the origin of the copula shi. see Yen (1986) and Guo (1990 ) .

The first NP is optional, and thus it is not uncommon that an equational sentence consists of the second NP with the particle ye. 59

(17b) Min da yue. (IB/ll)

people.-greatly-pleased ’The people were very pleased.’

Meng-Zi ’Mencius’ in (17a) is SUBJ of the narrative sentence, and Liang-Hui-Wang ’King Hui of Liang’is O of P xian ’to visit’. The whole VP of xian Liang-Hui-Wang are PRED.v of the narrative sentence.(17b) is a descriptive sentence with min

’people’ as SUBJ and vue ’pleased’ as PRED.a modified by ADV da ’greatly’. Although there is no numeral to specify the common noun min in (17b), it is obviously plural Judging from the contexts.

3. TIME AND PLACE WORDS. Time and place words are treated by former studies as subclasses of nouns (e.g., Liu 1958, Wang

1958). However, it is observed that time and place words possess syntactic features similar to ADV, esp. when they occur immediately before P. Moreover, there is one rarely mentioned issue about time and place words, i.e. the optionality of a preposition before a time or place word.

3.1. TIME AND PLACE WORDS IN TYPICAL NOMINAL FUNCTIONS. The obvious reason for classifying time and place words as nouns is that they naturally play the functional roles typical to

SUBJ and PRED in narrative sentences are connected by transitivity relationships. More information on the semantic connections will be provided in Chapter IV and Chapter V. 60 nouns. For example;

(18a) Zi-Tai-Shu zhi miao zai dao nan. (Zhao 18)

NAME-POSS-temple-be in-road-south *Zi-Tai-Shu’s temple was located to the south of the road.’

(18b) Wu ren zai qi nan. (Ding 4)

NAME-people-be in-its-south ’The Wu people were located to its south.’

(19a) Jin ren yi qi yi zhi lao, qing si ta nian. (Cheng 13)

NAME-people-because-DEMON-battle-NOM-weariness, request-wait-DEMON-year ’Due to the weariness from that battle, the Jin people requested to wait for another year.’

(19b) Ding wei san nian zhi sang. (3A/2)

make-be-three-years-POSS-mourning ’[It was] made to be a three-year mourning.’

In (18a), nan ’south’ is the head of the NP dao nan ’road’s south’ with a noun as its modifier, and in (18b) it is the head of the NP gi nan ’its south’ with a demonstrative as its modifier. In (19a), nian ’year’ is the head of the NP ta nian

’another year’ with a demonstrative as its modifier, and in

(19b), nian is modified by a numeral san ’three’.

On the other hand, when a time or place word occurs before the main predication, even if it occurs in the initial position of the utterance, it can hardly be recognized as 61

SUBJ:

(20a) Nan ru yu Chu. (lA/5)

south-humiliate-PBEP-NAME 'In the south, [our state was] humiliated by the Chu state.'

(20b) Nan ji Hua-Shan. (Xi 15)

south-reach-NAME 'To the south, [it] reaches the Hua mountains.'

(21a) Wu nian Chen jiang fu feng. (Zhao 9)

five-years-NAME-will-again-establish fiefdom 'In five years, the Chen state will re-establish its fiefdom.'

(21b) Ta nian qi er zi lai. (Wen 15)

DEMON-year-his-two-sons-come 'Several years later, his two sons came to him.'

Nan 'south' in (20a-b), wu nian 'five years' in (21a) and ta nian 'another year' in (21b) all occur in the pre-PRED position usually assigned to SUBJ. However, none of them are holding a transitivity relationship with the main predication.

Instead, they are linked to the PREDs circumstantially, providing temporal or spacial frames for the following predications. In other words, these words are 'topics' of the sentences in question, in terms of Li & Thompson (1981:85).

But in the present study, the relation will be realized as ADV

- S(entence) by virtue of syntactic functions. 62

Another noteworthy point about the pre-PRED ADVs in (20) is that a same place word may draw different spacial frames, mainly because of the main verbs. Thus, nan in (20a) locates where the event happens, while it in (20b) indicates the spacial scope to which the following event applies. Similarly, time words may differ in their semantic implications on account of the verbs. That explains why wu nian in (21a) sets up a temporal scope for an event to come, while ta nian in

(21b) pinpoints a time for the happening. (More discussions on semantic contrast in time words will be made in 3.2.3.)

Based on the modification function of time and place words, it may be hypothesized that these words in question represent special cases of the noun category. But according to Wang (1958), nouns other than time and place words can occasionally function as ADV. Cases of these atypical uses of nouns will be examined in Chapter V. For the time being, this evidence will be taken as the temporary uses of nouns. Our present interest in time and place words is to examine the cases that a time or place word occurs in a special syntaxeme.

3.2. NOUNS IN THE RELATIVE POSITIONS. Nouns can function in some syntactic positions identified as guanxi wei [ ]

’relative positions’ by Wang (1958). Nous in the relative positions are referred to as guanxi vu [ ] ’relative words’. Wang’s (p.388) definition of relative positions can 63

be translated as follows:

For any of those nouns (or NPs ) which either directly relate to a verb, or occur in the initial or ending place of a sentence to indicate time, place, range, manner, and instrument by which an action takes place, or cause of an action, the position it occupies is the relative position. (My trans.)

Relative words which indicate range, manner and instrument will be taken care of in Chapter V. In the remainder of this

chapter, the relative words which indicate time and place will

be the focus of interest.

3.2.1. PLACE WORDS IN THE YU PHRASES. Referring to relative words, Wang points out an interesting opposition between a

relative word and a preposition phrase. For example:

(22a) You qian niu er guo tang xia zhe. (lA/7)

have-drag-cow-CONN-pass-court-under-NOM ’There was one who led a cow passing in front of the court.’

(22b) Li yu tang xia. (Zhao 28)

stand-PREP-court-under ’[Ran-Ming] stood in front of the court.’

(22a) is Wang’s (p.390) example. As can be seen, the place NP

tang xia ’position below the court hall’ occurs immediately

after the verb guo ’to pass’ in (a), while it occurs after yu

in (b). It is Wang’s opinion that ’place’ can be expressed by

a PP or by a relative word alternatively (p. 389). Thus he 64

makes tang xia in (22a) an example of a place word in the

relative position. Unfortunately, however, (22a) is a

misleading example for relative words. That the place word

tang xia occurs immediately after guo is a result of the

transitivity requirement of guo. The verb guo requires a place

word to be its transitive goal. In contrast, the verb l_i is

intransitive when denoting ’to stand’. That is why yu is

required in the utterance of (22b).^®

Typical examples of place words in the sentence final position as relative words are listed in (26). It is worth noting that time words scarcely occur in the yu structure on the basis of data from Zuozhuan and Mencius. Time words in these texts are mostly relative words in the sentence initial or post-PRED position.As will be explained in the following discussion, this may imply a difference between time and place words.

The verb l_i in transitive uses denotes a causative sense ’to make sb. stand’. In an extended denotation, it means ’to make sb. become a king’.

In Wang’s examples of time words, there is a rare use of time word in a yu phrase:

(a) Zi yu shi ri ku, ze bu ge. (Lunyu)

master-PREP-DEMON-day-cry, CONN-NEG-sing ’In the same day if the master has cried, then he won’t sing.’

This sentence comes from Lunvu [ ] ’Analects of Confucius’ and the translations on the demonstrative shi are adjusted for English. 65

The preposition £u is one of the most widely used in

Classical Chinese.** According to Yang (1954) and Dobson

(1974), yu phrases may link with the main predication in many different senses. Compared with the English prepositions, vu may range from ’at ’, ’in’, ’on’, ’upon*, ’from’, ’among’, ’by ’ to ’to’ and ’toward’ (of. Dobson p.862-4). Yu may indicate a starting or an ending point of an action; it may indicate a time, a place or a goal of an action, or other relationships between P and 0 (cf. Yang p. 571-6). A number of the common usages of yu to indicate location are listed as follows:

(23a) Wang li yu zhao shang. (lA/2)

king-stand-PREP-pond-above ’The king stood by the pond.’

(23b) Wen-Wang sheng yu Qi-Zhou. (4B/1)

NAME-be born-PREP-NAME ’King Wen was born in Qi-Zhou.’

(23c) She Zi-Fu-Zhao-Bo yu ta yi (Zhao 23)

dwell-NAME-PREP-another-city ’[Shi-Bo] forced Zi-Fu-Zhao-Bo to dwell in another city.’

(23d) Dai ye, zhi yu Qi . (Ai 6)

reach-night, arrive-PREP-NAME ’By night, [Yang-Sheng] arrived at Qi.’

** There are different opinions regarding the treatment of functional words like yu and yi (p.c. Prof. Hsueh). For the purposes of discussion, I will refer to them as prepositions. 66

(23e) Nai xian gong yi chu yu zhao. (Cheng 16)

then-lift up-duke-PREP-go out-PREP-pond ’[Zhen] then lifted up the duke to make him out of the pond.’

(23f) Shi chu yu Chen Zheng zhi jian. (Xi 4)

troops-go out-PREP-NAME-NAME-GENI-middle 'The troops went through Chen and Zheng.’

The above examples display how yu introduces location of an action. All the yu phrases occur after the main predication.

Exactly how the main predication is semantically related to the yu phrase depends on the lexical properties of the verb.

3.2.2. THE OPTIONALITY OF YU. To a certain degree the presence or absence of the preposition yu is controlled by the main verb. For example:

(24a) Zhao-Meng mou yu zhu dai-fu. (Xiang 27)

NAME-plan-PREP-group-ministers 'Zhao-Meng consulted with the ministers.’

(24b) Shei xia mou ren? (Wen 16) liPgmA? who-leisure-plan-people 'Who would have time to scheme [how to attack] people (us)?’

Mou ’to plan’ in (24) is a verb which requires a goal object in its transitive usages. When a noun holds a comitative relationship with mou. the preposition yu has to show up. In 67

(24a), the NP zhu dai-fu ’the ministers’ is of a comitative relationship with mou. and therefore yu appears to head the

PP. In (24b), ren ’people’ is the goal of mou and thus yu is not allowed to occur between the verb and the noun.

Another example concerning the requirement of yu involves the verb ru ’to enter’. Ru is a verb which requires a locative

PP headed by yu. When the verb takes a goal noun without the marking of yu, it must imply that the action is carried out by force;!*

(25a) Chu-Zi ru yu Ying. (Ding 5)

NAME viscount-enter-PREP-NAME ’The viscount of Chu entered the Ying city.’

(25b) Wu qi ru Ying hu! (Zhao 31) ! NAME-probably-enter-NAME-PAR ’Wu will probably forcedly enter the Ying city!’

The Ying city was the capital city of the Chu state. Thus when the viscount of Chu entered the city, he did not need to enter forcedly. Therefore, yu is required in (25a). But the state of Wu presented a constant threat to Chu. When the Wu troops attempted to enter the city, they had to enter forcedly.^*

!* Of course, it is possible that these are polysemous ru, one for the meaning of ’to enter’, the other for ’to enter by force ’.

This analysis is on the basis of the ru occurrences in Zhozhuan. In the text in question, ru in the transitive sense regularly means ’to enter by force’, while in the intransitive sense it regularly means ’to enter’. (Also see Note 19.) Prof. Hsueh suggests that this distinction does not exist 68

That is why zu is not allowed to appear in (25b).

On the other hand, %u may be optional in certain

predictable conditions. One rule states that when direct

object of the verb is the pronoun zhi. the following £u will

be omitted. Take the verb mou as an example. Although the

comitative NP is required to occur in a yu phrase by the verb mou as shown in (24), when the goal NP is substituted by the pronoun zhi. yu can be dropped, as can be seen in (26a):

(26a) Mou zhi duo zu. (Xiang 8)

plan-it-quite a few-families ’[Zi-Si] made a plan for it by consulting with a number of family leaders.’

More parallel examples are listed as follows:

(26b) Gong-Ruo xian gong yu Gong-Wei. (Zhao 25)

NAME-offer up-bow-MARKER-NAME ’Gong-Ruo offered up a bow to Gong-Wei.’

(26c) Qie ma er xian zhi Zi-Chang. (Ding 3)

steal-horse-CONN-offer-it-NAME ’[The Tang people] stole the horse and offered it to Zi-Chang.’

in Lunyu. For example, there is an utterance Zi ru tai miao. mei shi wen [ ] , in which ru has no chance to indicate entering by force. If ru systematically makes a force implication in Zuozhuan. but does not in Lunvu. then the languages of the two texts may have existed at different times. Or the inconsistency may have resulted from dialectal or stylistic differences. 69

(26d) Mai qi shou yu Zi-Ju-zhi-Men. (Wen 11)

bury-h,is-head-PREP-NAME ’[Fu-Fu-Zhong-Sheng] buried his head at the gate of Zi-Ju.*

(26e) Sha er mai zhi ma shi zhi zhong. (Wen 18)

kill-CONN-bury-him-horse-manure-POSS-middle ’[Zhong] killed him and buried him inside horse manure.’

(26f) Gu feng Zi-Guo yu X i . (Ai 18)

therefore-fief-NAME-PREP-NAME ’Therefore, [the king] fiefed Zi-Guo in Xi.’

(26g) Xiang zhi bu ren, feng zhi You-Bi. (5A/3) itZWlMo NAME-NOM-NEG-humane, enfeoff-him-NAME ’Xiang was the most unhumane person, [but Shun] enfeoffed him in You-Bi.’

It is likely that VP-zhi-NP. with the NP to be a person or a place, is an idiomatic pattern in Classical Chinese. As shown

in (26b), xian ’to offer’ is a P taking the offered object as

0 and a %u phrase to indicate the ending point of the action.

But in xian zhi Zi-Chang ’to offer it up to Zi-Chang’ of

(26c), however, because the pronoun zhi replaces ma ’horse', yu is allowed to be dropped. Similarly, mai ’to bury’ is a P taking the buried object as O and requiring the place of the action to be in a yu phrase, as shown in (26d). But when the

O is replaced by zhi. yu before the place is allowed to be dropped, as in (26e). The same opposition holds in (26f-g) for the verb feng ’to enfeoff’. 70

In addition to the above factors on the optionality of the preposition %u, there are still other factors which have an influence on the presence or absence of yu.

Diachronically, as Wang (1990:148-52) points out, %u after the verb zhi ’to reach/arrive’ gradually disappears when the verb drifts from an intransitive to a transitive. In

Zuozhuan. zhi can alternatively take a location word as its object without vu :

(27a) Zhi He nai fu. (Zhao 2)

reach-NAME-then-return ’[Lord Zhao] got to the River and then returned.’

(27b) Xi zhi yu He. (Xi 4)

west-reach-MARKER-NAME ’[It] reached the River in the west.’

Zhi He in (27a) and zhi vu He in (27b) mean ’to reach the

(Yellow) River’, having He ’the River’ with or without yu.^i

But based on the data, the alternative without yu is of higher frequency in occurrence. We may therefore suggest that zhi is developed to have a fully transitive sense by the time of

Zuozhuan.

Another factor on the optionality of yu is the semantic role of the noun after yu,. When yu is utilized as indicating

It is possible that zhi He and zhi vu He have subtle semantic differences by virtue of transitivity relationship (cf. Hopper & Thompson 1980). 71 an agent of passive, it is never omitted. For example:

(28a) Qiu zhuan Chen-Guo. (Xiang 20)

seek for-monopolize-NAME state '[Qing-Shi] sought for full power to govern the Chen state.’

(28b) Si wu fu er zhuan yu guo. (Xiang 21)

die-my-father-CONN-monopolize-PREP-state ’[Fan-Shi] made my father die and thus became an absolute head in the s t a t e .

(28c) Yi Fan-Shi wei si Huan-Zhu er zhuan zheng yi. (Xiang 21)

with-NAME-is-die-NAME-CONN-monopolize-government- PAR ’The head of the Fan clan was the cause of Huan- Zi’s death, and he afterwards monopolized the government.’

(28d) Xu bu zhuan yu Chu. (Zhao 18)

NAME-NEG-monopolize-PREP-NAME ’The Xu state is not monopolized by the state of Chu.

This use of ^ is a particular case in Classical Chinese. The regular pattern for si. ’to die’ is that the verb takes a direct object as a reason to die. In the example at issue, ^ is rather in a causative pattern in which the object wu fu ’my father’ is the victim of the causative verb.

This use of zhuan in the passive sense is uncommon to the text. However, zhuan here has to be interpreted this way (cf. Yang 1982a; 1399). According to the context, a Chu’s minister urged the prince to move the Xu people out of Chu territory. Thus, he said, the people of Xu would not be monopolized by the Chu state, and even if one day Xu was attacked by its enemy Zheng, the Chu state would have no responsibility to fight Zheng and save Xu. 72

In (28a-b), zhuan 'to monopolize’ is followed by a direct object Chen-Guo ’the Chen state’ or by a yu phrase yu guo ’in the state’. There is no yu for the former case, because Chen-

Guo is understood as a goal NP indicating the government of the Chen state.In the latter case, because of the presence of yu, the object is in a locative case. Suppose yu is added before Chen-Guo in (28a), the sentence will probably still be grammatical, even though we do not have the actual example.

Since it is now in a locative PP, Chen-Guo refers to ’the state’ rather than ’its government’.

On the other hand, if the yu is dropped out of (28b), guo will be blocked to be a goal noun, because another noun zheng ’government’ has already been in existence for the goal noun of zhuan. as in zhuan zheng ’to have full power in the government’ of (28c). Zhuan guo seems to have been pre-empted by zhuan zheng. since they are synonymous expressions, (cf.

Clark & Clark 1979:798)

Zhuan in (28d) is required by the passive voice rather than the verb zhuan (cf. Note 23). The object of yu indicates an agent of the event rather than a locative case. If yu is dropped out, the agent will inevitably become a goal of the verb chuan and result in a meaning contrary to the original.

Such a subtle semantic shift is fairly common in the language. 73

3,2.3. SYNTACTIC POSITIONS OF THE RELATIVE EXPRESSIONS OF

TIME AND PLACE. As defined in 3.2, when a place word is subordinate to the main predication without a preposition, this word is in a relative position. Moreover, as shown in

(26e) and (26g), the only condition for a place word to be a relative word after the main predication without yu is when the direct object is replaced by the pronoun zhi. Otherwise, a yu plus place PP will be expected in a post-PRED position, as shown in (23). But, this restriction does not apply to those verbs which require a place word as the direct object, as in (22a), (25b) and (27a).

When a place word occurs before the main predication, as in (20), it is always a relative word. When it occurs after the main predication, it usually occurs in a yu phrase. The contrast of a place word before and after the main preidcation is very transparent in (29):

(29a) Tian-xia wu di yan. (7B/4)

world-NEG-enemy-PAR ’There is no equal in the world.’

(29b) Ren ren wu di yu tian-xia. (7B/3)

humane-people-NEG-enemy-PREP-world ’Humane people have no equals in the world.’

Both sentences in (29) are existential sentences. Therefore, there is no verb controlling NP. The presence or absence of yu depends on the position of the place word. The place word 74

tian-xia ’world’ in (29b) shows up in a %u phrase, since it

occurs after the PRED wu di ’there is no equal’. But tian-xia

stands by itself in (25a), because in this case, it functions

as a predicative modifier, thus is a relative word.^®

Time words, on the other hand, scarcely occur in a PP.

This by no means implies that time words are not allowed to

occur in a PP. As Wang (1958:388) displays, time words are

sometimes marked by yu or yi. [ lil ] (another frequently used preposition, usually indicating insturment or manner). But it

seems unnecessary for a time word to be in a PP. Time words can occur, directly before or after a predication. In the

former case, they can be a relative word subordinate to the predication, or a nominal clause subordinate to the main clause. In the latter case, time words serve as the main predications. For example:

(30a) Wu ren ju qi jian qi ri. (Xiang 25)

NAME-people-dwell-its-space-seven-days ’The Wu people stayed in the middle of [the Chu troops] for seven days.’

However, there is a debatable point. In (29a), the particle van appears in the sentence final position. This particle is widely recognized as a fusion of yu ’in/at’ and zhi ’it’. If yan actually functions to indicate location, the sentence should be translated word by word as ’world-NEG- enemy-in-it’ or ’The world, there is no equal in it.’ Such an interpretation will make the place word tian-xia less like a predicative modifier by itself. 75

(30b) Yue shi nian sheng ju er shi nian jiao xun. (Ai/1)

NAME-ten-years-multiply-assemble-CONN-ten-years- teach-train ’The state of Yue multiplies and assembles [its people] for ten years and teaches and trains [its people] for another ten y e a r s .

The time expressions ai ri 'seven days’ in (30a) occurs after

■iu qi iian ’to dwell in between’, whereas shi nian ’ten years’ in (30b) occurs before sheng iu ’to multiply and assemble’ and jiao xun ’to teach and train’. Both time NPs indicate the duration of time of the events to which the main predication refers. According to a general principle in Classical Chinese, as well as in word order in general, qi ri in (30a) is the information focus of the sentence. (For the impact of word order, see Norman (1988:94) and Tai (1989:210)). Similarly, sheng iu and jiao xun are the focuses of (30b).

When a time word occurs before the predication, the word can be a relative NP subordinate to the main predication or a nominal clause subordinate to the main clause. As a result, time words in the initial position should become potentially ambiguous. For example:

(31a) San ri ku. (Cheng 3)

three-days-cry ’[The people] mourned for three days.’

According to the following context, Er-shi nian zhi wei, Wu qi wei zhao hu ! [ ! ] ’Wu would probably become a pond after twenty years’, (30b’) should be understood as two ten years in sequence. 76

(31b) San ri Qi-Ji qing shi. (Xiang 22)

three-days-NAME-request-corpse ’Three days later, Qi-Ji asked for the corpse.’

(31c) San nian zhi zhi. (Ding 3) E^±Zo three-years-stop-him ’[Zi-ChanJ stopped him [from returning] for three years.’

(31d) San nian nai bao Yue. (Ding 14)

three-years-then-revenge-NAME ’Three years later, [Fu-Chai] revenged on Yue.’

San ri ’three days’ and san nian ’three years’ can mean ’for three days’ and ’for three years’ as in (31a) and (31c) respectively, or ’being three days later’ and ’being three years later’, as in (31b) and (31d) respectively. In the former case, these time NPs are PRED modifiers indicating the duration of the event designated by the main predication. They themselves function as PRED.n in the latter case, indicating a period of time that has passed before the other event happens. Thus, they serve as subordinate clauses to the following main clauses.2?

This analysis on the time NPs as nominal PRED in a subordiate status follows an analysis of NPs utilized before

This is also in line with Tai’s temporal sequence principle (cf. Tai 1985). 77 the connective er by Hsueh (1990).^® This analysis can be supported by two more examples in Zuozhuan which show that time NPs in the initial position may be analogous to PREDs marked by PRED markers er and b u ;

(32a) Liu nian er Li-Gong ru. (Zhuang 14)

six-years-CONN-NAME-enter 'Six years afterwards, Lord Li entered [Zheng’s state capital].’

(32b) Bu shi nian wang fu zhao ye. (Xi 13)

NEG-ten-years-king-NEG-summon-PAR 'Not until ten years later, would the king summon [him] back.'

Liu nian 'six years' before er in (32a) and shi nian 'ten years' after the negative bu in (32b) are PREDs which are identifiable by the marking of er and (These PRED markers will be discussed in Chapter V.) Both time NPs indicate the duration before the second event happens.

Finally, there are some other conventionalized usages of time words in the language. Take ni 'day' as an example. When ri occurs in the sentence initial place, it means 'in the past', just like wang ri [ B ]. When ri occurs immediately before PRED, it means 'every day' or 'day by day' . For example :

About er as a connective for two PREDs, see discussion in Chapter V. 78

(33a) Ri jun you hui. (Zhao 4) H S WMo in the past-monarch-have-favor ’In the past, Your Majasty did a favor [to us].’

(33b) Gong shan ri shuang ji. (Xiang 28)

common-meals-every day-couple-chickens ’There are two chickens for daily common meals.’

(33b) Wu ri bi yu bing. (Ai 1)

NAME-day by day-wear out-PRED-military action ’The state of Wu is gradually worn out because of military actions.’

3.2.4. A DISCUSSION ON RELATIVE POSITIONS. In Wang’s definition on relative words, as translated in 3.2, the key to justifying a relative word is its modification function relative to the main verb. According to conventional studies on the linear order of syntactic slots by Chinese scholars, the modifier before PRED is ADV and that after PRED is COMP.

We may thus say that relative positions are concerned with the positions of ADV and COMP. Obviously, the above discussion on relative positions is limited to ADV.

As illustrated in the above analysis, relative time and place words usually occur before PRED when not being marked by a preposition. When a time word occurs after a predication and not in a PP, it actually functions as the main predication

(cf. (30a)). The condition for place words to occur after PRED without being in a PP is highly restrictive (cf. the analyses 79 on (26)).

Wang’s proposal on ’relative positions’ has its root on his idea about temporary functions. When a noun occurs in the pre-PRED and non-SUBJ position, i.e., a relative position, it must temporarily serve as ADV. From the present point of view, the noun has converted into an adverb.

Regarding the higher frequency in occurrences for time and place words as ADV than that for other nouns, it may be suggested that time and place words are more like modifiers than typical nouns. Therefore, time and place words are less than prototypes of the noun category. In addition, according to the above discussions, time words are less typical than place words.

At the same time, it has been argued in Chapter II that when a lexeme functions atypically, i.e., the lexical category of the lexeme correlates to a syntactic function unnaturally, a semantic drift should be expected. It is transparent in (31) that the drift does happen. More importantly, the created new meaning depends on the properties of main verbs and discourse contexts as well. (Context-dependent semantic drifts will be discussed in detail in the following chapters.) CHAPTER IV

NOMINALIZATION

0. Nominalization in Classical Chinese is often overtly marked by functional words such as zhi [ 2 ] , gi. [ ^ ] and zhe

[ # ]. These words are nonetheless hardly pure markers of

nominalization. For example, zhe may carry a much broader

grammatical sense than a nominalizer normally does, because

zhe can also mark off a typical nominal unit like a proper

noun (cf. Zhu 1983). Although the basic functions of these words are still in need of more careful investigation,

scholars agree that words or phrases must become nominals, when marked by them. In the following discussion, zhe is

tentatively labeled NOMINALIZER, zhi. GENITIVE, and oi.

GENITIVE ATTRIBUTE (which substitutes for a noun + a GENITIVE

marker ).

The present focus of interest in nominalization centers

on nominalization without overt markers. A verb or VP does

not have to be marked to play nominal roles. It can directly

occur in the SUBJ/0 positions and play those functions. This

is an interesting scenario shared by both Classical Chinese

and Mandarin Chinese, which gave rise to a debate on mingci-

hua [ ] (nominalization) in the 50s among Chinese

80 81

scholars.* The key issue of mingci-hua is whether a typical

verb is really converted into a noun when it occupies the

SUBJ/0 positions.

Without available phonetic transcriptions, it is

difficult to discern whether there were consistent segmental

or suprasegmental markers for word derivation in Classical

Chinese. Even so, it is possible to determine whether semantic drift occurs. An interesting but not fully developed point in the 50’s debate involves the semantic properties of verbs as

they function nominally. It is agreed that verbs in the nominal functions preserve only part of their verbal capacity.

Moreover, it is agreed that verbs in these functions are similar to verbs in the gerund form in English.

An unsolved issue of the 50’s debate is the concept of

’zero conversion’ as justifiable in the Chinese language’s

grammar. The most obvious obstacle to zero conversion is a common effort to avoid lexical overlapping. It seems to be a

shared opinion at that time that to admit zero conversion is

to admit that many lexical items belong to more than one

category. Under such circumstances, Wang’s proposal of

’temporary functions’ appeared as a compromise. (For details

about Wang’s theory, see Chapter II.)

However, Wang’s proposal does not completely solve the

problem of zero conversion, since the question remains whether

* For more information about the debate, see Hanvu de Cilei Wenti published by Zhongguo Yuwen in 1956. 82 lexical conversion is involved in the process of legitimizing the temporary functions. If lexical conversion, or categorial neutralization, is not assumed, it would be difficult to see how verbs can be allowed to play nominal roles.

Admitting zero conversion does not necessarily mean admitting lexical overlapping. When zero conversion happens, it may mean a contextual actuation and the actuation will become obliterated out of the context.% It is worth noting that not only verbs can be converted into deverbal entities by virtue of mingci-hua. nouns can similarly be converted to verbs or adverbs by virtue of fei-mingci-hua [ ]

(denominalization).s Nominalization is focused on in this chapter. Denominalization will be the topic of Chapter V.

Because of the limitation of the present study, only a small portion of ambs and verbs are examined in detail. The examined items share an important dénotâtional property: They denote something about human qualities or actions. Moreover, the selected items in the study, even though small in number, are highly representative. They will probably reveal general

^ Or this actuation may be conventionalized by and large. Then, a lexical split will eventually happen. If so, lexical overlapping is inevitable.

^ Deverbal entities are non-equivalents of nouns. This statement should be recognized from two aspects. First, the category of ’abstract nouns’ may probably not exist in the lexicon of the language. (For more information, see Chapter VI.) Moreover, evidence shows that conditions for deverbal entities to play the nominal roles are highly restrictive. This will be discussed in this chapter. 83 principles which have braod application.

As can be observed in the following discussions, the semantic drifts of de-ambs and deverbal entities are context- dependent. The context-dependent characteristic is associated with decategorization or category shift, and yet the size of the contexts vary from case to case. The restrictive force of the main verb is an essential factor in determining a context- dependent meaning, but sometimes a larger discourse context has to be taken into account to specify the innovation.

1. AMBS IN NOMINAL FUNCTIONS. By definition, ambs are ambiguous in terms of categoriality, since they possess both nominal and verbal features. As observed by Kennedy (1964), an amb can surface like a noun or a verb, judging by the adjunct it takes or the phrase structure it is in. That is,

’ambs’ by themselves have two semantic phases, viz. a static phase and a dynamic phase. In a nominal function, the static phase of ambs will be instantiated. A static instantiation usually refers to an abstract entity. Since this is a case of natural correlation between an amb and a nominal function, no category shift will be expected. A category shift occurs only when an amb is used to indicate a physical object.*

* This category shift has nothing to do with unnatural correlations. It may result from a context-sensitive process which makes a descriptive unit correspond to a context- specific referent. This process will be discussed in detail in 1.2. 84

1.1. AMBS AS PRED OF EQUATIONAL SENTENCES. Whereas an unnatural correlation between a lexeme and its function is regularly marked by semantic drift, a natural correlation gives rise to no semantic change in the denotation of the lexeme. When an amb functions as PRED in an equat ional sentence, the static phase is naturally instantiated. For example :

(1) Ying-Kao-Shu, chun xiao ye. (Yin 1)

NAME, pure-filial piety ’Ying-Kao-Shu is [a person] of pure filial piety.’

(1) is a typical equational sentence, marked by the particle ye. Xiao ’filial piety’ is an amb. The phrase chuen xiao ’pure filial piety’ serves as an assertive judgement on Ying-Kao-

Shu . Since the static phase of xiao is naturally instantiated, there is no semantic drift involved in the instantiation.^

1.2. AMBS AS SUBJ/0. When an amb surfaces in its static phase, it is usually treated as an entity denoting an abstract notion. Xiao in (1) is a typical example. However, an amb may

® Harbsmeier (1985;80ff) makes a good point on the semantic characteristics of ambs in the static phase. He explains that the amb ren [ {H ] can be ambiguous in English, since English lexically distinguishes an abstract noun from an abstract attribute. Thus, when translated into English, ren can be ’humane’ or ’humaneness’. On the other hand, ren may never obtain this semantic distinction in Classical Chinese, because the language does not distinguish the category of abstract nouns from that of abstract attributes. 85 refer to physical objects in a collective sense at times. This means that the amb is converted into a real noun.

In terms of the ways they are converted, ambs can be subcategorized into a labelling group and a describing group.

The describing ambs can undergo nominalisation without overt marking, whereas the labelling ambs are not allowed to be nominalized directly. When nominalized, part of the innovated meaning will be determined by the restrictive forces from the main verb or the discourse contexts.

1.2.1. DESCRIBING AMBS VS. LABELLING AMBS. ’Describing’ corresponds to the group of ambs whose members can function directly as PRED.a, while ’labelling’ corresponds to those ambs which cannot directly function as PRED.a. Labelling ambs are required to co-occur with vou or wu to function as PRED.a.

Most of ambs belong to the describing group, and they can be nominalized through zero derivation. The labelling group is extremely small in number, and its members cannot undergo a zero derivation. For example:®

(2a) You-li

HAVE-polite ’the polite one(s)’

® It is for brevity that the following examples are listed in a context-free situation. As a matter of fact, the exact objects the pronoun corresponds to will be recognized by discourse contexts or the main verbs. Examples about the impact of contextual factors will be presented later. 86

(2b) Wu-li mm NOT HAVE-polite ’the impolite one(s)’

(2c) You-zui

HAVE-guilty ’the guilty one(s)’

(2d) Wu-zui

NOT HAVE-guilty ’the innocent one(s)’

(3a) Da / xiao iz'h big / small ’the big/small one(s)’

(3b) Lao / you

old / young ’the old/young one(s)’

(3c) Xian R wise ’the wise one(s)’

(3d) Bu-xian

NEG-wise ’the unwise one(s)’

Li ’propriety’ and zui ’guilt’ in (2) belong to the labelling group. Da ’big’ / xiao ’small’, lao ’old’ / you ’young’ and xian ’wise’ in (3) belong to the describing group. 87

There is no way for labelling ambs to be nominalized directly. Only after going through an adjunct process will a labelling amb further be shifted to refer to concrete objects.

As shown in (2a) and (2c), the ambs M and zui are adjuncted by vou ’have*. It is the outcome of the process, viz. vou li and you zui. that can be nominalized. In negative forms, vou will be replaced by wu ’not have’, as in (2b) and (2d).’

The labelling group is a remarkably small group. In

Zuozhuan. in addition to vou-zui and vou-li. vou-de [ ]

’virtuous’ is another unit of high frequency in occurrence.

De ’virtue’ is one of the prototypes of ’abstract nouns’ in

Classical Chinese. According to Liu (1958:20), the number of pure abstract nouns is extremely limited, and de is one of them. Wang (1955b:24-6) observes that pure abstract nouns in the Chinese language originated from philosophical terms. The fact that li. and zui are analogous to ^ in their adjunct process seems to suggest that the labelling ambs are also abstract nouns.®

’ Sometimes when the quality that an amb denotes is limited to certain types of objects, the innovated senses are predictable. For example, since l_i and zui in (2) are human qualities, the de-ambs will denote ’people’. Even so, the ultimate factor for the innovation is the context.

® Up to this point, the status of abstract nouns remains vague. It is undoubtable, however, that abstract nouns are not core members of the noun category. They may be marginal units between the borderline of ambs and nouns. (For the status of abstract nouns, see Chapter VI.) 88

Describing ambs are nominalized without overt markers,

as shown in (3a-c). There are many antonymous pairs in this

group, as da vs. xiao in (3a). Those describing ambs which do not have an antonymous counterpart are usually adjuncted by bu in their negative form before being decategorized to refer

to concrete objects, as shown in (3d).

It seems that the effect of adjuncting you / wu to the labelling ambs is to make them more descriptive. This can be

supported by the fact that labelling units need to co-occur with you or wu to function as PREDs in descriptive sentences.

In comparison, members of the describing group can naturally

function as PRED.a without further modification. The contrast

seems to indicate that de-amb nouns are derived from items of descriptive nature.

1.2.2. CONDITIONS FOR DE-AMBS. As explained in Chapter III, nouns of Classical Chinese make no distinction in number,

unless a numeral appears before a head noun. A noun by itself denotes objects of the same kind in a collective sense, and

its number, singular or plural, will be implied in discourse contexts. This is also true for de-amb nouns. De-ambs always

correspond to context-specific referents. Thus, they obtain

different degrees of specificity from context to context. For

example : 89

(4a) Wu jian xin gui da, gu gui xiao. Xian da hou xiao, shun ye. (Wen 2)

I-see-new-spirit-large, old-spirit-small. place ahead-large-place behind-small, proper-PAR ’I saw that the spirit of the newly dead one (Lord Xi ) is large, and the spirit of the formerly dead one (Lord Min) is small. To place the large ahead of the small, this is properness.’

(4b) Xiao jia da. (Yin 3) 'j'AO* small-intimidâte-big ’The inferior (Zhou-Xu) intimidates the superior (Wan)’

(4c) Da wu qin xiao. (Xiang 19)

big-not-invade-small ’The big states must not invade the small states.’

The de-ambs ^ and xiao correspond to different referents. The number and specifity of the referents differ as well. With respect to the discourse contexts these utterances are within, the de-ambs in (4a) and (4b) refer to singular persons respectively, while in (4c), the de-ambs refer to ’states’ in general. As shown in the preceding context of (4a), for example, the de-amb da corresponds to xin gui ’the spirit of a newly dead’ and xiao corresponds to gu gui ’the spirit of a formerly dead’. A larger context indicates that the newly dead refers to Lord Xi and the formerly dead refers to Lord

Min of the Lu state.

Moreover, the restrictive forces of main verbs on the transitive goal are critical in determining the conversion of 90 de-ambs. When a verb requires a goal which designates concrete objects, the amb after it will become de-ambed and designate what is required by the verb. If a verb is allowed to take a goal in a circumstantial relation, then the amb will be understood by its static phase. For example:

(5a) Jiu yue, ru Q i . Tao bu jing ye. (Huan 2) xn, Afôo nine-month, enter-NAME. make war-NEG-respect-PAR 'In September, [the troops of Lu] forcedly entered the Qi state. [Lu] made war on Qi ’s disrespectful manner [at the Lu court].’

(5b) Qiu, fa Chen. Tao bu zhong ye. (Xi 4)

autumn, attack-NAME. make war-NEG-loyal-PAR ’In the autumn, [Qi] attacked Chen. [Qi] made war for Chen’s disloyalty.’

(5c) Gu zhe ming-wang fa bu-jing. (Xuan 12)

ancient-NOM-wise king-attack-the disrespectful ’In old days, wise kings attacked the disrespectful warlords.’

(5d) Yi zhi ren fa zhi bu-ren (7B/3)

with-extreme-humane-attack-extreme-unhumane ’With the very humane leaders to attack the very unhuman leaders’

Tao ’to make war on’ in (5a-b) and fa ’to attack’ in (5b-d) differ in their restrictive forces on the goal. The action verb tao allows the following unit to hold a circumstantial relation. As a result, bu jing ’being not respectful’ in (5a) and bu zhong ’being not loyal’ in (5b) are given as excuses 91 for starting a war. However, the action verb fa requires a concrete aim of action. Because of the restrictive forces from the verb fa, bu-jing in (5c) and bu-ren in (5d) have to be de- ambs. The contexts indicate that these de-ambs correspond to

’political leaders’. Thus, bu-.iing means ’the disrespectful leaders’ and bu-ren means ’the unhumane leaders’.

The extent to which main verbs have influence on the object is even more transparent in the following examples:

(6a) Zhu bu-ren ye. (5A/3)

ki11-unhumane-PAR ’[It is that he] killed the evil people.’

(6b) Guo-jun hao ren, tian-xia wu di yan. (7B/4)

king-fond of-humane, world-NEG-enemy-PAR ’If a king is fond of humaneness, he will have no equals in the world.’

(6c) Yan bao hu ren yi ye. (6A/17)

say-full-PAR-humane-righteous-PAR ’It means full of humaneness and righteousness.’

(6d) Bu xin ren xian, ze guo kong xu. (7B/12)

NEG-believe-humane-wise, then-country-empty-empty ’If [the king] does not believe the humane and wise men, then the country will become empty.’ 92

In (6a), zhu ’to kill’ is a verb requiring a human object as

its transitive goal.* Meanwhile, ren is a quality normally

limited to human beings. Therefore, bu-ren are understood as

’unhumane (cruel) people’. This is confirmed by the preceding context which mentions about four barbarous leaders who were

killed by a kind and loving leader. In (6b), ren denotes the notion ’humaneness’ after hao ’to have a special liking for’.

Based on the occurrences of hao in the texts at issue, this verb regularly takes entities of ’quality’ or ’activity’ as

its transitive goal. Accordingly, the static phase of ren is naturally instantiated after hao. In (6c), ren vi should be understood as ’humaneness and righteousness’. This is because the verb bao ’to be full of ’ exempts human objects from its goal.

The case of (6d) differs from the above cases. The verb xin ’to believe’ does not enforce a human goal, although ren xian is understood as ’humane and wise people’. It is rather kong XU ’empty’ in the following context that helps ease the decoding of ren xian. The emptiness of a state is more likely

to result from the lack of humane and wise people than from the lack of the humane and wise qualities. Thus, ren and xian are realized as de-amb nouns.

* According to the data, there are two fixed denotations for the form zhu. The one denoting ’to kill’ requires a human object as its transitive goal. The other denotes ’to blame’ as in zhu iu yu tu-ren Fei [ ] ’ to blame the loss of shoes on his servant Fei’ (Zhuang 8). In this denotation, the human object has to occur in a yu phrase. 93

In addition, de-amb nouns are usually found in pairs.

This may have more rhetoric impact than grammatical impact.

However, juxtaposing or parallelism should have, more or less, some influence on the decoding processes.

(7a) Gui lao you (Xiang 9 / Xiang 26)

return-old-young ’Return the old and the young’

(7b) San ming yue: Jing lao ci you, wu wang bin lu. (6B/7)

three-command-say: respect-old- kind-young, NEG- forget-guests-travel ’The third command says: Respect the old and love the young, and don’t neglect the guests and the travelers.’

Coincidentally, juxtaposing two ambs sometimes facilitates the decoding process of unmarked de-ambs. In (7a), the de-amb nouns after the verb gui ’to return’ appear in a pair. But the reason for identifying them to denote ’people’ is by no means the juxtaposition. On the other hand, the parallelism of jing lao ’to respect the old’ and ci you ’to love the young’ in

(7b) indeed helps us recognize the amb ci. to be instantiated in its dynamic phase, denoting ’to treat kindly’, because çi is parallel to the transitive verb jing ’to respect’ and its following unit you is in parallel to lao. the direct object of jing. 94

2. VERBS IN NOMINAL FUNCTIONS. Verbs may carry out the nominal functions, denoting a palpable object or an abstract notion, without being overtly marked. In the former case, a verb is converted into a noun. In the latter case, a verb is decategorized and becomes a deverbal entity.*-® Nominalisation is used to term both cases, because verbs in these situations are conventionally recognized as being ’nominalized’. It will be demonstrated that semantic drifts consistently occur with the nominalisation.

2.1. VERBS AS PRED OF EQUATIONAL SENTENCES. When serving as

FRED of an equational sentence, a verb or VP will lose part of its inherent capacity and obtain some property innovated to practice the syntactic function in question. This is just as Dawson (1968:49) notes:

Here perhaps the best way of thinking of [a ye marked verbal unit] is that it partakes of the quality of the non-verbal ('X is Y ’) type of sentence, the essence of which is that it does not record acts, but judgements, does not describe momentary events but continuing states.

In Dawson’s classification, sentences are either verbal or nominal. The marking of ye indicates that verbal sentences

*® ’Converting’ has to do with category shift. When a verb is converted into a noun, for example, it is not a verb any more, and it should be able to play nominal role like a noun is able to. ’Decategorizing’ means demotion of categorial property. When a verb is decategorized into a deverbal entity, the verbhood is demoted, but the entity is not converted into another category. 95 are converted into nominal sentences. This can be seen in an example of Dawson's (p.50):

(8a) Gu yi yang yi zhi ye. (lA/7)

therefore-with-sheep-change-it-PAR 'That was the reason why I changed it for sheep.'

Dawson thinks that a predicate should inevitably lose the capacity of narration when it is marked by the final particle ye. Since the predication vi yang vi zhi 'to change it for sheep' in (8a) ends with ye, it is to express an assertion about an incident rather than to narrate the incident itself.

Moreover, because of thé adjunct of gu 'therefore', this predication is explanatory in nature.

When the ye marked predication is not modified by an explanatory ADV like gu in (8a), the predication is usually a Judgement:

(8b) Shi fu Jie ye. (6B/9)

ANAPHORIC-assist-NAME-PAR 'It is to assist (the tyrant) Jie.'

ANAPHORIC means an anaphoric word indicating that the antecedent complex predication serves as SUBJ of the ye predication.^^

Due to the restriction of ye, it is inappropriate to treat fu

Jie 'to assist Jie’ as a narrative element.

The antecedent predication sets up a situational or hypothetical event for the following judgement. 96

The fact that £e and zi [ ^ ] are mutually exclusive

further supports Dawson’s analysis of ye. Yi is a particle

involving the concept of time. Cikoski (1976:75-84) thinks

that yi can imply a real time, a definite time, or a cognitive or judgement time for a nuclear (an equivalent to non- equational) sentence.*2 The particle ye, on the other hand, never indicates temporal aspect (cf. LU 1942:278ff). The effect of ye on a verbal unit is simply to demote its narrtive capacity and the verbal unit will therefore become a deverbal entity.

2.2. VERBS AS SUBJ/0. On the surface. Classical Chinese verbs or VPs seem free to carry out nominal functions. This is especially true when a reported event is mentioned or repeated later in the same discourse. There are nevertheless some restrictions which block verbs to carry out the nominal functions. One important restriction states that verbs do not occur in the nominal slots which lexical nouns are available to fill in. When a verb plays a nominal role, it needs to be converted into a noun or decategorized into a deverbal entity

(cf. Note 10 and 2.2.2). Even without overt markers, the

Cikoski’s analysis on yi is similar to LU’s (1942:275) analysis, in which Lii claims that yi usually indicates an event which happened, will have happened or is supposed to have happened. Both authors consider that the grammatical function of yi is close to the final particle le [ 7 1 in Mandarin. Both also assumes that yi implies ’change of state’. Therefore, if an NP is marked by yi, it means that the NP is an outcome of a change. 97

conversion or decategorization should always be recognizable

by a semantic drift. Moreover, when a series of verbs occur

consecutively, P, the center of predication, is determined by

subcategorization of verbs.

2.2.1. VERBS WHICH DO NOT SERVE AS SUBJ/0. In Classical

Chinese, verbs enjoy quite a flexibility to be employed as nouns. However, deverbal innovation is actually blocked when a lexical noun is ready for a certain job. For example:

(9a) Zheng-Bo ru Jin. ... Shi xing ye, Zheng-Bo chao Jin. (Xiang 24) mem, NAME-leave for-NAME ... DEMON-movement-PAR, NAME- court-NAME 'Count Zheng left for the Jin state. ... In this travel. Count Zheng visited the Jin court to pay his respects.’

(9b) Zheng Bo-You shi Gong-Sun-Hei ru Chu. ... Bo-You yue: Shi xing ye.

NAME-NAME-make-NAME-leave for-NAME ... NAME-say: generation-movement-PAR 'Bo-You of the Zheng state made Gong-Sun-Hei go to the Chu state. ... Bo-You said: (It is) a mission (your family carried out) for generations.'

The verb ru 'to leave for' requires a transitivity goal to

indicate destination of the action. In Zuozhuan. no evidence

demonstrates that this verb by itself can be nominalized and

serve as SUBJ or 0. Rather, two cases are found in which the

noun xing 'movement' is used to label the event of 'leaving 98

for'.As shown in (9a), ru, which reports an event in the antecedent context, is repeated by xing. Ru in (9b) is also replaced by xing. That the antecedent verb ru is not nominalized to refer to the event is because xing is conventionalized to carry out this job.

Similarly, van 'words’ is a noun which normally blocks redundant deverbal innovation of the verb vue 'to say':

(10a) Wang wei Shi-Zi yue: .... B: king-tell-NAME-say ... 'The king told Shi-Zi that ....'

Shi-Zi yin Chen-Zi er yi gao Meng-Zi; Chen-Zi yi Shi-Zi zhi yan gao Meng-Zi. (2B/10)

NAME-by-NAME-CONN-with-tell-NAME;NAME-with-NAME- GENI-words-te11-NAME 'Shi-Zi entrusted Chen-Zi to tell Mencius; Chen- Zi told Shi-Zi's words to Mencius.'

(10b) Qin xing-ren ye jie Jin shi yue: ....

NAME-emissary-night-warn-NAME-troops-say ... 'Qin's emissary warned the Jin troops by night, saying: ....'

Shi-zhe mu dong er yan si, ju wo ye. (Wen 13)

emissary-eyes-move-CONN-words-exert utmost, be afraid-us-PAR 'The emissary's pupils kept moving and his words were exaggerative, [it is that he] feard us.'

13 The form xing [ fr ] is polysemous in Classical Chinese. While the nominal alternant denotes 'movement', the verbal alternant denotes 'to move'. 99

In (10a), van ’words’ refers to what Shi-Zi was told by the king to tell Mencius. In (10b), van refers to the warning that the Jin troops received from the emissary. Whatever content is reported, a telling or a warning, it is mentioned by van.^*

On the other hand, when a conventionalized noun is ineffective in achieving the speaker’s purposes in specific occasions, nominalziation of a verb from a previous context will be expected;

(11a) Xia, zhu-hou zhi dai-fu cong Jin-Hou fa Qin, yi bao Li zhi yi ye. ... Ci yi ye, bao Li zhi bai ye. (Xiang 14) 3C.

summer-warlords-GENI-ministers-f ollow-NAME-attack- NAME, to-revenge-NAME-GENI-battle-PAR. . . . DEMON- battle-PAR, show revenge-NAME-GENI-defeat-PAR ’In the summer, the ministers of warlords followed the duke of Jin to attack Qin to revenge for the battle of Li. ... This battle is to show revenge for the former failure of Li.’

(lib) Yu Shan-Xiang-Gong yu, zou cheng qi fa. (Cheng 16)

with-NAME-speak, frequently-mention-his-attack ’[Xi-Zhi] spoke with Lord Xiang of Shan, mentioned his attack from time to time.’

Yi ’battle’ is the noun usually referring to events reported by the verb fa ’to attack’, as exemplified in (11a). However,

One possible explanation for verbs like ru and yan not generally found in the nominal functions is that these verbs inherently possess absolute dynamic senses. If it is correct, it should be expected that there exists a sub-category of verbs in the language, the members of which have similar absolute dynamic senses. 100 fa is nominalized as in qi fa ’his attack’ in (lib). This verb is nominalized in this case by the speaker’s intention to put emphasis on a specific attack. According to the explanation of fa by Zuozhuan (Zhuang 29) and an additional note by Yang

(1982b:244), fa conventionally reports on attacks justifiable by vindicative reasons. Therefore, in this special situation, this verb may probably be nominalized so as to exaggerate the speaker’s military merits to impress a high-ranking official of the king.^®

2.2.2. DEVERBAL ENTITIES. The above discussion of de-amb nouns suggests that ambs of a descriptive nature can directly be converted into nouns to indicate physical objects, while ambs of a labelling nature tend to be instantiated to refer to abstract entities.

This speculation on the difference of describing vs. labelling applies to deverbal units. According to Hopper &

Thompson (1984), when a verb is nominalized, it means that an event is conceptualized as an entity. Since deverbal units are to label ’actions’ or ’states’, they are in general of a labelling nature. Thus, it should be expected that deverbal units normally denote abstract entities.

The cases in (11) can be interpreted by virtue of Clark & Clark’s (1979:797) principle of pre-emption. In (11a), the verb fa is not nominalized for the existence of the well- established synonymous noun xing. But when fa is instantiated in a specialized meaning, it is not synonymous with xing any more. Thus, its nominalization should not be pre-empted. 101

In Zuozhuan and Mencius. deverbal nouns, i.e., nouns

which refer to physical objects, are small in number. More

importantly, deverbal nouns are usually found in certain

highly idiomatic expressions. In (12) are representative

examples :

(12a) Nan nu tong xing, qi sheng bu fan. (Xi 23)

man-woman-same-surname, their-grow-NEG-prolific ’[If] the husband and wife bear the same family name, their offsprings won’t multiply well.’

(12b) Shi shi min yang sheng sang si wu han ye. (lA/3)

ANAPHORIC- make-people- raise-live-bury-die-NEG- regrets-PAR ’It is to make people raise the alive and bury the dead without regrets.’

Sheng in (12a) means ’offsprings’ which is derived from the parent verb ’to grow’, while sheng in (12b) means ’people who

are alive’ derived from the parent verb ’to live’. Similarly,

si in (12b) means ’the dead’ from the parent verb ’to die’.

Some more examples of deverbal nouns analogous to these concrete uses of sheng and sj, are listed in (13):

(13a) Qi sheng bu zhi. (Zhao 1)

their-grow-NEG-prolific ’Their offsprings won’t multiple well.’

(13b) Qi zi-sun bi fan. (Xuan 3)

their-offsprings-def initely-prolific ’Their offsprings will definitely multiply well.’ 102

(13c) Ai si shi sheng (Zhao 27)

mourn-die-serve-live ’To mourn the one who has died and to serve the one who is still alive’

(13d) Song wang shi ju (Xi 9)

send off-leave-serve-live ’To send off the one who has died and to serve the one who is still alive’

The deverbal use of sheng in (12a) is parallel to that of

(13a). Both uses of sheng actually mean zi-sun ’offsprings’, as zi-sun in (13b). The uses of sj^ and sheng in (13c) are parallel to those in (12b) and also to wang and ju in (13d).

Although the parent verbs wang ’to leave’ and jja ’to live’ differ from s% and sheng. both sheng and ju denote ’the one who lives’ and ^ and wang denote ’the one who died’.^®

The deverbal nouns in (12-13) are in fixed uses of some idiomatic patterns. Therefore, it is probable that these uses do not result from the productive process of nominalization.

Normally, deverbal units are labels of ’actions’ or ’states’ of the parent verbs. The deverbaled sheng and sj, for example, often designate abstract notions of ’being alive’ and ’being dead’ in the texts:

Each of the deverbal nouns in (13c-d) corresponds to a context-specific referent. 103

(14a) Jian qi sheng, bu ren jian qi si. (lA/7)

see-its-live, NEG-bear-see-its-die 'Since you see its being alive, you won’t bear to see its being dead.’

(14b) She sheng er qu yi zhe ye. (6A/10)

abandon-1ife-CONN-take-righteousness-NOM-PAR ’(This is to) sacrifice life in order to choose righteousness.’

The deverbal sheng and si which denote physical objects in

(12-13) denote abstract notions in (14). Unlike the idiomatic uses, sheng and si in (14) have context-dependent denotations.

Interestingly, the occurrences of sheng in (14) come from a same parent verb but denote entities with different degrees of abstractness. Sheng in jian ai sheng ’to see its being alive’ denotes the state of ’being alive’, while that in she sheng ’to sacrifice life’ denotes ’life’ itself. For a context-dependent denotation, the degree of abstractness is determined by the verb, and discourse contexts as well.^’

It is important to distinguish the uses of sheng and ^ in (14) from those in (12-13). Whereas the latter are typical deverbal nouns, the status of the former cases may be subject to interpretation. Some scholars such as Zhu (1985, 1990)

This notional difference between ’being alive’ and ’life’ is lexically distinctive in English but derivational through restrictive forces of the main verbs and discourse contexts. More precisely, this difference on the degree of abstractness is unlikely to be detected if not through the English senses. 104 suggest that verbs in the SUBJ/0 positions should remain verbs. However, from our point of view, not to treat sheng and si in (14) as outcomes of nominalization will create doubt about the necessity of full words categorization. If verbs remain as verbs in the nominal functions, why should verbs be distinguished from nouns in the first place.

The assumption on ’nominalization* is that this process legitimizes verbs to carry out nominal functions, given that nouns and verbs are distinct in the lexicon. The concern here about the uses of sheng and ^ in (14) is in what sense they are nominalized. Perhaps the status of verbs in the nominal functions is just as Hopper & Thompson (1984:746) describe:

The nominalized verbs are ’conceptualized metaphorically as objects’ rather than ’bona fide’ nouns.

In other words, verbs in the nominal functions will lose part of the verbal property and thus be decategorized. At the same time, these decategorized verbs will preserve part of the verbal property and consequently are not analogous to typical nouns. Deverbal items of this kind are labelled ’deverbal entities’.

As Prof. Hsueh points out, traditional philologists before Mashi Wentong treated full words as a single category, since they could not see any need to distinguish those lexemes which naturally occur in certain positions from those which unnaturally occur in the same positions.

Hopper & Thompson make this suggestion in their effort to explain the morphological asymmetry of ’nominalization’ and ’verbalization’. 105

2.2.3. RELATIVE VERBHOOD. Relative verbhood is a eue for identification of P in a series of consecutive verbs. Chan

(1984) examines a group of verbs in Mandarin Chinese which display absolute verbhood. Following Chao (1968), Chan labels these verbs as pro-verbs. According to Chan (p.60), the pro­ verbs occupy 'the central nucleus-like position’ and the emergency of a pro-verb will demote the verbhood of other verbs in the same context.^® Since these verbs consistently show strong verbhood relative to the other co-occurring verbs, they should always act as P, while the following unit, whether nominal or verbal, cannot but be 0.

There is no systematic evidence for the existence of pro­ verbs in Classical Chinese.The 'central nucleus-like position’ is determined by a sub-category of verbs which regularly serve as P for their verbal propertiesFor examples :

Chan asserts that the verbhood of verbal units after a pro-verb is demoted. However, he does not make explicit if this demotion has to do with nominalization.

The only unit which may be comparable to pro-verbs in Mandarin is the verb you [^] 'to have’. The verbal element after you seems to become static. But more investigation is in need to explain the essential function of you.

Unlike pro-verbs in Mandarin Chinese, verbs of this sub-category never imply full sense of action. They are not ’stylistically useful for converting any verb into a fuller verb — more aspectual involvement’ (Chan, p.62). 106

(15a) Mou jiu Song ye. (Cheng 18)

plan-rescue-NAME-PAR ’[The warlords] made plans for rescuing Song.’

(15b) Qun chen yuan feng Feng ye. (Yin 3)

group-ministers-hope-show loyality to-NAME-PAR ’The ministers hope to show loyality to Feng.’

(15c) Zi-Yu qing sha zhi. (Xi 23)

NAME-request-kill-him ’Zi-Yu requested to kill him.’

The verbs mou ’to plan’, yuan ’to hope’ and oing ’to request’ in (15) belong to a set of verbs which have strong narrative force in making the following verbal unit be conceptualized as a deverbal entity. VPs after these verbs are inevitably decategorized and thus function as O.

Moreover, there is a small set of adverbs in Classical

Chinese which frequently occur before a verbal series. These adverbs are split from original verbs such as lai ’to come’, wang ’to go’ and chu ’to go out’. It will be assumed that these forms have become context-sensitive polysemous words, each of which contains a verbal alternant and an adverbial alternant. When a polysemy of this kind occurs immediately before a series of verbs, the adverbial alternant will be instantiated.^^

In addition to the directional modifiers, there are other polysemous words, the adverbial alternant of which will be instantiated before another verb. The form hui is a good example : 107

The occurrence of lai in its adverbial alternant is as exemplified in (16):

(16) Qiu, Zheng ren yi wang ming lai gao fa Song. (Yin 9)

fall, NAME-people-by-king-order-come-tell-attack- NAME ’In the fall, a Zheng’s messenger by the order of the king came to report on the attacking at Song. ’

Conventionally, lai is treated as a verb, meaning ’to come’.

If this is the case, there are three consecutive verbs in (16) with two serial predications. The first P is the verb lai and the second P is gao ’to tell’. However, it is more likely that lai in this case is an adverb and functions as a directional modifier to the predication headed by gao. This analysis can easily be attested in many other examples:

(a) Qi ren, Chen ren, Cao ren fa Song. (Zhuang 14) #A, m x , WAffe^o NAME-people-NAME-people-NAME-people-attack-NAME ’The Q i , Chen, and Cao troops attacked Song.’

(b) Xia, Shan-Bo hui fa Song. (Zhuang 14) X, # # # % % . summer, NAME-meet together-attack-NAME ’In the summer, Earl of Shan joined to attack Song.’

(a) and (b) are two events in sequence. In both events, the VP fa Song ’to attack the Song state’ functions as the dynamic center of predication. In the second event, hui has a function analogous to lai in (16). Therefore, to be more specific, the English translation for (b) should be ’In the summer, Earl of Shan attacked the Song state by joining [the Q i , Chen and Cao troops] ’ . 108

(17a) Qi Wei Cheng lai zhan yu Lang. (Huan 10)

NAME-NAME-NAME-come-fight-at-NAME ’The states of Qi, Wei and Zheng came to fight [with us] at Lang.’

(17b) Ju ren lai qiu lu. (Xi 1) BA###. NAME-people-come-demand-bribe ’The Ju people came to demand bribes.’

(17c) Guo-Ruo lai ben. (Cheng 18)

NAME-come-run into exile ’Guo-Ruo came to seek asylym.’

There are a large number of occurrences of lai serving as ADV of PRED. In (17a), lai modifies the VP zhan yu Lang ’to fight at Lang’; in (17b), it modifies oiu lu ’to demand bribes’; and

in (17c), it modifies ben ’to run into exile’.

This modification use of lai is parallel to the use of wang;

(18a) Chen-Hou zi bi-yi wang chao yu jun. (Wen 17)

NAME-from-my state-go-visit the court-to-king ’The marquis of Chen went from my humble state to pay respect to His Majesty.’

Unfortunately, this use of lai is not reflected in the English translations. If translated in accordance with the function of lai. (18a) should be ’The states of Q i , Wei and Zheng by coming fighted with us at Lang.’ This problem also occurs with the translations in (18b-c), (19) and (20). 109

(18b) Cao-Wen-Gong lai chao. (Wen 11)

NAME-come-visit the court ’Lord Wen of Cao came to pay respect to our Lord’.

The verb chao means ’to visit the court of a ruler to show

one’s respect’. The contrast of lai and wane indicates directional opposition in the action of chao. Lai indicates direction toward the speaker and wang indicates a direction away from the speaker. Moreover, the word chu has a similar directional effect on the predication:^®

(19a) Tai-Shu chu ben Gong. (Yin 1)

NAME-go out-run into exile-NAME ’Tai-Shu went out to seek asylum at Gong.’

(19b) Shi Gong-Zi-Ping chu ju yu Zheng. (Yin 3)

make-NAME-go out-live-in-NAME ’[Lord Mu of Song] made his son Ping to leave the state and live in the Zheng state.’

Chu implies that the agent leaves his own country. Therefore, the point of reference for the direction of chu is one’s own country rather than the speaker. Thus, chu in (19a) indicates a direction away from Tai-Shu’s home state Zheng and that in

(19b) indicates a direction away from Gong-Zi-Ping’s home state Song.

Although wang and chu bear a similar directional function, they differ in their subtle semantic contents. 110

If lai in (16) is accordingly treated as an adverb, the

sentence should have only two verbs gao ’to tell’ and fa ’to

attack’. Gao is the one with the absolute verbhood and thus

P of the sentence. Fa with its transitivity goal Song ’the

Song state’ serves as O of P.

3. VERBS AND THE SENTENCE-INITIAL POSITION. As explained in

2.2.2, deverbal entities are different from deverbal nouns in two aspects. First of all, deverbal nouns mostly occur in

idiomatic expressions, whereas deverbal entities result from a productive process. Secondly, deverbal nouns are analogous to typical nouns, but déverbal entities are in part still verbs.

Under such circumstances, how deverbal entities are allowed to associate with the nominal functions will be the question to pose. It should be argued that although deverbal entities preserve part of verbal capacity, the instantiation

is static relative to the dynamic center of predication, P.

When a deverbal entity is instantiated to correlate with a nominal function, it will denote a sense compatible with the

static phase of ambs. The static instantiations of ambs and

the static entities of verbs result in a set of ’abstract

nouns’ Most of them may probably remain to be contextual

As mentioned above, there are a small number of pure abstract nouns in Classical Chinese which originated from philosophical terminology. Most of the abstract nouns are derived from ambs and verbs. Some of them may be, more or less, lexicalized. But evidence shows that there is scarcely Ill

innovations, like sheng and si. in (14).

There is strong evidence to prove that deverbal entities

indeed preserve part of verbal capacities. Examples in Chapter

III display that when a typical noun acts as SUBJ, its PRED

can be equational, descriptive or narrative. But when a

deverbal entity acts as SUBJ, its PRED is most possibly

descriptive. For example:

(20a) Li ke bi hu? (Min 2) it ? ascend the throne-can-definite-PAR 'Can [your] ascending the throne be for sure?’

(20b) Qi qiu bu duo. (Zhao 22)

hi s-demand-NEG-much 'His demand was not much.'

Li and oiu in (20) are deverbal entities converted from the verbs l_i 'to ascend the throne' and qiu 'to demand/beg'. As expected, bi, 'to be for sure’ and duo 'much' are centers of descriptive predications.

Sometimes, deverbal entity SUBJ may be re-interpreted as a conditional clause. The marginality of interpretation on the

'verbs' in the sentence-initial position should imply that the property of deverbal entities differs from that of a typical

noun. For example:

a category of 'abstract nouns' in the language. (For more information, see Chapter VI.) 112

(21) Bel meng er qi da guo. Ci bi bai. Bei meng bu xiang, qi da guo bu yi. (Cheng 1)

abandon-covenant-CONN-deceive-large-state. it-for sure-defeat. abandon-covenant-NEG-omens, deceive- large-state-NEG-righteous ’(This military movement) is to abandon a covenant and to deceive a big state. It will fail for sure. Abandoning a covenant is a bad luck, and deceiving a big state is not a right thing to do.’

Compared with (20b), bei meng bu xiang ’to abandon a covenant is a bad luck’ in (21) can be understood as a deverbal entity

SUBJ bei mena ’to abandon a covenant’ followed by a PRED.a bu xiang ’to be a bad luck’. This reading fits the.observation on the general semantic connection between a deverbal entity and its predication. However, there is another possible reading for this clause.

As Hsueh (1990) points out, the subject of a sentence is frequently omitted in Classical Chinese, and yet the concept of subject remains essential in the grammar. The condition for a subject to be omitted is that it is retrievable from either discourse contexts or extralinguistic contexts. In order to read accurately, omitted subjects need to be recognized. From this aspect, the clause in question can be re-interpreted as two clauses in a cause-effect chain with an omitted subject.

Therefore, bei meng. bu xiang will be re-analyzed as ’if you abandon the covenant, you will have bad luck.’

Similarly, the clause gi da guo bu vi ’to deceive a big state is not a right thing to d o ’ with a deverbal entity SUBJ 113

followed by a descriptive PRED can alternatively be understood

as two clauses gi da guo. bu yi ’if you deceive a big state,

you are wicked’. In this case, the conditional sense is even

stronger.

To realize the possibilities of interpretation on a verb

in the sentence-initial position helps analyze the complex use

of shi ;

(22) Shi shu wei da? Shi qin wei da! Shou shu wei da? Shou shen wei da! (4A/20)

serve- what-be-great? serve-parents- be-great! protect-what-be-great? protect-body-be-great ’To serve, what is the most important? To serve parents is the most important! To protect, what is the most important? To protect your body is the most important!

The form shi has a verbal and a nominal alternants. The noun denotes ’affairs’, as in rone shi [ 1%$ ] ’military affairs’ and .ii shi [ ] ’sacrifice affairs’. The verb denotes ’to serve’. In (22), the first shi is a deverbal entity, meaning

’to serve/serving’. It seems that this deverbal use of shi will give rise to a question about the nature of polysemy, since by definition, the nominal alternant of a polysemy is naturally instantiated when the form occurs in the nominal

functions.

However, this deverbal use of shi is by no means in conflict with the definition of polysemy. In this particular case, shi is rather a sentential modifier than SUBJ. There is 114 a strong verbal sense revealed by the contexts. The following phrase shi gin 'to serve parents' indicates that shi is not purely nominal. The parallel shou 'to protect' and shou shen

'to protect your body' strengthens the feasibility of a verbal interpretation. Accordingly, the initial shi and shou should be properly understood as ADV, meaning 'about serving' and

'about protecting' respectively. CHAPTER V

DENOMINALIZATION

0. Denominalization means zero conversion of a noun to carry out syntactic functions atypical to its inherent property.

Denominalization drew less attention than nominalization did in the debate of the 50s on word categorization among Chinese scholars, because it is a relatively rare phenomenon in

Mandarin. However, it still is observable, as exemplified by

LÜ (1954b:17):

( 1 ) Wo shi he huang-jiu de. Ke ru-guo ni-men yi-ding yao bai-gan, wo ye ke-yi BAI-GAN yi-xia.

I-EMP-drink-yellow wine-EMP. but-if-you-certainly- want-1iquor, I-also-can-1iquor-once 'I usually drink yellow wine. But if you insist to drink bai-gan liquor, I will [DRINK] BAI-GAN liquor for this occasion.’

EMP indicates the shi...de pattern.

The unit from shi to de is a focus of emphasis. The speaker stresses that he usually drinks yellow wine. On this special occasion, however, because his friends wanted white wine, he could keep them company and drink white wine. The noun BAI-

GAN ’white liquor’ is used as the center of predication in wo

115 116

ve ke-yj bai-gan yi-xia ’I will [drink] bai-gan liquor for

this occasion’. Since bai-gan is a noun (bisyllable word) but carries out a verbal function, Lu thinks that it should be

treated as a temporary verb.

Despite Mandarin Chinese acceptance of innovate denominal verbs like bai-gan in (1), the language is unable to generate denominal verbs actively (cf. Tai 1983). On the other hand, denominalization is a far more productive process in Classical

Chinese. It deals with category shift in nouns which act as

ADV of a predication or as P/PRED.v. Denominals used as ADV are analogous to adverbs. Denominals used as P/PRED.v are well known as ’verbalization’, since denominals of this kind are analogous to verbs.

Denominalization is a process without overt marking. The evidence for its existence is nevertheless conspicuous, since

the denotations of denominals are the result of contextual

innovation. The context-dependent feature is prominent in both denominalization and nominalization. As mentioned in Chapter

IV, the restrictive forces of the main verbs and the contexts will critically determine the denotations of de-ambs and deverbal entities. Similarly, restrictive forces of some kind will be the factors which determine the innovative denotations

of denominals.

The restrictive forces which determine the denotations

of denominal verbs are special in the sense that they serve

as center of predication, P. P is the syntactic componant of 117 predominant power over other components. In nominalization,

P helps facilitate the recognition of a deverbal entity or a de-amb. In verbalization, a denominal verb will eventually possess the power to control other components. Therefore, it is impossible to use P as a clue to recognize category shift of nouns.

To identify denominal verbs, the concept of 'parent' vs.

'surface' used in Clark & Clark (1979) is essential, because the parent noun of a denominal verb takes part in the process of verbalization actively and because the syntactic/semantic cooperation of the parent noun and other surface arguments is key to a contextual meaning. Similarly, the cooperation between parent noun of a denominal adverb and other arguments is important, although P will function to ease recognition of a denominal adverb.

'THE INNOVATIVE DENOMINAL VERB CONVENTION' of Clark &

Clark suggests that the denotations of denominal verbs are determined by a pragmatic factor and a syntactic(/semantic ) factor.* The syntactic factor has to do with the cooperation of a parent noun and the other surface arguments, and the

* 'THE INNOVATIVE DENOMINAL VERB CONVENTION' proposed by Clark & Clark's (1979:787) is quoted as follows: 'In using an innovative denominal verb sincerely, the speaker means to denote (a) the kind of situation (b) that he has good reason to believe (c) that on this occasion the listener can readily compute (d) uniquely (e) on the basis of their mutual knowledge (f ) in such a way that the parent noun denotes one role in the situation, and the remaining surface arguments of the denominal verb denote other roles in the situation.' 118 pragmatic factor is concerned about general communication principles of cooperation between the speaker and the hearer

(also cf. Aronoff 1980).

In the present study, it will be hypothesized that the pragmatic factor which helps determine the innovative verbs is identical with the contextual factor which determine the innovative de-ambs and deverbal entities.% It will be assumed that the syntactic factor which has impact on denominal verbs has a similar impact on denominal adverbs.^

Since denominalization in Classical Chinese, as is the case in Modern English, does not involve overt morphological markings, it will be recognized primarily by semantic drifts.

In order that semantic drifts are unambiguously illustrated, the exemplifing denominals will be chosen from those which have typical nouns as their parent nouns. The guideline for choosing parent nouns is that the parent nouns denote concrete physical objects. At the same time, the parent nouns have to generate denominals which ’have a non-metaphorical concrete use as far as possible* (Clark & Clark p.768-9).*

^ Aronoff (1980) points out that the pragmatic factor in Clark & Clark’s CONVENTION should correspond to general rules of communication.

^ This assumption is based on the fact that parent nouns of denominal adverbs participate in the process actively, just as parent nouns of denominal verbs do.

* Clark & Clark establish four guidelines for denominal verbs of English. But it will be a problem to follow all the guidelines in the present study for the lack of inflectional morphology in Classical Chinese. 119

1, NOUNS AS ADV. A noun serving as a modifier of another noun is a universally natural function of the noun category.

But when a noun serves as a modifier of PRED.v, which by definition is ADV, it has to go through zero conversion and become a temporary adverb.

When a noun functions as a modifier of another noun, its denotation will be peripheralized to denote an attribute that the head noun possesses. For example, when the noun ren t A l

’person' is used as a modifier of xin [

The reason for not treating this peripheral use of nouns as a result of zero conversion is that the function of nominal modification is regularly carried out by nouns and ambs in the language.s At the same time, this function does not correspond

® It is noticeable that a verb can sometimes function as a nominal modifier. For example:

(a) Lai ren shi Zheng-Yu-Zi lu Su-Sha-Wei yi suo niu ma. (Xiang 2)

NAME-people-make-NAME-bribe-NAME-with-select-cows- houses ’The Lai people made Zheng-Yu-Zi bribe Su-Sha-Wei with selected cows and horses.’

Suo ’to select’ in (a) is used as an attrubute of the head NP niu ma ’cows and horses’, meaning ’selected’. Compare ren xin with suo niu m a . it is evident that ren functions as a nominal modifier more naturally than suo does. This can be attributed to inherent differences between nouns and verbs. 120 to any specific category in the lexicon.®

On the other hand, when a noun functions as a modifier of PRED.v, the concept of category shift has to be involved, as the modifier of PRED.v is by definition an adverb. More importantly, when a noun functions as a modifier of PRED, its inherent denotation alone is insufficient for generating a novel denotation. Under such circumstances, the factors proposed by Clark & Clark need to be taken into account to assign a denotation to the contextually innovated use.

Occurrences of denominal adverbs in Zuozhuan and Mencius are quite limited. In the following are some examples:

(2a) Guan-Qi zhe lie. (Xiang 22)

NAME-cart-split 'Guan-Qi was dismembered by cart-dragging.’

(2b) Shi ren li er ti. (Zhuang 8)

boar-person-stand-CONN-cry 'The boar stood like a man and cried.'

(2c) Si er fu ai, shi jiao zhi ye. Ai er bu jing, shou XU zhi ye. (7A/37) mezA. feed-CONN-NEG-love,pig-contact-him-PAR.love-CONN- NEG-respect, beast-raise-them-PAR 'To feed but not love [someone], it is to treat him as a pig. To love but not respect [someone], it is to raise him as a beast.'

® The 'nouns’ category is established in terms of nominal functions reference/SUBJ/0. The 'ambs' category is established for lexemes which carry out both nominal and verbal functions naturally (cf. 3.1 of Chapter II). 121

Che 'cart' is a typical noun denoting a concrete object. But in (2a), che indicates how Guan-Qi was dismembered.? In (2b), ren is to indicate the manner of how the pig stood up. The noun shi 'pig'in (2c) is used to indicate the way of treating a person. So is the noun shou 'beast'.

The semantic roles the parent nouns play in the above example differ in each case. Che is in the instrumental case, which indicates the instrument used in the dismemberment. Ren in (2b) is in the agentive case, which indicates the agent of the action JLi 'to stand' in the parent phrase. Shi and shou in (2c) açe in the objective case, which indicate the parent objects of iiao 'to contact' and 'to raise'.®

The understanding of innovated denotations relies on the understanding of parent nouns' semantic roles. For example, in (2b), shi 'pig' in the initial place has to be understood as agent of the actions l_i and ti 'to cry'. As a consequence, ren, the otherwise agent of the action ü , cannot but be understood as a modifier of ü . Ren li thus means 'to stand like a man'. In (2c), shi and shou are transitivity goals of

■iiao and jcu in the parent phrases. Since iiao and 2Ü1 take the pronoun zhi as goal, shi and shou cannot but be realized as

? To dismember a person by pulling carts into different directions was an unusual form of execution.

® The term of 'parent' is here adapted from Clark & Clark's discussions on the parent nouns of denominal verbs. 'Parent phrase' refers to the phrase that a parent noun is in before verbalized. More about 'parent' will be mentioned in Section 3. 122

pre-PRED modifiers. Shi iiao zhi thus means ’to treat him as

a pig’ and shou xu zhi means ’to raise him as a beast’.*

2. NOUNS AS PRED IN NON-EQUATIONAL SENTENCES. When nouns

function as PRED in non-equational sentences, they primarily

serve as PRED of narrative sentences. The opposition of a noun as PRED.n in an equational sentence and PRED.v in a narrative sentence is as follows:

(3a) Chu, Jiu-Ji shi, guo Ji. (Xi 33) m, a*. once, NAME-emissary, pass through-NAME’ ’Once Jiu-Ji was serving as an emissary and passed through the place of Ji.’

(3b) Jiu-Ji shi ye. a ? , NAME-emissary-PAR ’Jiu-Ji is an emissary.’

The expression in (3b) is made up to contrast with the one in

(3a). That shi ’emissary’ in (3b) is followed by the particle

ye indicates that the noun functions as PRED in an equational

sentence. In (3a), on the other hand, Jiu-Ji is understood as an agent of the denominal verb shi ’to act as an emissary’.

* Without recognizing that shi and shou are denominal adverbs in (2c) and that their parent nouns are actually in the objective case, shi and shou may mistakenly be understood as agents of iiao and 2£M respectively.

In the lexicon there are two words represented by the character [ ^ ]. The verb means ’to make sb. do sth.* and the noun means ’emissary’. The verb requires a transitive goal and an agent which is at the same time a causer. The use of shi here a denominal verb converted from the noun ’emissary’ 123

This agent-action analysis is supported by the agent-action relation between Jiu-Ji and the following action guo 'to pass through'.*!

2.1. NOUNS AS PRED IN DESCRIPTIVE SENTENCES. Liu (1958) observes that nouns can function as PRED in descriptive sentences when negated by However, this observation may raise questions. (4a-b) are two examples of Liu's (p.143):

(4a) Wui er zi zhe, Chu bu guo yi. (Ai 16)

without-two-persons-NOM, NAME-NEG-state-PAR 'Without these two men, the Chu state would never operate as a state any more.’

(4b) Jin-Ling-Gong bu jun. (Xuan 2)

NAME-NEG-king 'Lord Ling of Jin does not behave like a king.’

(4c) Fu fu, fu fu, suo wei shun ye. (Zhao 1)

husband-husband, wife-wife, so-called-proper-PAR 'A husband behaves like a husband, a wife behaves like a wife; [it is] so-called appropriateness.'

Liu thinks that bu guo in (4a) and bu iun in (4b) serve as

PRED of a descriptive sentence respectively. However, there

is transitivity relationship between SUBJ and PRED in these

sentences. As can be seen in (4b), Jin-Ling-Gong 'Lord Ling

rather than a typical use of the verb.

** The role of agent is as defined in Teng (1975). 124 of Jin* is agent of bu iun. Consequently, bu .iun should be better understood as the ruler ’did not behave like a king*.

Similarly, Chu acts as agent of bu guo in (4b), and thus it should be understood as the state ’did not operate as a state*. The transitivity relationship proves against the feasibility of the predications in (4a-b) as being descriptive.

Without the modification of the narrative nature is transparent. The expression fu fu. fu fu in (4c) is apparently narrative. Since the units in the PREDs are identical with those in the SUBJs, the expression would make little sense, unless actions are involved in the predications. Therefore, fu should mean * to behave like a husband* with the preceding fu ’husband* as an agent, and fu should mean * to behave like a wife* with the preceding fu ’wife* as an agent.

2.2. NOUNS AS P/PRED IN NARRATIVE SENTENCES. Denominal units in Classical Chinese are mostly denominal verbs. When being converted to be a transitive verb, it functions as P. If being converted to be an intransitive verb, it functions as PRED.v.

2.2.1. NOUNS AS PRED. In the surface formation, a denominal verb as PRED.v is sometimes indistinguishable from the PRED 125 of an existential sentence.Semantically, the former case is a noun used dynamically and the latter case is a noun used statically. Therefore, whether a noun is used statically or dynamically depends on the semantic relation between the noun and its preceding element. For example:

(5a) Yi guo san gong, wu shei shi cong? (Xi 5)

one-state-three-duke, I-who-go-follow ’There are three masters in one state and which one should I follow?’

(5b) Gong ji zhi di , di fen. Yu quan, quan bi. (Xi 4) ife«o 2 : % . lord-sacrifice-it-ground, ground-mound. give-dog, dog-die ’The lord sacrificed it (by pouring) to the ground, the ground bulged like a mound. [He] fed it to a dog, the dog died.’

(5c) Ji-Chou hui, gong gong huo. (Xi 24)

DATE-last day of the lunar month-duke palace-fire ’On the last day of the month, the Ji-Chou day, there was a fire in the duke’s palace.’

In (5a), the element before the nominal predication serves as

ADV to set up a spacial frame for the predication. Thus, vi guo ’onestate’ locates san gong ’three masters’.

In addition to the equational sentence pattern, nouns occur in the existential sentences as center of predication without undergoing a category shift. The normal formation of an existential sentence is a pre-PRED element setting up a temporal or spacial frame for the following predication. There is a common application of the existential pattern, viz. the first predication in an er construction. For details on the er construction, see Hsueh (1990). 126

But this existential relationship in (5a) does not hold

in di fen of (5b), because the phrase does not mean ’there is

a mound on the ground’. Fen ’to bulge like a mound’ is rather

a denominal verb, which indicates an outcome of the preceding

action: Immediately after the lord poured the poisonous wine

to the ground, it bulged like a mound. Di in this case serves

as patient of fen.T h i s transitivity relationship is

determined by the contextual denotation of fen, which in turn

is determined by its preceding and subsequent events. Without

the contexts, there is no way to exclude the other possible

semantic relationship between di. and fen, i.e., di sets up a

spacial frame for fen.

Whether a transitivity relationship holds in (5c) is hard

to discern. On the surface, it seems that gong gong ’the duke’s palace’ and huo ’fire’ maintains a semantic relation

similar to the relation in (5a). That is, gong gong sets up a spacial frame for huo. However, the existential sense in

(5c) is less explicit. It is actually possible that gong gong

serves as a patient of a state verb huo ’to be on fire’.

There is an interesting example to support this latter

analysis: Cheng bi bu huo [ ] ’the Cheng state will

definitely not suffer from a fire’ (Zhao 17). If the relation

between Cheng and huo is existential, the negative should be

The role of patient will be defined as an affected NP in the SUBJ slot. This differs from Teng’s (1975) definition (cf. Zhang & al. 1989:68). 127

wu huo [ ] rather than bu huo.

Moreover, as in (5b), the transitivity relationship of

patient-verb between the pre-PRED NP and the PRED critically determines the zero conversion in the inherent noun fen. It

is more common that the transitivity relationship is agent- verb;

(6a) Qi-Hou qin gu. (Cheng 2)

NAME-personally-drum ’The marquis of Qi beat the drum himself.’

(6b) Song cheng, Hua-Yuan wei zhi, xun-gong. (Xuan 2)

NAME-city, NAME-do-supervisor, inspect ’Song was building the city wall, Hua-Yuan served as the supervisor, and made a tour of inspection. ’

In (6a), gu ’drum’ is the parent noun of a converted verb,

indicating the action of ’beating a drum’, and Qi-Hou ’the marquis of Q i ’ is unambiguously the agent of the action. In

(6b), Song ’the Song state’ also serves as the agent of the denominal verb cheng ’to build the city wall’. However, it deserves noting that the agent-verb relationship in (6b) is

less explicit than it is in (6a). This is because Qi-Hou is animate and more likely to play an agent role.

Without overt marking, a two-noun cluster can possibly be a modifier-head NP or a SUBJ-PRED phrase. Thus, Song cheng

potentially acts as NP if context-free, meaning ’a city of

Song’ with Song being an attribute to the head noun cheng. It 128 is due to the following context which says that Hua-Yuan served as a supervisor and made a tour of inspection that cheng is recognized as a denominal unit. Consequently, Song is understood as an agent of cheng.

2.2.2. NOUNS AS P. Evidently, in a two-noun cluster that serves as PRED of a narrative sentence, the first noun should be recognized as a denominal verb, and the second, its object.

The problem is rather to verify its V-0 status in the first place. This task will be difficult if out of context. (6b) shows that a denominal verb may be misunderstood as a head of the NP without the context. A denominal verb which takes an

NP as a direct object will also need contexts to verify its verbal status. Otherwise, it may be misunderstood as a modifier of the following NP:

(7a) Hui-Gong yuan-fei Meng-Zi. (Yin 1)

NAME-first wife-NAME ’Lord Hui’s first wife Meng-Zi.’ 129

(7b) Song zu Di-Yi, Zheng zu Li-Wang. Yiu shang zu ye. (Wen 2)

NAME-ancestor-NAME, NAME-ancestor-NAME. just like- honor-ancestor-PAR ’The Song state regards Emperor Yi as its topmost ancestor; the Zheng state regards King Li as its topmost ancestor. [This is because the founding fathers of these states] regarded their fathers or grandfathers as the states’ ancestors.’i*

(7a) consists of three juxtaposed units, i.e., Hui-Gong ’Lord

Hui’, yuan-fei ’the first wife’ and Meng-Zi ’NAME of the first wife’. There are two possible analyses for this utterance. One analysis is that the three are NPs forming a larger NP to act as TOPIC of the following discourse. The other analysis is that this utterance is a narrative expression, with the unit yuan-fei Meng-Zi to be PRED. If so, it will have the denominal verb fei ’to wife’ as P, Meng-Zi as 0 and yuan ’first’ as ADV.

■Thus, this utterance will be understood as ’Duke Hui first took Meng-Zi as his wife’.

Since there is no other verbal use of fei in Zuozhuan. the second analysis is less feasible than the first one. Yuan- fen in the text is normally an NP with a fixed denotation of

’first wife’, which is analogous to er-fei [ 1 ’second wife’ (Zhao 8).^®

See the notes of Yang (1982a:524).

The verb for ’taking first wife’ is gu ’to marry’ as in gu yuan-fei yi feng zi-sheng [ ] ’to marry his first wife so as to offer rice and other kinds of grains to Gods’ (Wen 2). 130

Compared with (7a), Song zu Di-Yi in (7b) can be analyzed

as a larger NP formed by three juxtaposed NPs, i.e.. Song ’the

Song state’, zu ’ancestor’ and Di-Yi ’Emperor Y i ’. and thus

it will mean ’the Song state’s ancestor Di-Yi’. Judging from

shang zu ’to regard their fathers or grandfathers as the

states’ ancestors’ in the following context, however. Song zu

Di-Yi should be better understood as a narrative expression with the denominal verb su ’to regard as ancestors’ as P.^®

There is an interesting example in Zuozhuan. which displays that pragmatic knowledge is the essential determiner of denominal uses:

(8) San yue, Wu fa wo, Zi-Xie shuai, gu dao xian, cong Wu-Cheng. (Ai 8) EM, ikmm, march, NAME-attack-us, NAME-leader, on purpose-road- risky position, by-NAME 'In march, the Wu state attacked us, Zi-Xie was the leader who led the army to pass through the risky road on purpose and went by Wu-Cheng.’

Gu dao xian can only be understood to mean ’to pass through a risky road on purpose’, with dao ’to pass through a road’ a denominal verb. This reading is specified by the knowledge about the place Wu-Cheng.

This use of shang [_b] is analogous to shang in shang lun [(=)%] ’to trace and comment o n ’ and shang you ’to trace and make friends with’ (5B/8), with the first element to be ADV and the second element to be P (cf. Yang 1954:321- 2 ).

This reading is noted by Yang (1982a:1648). 131

There is otherwise a more natural reading for gu dao xian. with xian ’dangerous’ as PRED and gu dao ’old road’ as

SUBJ. This latter reading is in accordance with a phrasal principle of gu, which usually acts as a modifier of NPs such as gu gui [ ] ’old spirits’ (Wen 2) and gu chen [ i%S ]

’old minister’ (Ai 14).

The problem in the latter reading is that geographical features of Wu-Cheng prove against it. If gu dao xian means

’the old road was dangerous’, it must imply that the path near

Wu-Cheng is safer to pass through since the following context says that the troops went by way of Wu-Cheng. But the fact is that Wu-Cheng is a mountainous area in a strategic position.

Consequently, the reasons why the troops chose Wu-cheng as a route are all but safety, and gu dao xian has to be understood as ’the Wu army passed through the risky road on purpose’, with gu as ADV and dao xian as VP with dao as P.

3. CATEGORIES OF DENOMINAL VERBS. Clark & Clark’s denominal verb convention says in (f) that a parent noun will denote a semantic role which is in cooperation with the roles of other surface arguments to help achieve the communication purpose

(see quotation in Note 1). As shown in Clark & Clark’s data, a parent noun may end up representing different verbs due to the different roles it plays in different contexts.

For example, in the phrases ’tree the shoes’, ’tree the avenue’ and ’tree the possum’, ’tree’ is the shared parent 132 noun.i* But 'tree' in fact plays two different roles in these phrases. It is in the objective case in the first two phrases, while in the locative case in the last phrase. Thus, 'tree the shoes' means 'to locate trees in the shoes’, 'tree the avenue' means 'to locate trees along the avenue', and yet 'tree the possum' means 'to locate the possum in the tree'.

Therefore, it is a feasible strategy to present the semantic roles that denominal verbs can possibly play in

Classical Chinese for further analysis on denominal verbs.

Denominal verbs found in Zuozhuan. whose parent nouns denote palpable objects and which are denominally in a non- metaphorical concrete use, can be categorized into four sets by virtue of the semantic roles the parent nouns play.In this section, each of the categories will be illustrated by representative denominal uses from Zuozhuan.

3.1. IN THE AGENTIVE CASE. When a denominal verb has a parent noun in the agentive case, it should imply a referent which is identical with the surface agent. The verbs in (3a) and (4) are agent verbs. As is expected for (4b), Jin-Ling-

These examples are from Clark & Clark (1979, p.771 and p.772) respectively.

A metaphorical use of denominal verb found in Zuozhuan is gou 'hook' in yi gou Yue [ ] 'to make a connection with the Yue state' (Ai 25). In this case, the instrument gou has nothing to do with the action. It metaphorically indicates the notion of 'connecting' like hooks usually do to connect things. 133

Gong and .iun point to the same referent. Following are two more examples of agent verbs in (9):

(9a) Gong-zi ju, xiang fu er qiu. (Xi 23)

prince-fear, put off-garment-CONN-prisoner ’The prince got scarced, (he) took off his garment and made himself a prisoner.’^®

(9b) You Chu-Guo zhe, qi Qi-Ji hu! Jun Chen Cai, cheng wai shu yan. (Zhao 13)

have-NAME-NOM, indeed-NAME-PAR! king-NAME NAME, city-outside-be in charge-PAR ’The person who will own the state of Chu would be Qi-Ji! He governs the states of Chen and Cai like a king, and he is in charge of outskirts of the capital city [of Chu].’

The agent of qiu ’to act as a prisoner’ is gong-zi ’prince’

in (9a). The referent implied in the agent verb ’prisoner’ is

identical with the prince. In (9b), the surface agent is Qi-

Ji. Since the parent noun .iun ’king’ refers to Qi-Ji who governs the states Chen and Cai like a king, .iun ’to act as a king’ is an agent verb.

An agent verb may or may not take a transitive goal.^i

Given that a verb does not take a goal, as qiu in (9a), its goal of action should be implied in the verb. Then, the agent

This explanation on iiang fu follows Yang (1982a) which quotes Du-Yu’s [ ttffl 1 note.

The role of goal refers to the direct object of a transitive verb. Whether the object is affected or not will for the time being be omitted. The concern for this treatment of goal will be discussed in Section 4. 134 will at the same time be a patient. Consequently, the agent

verb is also a result verb. (For result verbs, see 3.4.3.)

3.2. IN THE INSTRUMENTAL CASE. Denominal verbs can have parent nouns in the instrumental case. For example:

(10a) Kong-Qiu yi gong tui, yue: Shi bing zhi! (Ding 10)

b : ± ^ z \ NAME-with-duke-retreat, say, warrior-weapon-them ’Kong-Qiu moved back with the duke, saying: You warriors, attack them with your weapons!’

(10b) Qi-Xuan-Gong wei fu-ren, shi guan shi. (Xuan 17)

NAME-curtain-woman, cause-watch-him ’Lord Xuan of Qi hid the woman behind a curtain, and let her watch him.’

The parent nouns bing ’weapon’ and wei ’curtain’ of (10) are instruments. In their denominal uses, both mean ’to impose the

■instrument upon somebody’. The motion is specified by a general function of the instrument and the agent’s intention of using the instrument, as explicitly indicated in the contexts. The weapons in (10a) were for the soldiers to attack the enemy, and the curtain in (10b) was used by the lord to cover up the woman (his mother).

Like the agent of agent verbs, the agent of an instrument verb needs not to co-occur with the verb. It should, however, be retrievable from the contexts, as in (11): 135

(11) Cong zuo you, jie zhou zhi, shi li yu hou. (Cheng 2 )

take up-left-right, all-elbow-him, cause-stand-at- back ’No matter [Qi-Wu-ZhangJ took up the left or the right side (of the chariot], [Han-Jue] stopped him by his elbows, and made him stand in the back.’

According to the preceding context of (11), Han-Jue dreamt in the previous night that his late father warned him not to take either the left or the right position in his chariot. Earlier

in the battle, Han’s two assistants who stood at his left and right sides had been shot dead. Ji-Wu-Zhang lost his chariot

in the battle and afterwards tried to get on Han’s chariot.

When Ji-Wu tried to stand at Han’s side, Han used his elbow to prevent him from taking the side positions but made him stand in the back.

It is natural for verbs to have a contextually implied agent.As shown in (11), no overt agent co-occurs with any of the four consecutive verbs cong ’to take u p ’, zhou ’to elbow’, shi ’to make’ and li. ’to stand’. However, two agents can be retrieved from the contexts. Ji-Wu-Zhang is the agent of the first and fourth actions cong and li. Han-Jue is the agent of the second and third actions zhou and shi.

When an instrument verb does not take a transitive goal, it means that the goal is the instrument itself. In this case.

In normal conditions, an agent of the first person, singular or plural, is omitted in Classical Chinese. 136 the instrument verb is also a target verb. (For target verbs, see 3.4.2. )

(12) Qi-Hou qin gu. (Cheng 2)

NAME-personally-drum ’The duke of Qi beat the drum himself.’

In (12), gu is an instrument verb with the instrument implied to be the action goal.^^

3.3. IN THE LOCATIVE CASE. With parent nouns in the locative case, location verbs naturally take a transitive goal which indicates the object to be located in locations implied in the verbs ;

(13a) Fan xing, yin zhi, she jue, ce xun yan. (Huan 2) jgft, ikm, return-movement, drink-arrive, locate-cup, strip- meritorious service-PAR ’[When the duke] returned from a military action, [he] drank for being home, put wine cups [for his followers], and recorded meritorious services on the bamboo strips.’

(13b) Gai guan Jin-Hou. (Xi 15)

change-dwelling house-Name ’[The earl of Qin] changed over and placed the marquis of Jin in a guest house.’

It is possible that gu has been conventionalized to be a lexical verb. In the VP of gu gin [& ^ ] ’to beat a stringed instrument’ (Xiang 24), for example, gu denotes the action of beating, while the goal is the stringed instrument gin. 137

In (13a), the verb ce ’to record on the bamboo strips’ takes

a goal xun ’meritorious service’, which is what is to be written on the strips. The semantic relation between the parent noun ’bamboo strips’ and the goal is evident.In

(13b), guan ’to place in a guest house’ takes Jin-Hou ’the marquis of Jin’ as its goal; the marquis is the object to be placed in the guest house.

On the other hand, location verbs may not take a goal object. Then, the action goal or result will be implied in the verb, as in (14):

(14a) Xi Zhao-Shuai yi hu-sun cong, jing, nei er fu shi. (Xi 25)

in former days-NAME-with-meals-follow, footpath, hungry-CONN-NEG-eat ’In former days, Zhaoshuai taking meals followed [Your Majesty], walked by footpath, and didn’t eat even when hungry.

(14b) Qian er si. (Zhao 29)

sunken hole-CONN-die ’[The horse] fell into a sunken hole and died.’

There is a similar use of ce to be a location verb in Zuozhuan. which is ce ming ] (Xi 23), meaning ’to record names on the bamboo strips’. It is difficult to tell to what extent ce is lexicalized to be a verb in the language, and yet it seems that the uses of ce xun and ce ming are idiomatic.

This punctuation follows Yang (1982a:436). Prof. Hsueh points out an alternative punctuation would be .ling nei er fu shi. meaning Zhao-Shuai ’became hungry on the road but did not eat’ with iing ’road’ as a denominal adverb modifies nei ’to be hungry’. 138

In (14a), the location .ling ’footpath’ is to indicate that

Zhao-Shuai placed himself as following his master, Lord Wen

of Jin, during the lord’s exile. While the lord walked along

the road, Zhao walked by footpath to show his respect. Thus,

the location verb iing means ’to walk by footpath’ in this context and Zhao is an agent.

In (14b), the parent noun oian ’sunken hole* denotes where a horse fell. Hsueh (1990) proposes that the connective er connects two predications. Accordingly, oian is recognized as a denominal verb functioning as PRED. But the denotation of this verb is ambiguous. In a sense, oian should mean ’to fall into a sunken hole’, indicating a resultative state which

is related to the horse’s death. If so, the horse is an affected object and thus a patient. On the other hand, there

is no reason for the verb not to hold an agent-verb relation with the horse. In this sense, the horse will be a volitional agent and oian should mean ’to jump into a sunken hole’

Since the co-occurring argument of iing. Zhao-Shuai. is an agent, the verb can alternatively be classified within the

target group (see 3.4.2). Similarly, oian can be classified

into the target group, if the horse is an agent. If a patient,

oian should alternantively be classified into the result group

(see 3.4.3). In contrast with (14), ce and guan in (13) have

Unfortunately, the vagueness of semantic relation between the denominal verb and its argument is inevitable. Even typical verbs may result in potential vagueness similar to this. For more comments on this problem, see Section 4. 139

to be recognized as location verbs for their co-occurring with an agent and a transitive goal simultaneously.

3.4. IN THE OBJECTIVE CASE. This pattern consists of verbs with parent nouns in the objective case. Its members can be subclassified into verbs of locatums, verbs of targets, and verbs of results, in terms of differences in the semantic roles of parent nouns,

3.4.1 LOCATUM VERBS. Parent nouns of locatum verbs denote an object which is to be placed in a location. In the parent phrase, while the locatum noun is in the objective case, the other two arguments are in the agentive and locative cases respectively. For example:

(15a) Cui-Shi die qi gong er shou zhi. (Xiang 27)

NAME-parapet-its-palace-CONN-defend-it ’The Cuei clan added parapets to their palace to defend it (the place).’

(15b) Ling-yin Dou-Qi, mo-ao Qu-Chong chu dao liang Zha. (Zhuang 4 )

magistrate-NAME, minister of war-NAME-open-road- bridge-NAME ’The magistrate Dou-Qi and the minister Qu-Chong opened up a road and constructed a bridge over the Zha river.’

In the parent phrase of (15a), Cui-Shi ’the Cui clan’ is in

The term ’locatum’ is used as in Clark & Clark (1979). 140 the agentive case, die ’parapets’ is in the objective case and oi gong ’their palace’ is in the locative case. In the surface phrase, the parapets are implied in the locatum verb die and qi gong becomes a transitive goal. Similarly, in (15b), the parent phrase has Dou-Qi and Qu-Chong in the agentive case, liang ’bridge’ in the objective case and Zha ’the Zha river’ in the locative case. In the surface, the locatum verb liang

’to construct a bridge’ implies the object to be located and

Zha ’the Zha river’ becomes a transitive goal.

3.4.2. TARGET VERBS. There are two kinds of target verbs.

In one kind of target verbs, ’target’ indicates the object offered to a recipient, as shown in (16):

(16a) Qi ren jiang qi zhi. (Huan 11)

NAME-person-will-wife-him ’The Qi people were about to give him a wife.’

(16b) Song bai Lao wo. (Ai 7)

NAME-hundred-Lao-we ’Song bribed us with one hundred Lao.’

As can be observed by the English translations, target verbs of this kind consistently imply the action of ’giving’. The parent phrase of (16a) has gi ’wife’ in the objective case,

Qi ren ’the Qi people’ in the agentive case, and zhi ’him’ in the benefactive case. In the surface phrase, gi. ’to give a wife’ becomes a target verb and zhi is a transitive goal. Qi 141 zhi thus means ’to give him a wife’. In (16b), the parent phrase has bai-Lao ’one hundred sets of an ox, a sheep and a pig’ in the objective case and wo ’us ’ in the benefactive case. Bai-Lao wo thus means ’to give us one hundred sets of an ox, a sheep and a pig’.

The other kind of ’target’ indicates the action goal.

Target verbs of this kind imply the action goals in the verbal denotations. Since the concept of action is involved, they are expected to co-occur with an agent. For example:

(17a) Yu-shi hu sou yu Pi-Lu. (Xi 27)

therefore-PAR-spring hunting-at-NAME ’Therefore, [the Jin troops] went spring-hunting (as a military training) at Pi-Lu.’

(17b) Guan yu Zi-Lu-Shi. (Zhao 22)

dwelling house-at-NAME ’[King Jing] stayed in the Zi-Lu clan’s house.’

Although the agents of sou ’to go spring-hunting’ in

(17a) and guan ’to stay in a dwelling house’ in (17b) are omitted, they are retrievable from the contexts. Due to the existence of agency, sou and guan have to be identified as denominal verbs, and to be more specific, goal verbs, since they do not take a transitive goal.

Lao has at least two meanings. It is a label for stables like cattles, sheep, and pigs. Sometimes the term is used for a collection consisting of one ox, one sheep, and one pig. The latter is the meaning for Lao in (16b). 142

That the goal of action is implied in the verb results in the content of the action being critically determined by the goal itself. In (17a), the parent noun sou means ’spring- hunting’ and the verb means ’to go spring-hunting’. In (17b), the parent noun guan means ’dwelling house’ and the verb means

’to stay in a dwelling house’. Target verbs of this kind rely heavily on the inherent denotations of the parent nouns. In other words, the contexts have less impact on denotations of target verbs than on those of other types.

3.4.3. RESULT VERBS. As opposed to one-argument denominal verbs which require an agent, viz. the goal verbs, there are other one-argument denominal verbs which require a patient.

These are result verbs. Result verbs imply some resultative states about the patients. Fen in (5c) is a result verb and nue in (18) is another:

(18) Xu-Dao-Gong nue. (Zhao 19)

NAME-malaria ’Lord Dao of Xu suffered from malaria.’

Xu-Dao-Gong ’Lord Dao of X u ’ is an affected NP, a patient, of the verb nue ’to suffer from malaria’. The sense of suffering is mostly determined by the state of ’having malaria’.

4. SOME REMARKS ON THE CATEGORIES. The categorization of

Classical Chinese denominal verbs displays how a parent noun 143 may cooperate with the arguments controlled by the verb to bring up a contextual meaning. But there are important points somehow overlooked in the above analyses.

In every category, there are verbs which have only one dependent argument, as exemplified in (9a), (12a), (14a-b),

(17a-b) and (18). It seems a disadvantage for these verbs to have only one argument to help specify its innovative meaning.

In fact, innovated meanings for the one-argument verbs are as predictable as meanings of two-argument verbs, because one- argument verbs automatically imply action goals of agents or resultative states of patients in the verbs.

It is a strong tendency that denominal verbs of the language act as one-argument verbs. As can be observed in

Zuozhuan. there are a large number of denominal verbs derived from ceremonial dresses, ceremonial activities, military activities, health conditions, and etc. to act as one-argument verbs.For example:

Since the cooperation of arguments in the parent phrase is only one factor for the determination of innovated denotations, the parent nouns of the one-argument verbs will never predetermine the associated actions.

The categories of nouns which are allowed to be used denominally should be language-specific. It is interesting that weather terms in Classical Chinese are in strong dynamic senses. For example, ’snowing’, ’hailstorming’ and ’raining’ are expressed as yu xue [ ] (Huan 8), vu pao [ ] (Xi 10) and yu lin [ ] (Yin 9), following the verb yu [ i# ] rather than being verbalized like many terms of ceremonial activities. 144

(19a) Yan-Ping-Zhong duan-wei li yu Hu-Men zhi wai. (Zhao 10)31

NAME-ceremonial dress-stand-at-NAME-GENI-outside ’Yan-Ping-Zhong wore his ceremonial dress and stood outside the Hu gate.’

(19b) Qi zhong Wan yu Ji-Shi. (Zhao 25)

its-many-NAME-at-NAME ’The majority (of the dancers) danced Wan at the Ji clan’s temple.’

(19c) Chu-Wu-Wang Jing-Shi. (Zhuang 4)

NAME-NAME ’King Wu of Chu deployed soldiers to practice the Jing-Shi formation.’

(19d) Qi-Hou jie, sui dian. (Zhao 20)

NAME-skin disease, at last-malarial fever ’The marquis of Qi suffered from a skin disease, and then a malarial fever.’

In (19a-c), the verbs are target verbs with one argument, viz. duan-wei ’to wear ceremonial dress’ in (a). Wan ’to dance the

Wan dance’ in (b), and Jing-Shi ’to practice the Jing-Shi

formation’ in (c). To a large extent, the associated dynamic parts of the verbs are determined by their parent noun. The verbs in (19d) are result verbs, viz. .iie ’to suffer a skin disease’ and dian ’to suffer from a malarial fever’. It seems

that the semantic role of pre-verbal argument, an agent or a

31 Yang (1982b) notes that nouns of dressing can naturally function as PRED in a nominal sentence. From our view, it represents a special use of target verbs. 145 patient, is closely related to the nature of the parent noun.

In marginal cases between target verbs and result verbs, the senses of result or action may be vague, and have to be determined by the contexts. For example:

(20) Qin shi sui dong. (Xi 32)

NAME-troops-consequently-east ’Then, the Qin army headed for the east.’

The English translation shows that Qin shi ’the Qin army’ is treated as an agent and dong. a target verb. This treatment is in accordance with the following context which indicates that the army was on its way moving eastward. However, it is undeniable that in certain situations, dong can indicate a resultative state that the army has completed its journey to the east.32

Moreover, it is evident that in addition to result verbs, verbs of other categories may also imply a resultative state in the verb. Association of a resultative state and an agent usually suggests a sense of causativity.33 The following are some examples:

32 This potential vagueness exists in events reported by inherent verbs. It may have to do with a more general process than verbalization.

33 ’Causative’ here is used as in Wang (1958), Dobson (1959), Li & Thompson (1976) and Cikoski (1978). 146

(21a) Gu wang chen gong, gong chen dai-fu. (Zhao 7)

therefore-king-servant-duke, duke-servant-minister ’Therefore, kings made dukes their servants, and dukes made ministers their servants.’

(21b) Cheng-Zhou Xuan-Xie huo, ren huo zhi ye. (Xuan 16) X ‘kZiko NAME-NAME-fire, person-fire-it-PAR ’There was a fire at the palace of Xuan-Xie in Cheng-Zhou; someone set it on fire.’

Chen in (21a) is a special case of target verbs. Target verbs of this kind normally take one dependent argument, because the action goal has already been implied in the verbs (cf. (17) in 3.4.2).* In the parent phrases of (21a), the first chen and gong refer to an identical referent, i.e., ’dukes’ and the second chen and da-fu refer to an identical referent, i.e.,

’ministers’. In the surface, the verb chen ’to treat as a servent’ takes gong and da-fu respectively, as goals. The denotation of the parent noun, now part of the denotation of the denominal verbs, indicates a resultative state about the goals. Therefore, a causative sense is involved in the verb chen.

In (21b), huo ’to set on fire’ includes a result, since the first phrase has indicated the existence of a fire. At the same time, since ren ’someone’ contributes a strong sense of agency, the causativity relation is obvious.

As for two-argument verbs, the pre-verbal argument is usually an agent and the post-verbal argument, a transitive 147 goal. The cooperation of a verb and different goals may end up having different dynamic senses, as shown in the following exmaples for the verb ming;

(22a) Shan-Yue sheng zi, jiang dai shi er ming zhi. (Ding 8)

NAME-give birth to-child, about to-wait for-event- CONN-name-him ’Zhan-Yue got a son, and was waiting for an event to name the child after.*

(22b) Ming qi er qing, zun Jun ye. (Cheng 9)

name-its-two-off icial, respect-king-PAR ’[The prisoner] mentioned two of his state’s high- ranking officials by their names; [it is that] he showed his respect to his king.’

In (22), each ming is a target verb, derived from the same parent noun ’name’. But the actions involved are different.

In (22a), ming means ’to give a name’, while in (22b), it means ’to mention a name’. The associated action is equally natural for the parent noun. It is because of the difference in goals that makes the actions differ. For a new-born child, ming has to denote ’to give a name’. For officials, the most likely meaning will be ’to mention the names’.3*

Finally, it should be noted that the cooperation of the surface arguments and the parent noun are insufficient for the the determination of novel denotations from time to time. This is especially true when the goal of action is the pronoun zhi.

34 This can be supported by the contexts. 148

The following examples on iiao is quite representative;

(23a) Pi-ru pu lu, Jin ren jiao zhi, zhu Rong ji zhi. (Xiang 14)

take as an instance-catch-deer, NAME person-horn- it, these-NAME-drag-it ’(It’s] like catching a deer, the Jin people held its horns, and those Rong people dragged it (from the back). ’

(23b) J in ren zhu zhi, zuo you jiao zhi. (Xuan 12) ^A'Æz, tamz. NAME-people-pursue-it, left-right-horn-it ’The Jin troops pursued them (the Chu troops), and the left and the right divisions chased them from two corners.’

The formation of Jin ren iiao zhi of (23a) and zuo you iiao zhi of (23b), and given the knowledge that the pre-verbal element is probably an agent, it will still be difficult to figure out the meanings of j iao. Only when the referents of the pronouns are retrieved from the contexts can the category of the verb be identified.

In (23a), zhi replaces lu ’deer’, which can be specified by pu lu ’to catch a deer’. Therefore, the parent noun of jiao means ’deer’s horns’. In addition, the co-occurring actions of Bu ’to catch’ and j_i ’to drag’ also have an impact on the denotation of the verb jiao. Because of the contexts, it is understood that jiao zhi means ’to hold the deer’s horns’ and thus j iao is a target verb.

In (23b), zhi replaces the Chu troops. It is possible that the parent noun of j iao ’horns’ has little to do with the 149 transitive goal. At the same time, the pre-verbal unit zuo you

'the left and the right (divisions)’ is likely to resemble the horns. Therefore, iiao should be identified as a metaphorical denominal use of the parent noun.

5. THE PRAGMATIC FACTOR. In Clark & Clark’s convention, the pragmatic factor of ’contextuals’ is concerned about general communication principles. The above examples of denominal verbs demonstrate clearly that the verbs tend to use the most salient properties of the parent nouns. In other words, the mutual knowledge of the speaker and the listener necessary for computing the innovated denotations usually remains at a generic level of abstraction. The dynamic part of the contextual verbs is highly associated with what the parent noun is for or about.

Due to this characteristic of verbalization, contextual innovations which do not make use of the core knowledge about the parent nouns especially need contexts to be assigned a novel meaning. In fact, this kind of innovations are extremely rare in Zuozhuan and the use of j_i ’halberd’ in (24) is a good example :

(24) Gong ji qi shou. (Ai 25)

duke-halberd-his-hand ’The duke made his hands the shape of halberd.’ 150

Although the parent noun denotes an instrument, it is not the instrument that the duke is holding in his hand, neither does it indicate that the duke made use of the instrument to do something to his hands or to someone’s hands. The innovative meaning in fact has little to do with the function about the instrument. The shared knowledge about the instrument is the shape, i.e., the duke held his hands in the shape of a halberd.

The reason why j_i tends to be identified as an instrument verb rather than a target verb as it actually is is that the shape of the weapon is beyond ordinary expections on denominal innovations. Therefore, its denotation is relatively difficult to decode.

Moreover, the pragmatic factor is critically needed to distinguish the agent verbs from a type of target verbs. For example :

(25a) Shun chen Yao. (Wen 18)

NAME-servant-NAME ’Shun served as Yao’s servant.’

(25b) Dong-Guo-Yan chen Cui-Wu-Zi. (Xiang 25)

NAME-servant-NAME ’Dong-Guo-Yan made Cui-Wu-Zi his servant.’

It should be difficult for modern readers to figure out what the gesture of ’halberb’s shape’ looks like. Based on Yang (1982a:1724), the duke was putting his hands on both sides of his waist. 151

Both chen come from the same parent noun ’servant*. In (25a), chen is an agent verb, meaning ’to serve as a servant’. The referent of the verb is identical with the surface agent Shun.

In (25b), chen is a result verb, meaning ’to treat sb. as a servant’ and the referent of the verb is identical with the surface argument Cui-Wu-Zi. In the surface formation, it is impossible to detect the difference between the two verbal uses of chen. These interpretations need to be built upon knowledge about the relative social status of Shun and Yao. and Dong-Guo-Yan and Cui-Wu-Zi.

The same is true for the interpretation of a very unique example in Zuozhuan;

(26) Er yu Wu-Wang wo hu? (Ding 10)

you-want-NAME-I-PAR ’Do you want to make me "the king of Wu"?’

Wu-Wang ’to treat as a king of W u ’ in this case is a result verb. Its parent noun refers to a referent identical with the first person wo. Moreover, Wu-Wang does not mean ’any king of

W u ’. It rather has a fixed referent whose name is Liao [ ].

King Liao was trapped and killed by his servant. In a similar situation, when the speaker found that his disloyal servant was trying to trap and kill him, he referred to the king, and used the king, i.e., Wu-Wang. in a verbal sense. 152

6. MORE EXAMPLES ABOUT TARGET VERBS. Through the above

analyses, it seems clear that the range of semantic drifts

from nouns to denominal verbs is quite limited in Classical

Chinese, compared with the semantic drifts in Modern English

presented in Clark & Clark (1979). Moreover, it seems a

general tendency that when a noun is verbalized, the outcome

is a verb with the object the noun refers to as target of a

context-dependent action. Thus, it may be true that target

verbs are the norm of denominal verbs.

The following are more instances about how different

target verbs are generated from the same parent nouns. Based

on the data from Zuozhuan. it is likely that cheng has become

a polysemy of the verb 'to build (city) walls’ and the noun

'(city) walls'. But this does not hinder contextual innovation

of denominal verbs;

(27a) Bi-Bo shuai shi cheng Lang. (Yin 1) grngmmimAP. NAME-lead-troops-city wall-NAME 'Earl of Bi led troops to build the city walls of Lang.'

(27b) Xia, cheng Zhong-Qiu. (Yin 7)

summer, city wall-NAME 'In the summer, (we) built the city wall of Zhong- Qiu. '

This co-incides with the fact that the Mandarin example of bai-gan in (1) is also a target verb. 153

(27c) Cheng qi xi-bei er shou zhi. (Ding 10)

city wall-its-northwest-CONN-guard-it ’(The duke of Wei) attacked the city’s northwest, [took] and guarded it (the corner).’

In the uses of (27a-b), cheng means ’to build the city wall’.

But the verb cheng in (27c) is contextually innovative, which means ’to attack the city’.3? The sense of attacking must be created through the preceding verb fa ’to attack*.

Cheng and guo ’outer (city) walls’ are a pair of nouns indicating inner and outer areas of a city. While cheng had obviously become a polysemy, quo probably did not. But in one occasion, quo is found in a verbal use analogous to the verbal use of cheng;

(28) Chu-Zi shi Wei-She cheng Zhou-Qu. ..Shi Xong-Xiang- Mei guo Chao, Ji-Ran guo Juan. (Zhao 25)

NAME-make-NAME-city wall-NAME. ...make-NAME-city wall-NAME, NAME-city wall-NAME ’The viscount of Chu ordered Wei-She to build walls for Zhou-Qu. ... [He] made Xong-Xiang-Mei build walls for Chao and Ji-Ran build walls for Juan.’

The following set of examples with the parent noun men

’gate’ will illustrate how easily a noun can be converted to be target verbs:

See the note of Yang’s (1982a:1579) 154

(29a) Yi liang shi men zhi, zhong mo gan Jin. (Ai 4)

by-two-arrow-gate-it, people-NEG-dare-enter *[Gong-Sun-Yan] guarded the gate by two arrows; none dared to enter.’

(29b) Gong-Sun-Gan men yan. (Ai 15)

NAME-gate-PAR *Gong-Sun-Gan guarded the gate.’

(29c) Wu-Zi men yan. (Xiang 25)

NAME-gate-PAR ’The prince of Wu entered the gate.’

(29d) Jiang men. (Ai 27) MM. will-gate ’[The Jin army] is going to attack the gate.’

(29e) Men qi san-men. (Xiang 9)

gate-its-three gate ’[The allies’ army] attacked its (Zheng’s) three gates.’

Men in (29a-b) means ’to guard a gate’. In (29c), it means

’to enter a gate’. In (29d-e), it means ’to attack gate(s)’.

All of the verbs are target verbs. In each case, the parent noun contributes its core property to all the verbs. These verbs differ in the actions toward the gate. According to the data, the action of attacking is the one of high frequency in occurrences. Therefore, it relies on the contexts least for its meaning compared with the other actions. 155

Interestingly, denominal verbs which refer to the same action can be used transitively or intransitively, as shown

in (29d) and (29e).^® Although 'transitivity' is not a major concern here, further investigation on this difference will hopefully lead to better understanding on denominal verbs of

Classical Chinese.

This difference may be interpreted in terms of the theory of 'transitivity' proposed by Hopper & Thompson (1980). CHAPTER VI

CONCLUSION

0. The present study begins with some compositional principles of the Chinese language in conjunction with some premises which are deduced from general knowledge about natural languages. Its purpose is to establish a foundation for the analysis of word classes in Classical Chinese. To conclude the study, a summary is provided here, and some related issues will be briefly discussed as well.

1. SUMMARY. By two phrasal constructions and three types of

PREDs, three classes of full words, viz. nouns, ambs and verbs are postulated. The unmarked/natural correlations between nouns and their nominal functions in reference/SUBJ/0 are presented in Chapter III. Chapter IV shows how verbs which play the nominal role are decategorized by nominalization without overt markers. Chapter V displays how nouns are verbalized to carry out the verbal functions, viz. nucleus/P/

PRED. Chapters IV and V also display how contextual factors operate in category shift.

156 157

1.1. THE FULL-WORD CATEGORIES. The distinction of nouns and verbs is based on two predominant phrasal constructions, viz. the [head-object] VP and the [modifier-head] NP.^ The [head- object] VP construction is the most crucial and useful for the distinction. Full words which consistently serve as head of this VP are by definition verbs. Among non-verbs, those which consistently serve as head of the [modifier-head] NP are nouns. The category of ’ambs’ first conceived by Kennedy is recognized as within the fuzzy boundary between nouns and verbs, and closer to nouns. The reason for ’ambs’ belonging to a category independent of nouns is that ambs naturally function as PRED.a. In contrast, nouns naturally function as

PRED.n and verbs naturally function as PRED.v.

1.2. CORRELATIONS BETWEEN LEXICAL AND FUNCTIONAL CATEGORIES.

Croft’s theory of natural correlations predicts that typical nouns naturally correlate with the nominal functions, while typical verbs naturally correlate with the verbal functions.

Based on the data from Zuozhuan and Mencius. nouns which are identified by the [modifier-head] NP naturally corrspond to reference/SUBJ/0 in all the major sentence types, as the theory predicts. Unnaturalness in the associations of nouns and P/PRED.V results in semantic drift.

^ Since there is no morphological marker for the head, the concept of ’profile’ from Langacker’s semantic frame is adopted as an essential criterion for the phrasal head (cf. 1.3.1 of Chapter II). 158

Moreover, unnaturalness in the associations of verbs identified by the [head-object] VP and the nominal functions has two consequences. In the first place, when verbs (or VPs) play the nominal role, semantic drifts will co-occur. In the second place, when verbs (or VPs) serve as SUBJ, the condition is highly restrictive and ambiguity will sometimes come into existence.

1.3. UNNATURAL CORRELATIONS AND SEMANTIC EVIDENCE. The markedness of unnatural correlations is summarized as follows:

reference nucleus SUBJ/0 P/PRED

NOUNS unmarked marked AMBS unmarked (unmarked) VERBS marked (unmarked)^

Figure 6.1 Markedness of Unnatural Correlations

The ’marked’ correlations in Figure 6.1 indicates unnatural correlations. When a typical noun unnaturally correlates with P/PRED.v, there will be a meaning change from its static notion to a contextually actuated dynamic notion.

The semantic change indicates that the noun goes through a zero conversion and becomes a denominal verb. The zero

^ In Figure 6.1, the correlations in the parentheses are assumed to be natural/unmarked by virtue of Croft’s theory. But because this study does not take these correlations into account, no discussion about them will be provided. 159

conversion from nouns into verbs can be generalized as:

(1) N — > V / (SUBJ) (ADV) ___ (0)

Rule (1) says that when a noun occurs in the P/PRED slot, it

has been converted into a verb and is thus able to serve as

a verb.

When a typical verb unnaturally correlates with SUBJ/0,

the meaning change will begin from the inherent dynamic notion

to a contextually actuated static notion. The semantic change

indicates that the verb goes through a decategorization process and becomes a deverbal entity. The decategorization

from verbs to deverbal entities can be generalized as:

(2) V --- > non—V / P // ___ PRED (Note: the pre-PRED unit f ADV)

Rule ( 2 ) says that a verb has been decategorized when it acts

as 0, or when it occurs before PRED and does not function aS

an adverbial element, i.e., when it functions as SUBJ.^

1.4. CONTEXT-DEPENDENT INNOVATIONS. When a noun is converted

into a verb, as specified in rule (1), its verbal meaning will

be formed in a predictable manner: an additional dynamic

situation associated with the parent noun will be referred to.

^ According to Kennedy, ambs have nominal and verbal capacities. When an amb naturally correlates with SUBJ/0, it will refer to an abstract entity through instantiating the static phase. 160

Exactly what the dynamic situation is referring to depends on

uiiC# % uxs^uuxac; ^ ^ —- — — — ^ uvMueaua.

When a verb is decategorized into a deverbal entity, as rule (2) states, it dynamic imagery is demoted because of the way it is conceptualized. The degrees of abstractness of deverbal entities from the same parent verb may differ from context to context.®

The context-dependent category shift is not restricted to unnatural correlations of nouns and the verbal functions.

Ambs are sometimes converted into nouns to refer to objects.

When a describing amb is converted into a noun, it will have a referent retrievable from the context to be associated with the quality it denotes. This may be considered as a shorthand expression generated by a pragmatic process.®

1.5. SIGNIFICANCE OF THE CATEGORIZATION. As explained in

Chapter II, full-word categorization has long been an issue since Mashi Wentong. For those who believe that full words should be able to be sub-classified in terms of functional categories, their task is always to provide a sound theory to set up regular correlations between lexical and functional categories on the one hand, and to level off inconsistency

* See examples in Chapter V.

® See examples in Section 2 of Chapter IV, esp. (14).

® Seeexamples in 1.2.2 of Chapter IV. 161

revealed by irregular or unexpected correlations on the other

hand. Txiia is exactly the task this present study is focused

on.

Chapter III shows that nouns can play the roles of SUBJ/0 naturally in all three major sentence types. Chapter V shows

that when nouns function as P/PRED.v, semantic drift can be predicted. The theory of markedness properly accounts for the occurrences of irregular correlations with semantic drifts, and claims that the former are marked by the latter. Because of the markedness of semantic drifts, zero conversion of nouns into verbs can be consistently realized. As displayed in

Chapter VI, verbs which carry out the nominal functions and shift into deverbal entities are also marked by predictable semantic drifts.

Therefore, the basic distinction of nouns and verbs can be maintained by the theory of natural correlations, and the

irregularities can be sorted out by the theory of markedness.

This categorization proves that full words are categorizable even without overt morphological marking.

2. RELATED ISSUES. During the course of discussion from

Chapter III to Chapter V, there are a number of interesting

issues which have not been elaborated because of the limitation of the present study. Five of them will be briefly discussed here: 1) semantic features in the SF categories, 2)

the relation between lexicon and syntax, 3) the pre-PRED 162

position, 4) abstract nouns, and 5) denominal verbs in

Classical Chinese.

2.1. SEMANTIC FEATURES OF THE SF CATEGORIES. It seems

evident that SF categories in Classical Chinese impose

semantic characteristics on lexemes. Thus, a full word which carries out a syntactic function will be restricted by the

semantic force of the function. This view on the SF categories has made some scholars believe that full words in the langauge share basically the same lexical status. It is when a lexeme

is associated with a certain SF category that the lexeme will be recognized as a member of a certain category. Even scholars who acknowledge the existence of lexical categories suggest

that temporary syntactic functions should override more permanent lexical category of a lexeme.?

For those scholars who believe that full words share

basically the same lexical status, they face difficulty in

explaining why certain kinds of words tend to correlate with

certain functions and why some words rarely correlate with

certain functions. For those who acknowledge the existence of

lexical categories, assuming a direct impact of syntactic

functions on changes in lexical meaning is not completely

sufficient. Given that there is no lexical process involved,

it will be difficult to see how lexemes are able to correlate

^ This is exactly the basic idea of Wang’s theory of ’temporary functions’ (cf. 2.2 and 2.3.2 of Chapter II). 163 with functions which do not fit their lexical properties in the first place.

Following the prototype theory of categorization, it is proposed that natural correlations between nouns and the nominal functions on one hand, and between verbs and the verbal functions on the other hand, define essential distinctiveness of nouns and verbs. It is argued that unnatural correlations between verbs and the nominal functions and those between nouns and the verbal functions result from lexical processes without morphological markers. The lexical processes cause nouns to become denominal verbs and verbs to become deverbal entities respectively so as to play the roles atypical of their lexical properties.

From this point of view, it is not syntactic functions that override lexical features and result in semantic drifts.

It is rather lexical processes that entail semantic drifts to make lexemes legitimate to play atypical roles. The unmarked lexical processes are systematic aspects in the lexicon which overlap with certain syntactic aspects. (For details about overlapping of lexicon and syntax, see 2.2.)

The semantic features imposed by the SF categories should be considered as a compensation for the lack of morphological trappings in Classical Chinese. For a language whose lexical expansion processes are marked by morphological trappings, information regarding meaning change in lexical roots is explicitly provided by the trappings. In this situation, it 164

is redundant to have a set of SF categories assigned to the

lexical root to yield similar effect. For a language without

formal trappings of this kind, it is then reasonable to have

meaning change triggered and characterized by an unnatural

correlation between certain lexical property and functional

property.

2.2. LEXICON AND SYNTAX. As pointed out in Chapter II, the present study assumes that the lexicon has two components. One component of the lexicon consists of an open-ended list of primitive conceptions under an infinite number of lexemes with

ideosyncratic semantic and syntactic information. The other component contains systematic lexical rules which convert

lexical items from one lexical category into another.

The above analyses illustrate that the most prominent

systematic aspects in the lexicon of Classical Chinese are semantic drifts between static and dynamic phases. The process of nominalisation renders lexemes of static nature to be associated with dynamic senses. Verbalization, on the other hand, impels lexemes of dynamic nature to associate with static senses.

Particulars on semantic drifts of individual lexemes are contextually related. Certain subtle semantic oppositions marked by morphological trappings in Modern English, e.g., explore vs. exploration, humane vs. humaneness, and alive vs.

life, are conceived through contextual factors in Classical 165

Chinese. The contextual factors may compensate for the simple and direct relationship between lexicon and syntax. Assumedly, contextual factors in the langauge play a more important role than they usually do in languages with more overt syntactic and lexical categories. This assumption has been supported by some findings in the analyses. But it needs to be verified by more detailed studies on the subject.

2.3. THE PRE-PRED POSITION. In 3.2.3 of Chapter III is explained that a time word in the pre-PRED position can be a relative word functioning as ADV or a relative clause serving as a subordinate clause, depending on the temporal frame it sets up. In the former case, the time word is converted into an adverb and indicates duration of the event in the following predication. In the latter case, the time word remains a noun and indicates a period of time that has passed before the event of the following predication occurs. It serves as a nominal PRED (or PRED.n), which is analogous to the nominal

PRED before the connective er (see (24-25) of Chapter III,

Note 12 of Chapter V, also cf. Hsueh 1990).

Moreover, it is suggested in the last part of Chapter IV that deverbal entities are potentially ambiguous when they occupy the pre-PRED position. As is explained in 2.2.2 of the same chapter, deverbal entities are partially verbs. They are recognized as decategorized, because their verbhood is relatively weak compared with the verb in the P position. It 166

is noteworthy, however, that only when a verb or a VP

functions as SUBJ/0 will it become a deverbal entity. If a verb in the pre-PRED position is not SUBJ, it will not be decategorized. In this case, the verb serves as PRED of a clause subordinated to the main clause and it has absolute verbhood in its own clause. Sometimes, there is a conditional relation in between a PRED and its antecedent element, as in

(21) of Chapter IV. This relation may suggest that the antecedent element may be a clause rather than SUBJ.

The possibility of reinterpreting the units in the pre-

PRED position as relative words or PREDs implies a necessity to set up a criterion for SUBJ. Further investigation on the semantic relationships between PREDs and the pre-PRED units may be helpful for this task.

2.4. ABSTRACT NOUNS. In Classical Chinese, there are a very small number of inherent abstract nouns which originated from philosophical terms such as dao ’the way of nature’ and de

’virtue’. Moreover, as mentioned in Chapter Four, Wang and

Liu assume that abstract nouns in the langauge are usually derived from verbs and adjectives.® It seems that ambs, which are instantiated by the static phase, and decategorized verbs, which denote abstract entities, may probably be the ’abstract nouns’ that Wang and Liu have in mind.

® For the difference between adjectives and ambs, see 3.1.1 of Chapter II. 167

The difference between an inherent abstract noun like de

and a labelling amb such as li. and zui is marginal. As shown

in 1.2.1 of Chapter IV, ^ and ü undergo the same adjunct process to function as PRED.a. Harbsmeier’s example about the word ren (a descriptive amb) suggests that the language never

lexically distinguishes 'humaneness' from 'humane' (cf. Note

5 of Chapter IV). Therefore, it is difficult to recognize

'abstract nouns’ as a sub-group of the noun category.

2.5. DENOMINAL VERBS IN CLASSICAL CHINESE. Clark & Clark

(1979:804-5) postulate six stages of historical .change for

idiomatization of denominal verbs.® As revealed in Chapter V,

genuinely innovative denominal verbs are quite rare in

Classical Chinese, and Wu-Wang in (26) is one of the few

examples. Opaque idioms are not found in the texts.

Most of the denominal verbs in the language seem to be

assimilated transparent idioms, Clark & Clark's stage four.

Cheng in (27a), for example, is ready to be computed as a

target verb, meaning 'to build city walls' (also cf. (27b) and

(6b)). The same is true of verbs in other categories such as

gu in (6), bing in (10a), ce in (13a), ai in (16a) and sou in

(17a). Presumably, this results from historical development

of the denominal verbs.

® The six stages of idiomatization are 1 ) complete innovations, 2) near-innovations, 3) half-assimilated transparent idioms, 4) assimilated transparent idioms, 5) partly specialized idioms, and 6) opaque idioms. 168

Moreover, compared with the range of drifts in Modern

English, the range of semantic drifts for denominal verbs in

Classical Chinese is quite limited. For example, denominal verbs with parent nouns of human roles are limited to either agent verbs or target verbs. A parent noun like guan may be used as a target verb, as in (17b) or a location verb, as in

(13b). As proposed in the last section of Chapter V, target verbs are the norm of verbalization in the language. It is the general tendency to form a verb with the object that the parent noun refers to as target of a context-dependent action

(cf. (29a-e)). Therefore, when cheng is used as an innovation occasionally, ’city walls’ are readily interpreted as target of a certain action determined by its context (cf. (27c)). REFERENCES

Anderson, S. R. "Typological Distinctions in Word Formation," Language Typology and Syntactic Description III, ed. by T. Shopen, 3-56. Cambridge: University Press.

Andrews, A. 1985. "The Major Functions of the Noun Phrases," Language Typology and Syntactic Description I . ed. by T. Shopen, 62-154. Cambridge: University Press.

Aronoff, M. 1976. Word Formation in Generative Grammar. Cambridge: MIT Press.

. 1980, "Contextual," Language 56.744-58.

Bodman, N. C. 1950. "Karlgren’s ’The Chinese Language’ (a review)," Language 26.339-46.

Chafe, W. 1976. "Givenness, Contrastiveness, Definiteness, Subjects and Topics," Subject and Topic, ed. by Charles Li, 25-55. New York: Academic Press.

Chan, N. P. 1984. The Nouny Alternative; Crossing the Verb/ Noun Borderline in Modern Chinese. Ph.D Dissertation. University of California, Berkeley.

Chao, Y. R. 1968a. Yuyan Wenti (The Problems of Languages). Taipei: Taiwan Shangwu Yinshuguan.

_____ • 1968b. A Grammar of Spoken Chinese. Berkeley: University of California Press.

C h ’ien, Mu. 1956. Xiangin Zhuzi Xinian (Chronological Studies of the Pre-Chin Philosophers). Hong Kong: University Press.

Chomsky, N. 1965. Aspects of the Theory of Syntax. Cambridge: MIT Press.

• 1970. "Remarks on Nominalisation," Readings in English Transformational Grammar, ed. by R. A. Jacobs and P. S. Rosenbaum, 184-221. Waltham: Ginn and Company.

169 Chou, F. K. 1953. "Yuyin Qubîe Cilei Shuo (On Tonal Variations which Differentiate Word Classes)," Zhongguo Yuvanxue Lunwen.ii. 1968, 349-64. Hong Kong: Chongji Shudian.

. 1961a. Zhongguo Gudai Yufa: Zao.iu Bian. BIHP, Monograph Series 39.

. 1961b. Zhongguo Gudai Yufa: Gouci Bian. BIHP, Monograph Series 39.

Cikoski, J. 1978. "An Outline Sketch of Word-classes and Sentence Structure in Classical Chinese," CAAAL 8.17-152.

Clark, E., and H. Clark. 1979. "When Nouns Surface as Verbs," Language 55.767-811.

Comrie, B., and S. Thompson. 1985. "Lexical Nominalisation," Language Typology and Syntactic Description III, ed. by T. Shopen, 349-98. Cambridge: University Press.

Croft, W. 1984. "Semantic and Pragmatic correlates to Syntactic Categories," Papers from the Parasession on Lexical Semantics, ed. by D. Testen, V. Mishra, and J. Drogo, 53-70. Chicago Linguistic Society.

. 1990. Typology and Universels. Cambridge: University Press.

Dawson, R. 1968. An Introduction to Classical Chinese. Oxford: University Press.

Dobson, W. A. C. H. 1959. Late Archaic Chinese. Toronto: University Press.

. 1974. A Dictionary of the Chinese Particles. Toronto: University Press.

Downer, G. B. 1959. "Derivation by Tone-Change in Classical Chinese," BSOAS 22.258-90.

Engelkamp, J. 1988. "Nouns and Verbs in the Mental Lexicon," Understanding the Lexicon: Meaning, sense, and world Knowledge in Lexical Semantics, ed. by W. Hullen and R. Schulze, 303-13. Tubingen: Niemeyer.

Fillmore, C. 1968. "The Case for Case," Universals in Linguistic Theory, ed. by E. Bach and R. Harms, 1-88. New York: Holt, Rinehart and Winston.

170 Fraser, B. 1970. "Some Remarks on the Action Nominalisation in English," Readings in English Transformational Grammar, ed. by R. A. Jacobs and P. S. Rosenbaum, 83-98. Waltham: Ginn and Company.

Gao, Mingkai. 1953. "Guanyu Hanyu de Cilei Fenbie (On Word Classification in Chinese)," ZGYW 1953.10.13-6.

. 1954. "Zailun Hanyu de Cilei Fenbie (On Word Classification in Chinese, The Sequel)," ZGYW 1954.8.13- 7.

. 1955. "Lun Hanyu Yufa de Lishi Jichengxing (On the Historical Heritage of Chinese Grammar)," BJDXXB 1955.1.167-83.

Guo, Xiliang. 1990. "Guanyu Xici 'Shi* Chansheng Shidai he Laiyuan Lunzheng de Ji-dian Renshi (On the Origion of the Linking Verb ’Shi’ in Chinese," Wangli Xiansheng Jinian Lunwen.ii. 222-39. Peking: Shangwu Yinshuguan.

Halliday, M. A. K. 1967a. "Notes on Transitivity and Theme in English: Part I," Journal of Linguistics 3.37-81.

. 1967b. "Notes on Transitivity and Theme in English: Part II," Journal of Linguistics 3.199-244.

. 1968. "Notes on Transitivity and Theme in English: Part III," Journal of Linguistics 4.179-215.

Harbsmeier, C. 1985. "Where Do Classical Chinese Nouns Come From?" Early China 9-10.76-126.

Him, K. T. 1977. "Derivation by tone change in Cantonese: A Preliminary Survey," JCL 5.186-210.

Honey, D. B. 1989. "The Word Behind the Graph in Classical Chinese: Three Notes on the Logographic Writing System," JCLTA 24.3.15-26.

Hopper, P., and S. Thompson. 1980. "Transitivity in Grammar and Discourse," Language 56. 251-299.

, _____ . 1984. "The Discourse Basis for Lexical Categories in Universal Grammar," Language 60.703-52.

Hsueh, F . S. 1989. "The Structural Meaning of the Ba and the Bei Constructions in Mandarin Chinese: Do They Really Mean Disposal and Passive?" Functionalism and Chinese Grammar, ed. by J. Tai and F. Hsueh, 1-32. Chinese Language Teachers Association, Monograph Series 1.

171 ______. 1990. "On the Conjunction Er and the Formation of Compound Sentences in Classical Chinese," Proceedings of the International Seminar on Chinese Language and its Teaching in the World. Chinese Language Society of Singapore, 124-30. (Also in YYYJ. 1991, No. 1.)

Jacobson, W., Jr. 1979. "Noun and Verb in Nootkan," Heritage Record 4, 83-155. Victoria: British Columbia Provincial Museum.

Jespersen, 0. 1942. A Modern English Grammar: Part VI. Morphology. Copenhagen: J. Jorgensen & Co.

Karlgren, B. 1933. "Word Families in Chinese," BMFEA 5.9-120.

______. 1949. The Chinese Language: An Essay on its Nature and History. The Ronald Press.

. 1956. "Cognate Words in the Chinese Phonetic Series," BMFEA 28.1-18.

. 1960. "Tones in Archaic Chinese," BMFEA 32.113-42.

Kennedy, G. 1964. "Word-Classes in Classical Chinese," Selected Works of George A. Kennedy. 323-433. New Haven: Far Eastern Publications.

Langacker, R. 1987. "Nouns and Verbs," Language 63.53-94.

______. 1988. "A View of Linguistic Semantics," Topics in Cognitive Linguistics, ed. by B. Rudzka-Ostyn, 49-90. Amsterdam: John Benjamins.

Legge, J. 1960. The Chinese Classics II: The Works of Mencius. Hong Kong: University Press.

Lehmann, C. 1990. "Towards Lexical Typology," Studies in Typology and Diachrony, ed. by W. Croft, K. Denning and S. Kemmer, 161-85. Amsterdam: John Benjamins.

Li, C. , and S. Thompson. 1976a. "Subject and Topic: A New Typology of Language," Subject and Topic, ed. by C. Li, 457-89. New York: Academic Press.

, ______. 1976b. "Development of the Causative in Mandarin Chinese: Interaction of Diachronic Processes in Syntax," Syntax and Semantics V I . ed. by M. Shibatani, 477-92. New York: Academic Press.

, ______. 1981. Mandarin Chinese: A Functional Reference Grammar. Berkeley: University of California Press.

172 Li, Jinxi. 1957. Xinzhu Guoyu Wenfa (A New Mandarin Grammar). Shanghai: Shangwu Yinshuguan.

Li, Jinxi, and Shiru Liu. 1960. "Yufa Zai Taolun: Cilei Qufen he Mingci Wenti (An Over-Again Discussion on Grammar: Word Classification and the Issue of Nouns)," ZGYW 1960.5-8.

Li, Peiyuan, and Shiguang Lu. 1980. "Tan Ci de Xu-shi (A Discussion of Functional Words vs. Full Words)," LTS 1980.3.30-5.

Li, Yiping. 1983. "Fuci Xiushi Mingci huo Mingcixing Chengfen de Gongneng (The Function of Adverbs as Modifiers of Nouns or NPs)," LTS 1983.3.40-5.

Lin, Yutang. 1982. Chinese-Enalish Dictionary of Modern Usage. Hong Kong: The Chinese University.

Liu, Jingnong. 1958. Hanvu Wenvan Yufa (A Grammar of Classical Chinese," Peking.

Lu, Jianming. 1983. "’De ’-zi Jiegou he 'Suo’-zi Jiegou (The Phrasal Structures of ’De’ and ’Suo’ , " Yufa Yan.iiu he Tansuo. 57-68. Peking University Press.

LÜ, Shuxiang. 1942. Zhongguo Wenfa Yaolue (A Summary of Chinese Grammar)," Hong Kong: Shangwu Yinshuguan.

. 1954a. "Guanyu Hanyu Cilei de Yi-xie Yuanze-xing Wenti, I (Some Principles of Word Classification in Chinese, Part I)," ZGYW 27.6-14.

. 1954b. "Guanyu Hanyu Cilei de Yi-xie Yuanze-xing Wenti, II (Some Principles of Word Classification in Chinese, Part I D , " ZGYW 28.16-22.

Lyons, J. 1966. "Towards a Notional Theory of the ’Parts of Speech’," Journal of Linguistics 2.209-36.

. 1968. Introduction to Theoretical Linguistics. Cambridge: University Press.

Ma, Jianzhong. 1961. Mashi Wentong Jiaozhu (A Proofread Edition of the M a ’s Grammar), 2 Vols. Peking: Zhonghua Shuju.

Matthews, P. H. 1974. Morphology. Cambridge: University Press.

McCawley, J. D. 1982. "The Non-existence of Syntactic Categories." Thirty Million Theories of Grammar. 176-203. London: Croom Helm.

173 Mei, Zulin. 1980. "Sisheng Bieyi-zhong de Shijian Cengci (The Layers of Tonal Derivation)," ZGYW 1980.427-43.

Mukhin, A. M. 1985. "Lexical and Syntactic Semantics in Historical Aspect," Historical Semantics. Historical Word-Formation. ed. by J. Fisiak. Berlin: Mouton.

Norman, J. 1988. Chinese. Cambridge: University Press.

Robins, R. H . 1952. "Noun and Verb in Universal Grammar," Language 28.289-98.

. 1966. "The Development of the Word Class System of the European Grammatical Tradition," Foundations of Language 2.3-19.

Rosch, E. H. 1973. "Natural Categories," Cognitive Psvchologv 4.328-50.

. 1977. "Human Categorization." Studies in Cross-Cultural Psychology, ed. by N. Warren, 1-49. New York: Academic Press.

_____ . 1973. "Principles of Categorization," Cognition and Categorization, ed. by E. Rosch and B. Lloyd, 27-48. Hillsdale, N.J.: Erlbaum.

Sanders, G. 1988. "Zero Derivation and the Overt Analogue Criterion," Theoretical Morphology, ed. by M. Hammond and M. Noonan, 155-75. San Diego: Academic Press.

Sapir, E. and M. Swadesh. 1946. "American Indian Grammatical Categories," Word 2.107-12.

Schachter, P. 1985. "Parts-of-Speech Systems," Language Typology and Syntactic Description I . ed. by T. Shopen, 3-61. Cambridge: Uniyersity Press.

Shi, Zhenye. 1960. "Shilun Hanyu Dongci, Xingrongci de Mingci- hua (On Nominalization of Verbs and Adjectives in Chinese)," ZGYW 1960.422-5.

Smith, C. 1989. "Event Types in Mandarin," Proceedings of the Third Ohio State University Conference on Chinese Linguistics (May. 1988). ed. by M. Chan and T. Ernst, 215-43. Bloomington: Indiana U. Linguistics Club.

Tai, J. 1982. "Releyant Categorical Distinctions in Chinese," Papers from the Eighteenth Regional Meeting of Chicago Linguistic Society. 495-506.

174 ♦ 1983. "On the Absence of Denominal Verbs in Chinese," Paper Presented at 1983 Annual Meeting of Linguistic Society of America, Minneapolis, Minnesota.

. 1984. "Verbs and Times in Chinese: Vendler’s four categories," Papers from the Parasession on Lexical Semantics. ed. by D. Testen, V. Mishra, and Joseph Drogo, 287-96. Chicago: Chicago Linguistic Society.

. 1985. "Temporal Sequence and Chinese Word Order," Iconicity in Syntax, ed. by J. Hainan, 49-73. Amsterdam: John Benjamins.

. 1989. "Towards a Cognition-Based Functional Grammar of Chinese," Functionalism and Chinese Grammar, ed. by J. Tai and F. Hsueh, 187-226. Chinese Language Teachers Association, Monograph Series 1.

Tai, J. and L. Wang. 1990. "A Semantic Study of the Classifier 'Tiao'," JCLTA 25.1.35-56.

Teng, S. H. 1975. A Semantic Study of Transitivity Relations in Chinese. Berkeley: University of California Press.

Tsao, F. F. 1979. A Functional Study of Topic in Chinese: The First Step Towards Discourse Analysis. Taipei: Student Book Co.

Vendler, Z. 1984. "Agency and Causation," Midwest Studies in Philosophy 9.371-84.

Wang, Li. 1920. Zhongguo Guwen Fa (A Grammar of Classical Chinese). Wang Li Wenji 3, 1984, 1-85. Jinan: Shangdong Jiaoyu Chuban She.

. 1955a. "Guanyu Hanyu Youwu Cilei de Wenti (On the Presence or Absence of Word Classes in Chinese)," BJDXXB 1955.2.125-47.

. 1955b. Zhongguo Yufa Lilun (A Theory of Chinese Grammar). Shanghai: Zhonghua Shuju.

. 1958. Hanyu Shigao (A History of Chinese). Peking: Kexue Chuban She.

. 1959. "Hanyu Shici de Cide Fenlei (Classification on Full Words in Chinese)," BJDXXB 1959.2.53-67.

. 1984. "'Zhi', 'Qi' Goucheng de Mingci-xing Cizu (The NPs Marked by 'Zhi' and 'Qi')," YYYJ 7.1-9.

175 . 1990. "Hanyu Yufa Shi (A History of Chinese Grammar)," Wangli Wenii 11. 1-488. Jinan: Shandong Jiaoyu Chuban She.

Wang, Lianqing. 1991. A Semantic Study of the Chinese Classifier System. Ph.D. Dissertation. The Ohio State University.

Xing, Fuyi. 1989. "Cilei Panbie Si-yaodian (Four Points on Word Classification)," LTS 1989.3.60-8.

Yang, Bojun. 1956. "Cong Shanggu Hanyu Ji-zu Tongyici de Kaocha Shitan zai Cihui Fangmian Gujin Fenhe Xianxiang de Gueilu (A Study of Lexical Rules on Some Sets of Synonomous Words in Archaic Chinese)," BJDXXB 1956.2.81- 103.

_____ . 1982a. Chuenoiu Zuozhuan Zhu (A Commentary of Chuenqiu Zuozhuan)," Taipei: Yuanliu Chuban She.

. 1982b. "Gu Hanyu-zhong zhi Hanjian Yufa Xianxiang (Some Seldom Seen Phenomena in Classical Chinese)," ZGYW 171.401-9.

Yang, Shuda. 1920. Gaodeng Guowenfa (Advanced Chinese Grammar). Shanghai: Shangwu Yinshu Guan.

. 1954. Ci Quan (A Dictionary of Words). Peking: Zhonghua Shuju.

Yen, Sian. 1986. "The Origin of the Copula ’Shi’ in Chinese," JCL 14.2.227-42.

Zhang, Lili, Liping Zhang, Ruizhu Huang, Yaxia Zheng, Wenzhen Wei, and Quanshu Zha. 1989. Guovu de Cilei Fenxi (Word Classification in Mandarin Chinese). Taipei: Computing Center, Academia Sinica.

Zhou, Zumo. 1946. "Sisheng Bieyi Shili (Notes on Tonal Derivation)," Wenxue Ji. 1966, 81-119. Peking: Zhonghua Shuju.

Zhu, Dexi. 1983. "Zizhi he Zhuanzhi (Self-Designation and Transfer-Designation)," Fangvan 1983.16-31.

. 1985. Yufa Dawen (Questions and Answers in Chinese Grammar). Peking: Shangwu Yinshu Guan.

. 1988. "Guanyu Xianqin Hanyu-li Mingci-de Dongci-xing Wenti (On the ’Verbal Properties’ of Nouns in the pre- Qin Chinese)," ZGYW 203.81-6.

176 . 1990. "Guanyu Xianqin Hanyu Mingci he Dongci de Qufan de Yi-ze Zhaji (A Note on the Nouns/Verbs Distinction in the pre-Qin Chinese)," Wangli Xiansheng Jinian Lunwen.i i. 162-71. Peking; Shangwu Yinshu Guan.

Zhu, Dexi, Jiawen Lu, and Zhen Ma. 1961. "Guanyu Dongci, Xingrongci ’Mingwu-hua’ de Wenti (On ’Nominalization’ of Verbs and Adjectives)," BJDXXB 1961.4.51-64.

177