Newest Trends in the Study of Grammaticalization and Lexicalization in Chinese 9783110253009, 9783110252996

Citation preview

Newest Trends in the Study of Grammaticalization and Lexicalization in Chinese

Trends in Linguistics Studies and Monographs 236

Editor

Volker Gast Founding Editor

Werner Winter Editorial Board

Walter Bisang Hans Henrich Hock Heiko Narrog Matthias Schlesewsky Niina Ning Zhang Editor responsible for this volume

Volker Gast

De Gruyter Mouton

Newest Trends in the Study of Grammaticalization and Lexicalization in Chinese Edited by

Janet Zhiqun Xing

De Gruyter Mouton

ISBN 978-3-11-025299-6 e-ISBN 978-3-11-025300-9 ISSN 1861-4302 Library of Congress Cataloging-in-Publication Data A CIP catalog record for this book has been applied for at the Library of Congress. Bibliographic information published by the Deutsche Nationalbibliothek The Deutsche Nationalbibliothek lists this publication in the Deutsche Nationalbibliografie; detailed bibliographic data are available in the Internet at http://dnb.dnb.de. ” 2012 Walter de Gruyter GmbH & Co. KG, Berlin/Boston Typesetting: RoyalStandard, Hong Kong Printing: Hubert & Co. GmbH & Co. KG, Göttingen ⬁ Printed on acid-free paper Printed in Germany. www.degruyter.com

Table of contents Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Janet Zhiqun Xing

1

The development of the Chinese aspectual sentence-final marker yeˇ . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Qianrui Chen

21

The emergence of a definite article in Beijing Mandarin: The evolution of the proximal demonstrative zhe` . . . . . . . . . . . . . . Mei Fang

55

The grammaticalization of the directional verb ‘la´i’: A construction grammar approach . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Cheng-hui Liu

87

The degree-evaluative construction: Grammaticalization in constructionalization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Mei-Chun Liu and Chun Chang

115

The semantic historical development of modal verbs of volition in Chinese. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Alain Peyraube and Ming Li

149

Semantic change in the grammaticalization of classifiers in Mandarin Chinese . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Janet Zhiqun Xing

169

The repeater in Chinese and other languages. . . . . . . . . . . . . . . . . . Cheng Zhang

215

Lexicalization in the history of the Chinese language. . . . . . . . . . . . Xiufang Dong

235

Argument structure change, reanalysis and lexicalization: Grammaticalization of transitive verbs into ditransitive verbs in Chinese, Japanese and English . . . . . . . . . . . . . . . . . . . . . . . . . . . . Feng-fu Tsao Subject index . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

275 303

Introduction1 Janet Zhiqun Xing

In the last three decades, the study of grammaticalization and lexicalization has generated tremendous interest among Western researchers who study Indo-European and African languages. As a result, more than two dozen monographs and edited volumes have been produced (e.g. Heine and Reh 1984, Heine et al. 1991, Traugott and Heine 1991, Hopper and Traugott 2003 [1993], Heine 1993, Bybee et al. 1994, Lehmann 1995, Ramat and Hopper 1998, Heine and Kuteva 2002, Wischer and Diewald 2002, Traugott and Dasher 2002, Bisang et al. 2004, Fischer et al. 2004, Brinton and Traugott 2005, Echardt 2006, Lopez-Couso and Seoane 2008, Good 2008). Most of these works study grammaticalization on the premise that it is a unidirectional process or change whereby a lexical item or a construction has undergone a change and consequently serves a grammatical function. The major issues raised by those studies are related, but not limited, to the sources/origins, motivations, mechanisms, pathways, and targets/outcomes of grammaticalization. Many of these studies focus on the morpho-syntactic process of change, for instance, whether a lexical item has become cliticized, a‰xized, and then fossilized (e.g. Heine and Reh 1984, Heine et al. 1991, Lehmann 1995). Many others probe the patterns of semantic change that accompany morpho-syntactic change (e.g. Traugott and Dasher 2002, Visconti 2004, Echardt 2006). Nonetheless, what most of these studies have in common is that they are based on data from either Indo-European or African languages. Naturally, the patterns or tendencies 1. This collection of articles has been made possible in part by funding provided by the O‰ce of Research and Sponsored Programs, Western Washington University. On behalf of all the contributors, I would like to thank Walter Bisang and Sandra Thompson for their enthusiastic support and encouragement of the initial proposal for this project. We are all very grateful to the anonymous reviewer for his/her meticulous review of the entire manuscript and constructive suggestions for revision. Our gratitude also goes to Randi Hacker and the readers from Mouton for commenting on and proofreading all the chapters in the volume and to Birgit Sievert for her support throughout the process of this project. Without their help, it would have been impossible to complete this project.

2

Janet Zhiqun Xing

of semantic change, grammaticalization, and lexicalization generated from those studies reflect specific characteristics of the relevant languages, because even though the languages of these families (e.g. African languages vs. European Languages) may share some linguistic features, each language or language family has its own unique history and evolutionary process. Taking African languages as an example, we know that Heine and his associates’ earlier studies on grammaticalization were mostly based on the internal reconstruction of the languages in this family because there was no recorded history of their development. In comparison, English does not have the gap or lack of evolutionary history that African languages exhibit. However, throughout its history, it has been heavily influenced by other European languages. As a result it is di‰cult to determine whether certain changes in English are natural or coerced (e.g. via language contact, cf. Heine and Kuteva 2008). For instance, one of the major shifts in English occurred during the transition from Old English to Middle English when the language lost most of its agreement markers (e.g. case), which led English to change from an inflectional language to a more analytical one. Presumably, such a change would a¤ect the pathways and mechanisms of semantic change and grammaticalization. This type of unique history of a language or language family undoubtedly contributes to the typological characteristics of its later stages, like Middle English and Modern English. Therefore, we cannot say with any degree of certainty that those patterns and tendencies that are derived from English or African languages apply to other genetically unrelated languages, such as Chinese or Burmese, without studying grammaticalization in these languages in depth. Chinese, on the other hand, is clearly a language that has hardly been studied by western researchers with regard to grammaticalization and lexicalization. In the past three decades, we have seen only a few dissertations (e.g. Sun 1996, Xu 2006) published for English-speaking communities but no edited volume on this topic.2 The reason for this is probably twofold: 2. As far as I know, there are currently only two edited volumes on the market related to some of the issues discussed in this collection: Chinese Grammar: Synchronic and Diachronic Perspectives by Hilary Chappell (Oxford University Press 2001) and Space in Languages of China: Cross-Linguistic, Synchronic and Diachronic Perspectives by Dan Xu (Springer Science 2008). However, Chappell’s collection focuses on dialectal variations and Xu’s collection primarily discusses issues related to the syntactic aspects of spatial terms in Chinese. They do not overlap with the theme of our volume, namely, the characteristics of Chinese in the three areas of diachronic change: semantic change, lexicalization, and grammaticalization.

Introduction

3

Western researchers are not familiar with the Chinese language and their Chinese counterparts are not familiar with the theoretical framework developed in the West. Consequently, neither group feels comfortable in tackling such issues, even though both groups are aware that Chinese is typologically di¤erent from Indo-European languages, and moreover exhibits a rich and uninterrupted body of historical data (i.e. recorded history of more than 3000 years without major typological shifts such as those having occurred in English.) Because of the facts mentioned above, Chinese has a clear advantage over Indo-European languages and African languages when it comes to the study of grammaticalization. The first example of a study on grammaticalization in Chinese that attracted Western linguists’ attention is perhaps Li and Thompson’s (1974) study on the change of word order involving baˇ, a lexical item which underwent a change from a full-fledged transitive verb to an object marker in a serial verb construction (i.e. NP þ baˇ þ NP þ V). This study not only provides convincing evidence for syntagmatic change in grammaticalization, but also sheds light on the typological characteristics of syntagmatic change in Chinese. That is, in a language with serial verb constructions where verbs are not marked for tense, number, case, etc. verbs may become grammaticalized into function words more easily than they can in languages with agreement marking (i.e. tense, number, case, etc.). This assumption, as shown in the following section, is supported by the pattern of semantic change in grammaticalization observed in Chinese.

1. Regularity in semantic change According to studies conducted by Western researchers (Heine et al. 1991: 74 and Traugott and Dasher 2002: 11–12), semantic change triggered by metaphoricalization and metonymization in grammaticalization develops along the following cline: A > A,B > (B), where A stands for an early lexical semantic function from which a new meaning B – a polysemy (not necessarily a new grammatical category) – has been derived and coexists with the earlier meaning A. Over time, meaning A may gradually become obsolete leaving only the newly developed polysemy B in use, and later the polysemy B may become obsolete as well. Such a tendency for semantic evolution seems applicable to most cases of grammaticalization in English and other inflectional languages (Heine et al. 1991, Traugott and Heine 1991, Bybee et al. 1994, Heine and Kutiva 2002, Good 2008). In Chinese, however, study after study shows a di¤erent pattern (e.g. Liu 1989; Ma

4

Janet Zhiqun Xing

1993, 2002; Peyraube 1989a, 1989b, 1989c, 1992, 1994a, 1994b, 1996, 1998; Shi Z. 1989; Shi and Li 2001; Sun 1996; Wang 1980 [1956]; Xing 2003, 2009; Xu 1992; Zhang 1991; etc.). It has been observed that semantic change in the grammaticalization of Chinese lexemes undergoes a process of ‘accretion’ of meanings, i.e. A > A,B > A, B, C, in which the multiplication from a conceptual entity A to A, B, C first goes through an intermediate stage (A,B) where the older meaning (A) and the newer meaning (B) co-exist. Then the older meaning (A) or the newer meaning (B) may continue to develop and extend their meaning to (C), a new conceptual entity. As a result, all of the three entities may co-exist. This cline,3 or rather this tendency of semantic change in Chinese, is clearly di¤erent from the one reported by Western researchers, i.e. A > A,B > (B). The key difference between the two developments lies in the last stage, where Chinese allows a co-existence of multiple conceptual entities (i.e. polysemies), whereas Western researchers forecast the ‘‘recessive’’ nature and possible complete disappearance of the original meaning (Traugott and Dasher 2002: 11).4 I have argued in di¤erent case studies (e.g. Xing 2003, 2004, 2006, 2009) that the reason why semantic functions in Chinese evolve by way of an accretion of meanings over time is primarily attributable to the isolating nature of the Chinese language structure at three di¤erent levels: morphology, syntax and discourse. At the morphological level, all Chinese characters are represented by one immutable written form, beyond which they simply cannot be reduced

3. It should be noted that of the three generalizations (i.e. expansion, reduction, and metaphorical extension) of semantic change traditionally discussed by Chinese grammarians (e.g. Gao Mingkai 1942), reduction appears to provide evidence that contradicts the pattern of accretion discussed here. However, if we examine the examples used to demonstrate the reduction of meaning by those grammarians, it becomes obvious that almost all examples are solid words (i.e. nouns and verbs) which do not undergo grammaticalization, and are thus not comparable to the cases examined in this study. Therefore, they are not counter-examples. When discussing semantic change, Wang Li (1980: 537564) also indicates that the pathway of semantic change in Chinese is: A > A,B > B. Again, if we look at the examples, we find the same situation with the reduction just mentioned. 4. The English verb ‘have’ seems to behave more like Chinese lexemes that have undergone grammaticalization than those that follow the evolutionary cline suggested by Western researchers. In modern times, ‘have’ has multiple semantic functions including: 1) possessive verb – ‘‘I have a computer’’; 2) perfective aspect – ‘‘I have bought a computer’’; and 3) modal auxiliary – ‘‘We have to go’’.

Introduction

5

unless they are systematically simplified through a language reform. In addition, two or more characters/words, especially those that have undergone grammaticalization, cannot merge into one character/word5 as they can in alphabetical languages (e.g. ‘going to > gonna’; ‘you all > y’all’, etc.), nor is it likely that new characters will be created by rearranging di¤erent strokes, as can be done anagrammatically in alphabetical languages, e.g. ‘‘lead’’, ‘‘deal’’, ‘‘lade’’, and ‘‘dale’’ in English. But most importantly, there is no agreement marking in Chinese with respect to the grammatical categories like number,6 case, gender, tense, mood, etc. As a result, the same noun form (i.e. character) can be used as subject/agent or object/patient and the same verb form (character) can be used as a main verb, a serial verb, a complement, an adverb, an adjective, a conjunction, or a preposition. Furthermore, a verb can also be used as a noun and vice versa, e.g., huı` 會 and lia´n 連 illustrated in (1) and (2). In comparison, nouns and verbs in English are not likely to behave in the same way as shown in example (3) using ‘even’.7 It appears that the unmarked word structure in Chinese permits flexibility and relative freedom in the interpretation of nouns and verbs consequently leading to the co-existence of multiple polysemies. (1) huı` ’s 會 various functions a. NOUN: ‘meeting’ 至會所, . . . , 以遇禮相見。(2nd Century AD: Kongzi Jiayu) zhı` huı` suoˇ, . . . , yıˇ yu` lıˇ xia#ng jia`n. arrive meeting place, as-to meet ritual each-other meet ‘They arrived at the meeting place, . . . so as to meet them politely.’

5. This does not include those words that were created by combining two characters such as bu´ zhe`ng 不正 ‘not straight’ for wa#i 歪, or nuˇ¨ zıˇ 女子 ‘female child’ for haˇo 好 ‘good’. 6. The only plural marking in Chinese is the su‰x, me´n 們, which attaches to singular pronouns, woˇ 我 ‘I’, nıˇ 你 ‘you’, and ta# 他/她 ‘he/she’ among others (i.e. pe´ngyoˇumen 朋友們). 7. There are some lexemes in Modern English that can be used as both a verb and a noun, such as ‘report’, ‘cause’, ‘experience’ etc. In addition, lexemes that can be used as both nouns and verbs undergo a stress change e.g., PERmit for the noun, perMIT for the verb. However, when they are used as nouns, they have to be marked for number and/or definiteness. When they are used as verbs, they have to be marked for tense, aspect or mood.

6

Janet Zhiqun Xing

b. VERB: ‘to meet’

在禮，卿不會公侯，會伯子男. . . (1st Century BC: Zuo Zhuan) za`i lıˇ, qıng # bu` huı` go#ng-ho´u, huı` bo´-zıˇ-na´n for etiquette o‰cial not meet ranking-o‰cial meet low-ranking ‘For the sake of etiquette, the o‰cial could not meet the [high] ranking o‰cials but could meet the low-ranking o‰cials.’ c. SERIAL VERB: ‘meet þ verb’

知戰之日，則可千里而會戰。(4th Century AD: Sunzi) zhı # zha`n zhı # rı`, ze´ keˇ qia#n lıˇ e´r huı` zha`n. know battle poss day, then may thousand kilometer then meet fight ‘Knowing the date of the battle, we can go a thousand miles to meet and fight.’ d. AUXILIARY: ‘might’ 汝向後也會去住。(10th Century AD: Zutangji) ruˇ xia`ng ho`u yeˇ huı` qu` zhu`. 2sg toward late also possible to stay ‘You can also go and stay (there) later.’ e. SERIAL VERB/COMPLEMENT: ‘verb þ perceive’ 小學是直理會那事；(12th Century AD: Zuzi Yulei) xiaˇoxue´ shı` zhı´ lıˇ huı` na` shı`; basic-classics is direct understand perceive det issue ‘Primary learning means to directly understand and learn that issue.’ f. COMPLEMENT: ‘verb þ learnt’ 小尼姑也都學會了年紀卷經咒 (17th Century AD: Honglou Meng) `u xiaˇo nı´gu# yeˇ do#u xue´ huı` le nia´njı` jua`n jıng-zho # young nun also all learn get asp number volume Buddhist-doctrine ‘All the junior nuns have also learnt a number of volumes of Buddhist doctrine.’ (2) lia´n’s 連 various functions a. VERB: ‘to connect’ 根下相連 (7th Century AD: Bianwen) ge#n xia` xia#ng lia´n. root under each-other connect/join ‘Roots underneath connect with one another.’

Introduction

7

b. VERB/ADJECTIVE: ‘continuing/continuous’ 皇帝. . . 看之，連聲便喚。(7th Century AD: Bianwen) hua´ng-dı` . . . ka`n zhı,# lia´n she#ng bia`n hua`n. emperor see it connect/consecutive voice then call ‘When the emperor saw it, he repeatedly called . . .’ c. ADVERB: ‘continously’ 梵人連忙前來。(7th Century AD: Bianwen) Fa´n re´n lia´n ma´ng qia´n la´i. name people connect/consecutively busy forward come ‘The Turkish people came forward in a hurry.’ d. PREPOSITION: ‘including/with’ 久住則連肉爛也。(Liu 1989: 452, 12th Century AD) jiuˇ zhu` ze´ lia´n ro`u la`n ye. long stay then including flesh rotten asp ‘If it were kept long, it would become rotten, including the flesh.’ e. CONJUNCTION: ‘even’

衆人轟然一笑，連賈珍也撐不住笑了。 (17th Century AD: Honglou Meng) zho`ng-re´n ho#ng-ra´n yı´-xia`o, lia´n Jiaˇ-Zhe#n yeˇ che#ng-bu´-zhu` xia`o le. everyone suddenly laugh even name also neg-help laugh asp ‘Everyone burst into laughter; even Jia Zhen laughed.’ or ‘Everyone burst into laughter and Jia Zhen could not help laughing either.’ (3) even’s various (historical) functions (the examples are quoted from OED) a. ADJECTIVE: Eeuen/eauen: flat (of a land/ground), smooth, direct, exact, equal Me thinkes the ground is eeuen. (1605, Shakespeare’s Lear) b. VERB: Euenen: to level (ground), to even out Law, whose end is, to euen and right all things (1851, Sidney’s Apol ) He hath now evened all his reckonings. . . (1664, Pepys’ Diary) c. ADVERB: Euene/evene/even: evenly as the rest goes euen (1601, Shakespeare’ Twel )

8

Janet Zhiqun Xing

At the sentence level, since tense, number, gender and case are not marked in Chinese, the semantic and pragmatic relationships between syntactic units (i.e. NP þ VP þ NP þ NP) are not as evident as they are in languages with those markings. As a result, a NP preceding a verb (be`i 被 or jia`o 叫) can be interpreted as either an agent or a patient, depending on the meaning of the main verb (prototypical verbs in a passive construction expressing some kind of adversative meaning) and the overall meaning of the sentence, as shown in (4)–(5). Such flexibility of NP interpretation means that the verb in a sentence can be easily coerced into an interpretation that fits the contextual meaning of the whole sentence. This is not possible in languages with clear markings for case, tense and/or mood. (4) a. Noun preceding be`i as an agent 月被其光而明。(12th Century AD: Zhuzi Yulei) yue` be`i qı´ gua#ng e´r mı´ng moon receive its light then bright ‘The moon becomes bright when receiving its light.’ b. Noun preceding be`i as a patient

阿孃被問來由，不覺心中歡喜。(7th Century AD: Bianwen) a#-nia´n be`i we`n la´i-yo´u, bu`jue´ xınzho# # ng hua#nxıˇ nanny pass ask come-reason, not-feel heart-middle happy ‘When the nanny was asked for the reason, she could not help feeling happy.’ (5) a. Noun preceding jia`o as an agent

他爸爸叫他修理一下那輛自行車。 ta# ba`ba jia`o ta# xiu#lˇı yı´xia` na` lia`ng zı`xı´ngche#. 3sg father ask 3sg fix bit det cl bicycle ‘His father asked him to fix the bicycle.’ b. Noun preceding jia`o as a patient

他叫那條狗咬了一口。 ta# jia`o na` tia´o goˇu yaˇo le yı # koˇu. 3sg pass det cl dog bite asp one bite ‘He was bitten by that dog.’ or ‘He got bitten by that dog.’ At the discourse level, Chinese sentences are arranged by such logical relations as sequential order and cause/reason–result, as illustrated by

Introduction

9

jiu`’s functions in (6). Such ‘fixed’ discourse structures, a characteristic of isolating and analytic languages, coincide with Kiparsky’s (1997, 2008: 24) reflection on the characteristic of inflectional languages that ‘‘the loss of inflectional morphology entails fixed order of direct nominal arguments.’’ I argue that the fixed discourse structure in Chinese plays an indispensible role in the development of two functions of the lexeme jiu` 就, namely, its discourse function of connecting two sequential events as in (6a) and (6b), and its function of connecting two logically related events as shown in (6c)–(6e) (cf. M. Liu 1993, 1997). Certainly, jiu`’s lexical meaning ‘to approach’ may also be a factor contributing to its discourse function. However, the discourse structure appears to be instrumental in the development of its discourse function. (6) a. Sequential: . . . , 欲遣就師。(3rd Century AD: Sanguo Zhi) . . . , yu` qiaˇn jiu` shı.# want dispatch approach master ‘(Someone) wants to approach the master.’ b. Sequential:

虎賁舁上殿就坐。(3rd Century AD: Sanguo Zhi) huˇbe#n yu´ sha`ng dia`n jiu` zuo`. commander carry up throne approach/then seat/sit ‘The commander was carried up to the throne to sit down.’ c. Condition–Result:

施薪若一，火就燥也; . . . 。 (1st Century BC: Xunzi) shı # xın# ruo` yı,# huoˇ jiu` za`o yeˇ, . . . . add hay like one, fire approach/then dry part ‘(If you) add one piece of hay, the fire will die.’ d. Reason–Result/sequential

資復遜位歸第，就拜驃騎將軍，(3rd Century AD: Sanguo Zhi) zı #fu` su`nwe`i guı # dı`, jiu` ba`i bia´oqı´ jia#ngju#n, leader resign return home, then pay-visit horse-riding commander ‘(If/when) the leader resigns and returns home, then (he) will pay a visit to the commander.’

10

Janet Zhiqun Xing

e. Reason–Result

遇富貴，就富貴上做工夫；(12th Century AD: Zhuzi Yulei) yu` fu`guı`, jiu` fu`guı` sha`ng zuo` go#ngfu#; run-into rich-honorable, then rich-honorable on do diligently ‘(If you) run into someone rich and honorable, then work accordingly hard.’ We have seen that all three levels, morphology, syntax, and discourse, leave room for Chinese lexemes to be used/interpreted/reanalyzed somewhat di¤erently by speakers/listeners. As a result, various polysemies have developed and co-exist at the modern stage of language development, a conclusion also supported by Bisang’s (2008: 586) study of Archaic Chinese, which shows that a lexical item in a given position is coerced into a particular semantic interpretation associated with that position. Arguably, if Chinese had agreement markers, it probably would not be common for a noun to function as a verb, an agent to be interpreted as a patient, or for a verb to be interpreted as an adjective, adverb, preposition or conjunction. If that were the case, the semantic change in Chinese would probably follow the cline suggested by Heine et al. (1991) and Traugott and Dasher (2002). Another factor that appears to have accompanied and somewhat a¤ected the development of polysemies in Chinese is the process of lexicalization. It is well documented in Chinese linguistic literature (cf. Peyraube 1988, Feng 1999, Dong 2002, and Dong in this volume) that disyllabic words and serial verb constructions emerged during the Han Dynasty (2nd Century BC–2nd Century AD) and became well established in the Tang Dynasty (7th–9th Century AD). Xing (2009) reported that among the 23 lexemes that have undergone grammaticalization, all had a tendency to be paired up with other lexical items to form disyllabic words after the Six Dynasties period (4th Century AD). Taking guo` 過 as an example, we can easily find disyllabic lexemes or words built from the monosyllabic guo` 過 in the course of its development, as shown in Table 1. Notice that some of these words have inherited guo`’s original verbal meaning ‘pass’ or its earlier nominal function ‘mistake’,8 whereas others were derived from guo`’s later developed polysemies ‘over, celebrate’. 8. The nominal function of guo` 過 ‘mistake’, as pointed out by the anonymous reviewer, was already attested in Classical Chinese (11th Century BC–220 AD) and should be accounted for in some of the compounds developed later, such as zuı`guo ‘crime mistake’, with a loss of guo`’s lexical tinge.

Introduction

11

Table 1. Lexicalized words with guo` 過 Char.

pinyin

gloss

English

過火 過活 過去 過節 過門兒 過錯 過目 過失 錯過 難過 不過 罪過

guo`huo˘ guo`huo´

over fire

‘overdone’

pass life

‘to live’

guo`qu` guo`jie´

pass go

‘to go over’

celebrate festival

‘to celebrate a festival’

guo`me´nr guo`cuo`

pass door over mistake

‘to marry into a family’ ‘fault’

guo`mu` guo`shı #

pass eye

‘to look over’

pass miss

‘wrong doing’

cuo`guo` na´nguo`

miss pass

‘to miss’

di‰cult pass

‘sad’

bu´guo` zuı`guo`

not pass

‘but’

crime mistake

‘sin’

These lexicalized disyllabic words commonly used in modern texts not only provide evidence of guo`’s polysemous functions but also reinforce those functions in modern communication. In other words, once guo` is combined with another lexeme (be it a verb þ object or verb þ complement combination) and becomes a frequently used lexical item, it is likely to be in use for a long time before it undergoes further change. In this volume, more evidence will be provided to illustrate the patterns of semantic change, grammaticalization, and lexicalization in Chinese.

2. Summary of the contributions to this volume The purpose of this volume is to provide an overview of recent developments in the study of grammaticalization and lexicalization in mainland China and Taiwan for English-speaking communities in the west. There are nine articles included in this volume, all of which are empirical studies based on diachronic and/or synchronic data, and all of which discuss issues relevant to either the characteristics of grammaticalization or lexicalization in Chinese or the typological patterns of the Chinese language in comparison to other languages. The nine articles are divided into two parts: Part I centers primarily on issues of grammaticalization and Part II

12

Janet Zhiqun Xing

focuses on lexicalization. Following is a brief summary of each of the nine articles arranged alphabetically according to the author’s last name. In Chapter 1, Chen investigates the development of sentence-final yeˇ 也 based on historical data. According to him, yeˇ has had two aspectual functions in the history of Chinese literary works: static and dynamic. There are two di¤erent views on their relationship, viz., whether the latter represents an extension of the former or whether there is any relationship between the two at all. Using historical data as evidence, Chen first defines the static yeˇ 也 in non-judgment sentences as a stative element and then argues that the emergence of dynamic yeˇ 也 is derived from static yeˇ 也, consistent with an established pattern of the grammaticalization of perfect markers in Chinese and some other languages. In Chapter 2, Fang studies the discourse and pragmatic functions of the proximal demonstrative zhe` 這 and the distal demonstrative na` 那 in Contemporary Beijing Mandarin. The results of this study show that the definite article was derived from the demonstrative zhe` through its recognitional use, a process that is accompanied by the emergence of the use of yı # 一 ‘one’ as an indefinite article. She argues that such a functional shift from a demonstrative to a definite article is a clear case of grammaticization, a term she uses to ‘‘refer to a process whereby an item is entering the grammar of a language synchronically and may become fixed and constrained in distribution’’ (cf. Hopper and Traugott 1993: XVI). As a result, a new grammatical category, definiteness, has emerged in Contemporary Beijing Mandarin, even though this pattern, as she points out, is not yet observed in written Mandarin Chinese. In Chapter 3, Liu investigates the various syntagmatic functions of the directional verb la´i 來. By providing resolutions to confusing cases as well as preventing erroneous assumptions about la´i’s grammaticalization, Liu concludes that the constructionist view seems the only available perspective to explain why la´i VERB 來 has remained vital over such an extraordinarily long period of time (from the 6th Century BC to the present), without being obviously influenced by all the changes related to it. She argues that the reason for la´i’s long life may be that the form of the verb remains at all times a free morpheme and change takes place only after the form has stepped into specific constructions and deviated from its original form. In Chapter 4, Liu and Chang explore one type of attributive predication in Mandarin Chinese in which a degree modifier (e.g. heˇn 很 ‘very’) is normally required to precede an attributive predicate. Adopting a constructional approach, Liu and Chang suggest that the attributive pattern

Introduction

13

‘Degree þ X’ can be re-analyzed as a Degree-Evaluative Construction where the degree marker is taken to be the constructional operator and the following element X, whatever it is, contributes a quality or attribute that is inferred. Through discussion of the interaction between grammaticalization and constructionalization, the authors demonstrate that the adverbial element heˇn triggers the constructional interpretation and becomes grammaticalized into a constructional operator. In Chapter 5, Peyraube and Li investigate three categories of volitional verbs in Chinese written texts from as early as the Pre-Classical period (11th–2nd Century BC) to the Medieval (4th–6th Century AD) and Modern periods: Category I: expressing the meaning of yua`nyı` 願意 ‘be willing to’; Category II: expressing the meaning of xı #wa`ng 希望 ‘hope’ and Category III: expressing the meaning of yu`wa`ng 欲望 ‘intention’. By tracing their origins and following their processes of semantic change, the authors come to the conclusion that volitional verbs have evolved in three di¤erent ways: 1) from intentional to future meaning, 2) from weak volition to strong volition, and 3) from concrete physical meaning to abstract mental meaning. They argue that modals that originally expressed the meaning of ‘intention’ (i.e. Category III) are the only ones that can become grammaticalized into future markers. It is less likely, if not impossible, that the other two types will evolve in this way because of their unique semantic and syntactic properties. In Chapter 6, Xing studies the emergence, development, and disappearance of classifiers in Mandarin Chinese by investigating the semantic changes involved in the grammaticalization of 16 of the most commonly used Modern Chinese numeral classifiers. She provides historical evidence showing that three mechanisms – metaphor, metonymy, and semantic reanalysis – play an important role in the emergence and development of classifier meaning while in the disappearance of classifiers, loss of semantic function and high frequency have been shown to be major contributing factors. She argues that the numeral classifier meaning is derived from the ‘numeral þ NP’ construction and not the other way around and concludes that such an evolution of meaning provides evidence for the interaction between the construction’s form and grammaticalization. In Chapter 7, Zhang provides diachronic and synchronic evidence to refute the view that the repeater is the earliest numeral classifier in SinoTibetan languages. She argues that since the original meaning or the construction where the repeater is used is not compatible with that of the classifier, it is unlikely that the numeral classifier is derived from the repeater. In addition, she suggests that the repeater disappeared before

14

Janet Zhiqun Xing

1100 B.C. in Chinese, which provides little evidence to support the claim that the numeral classifier is the origin of the repeater. She claims that the reason why researchers believe the repeater and the classifier to be related is that the repeater happens to be the most convenient way to express the grammatical category of ‘classifier’. She also suggests that numeral classifiers emerged and developed quite abruptly in Late Archaic Chinese (200 B.C.–200 A.D.). In Chapter 8, Dong discusses various issues related to the characteristics and processes of lexicalization in Chinese. First she demonstrates di¤erent types of lexicalization observed in the history of the Chinese language (e.g. from lexical phrases to words, from functional phrases to words, from syntactically unrelated word strings to words), then she discusses the degree of lexicalization, the constraints on lexicalization, the relation between syntactic change and lexicalization, and the interaction between Chinese typology and lexicalization. She concludes that lexicalization, like grammaticalization, is a naturally occurring change and thus is pervasive which can be seen very clearly from data in the history of the Chinese language. She points out that lexicalization might have idiosyncratic and languagespecific features that are not, as yet, well understood by scholars and therefore, further research on the relationship between the characteristics of lexicalization and language typology is necessary. In Chapter 9, Tsao investigates the change of the argument structure, re-analysis and lexicalization of geˇi 給 ‘give’ from a transitive verb to a ditranstive verb in Chinese. By analyzing the relationship between the distribution of geˇi and the verbal clauses that occur in a ditransitive construction, Tsao singles out three patterns: (1) geˇi is optional in ditransitive constructions involving a verb of transference; (2) only three classes of transitive verbs, namely verbs of acquisition like maˇi ‘buy’, verbs of movement like re#n ‘throw’, and verbs of creation like za`o ‘build’, can enter into the ditransitive construction; and (3) in order for that to happen, a verb of the above-mentioned classes has to go through a process of grammaticalization or lexicalization such as serial-verb-construction condensation or adjunct incorporation. Then he compares geˇi’s development with its counterparts in Japanese and English and finds that its Japanese counterpart undergoes the serial-verb-construction condensation while its English counterpart undergoes adjunct incorporation. From the summaries of the nine articles given above, it is evident that all the studies rely heavily on empirical data for their analyses, generalizations, and conclusions. Notice that among the nine articles, five focus on the issue of a certain grammatical category, such as the emergence of the

Introduction

15

definite article (Fang), the development of modal verbs of volition (Peyraube & Li), the emergence of the classifier class (Xing), the disappearance of the repeater (Zhang), and the process of lexicalization (Dong), while the remaining four articles are case studies of unique grammatical words which have all undergone a complicated process of grammaticalization: the sentence particle yeˇ (Chen), the versatile directional verb la´i (Liu), the degree adverb heˇn (Liu and Chang), and the verb of giving geˇi (Tsao). Even though these studies do not have a uniform theoretical orientation or rely on the same implications (e.g. some prefer the framework laid out by Construction Grammar; others apply either syntactic pathways or semantic mechanisms of grammaticalization to their analyses), they all attempt to identify the characteristics of diachronic change in Chinese. In addition, some of them have revealed certain typological characteristics in Chinese and have compared them with the typological characteristics of other languages. All the articles are important contributions to the corpus of work on diachronic change in the Chinese language and all are designed to expand the understanding of Western scholars interested in the history of the Chinese language. We of course hope that the subjects discussed in these articles will inspire other researchers to do further research and thus extend the field. References Bisang, Walter 2008

Precategoriality and syntax-based parts of speech – the case of Late Archaic Chinese. Studies in Language 32: 568–589. Bisang, Walter, Nikolaus P. Himmelmann, and Bjo¨rn Wiemer (eds.) 2004 What Makes Grammaticalization? A Look from its Fringes and its Components. Berlin: Mouton de Gruyter. Brinton, Laurel J. and Elizabeth C. Traugott 2005 Lexicalization and Language Change. Cambridge: Cambridge University Press. Bybee, J., R. Perkins, and W. Pagliuca 1994 The evolution of grammar: Tense, Aspect, and Modality in the Languages of the World. Chicago: The University of Chicago Press. Chappell, Hilary (ed.) 2001 Chinese Grammar: Synchronic and Diachronic Perspectives. Oxford: Oxford University Press.

16

Janet Zhiqun Xing

Chappell, Hilary (ed.) 2008 Sinitic Grammar: Synchronic and Diachronic Perspectives. Oxford: Oxford University Press. Dong, Xiufang 2002 Cı´huı`hua`: Ha`nyuˇ Shua#ngyın# Cı´ de Ya´nshe#ng he´ Fa#zhaˇn [Lexicalization: The Development of Chinese Disyllabic Words]. Chengdu: Sichuan Mingzu Chubanshe. Eckardt, Regine 2006 Meaning Change in Grammaticalization: An Enquiry into Semantic Reanalysis. Oxford: Oxford University Press. Feng, Shengli 1999 Ha`nyuˇ Yu`nlu`¨ Ju`faˇ Xue´ [Prosodically Constrained Syntax in Chinese.] Shanghai: Jiaoyu Chuban She. Fischer, Olga, Muriel Norde, and Harry Perridon (eds.) 2004 Up and Down the Cline – The Nature of Grammaticalization. Amsterdam/Philadelphia: John Benjamins. Gao, Mingkai 1942 Zho#ngguo´ yuˇ de yuˇyı` bia`nhua` [Semantic change in Chinese.] Tia#nwe´nta´i [Almanac], No. 2. Good, Je¤ (ed.) 2008 Linguistic Universal and Language Change. Oxford, NY: Oxford University Press. Kiparsky, Paul 1997 The rise of positioning licensing. In Parameters of Morphosyntactic Change, Ans van Kemenade and Nigel Vincent (eds.), 260–294. Oxford: Oxford University Press. Kiparsky, Paul 2008 Universals constrain change; change results in typological generalizations. In Good (ed.), 23–53. Kuteva, Tania 2004 Auxiliation: An Enquiry into the Nature of Grammaticalization. Oxford: Oxford University Press. Heine, Bernd 1993 Auxiliaries: Cognitive Forces and Grammaticalization. Oxford: Oxford University Press. Heine, Bernd and Tania Kuteva 2008 Constraints on contact-induced linguistic change. Journal of Language Contact THEME II, pp. 57–90. Heine, Bernd and Mechthild Reh 1984 Grammaticalization and Reanalysis in African Language. Hamburg: Helmut Buske. Heine, Bernd, Ulrike Claudi, and Friederike Hu¨nnemeyer 1991 Grammaticalization: A conceptual Framework. Chicago: University of Chicago Press.

Introduction

17

Heine, Bernd and Tania Kuteva 2002 World Lexicon of Grammaticalization. Cambridge: Cambridge University Press. Hopper, Paul and Elizabeth C. Traugott 1993 Grammaticalization. 1st ed. Cambridge: Cambridge University Press. Hopper, Paul and Elizabeth C. Traugott 2003 Grammaticalization. 2nd ed. Cambridge: Cambridge University Press. Lehmann, Christian 1986 Grammaticalization and linguistic typology. General Linguistics 26, No. 1: 3–23. Lehmann, Christian 1995 Thoughts on Grammaticalization. Munich: Lincom Europa. Liu, Jian 1989 Shı`lu`n ‘he´’ zı` de fa#zhaˇn, fu`lu`n ‘go`ng’ zı` he´ ‘lia´n’ zı` [The development of he´ ‘and’ and the development of go`ng ‘together’ and lia´n ‘even’.] Zhongguo Yuwen [Chinese Language] 6. Liu, Mei-chun 1993 Discourse, Grammar, and Grammaticalization: Synchronic and Diachronic Analyses of Mandarin Adverbial Markers jiu` and ca´i. Ph.D. dissertation, University of Colorado, Boulder. Liu, Mei-chun 1997 From motion verb to linking element: discourse explanations for the grammaticalization of jiu` in Mandarin Chinese. Journal of Chinese Linguistics 25:2: 258–289. Ma, Beijia 1993 Jie`cı # ‘to´ng’ de chaˇnzhe#ng [The emergence of the preposition tong ‘same as’.] Zhongguo Yuwen [Chinese Language] 2. Ma, Beijia 2002 Jı`nda`i ha`nyuˇ jie`cı´ [Prepositions in Modern Chinese.] Beijing: Zhonghua Shuju. Mara Jose Lopez-Couso and Elena Seoane 2008 Rethinking Grammaticalization: New Perspectives. Philadelphia/ Amsterdam: John Bejamins. Peyraube, Alain 1988 Syntactic change in Chinese: On grammaticalization. Bulletin of the Institute of History and Philology, Vol. LIX, Part III, 617– 652. Taipei: Academic Sinica. Peyraube, Alain 1989a History of the comparative construction in Chinese from the 5th century BC to the 14th century AD. Proceedings on the Second International Conference on Sinology, 589–612. Taipei: Academic Sinica.

18

Janet Zhiqun Xing

Peyraube, Alain 1989b History of the passive constructions in Chinese until the 10th century. Journal of Chinese Linguistics, Vol. 17, No. 2: 335–371. Peyraube, Alain 1989c Zaˇoqı # baˇ zı` ju` de jıˇge we`ntı´. [Several questions on the early baˇ construction.] Zhongguo Yuwen [Chinese Language] 1. Peyraube, Alain 1992 History of some coordinative conjunctions in Chinese. Paper presented at the First International Conference on Chinese Languages and Linguistics. Singapore, June 24–26, National University of Singapore. Peyraube, Alain 1994a On the history of Chinese locative prepositions. Chinese Languages and Linguistics 2: 361–87. Peyraube, Alain 1994b New reflections on the history of the accusative forms in ba in Chinese. Cahiers de Linguistique Asie Orientale 23, 265–277. Peyraube, Alain 1996 Modal auxiliaries of volition in Ancient Chinese. Paper presented at the Symposium on Synchronic and Diachronic Perspectives on Grammar of Sinitic Languages. University of Melbourne. Peyraube, Alain 1998 On the modal auxiliaries of possibility in Classical Chinese. Selected Papers from the Fifth International Conference on Chinese Linguistics. Taipei: The Crane Publishing Co. Ltd. Ramat, Anna Giacalone and Paul J. Hopper (eds.) 1998 Limit of Grammaticalization. Amsterdam/Philadelphia: John Benjamins. Shi, Ziqiang 1989 The grammaticalization of the particle le in Mandarin Chinese. Language Variation and Change 1, 99–114. Shi, Yuzhi and Na Li 2001 Ha`nyuˇ Yuˇfaˇhua` de Lı`che´ng [The Process of Grammaticalization in Chinese.] Beijing: Beijing University Press. Sun, Chaofen 1996 Word Order Change and Grammaticalization in the History of Chinese. Stanford: Stanford University Press. Traugott, Elizabeth C. and Bernd Heine (eds.) 1991 Approaches to Grammaticalization, Volumes 1–2. Amsterdam/ Philadelphia: John Benjamins. Traugott, Elizabeth C. and Richard Dasher 2002 Regularity in Semantic Change. Cambridge: Cambridge University Press. Wang, Li 1980 Ha`nyuˇ Shıˇgaˇo [Chinese Grammar.] Revised ed. Beijing: Zhonghua Shuju.

Introduction

19

Wischer, Ilse and Gabriele Diewald (eds.) 2002 New Reflections on Grammaticalization. Amsterdam/Philadelphia: John Benjamins. Visconti, Jacqueline 2004 Conditionals and subjectification: Implications for a theory of semantic change. In Fischer, Olga et al. (eds.), 169–192. Xing, Janet Z. 2003 Grammaticalization of verbs in Mandarin Chinese. Journal of Chinese Linguistics, Vol. 31:1, 101–144. Xing, Janet Z. 2004 Grammaticalization of lia´n in Mandarin Chinese. Journal of Historical Pragmatics 5.1: 81–106. Xing, Janet Z. 2006 Mechanism of semantic change in Chinese. Studies in Language, Vol. 30:3, 461–483. Xing, Janet Z. 2009 Regularity in semantic change in grammaricalization in Chinese. Paper presented at the Linguistic Series, English Department, National Taiwan Normal University, March 30. Xu, Dan 1992 Ha`nyuˇ lıˇ de ‘za`i’ he´ ‘zhe’ [The aspect markers za`i and zhe in Chinese.] Zhongguo Yuwen [Chinese Language] 6. Xu, Dan 2006 Typological Change in Chinese Syntax. Oxford: Oxford University Press. Xu, Dan 2008 Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives. New York: Springer Science. Zhang, Wangxi 1991 ‘Baˇ’ zı` jieˇgo`u de yuˇyı` jı´qı´ yuˇyo`ng fe#nxi [Semantic and pragmatic functions of the baˇ construction.] Yuyan Jiaoxue yu Yanjiu [Language Teaching and Research] 3.

The development of the Chinese aspectual sentence-final marker yeˇ * Qianrui Chen 1. Introduction This paper studies the development of two aspectual functions of sentencefinal yeˇ 也 in the history of Chinese literary works. Both Lu¨ ([1942] 1982: 276) and Wang (1989: 306) suggest that in Classical Chinese, sentencefinal yeˇ is static as in (1), while sentence-final yıˇ 矣 is dynamic as in (2). The static function is realized as a judgment in sentence (1), whereas the dynamic function indicates a change of situation in sentence (2). (1) 是社稷之臣也。(5th Century BC, Lu´nyuˇ ) Shı` she`jı` zhı # che´n yeˇ. This country POSS o‰cial PART ‘This person is an o‰cial of the country.’ (2) 今日病矣 , 余助苗长矣。(3rd Century BC, Me`ngzıˇ ) Jı #nrı` bı`ng yıˇ, yu´ zhu` mia´o zhaˇng yıˇ. Today sick PART I help seedling grow PART ‘Today the seedlings are withered. I helped them to grow.’ Current studies have noticed that in Medieval Chinese and Early Modern Chinese,1 yeˇ 也 also has a dynamic function expressing the ‘‘result of a * This research is sponsored by the National Social Science Foundation of China (No. 08BYY050). I am most grateful to Professor Shaoyu Jiang who provided me with guidance, valuable comments and suggestions. In the paper, some constructive suggestions were also o¤ered by Professor Hongming Zhang, Professor Fuxiang Wu, Professor Bo Hong, Professor Rongxiang Yang, Professor Chirui Hu and the anonymous reviewers. The Chinese version, in which part 4 was deleted, was published in Zho#ngguo´ Yuuˇ¨ we´n [Studies of the Chinese Language] 2008(1). The author is solely responsible for all errors that remain. 1. This paper adopts Fang (2004)’s framework on the division of the history of the Chinese language: Archaic Chinese (prior to the 1st Century BC), Medieval Chinese (1st–6th Century), Early Modern Chinese (from 7th Century to 18th Century).

22

Qianrui Chen

change’’, as shown in the examples in (3), cited from Ohta ([1958] 1987: 353) and Cao (1987: 14). In (3) the functions of yeˇ 也 are similar to that of sentence-final le 了 in Modern Chinese and the present perfect pattern ‘‘have þ -ed ’’ in English. (3) a.

天下已有主也。(3rd Century, So#ushe´njı`) Tia#n xia` yıˇ yoˇu zhuˇ yeˇ. Heaven under already have master PART. ‘The kingdom already has its lord.’

b.

旦书至也 , 得示为慰。(3rd Century, Za´tieˇ ) Da`n shu# zhı` yeˇ, Morning letter arrive PART de´ shı` we´i we`i. receive read as comfort ‘The letter arrived in the morning and reading the letter comforted me.’

c.

石贤者来也 , 一别二十余年。(4th Century, Yo#umı´nglu`) Shı´xia´n zheˇ la´i yeˇ, name PART come PART yı # bie´ e`rshı´ yu´ nia´n. one apart twenty over year ‘Shixian has come; I have not seen him for over 20 years.’

From the examples given above, it can be seen that there exist two opposite uses of yeˇ 也 in Chinese history, namely, a static and a dynamic one, which raises the question of how the two usages developed. One view, the Sound-record Hypothesis, is represented by Ohta Tatsuo, who suggests that the two yeˇ s 也 are not in any way related to each other. Ohta (1987: 353) argues that dynamic yeˇ 也 perhaps originates from the dynamic particles yıˇ 矣 or yıˇ 已 ‘already’ because, in oral expression, the sound [i] 矣 or 已 changed to [ia] by attaching the popular particle [a] 阿. Therefore, yeˇ 也 was needed to express a dynamic function and was used to fill this gap. Following Ohta, Shimura (1995: 98–99) points out that it is not clear how the sound [ia] evolved to be used as a dynamic particle and therefore this issue needs to be explored further. The other view, the Extension Hypothesis, is represented by Luo (1994), Sun (1999: 46) and Dai (2006), who suggest that dynamic yeˇ 也 is an extension of static yeˇ 也. Luo (1994) claims that in Early Modern Chinese

The development of the Chinese aspectual sentence-final marker yeˇ

23

yeˇ 也, expressing static a‰rmation or judgment, extended its function to express a change of state. Sun (1999: 46) demonstrates that the scope of the use of yeˇ 也 has been expanding since Medieval Chinese (from the 1st Century to the 6th Century). However, neither Luo nor Sun comments on Ohta’s view. Comparing Shimura’s (1984) and Sun’s (1999: 46) views, Dai (2006) partially supports Sun’s idea, saying that yeˇ ’s 也 dynamic use is an extension of yeˇ ’s 也 earlier grammatical function. 1.1. The two hypotheses and their problems Yang’s study (1991) supports the Sound-record Hypothesis using evidence from historical phonology and the Minnan dialect in Chinese. She also points out (1991: 237) that Ohta does not give any reasons for the attachment of [a] 阿 to [i] 矣 or 已. In the same paper, Yang makes an attempt to explain the reason why yıˇ 矣 might have been read as [ia] in oral Chinese during the Medieval period, which might, in turn, have caused yeˇ 也 to be used to represent the sound of the particle expressing the dynamic function. Yang’s (1991) main argument is that, in the Minnan dialect, there is a similar sentence-final a 阿 which is said to have arisen from yıˇ 矣. However, Yang (1991) also recognizes that both yeˇ 也 and yıˇ 矣 in the Minnan Dialect contain the vowel [a], and are thus similar to sentence-final a 阿. Therefore, it is premature to interpret dynamic yeˇ 也 as having a phonetic origin.2 As for the Extension Hypothesis, Dai (2006: 209) demonstrates that this hypothesis is primarily based on the Tiantai dialect of the Zhejiang province, China. By examining the origin of the perfect marker [a] 啊 in the Tiantai dialect, Dai infers that [a] 啊 was the result of the reduction of yeˇ 也 and originated from Medieval Chinese dynamic yeˇ 也. Thus, a semantic relationship between the dynamic yeˇ 也 of Medieval Chinese and the aspectual marker a 啊 in the modern Tiantai dialect can be established, though a relationship between the dynamic yeˇ 也 of Medieval Chinese and the static yeˇ 也 of Archaic Chinese is not suggested. At this point, it is clear that the argument for the rise of dynamic yeˇ 也 based on historical phonology and dialectal analyses is not quite persuasive and that the support from semantic and grammatical functions is rather weak. Therefore, it is necessary to explore the issue more systematically and from a more theoretical point of view. 2. During interviews, Professor Bo Hong did not agree with the Sound-record Hypothesis and its explanation; Professor Fuxiang Wu confirmed that the two uses of yeˇ 也 are actually related.

24

Qianrui Chen

1.2. Inspiration from aspectual typology The theory of aspectual typology may shed some light on this study. According to Bybee, Perkins, and Pagliuca (1994: 105), the anterior or perfect aspectual marker has generally developed from three lexical sources: 1) auxiliary verbs like be and have; 2) verbs denoting the meaning of coming; and 3) verbs denoting the meanings of finishing and moving. The path of development of auxiliary verbs runs from resultative to anterior, then on to perfective or past tense. The resultative, commonly composed of an auxiliary verb and the past particle in Indo-European languages, denotes a state that resulted from some action in the past. Bybee, Perkins, and Pagliuca (1994: 63) illustrate how this works in English using the ‘‘be þ -ed ’’ construction. The sentence He is gone refers to the situation that he is no longer here. Therefore, the sentence He is gone and has come back already is not well accepted. In comparison, the anterior is formed by ‘‘have þ -ed ’’ in English. The sentence He has gone only indicates the current relevance of the action that happened in the past (perhaps he came back again). Thereby the sentence He has gone and come back already is acceptable grammatically and the anterior ‘‘have þ -ed’’ is extended from the resultative ‘‘be þ -ed ’’. The dynamic use of yeˇ 也 in Medieval Chinese and Early Modern Chinese is much more similar to that of sentence-final le 了 in Contemporary Chinese. With regard to the aspectual function of the latter, it has been labeled a ‘‘sentence-final aspectual particle’’, anterior or perfect (Cao 1995: 96, Li, Thompson, and Thompson 1982). In Archaic Chinese (prior to the 1st Century BC), the function of yeˇ 也 in its static use was mainly to express a judgment analogous to the original meaning of be in English. Other usages of static yeˇ 也 are similar to the resultative one and indicate a state which will be explained in the following section. Dynamic yeˇ 也, therefore, coincides with the perfect aspect in English not only in its lexical source but also in the path of its grammaticalization. When explaining Bybee and Dahl’s approach, Dahl (2000: 7) defines the basic units of investigation as grams and notions like tense, aspect and mood as ways of characterizing the semantic content of grams, therein observing the semantic content and diachronic change of the gram. Based on this approach, this study aims to demonstrate the typological characteristics of the aspectual marker yeˇ 也 in Archaic Chinese and to examine the process of development from static yeˇ 也 to dynamic yeˇ 也 in comparison with the grammaticalization of the perfect aspect in English. It will then go on to explore the typological significance of the aspectual uses of yeˇ 也.

The development of the Chinese aspectual sentence-final marker yeˇ

25

2. The aspectual use of yeˇ 也 in Archaic Chinese (prior to the 1st Century BC) The usage of yeˇ 也 in Archaic Chinese is rather complicated. This paper examines only the two uses of yeˇ 也 as an assertive-sentence-final particle defined as ‘‘static’’ and ‘‘dynamic’’. 2.1. Static yeˇ 也 in Classical Chinese Classical Chinese is the form of written Chinese that was in use from Archaic Chinese (prior to the 1st Century BC) to the early 20th Century when it was replaced by vernacular written Chinese. It follows the standards of classical works in Archaic Chinese, and has been studied thoroughly for a long time. Lu¨ ([1944] 2002: 226–227) classified yeˇ 也, when used at the end of assertive sentences or narrative sentences in Classical Chinese, into three types. The first type is defined as ‘‘the mood of judgment’’ to account for inclusiveness, as shown in (4)–(5). ((4)–(12) are cited by Lu¨ (2002) as examples in Classical Chinese): (4) 医者, 意也。(7th Century, Qıanjı # n# Yıfa# # ng) Yı # zheˇ, yı` yeˇ. Doctor PART awareness PART ‘One who is a doctor must be aware.’ (5) 孺子可教也。(Chinese Idiom) Ru´ zıˇ keˇ jia`o yeˇ. Child son be-able-to teach PART ‘This child is can be taught.’ The second use of static yeˇ 也 is defined as ‘‘the mood of explanation’’ to account for states of a¤airs or a cause, a result or a purpose involving some state of a¤airs as shown in (6) and (7). (6) 南方多没人, 日与水居也。(11th Century, Rı`yu`) Na´nfa#ng duo# mo` re´n, rı` yuˇ shuıˇ ju# yeˇ. South more down person day with water dwell PART ‘In the south, more people are drowned because they live close to water.’

26

Qianrui Chen

(7) 古之人与民偕乐, 故能乐也。(4th Century BC, Me`ngzıˇ ) Guˇ zhı # re´n yuˇ mı´n xie´ le`, Ancient POSS person with people together entertain gu` ne´ng le` yeˇ. therefore can happy PART ‘(Since) the ancient people always enjoyed life with the commoners, they were happy.’ The third use of static yeˇ 也 is defined as ‘‘the mood of resolution ( jia#njue´ 坚决)’’ to stress the meaning of the whole sentence, as shown in (8) and (9). (8) 环滁皆山也。(11th Century, Zuı`we#ngtı´ng Jı` ) Hua´n Chu´ jie# sha#n yeˇ. Surround city-name all mountain PART ‘Chu city is surrounded by many mountains.’ (9) 虽当世宿学, 不能自解免也。(1st Century BC, Shıˇjı` ) Suı # da#ngshı` su`xue´, Although current knowledgeable-person bu` ne´ng zı` jieˇ miaˇn yeˇ. not can self solve relieve PART ‘Even knowledgeable people cannot avoid being ridiculed.’ Yeˇ 也 in (4) expresses a typical mood of judgment. In the typical judgment sentence, the semantic function of yeˇ 也 coincides with shı` 是 ‘be’ in Modern Chinese as well as with the verb to be in English. According to her theory of situation types which include stative, activity, accomplishment, achievement and semelfactive, Smith (1991: 38) considers a construction with a judgment verb and its main arguments a stative situation with static and durative properties, but without telic properties. He (1992: 135–159) further subclassifies the stative situations into five states: 1) the absolute, such as have, belong to, etc.; 2) the non-absolute, such as more, snow-white, green, etc.; 3) the existential, like stand, lie, etc.; 4) the habitual, such as he smokes; and 5) the mental, such as love, believe, etc. Based on He’s classifications, static yeˇ 也 as in (4) belongs to the first type of stative situation – the absolute state. In (4) the predicate is nominal but in (5) it is verbal. He (2004: 421) suggests that yeˇ 也 following the verbal predicate conveys the speaker/ writer’s judgment of the characteristics, significance, and the intention of a person or issues represented by the subject of the sentence.

The development of the Chinese aspectual sentence-final marker yeˇ

27

As for the second use of yeˇ 也, the mood of explanation, it is reasonable to consider the relationship between a cause and a result clause one of relational judgment. If this is acceptable, it becomes easier to understand the property of stativeness of the cause-result clauses. It should be noted that a predicate followed by yeˇ 也 expressing the mood of explanation is primarily verbal. Existential sentences embody the stative property in many languages (e.g. the use of be and have in English.) As for the third type of yeˇ 也 in (8), it can be viewed as expressing redundant information and/or reinforcing information, but it does not express judgement of the predicate. Negative sentences expressed by bu` 不 ‘‘not’’ or we`i 未 ‘‘not yet’’ clearly refer to a stative situation. According to Pulleyblank ([1995] 2006: 128–129), the appearance of yeˇ 也 at the end of a we`i 未 ‘not yet’ sentence is a clear indication of its function as a marker of a continuing state. Examples given by Lu¨ (2002: 226–227) of the third type of yeˇ 也 sentence, those expressing resolution, instantiate stative situation types. According to He (1992), the sentences in (10), (11) and (12) correspond to the habitual stative, the mental stative and the non-absolute stative respectively. This seems to suggest that ‘‘the mood of resolution’’ is better described as ‘‘the mood of stressing a stative situation’’. (10) 今者项庄拔其剑, 其意常在沛公也。(1st Century BC, Shıˇjı` ) Jı #nzheˇ Xia`ngzhua#ng ba´ qı´ jia`n, Now name pull-out its sword qı´ yı` cha´ng za`i Pe`igo#ng yeˇ. his intention always towards name PART ‘Now Xiangzhuang pulled out his sword, his intention all along to assassinate Peigong.’ (11) 不患人之不己知, 患不知人也。 (5th Century BC, Lu´nyuˇ ) Bu´ hua`n re´n zhı # bu` jıˇ zhı #, Not worry people POSS not self understand hua`n bu` zhı # re´n yeˇ. worry not understand people PART ‘Don’t worry that others don’t understood us, worry instead that we don’t understand others.’

28

Qianrui Chen

(12) 子灿又尝见其写市物帖子, 甚工于楷书也。 (17th Century, Da`tie´zhuı # Zhua`n) Zıˇca`n yo`u cha´ng jia`n qı´ xieˇ shı` wu` tieˇzi, Name again once see him write market stu¤ poster she`n go#ng yu´ kaˇishu# yeˇ. very be-adept at standard-style PART ‘Zican had seen him write market posters. He was skillful with the Standard style.’ Three common uses of yeˇ 也 which include the mood of judgment, the mood of explanation, and the mood of stressing a stative situation, are all related to a stative situation and their functions appear to demonstrate di¤erent levels of grammaticalization. This is the reason why the analysis provided in this paper is based on Lu¨ (2002)’s classification of yeˇ 也 in Classical Chinese. Nedjalkov and Jaxontov (1988: 3–6) defined ‘‘state’’ as one of the basic properties for predicates, di¤ering from action and quality. However, the term stative is applied to forms that include both grammatical verb forms and derived verbs to denote states. States are not necessarily expressed by specially marked verb forms. The meaning of state may also be expressed by simple, non-derived verbs like to sit and to sleep, some adjectives like sick, open and ready, and predicative adverbs like afraid, aware and asleep etc. All those words whose base form or entire paradigm expresses a state may be termed ‘lexical statives’. Therefore, the properties of a predicate, in Nedjalkov & Jaxontov’s view (1988), are similar to the situation types described by Smith (1991: 38). In other words, Nedjalkov & Jaxontov’s (1988) state corresponds to Smith’s (1991: 38) stative situation, and the term ‘‘stative’’ in Nedjalkov & Jaxontov (1988) is used in a rather di¤erent sense. A resultative expresses a resultant state with a previous action being implied, while a stative expresses the state of a thing without any indication of its cause. As to the di¤erence between resultative and stative, Nedjalkov & Jaxontov (1988: 7) point out that it is not always easy to distinguish between the two because they share many important properties. Nedjalkov & Jaxontov (1988) suggest that all forms expressing states of either kind be united under the term ‘‘resultative’’ or ‘‘resultatives in a broad sense’’, covering both resultatives and statives. Whenever necessary, the form expressing a resultant state implying a previous action is called the ‘‘resultative proper’’ or ‘‘resultative in a narrow sense’’.

The development of the Chinese aspectual sentence-final marker yeˇ

29

Therefore, we conclude that the first type of static yeˇ 也 in a judgment sentence can be defined as a lexical stative, but not as an aspectual marker. The third type of static yeˇ 也 may be defined as a marker of stative aspect which developed from the first type of static yeˇ 也 (cf. Pulleyblank 1995). Furthermore, it is likely that the second type of static yeˇ 也 in an explanation sentence is used as a bridge between the first and the third type and is defined as a stative marker that helps to shift the predicative function to the causal or resultative function indicating a state. 2.2. Dynamic yeˇ in Archaic Chinese (prior to the 1st Century BC) Wang ([1798] 2000: 44) indicates that sometimes yeˇ 也 can be defined as dynamic yıˇ 矣; thus yeˇ 也 and yıˇ 矣 are interchangeable, as exemplified by Wang (2000: 42) and Yang ([1928] 1986: 333) in (13)–(16) (13) 从我于陈蔡者, 皆不及门也。(5th Century BC, Lu´nyuˇ ) Co´ng woˇ yu´ Che´n Ca`i zheˇ, jie# bu` jı´ me´n yeˇ. Follow I arrive country-name PART, all not reach door PART ‘All those who followed me from Chen to Cai did not reach the door (and meet the required standard).’ (14) 散军而郊射, 左射狸首, 右射驺虞, 而贯革之射息也。 (1st Century BC, Lıˇjı` ) Saˇn ju#n e´r jia#o she`, Dismiss army CONJ outskirt shoot zuoˇ she` Lı´shoˇu, yo`u she` Zo#uyu´, left shoot song-name, right shoot song-name e´r gua`n ge´ zhı # she` xı # yeˇ. CONJ pierce leather POSS shoot cease PART ‘Dismiss the army and hold a shooting competition on the outskirts (of the city); for those shooting on the left, play the song of Lishou; for those shooting on the right, play the song of Zouyu, thus the shooting which is to pierce the leather of the army can cease.’ (15) 刑罚行于国, 所诛者乱人也。如此则民顺治而国安也。 (1st Century BC, Lıˇjı` ) Xı´ng fa´ xı´ng yu´ guo´, Criminal law apply in country suoˇ zhu# zheˇ lua`n re´n yeˇ. PART kill PART disrupt person PART

30

Qianrui Chen

Ru´ cıˇ ze´ mı´n shu`n zhı` Like this therefore people obey govern eˇr guo´ a#n yeˇ. CONJ country peaceful PART ‘Apply a law to kill outlaws in a country; if this is done, the people will obey the government and there will be peace in the land.’ (16) 且夫栾氏之诬晋国久也 , . . . 若灭栾氏, 则民威矣。 (4th Century BC, Guo´yuˇ) Qieˇ fu# Lua´nshı` zhı # wu# Jı`n CONJ MODEL surname POSS deceive country-name guo´ jiuˇ yeˇ, . . . ruo` mie` Lua´nshı`, country long PART CONJ execute name ze´ mı´n we#i yıˇ. so people boost PART ‘Additionally Luanshi has deceived the country of Jin for a long time. . . . If the Luanshi family is executed, the people’s spirits will be boosted.’ However, this definition of yeˇ 也 and yıˇ 矣 is not completely correct. For example, Ma (2005: 453) disagrees that yeˇ 也 in (13) can be replaced by dynamic yıˇ 矣. While yeˇ 也 in (13) is static indeed, other sentences may have a dynamic reading. The properties of the verbs in (14) are the most important reason why Wang (2000: 42) and Yang (1986: 333) both agree on the interchangibility of yeˇ 也 and yıˇ 矣. In (14), the verb xı # 息 ‘cease’ expresses an achievement situation with the properties [–static], [–durative] and [þtelic]. In (15), although the verbs shu`n 顺 ‘obey’ and a#n 安 ‘stable’ can express a stative situation, the hypothesis clause pattern ru´cıˇ, ze´ . . . yeˇ 如此, 则 . . . 也 ‘like this . . . therefore’ often introduces the result of a change. What makes it clearer is that in (16) the temporal word jiuˇ 久 ‘long’ indicates an obvious change of time. Based on the aforementioned evidence, this paper argues that specific reasons that have led to the di¤erent understandings of yeˇ 也 should be examined when explaining the connections between them. One of the important parameters in examining the degree of grammaticalization of the aspectual marker is the analysis of the situation types of the verb or verb phrases which co-occur with the marker. In the following, the semantic characteristics of the predicate in the examples collected will

The development of the Chinese aspectual sentence-final marker yeˇ

31

be discussed.3 The Archaic Chinese corpus used for this paper includes Zuoˇzhua`n 左传, Lu´nyuˇ 论语, Guo´yuˇ 国语, Lıˇjı` 礼记, Zhua#ngzıˇ 庄子, and Shıˇjı` 史记. Due to the high frequency of yeˇ 也 in historical texts, it is impossible to examine all possible sources. 2.2.1. Dynamic yeˇ 也 in clauses whose verbs convey the meaning of death One third of the examples from the database used in this study contains verbs related to death (e.g. sıˇ 死 ‘die’ as in (17), zu´ 卒 ‘die suddenly’, bı`ng 病 ‘be sick’, or zı`sha# 自杀 ‘commit suicide’ as in (18), etc.) These verbs typically express an achievement situation, namely a process or result related to death. The property of these verbs ties in naturally with the stative property of yeˇ 也, thereby suggesting that dynamic yeˇ 也 is likely to have evolved out of this linguistic environment. (17) 秋, 雨螽于宋, 队而死也。(4th Century BC, Zuoˇzhua`n) Qiu#, yuˇ zho#ng yu´ So`ng, zhuı` e´r sıˇ yeˇ. fall, rain locust in country-name, fall CONJ die PART ‘In the fall, Song was attacked by a plague of locusts; the locusts fell down and died.’ (18) 夏六月庚申, 卒。鬻拳葬诸夕室。亦自杀 也, 而葬于绖皇。 (4th Century BC, Zuoˇzhua`n) Xia` liu`yue` Ge#ngshe#n, zu´. Yu`qua´n za`ng zhu# Xıshı # `. Summer June day-name die Name bury him-in place-name Yı` zı`sha# yeˇ, e´r za`ng yu´ die´ hua´ng. also suicide PART CONJ bury in dooryard underground-palace ‘In the summer, on the day of Gengshen, June, the king died. Yuquan buried him in the Xishi and then committed suicide, and then he was buried in the dooryard of the king’s underground palace.’ 2.2.2. Various other verbs with dynamic yeˇ 也 Various other verbs such as huo` 获 ‘obtain’, ‘go towards and enter’, she#ng 生 ‘bear’, and been discovered to have a clear dynamic and

jı`n 尽 ‘reach’, waˇngru` 往入 lı` 立 ‘confer’ etc. have also changing character that can

3. Following Prof. Hongming Zhang’s suggestion, the present paper analyzes the development of yeˇ 也 from the perspective of verb semantic types.

32

Qianrui Chen

force a yeˇ 也 sentence out of its static sense. In (20) the second clause we`i jı`n sha`n yeˇ 未尽善也 ‘have not yet reached the best’ confirms the negative state, and the first clause yo`u jı`n sha`n yeˇ 又尽善也 ‘and has reached the best’ confirms that it has reached the best state. The analogy from negative to a‰rmative triggers the shift from static to dynamic and the interchangeable use of dynamic yıˇ 矣 and yeˇ 也 in the first part of (20) is clear evidence of the shift. (19) 项羽既存赵 . . . 灭秦而立侯王也 , 乃徙齐王 . . . 。 (1st Century BC, Shıˇjı`) Xia`ngyuˇ jı` cu´n Zha`o . . . Name already reserve country-name mie` Qı´ e´r lı` ho´uwa´ng yeˇ, destroy county-name CONJ confer leud PART naˇi xıˇ Qı´ wa´ng . . . . then move country-name king ‘Since Xiangyu has already protected the Zhao country . . . destroyed the Qin Dynasty and conferred the title of leud on its governor, then he moved Qi king . . .’ (20) 子谓《韶》：‘‘尽美矣, 又尽善也。 ’’ 谓《武》： ‘‘尽美矣, 未尽善也。 ’’ (5th Century BC, Lu´nyuˇ ) Zıˇ we`i (Sha´o): ‘‘Jı`n meˇi yıˇ, Confucius comment song-name reach beauty PART yo`u jı`n sha`n yeˇ.’’ We`i (Wuˇ): again reach perfect PART comment song-name Jı`n meˇi yıˇ, we`i jı`n sha`n yeˇ. reach beauty PART not reach perfect PART ‘Confucius praised Shao: ‘‘It has achieved the best in melody, and has achieved the best in lyrics.’’ He viewed Wu: ‘‘It has achieved the best in melody but hasn’t achieved the best in lyrics.’’’ 2.2.3. Verbs of perception with dynamic yeˇ 也 Another type of verb commonly used in historical texts to express the nonstative meaning is we´n 闻 ‘heard’. In the Chinese language, verbs such as this convey the result of inherent perception, which di¤ers from the action of pure perception. Actually, they too express an achievement situation. Carry (1995: 95) suggests that in English the most important evidence for

The development of the Chinese aspectual sentence-final marker yeˇ

33

the resultative evolving into the perfect is the high frequency of its occurrence with verbs of perception and communication. Using the Old English word gehyred ‘heard’ as an example, Carry (1995: 90–91) also demonstrates that the subject of gehyred is not in the ‘‘heard’’ state – rather it is in an experiential/knowledgable state as the result of an anterior perception event. The perception event implies, rather than entails, the resultant outcome, which is cancellable and therefore shows the current relevance. This can be described as the use of the perfect in its early stage of grammaticalization. In (21), the action of we´n 闻 ‘hear’ has a great impact on the subject who performed the action of we´n 闻 ‘hear’. In (22), the action of we´n 闻 ‘heard’ leads the subject to express his opinion on what he has heard, which shows that there is current relevance between the two events, ‘‘hear’’ and ‘‘express’’. (21) 王闻群公子之死也 , 自投于车下. . . (4th Century BC, Zuoˇzhua`n) Wa´ng we´n qu´n go#ngzi zhı # sıˇ yeˇ, King hear all son POSS die PART ` ´ ´ zı tou yu che# xia` . . . self throw to vehicle underneath ‘The King heard of the death of all his sons, then threw himself under a vehicle . . .’ (22) 仲尼闻魏子之举也 , 以为义 . . . (4th Century BC, Zuoˇzhua`n) Zho`ngnı´ we´n We`izıˇ zhı # juˇ yeˇ, Confucius hear name POSS recommendation PART yıˇwe´i yı` . . . consider righteousness ‘When Confucius heard the recommendation of Weizi, he considered it a righteous action . . .’ 2.2.4. Di¤erent verbs in the clauses with yıˇ 已 ‘already’ Di¤erent verbs can appear in the sentence with yıˇ 已 ‘already’. In (23) and (24), both the verbs zhı` 治 ‘well-governed’ and che´n 陈 ‘exhibit’ embody the stative situation. In addition, in (25), the verb we´i 为 ‘become’ embodies an abstract achievement situation. Meanwhile, in (26) and (27), the predicates bu´ da`i 不逮 ‘not reach (standard)’ and guo` 过 ‘excess’ not only indicate the change of the state, but can also be considered adjectives indicating the non-absolute state. Note that all the verbs in (23)–(27) are

34

Qianrui Chen

accompanied by yıˇ 已 ‘already’, an indication that these verbs express some sort of dynamic characteristic, such as a change of situation or an event that has already occurred.4 The example in (26) is more intriguing: We see a construction in which the use of yeˇ 也 tells us that yeˇ 也 has the stative function to modify the whole negative sentence, and it might be possible that yeˇ 也 even has the dynamic function to modify the object clause, which might be considered an example of transition from the static function to the dynamic one. Consequently, in Archaic Chinese, the semantic function of verbs in the dynamic yeˇ 也 sentence corresponds with the common path of the grammaticalization of the perfect marker in its early stage, i.e., the perfect function is compatible with the typical dynamic achievement situation, especially with verbs that express the meaning to die and to hear. The following examples provide evidence as to how meaning shifts from static to dynamic. (23) 子治天下, 天下既已治也。(before 3rd Century, Zhua#ngzıˇ 5 ) Zıˇ zhı` tia#nxia`, tia#nxia` jı` yıˇ zhı` yeˇ. You govern country country since already well-governed PART ‘Since you have been governing the country, the country has been well-governed.’ (24) 师金曰：‘‘夫刍狗之未陈也, . . . 及其已陈也 , 行者践其首脊. . .’’ (before 3rd Century, Zhua#ngzıˇ ) Shı #jın# yue#: ‘‘Fu# chu´ goˇu zhı # we`i che´n yeˇ, . . . Name say PART weed dog POSS not exhibit PART Jı´ qı´ yıˇ che´n yeˇ, xı´ngzheˇ jia`n qı´ shoˇu jıˇ . . .’’ until it already exhibit PART, passerby step its head spine ‘Shijin said: ‘‘Before, the dog made of weeds was not o¤ered as a sacrifice to the gods . . . After the exhibition ended, the dog was discarded and stepped on by the passersby . . .’’ ’

4. Jiang (2004)’s study identifies the co-occurrence of yıˇ 已 ‘already’ and yeˇ 也 from (23) to (26). 5. Although the earliest annotated version of Zhua#ngzıˇ was complied in the 3rd Century, Zhua#ngzıˇ is commonly used as a data source for Archaic Chinese.

The development of the Chinese aspectual sentence-final marker yeˇ

35

(25) 今已为物也 , 欲复归根, 不亦难乎！ (before 3rd Century, Zhua#ngzıˇ ) Jı #n yıˇ we´i wu` yeˇ, yu` fu` guı # ge#n, now already become thing PART want again return root bu´ yı` na´n hu# ! not also hard PART ‘Now you are someone who is controlled by things, isn’t it hard for you to return to your original state?’ (26) 今太子闻光壮盛之时, 不知吾形已不逮也 , 幸而教之曰 . . . (1st Century BC, Zha`nguo´ce`) Jı #n ta`izıˇ we´n Gua#ng zhua`ng she`ng zhı # shı´, now prince hear name strong flourishing POSS time bu` zhı # wu´ xı´ng yıˇ bu` da`i yeˇ, not know I body already not reach PART xı`ng e´r jia`o zhı # yue# . . . luckily CONJ teach him say ‘Now the prince only knows who I was when I was young but does not know that I have become too old to be helpful. Fortunately, I told him that . . .’ (27) 臣不亲, 百姓不宁, 则忠敬不足而富贵已过也。 (1st Century BC, Lıˇjı` ) Che´n bu` qın, # baˇixı`ng bu` nı´ng, o‰cers not close people not peaceful ze´ zho#ng jı`ng bu` zu´ because faith respect not enough e´r fu` guı` yıˇ guo` yeˇ. CONJ richness luxury already excess PART ‘O‰cers are not close to the monarch and the people do not have peace, because loyalty and respect are lacking whereas wealth and luxury are excessive.’ The following sections (2.2.5–2.2.8) focus on the function of the perfect marker including current relevance, temporal reference, and information status.

36

Qianrui Chen

2.2.5. Dynamic yeˇ 也 used in cause-result context Clauses with dynamic yeˇ 也 mostly have a cause-result relationship with the context in which they are used. In (21) and (22) the events in the yeˇ 也 clauses present the causes for the succeeding events. In comparison, the yeˇ 也 clause in (28) serves as the succeeding event and also the indirect result of the causal event. And in (29) the causal event serves as a hypothesis, something that is found infrequently in Archaic Chinese. These examples are congruent with the function of other perfect markers, which generally occur in a cause-result discourse and strengthen the cause-result relationship. ’’ (28) 孔子曰：‘‘. . .太伯之吴, 遂生源也。 (1st Century BC, Shiji) Koˇngzıˇ yue#: ‘‘. . . Ta`ibo´ zhı # Wu´, suı` she#ng Yua´n yeˇ.’’ Confucius say name go country-name so born name PART ‘Confucius said: . . . Taibo went to Wu, so Yuan was born.’ (29) 宣子曰：‘‘秦获穿也 , 获一卿矣。秦以胜归, 我何以报？’’ (4th Century BC, Zuoˇzhua`n) Xua#nzıˇ yue#: ‘‘Qı´n huo` Chua#n yeˇ, Name say country-name gain person PART huo` yı` qı #ng yıˇ. Qı´n yıˇ she`ng guı,# gain one minister PART country-name with success return woˇ he´ yıˇ ba`o? ’’ I what with repay ‘Xuanzi said: If Qin captures Chuan, it has just captured a minister. If Qin wins the war, with what should I repay it?’ 2.2.6. Dynamic yeˇ 也 clauses functioning as temporal reference Besides indicating a change in the situation and providing a cause of the succeeding clause, the yeˇ 也 clause also functions as a temporal reference introducing old information. In (30), Wuˇ ho´u zu´ 武侯卒 ‘Wuhou died’ is new information, so yeˇ 也 is not attached to the end of this phrase, but when this event is mentioned again, it becomes old information, thus the yeˇ 也 in Wuˇho´u zu´ yeˇ 武侯卒也 ‘Wuhou has died’ is attached to refer back to the event. Such a pattern becomes more apparent in (31). The two clauses (Wu´wa´ng naˇi yuˇ qı´ huıxia # ` zhua`ngshı` shu` qia#n re´n ye` wa´ng qu` 吴王乃与其麾下壮士数千人夜亡去 ‘Wuwang fled with his commenders and soldiers’ and Wu´wa´ng zhı # qı` qı´ ju#n wa´ng yeˇ 吴王之弃其军亡也

The development of the Chinese aspectual sentence-final marker yeˇ

37

‘Wuwang abandoned his army and fled’ refer to the same event. The first clause describes the event of fleeing (new information) without yeˇ 也; when the second clause mentions the event again, yeˇ 也 is attached to the end of the clause because it is now old information. Notice that the fleeing event in the second clause serves as a temporal reference and the cause of the event in the following clause ‘‘his army collapsed’’. (30) 武侯卒, 子罃立, 是为惠王。惠王元年, 初, 武侯卒也 , 子罃与公中缓争为太子。(1st Century BC, Shıˇjı`) Wuˇho´u zu´, Zıˇyıng # lı`, shı` we´i Hu`iwa´ng. king-name die name enthrone this be king-name Huı`wa´ng yua´n nia´n, chu#, Wuˇho´u zu´ yeˇ, King-name first year early king-name die PART Zıˇyıng # yuˇ Go#ngzho`nghuaˇn zhe#ng we´i ta`izıˇ. name and name compete be crown-prince ‘Wuhou died. His son Ziying ascended the throne and became Hui wang. In the first year of Huiwang’s rule, when Wuhou died, Ziying competed with Gongzhonghuan to be the crown prince.’ (31) 于是吴王乃与其麾下壮士数千人夜亡去, . . .吴王之弃其 军亡也 , 军遂溃。(1st Century BC, Shıˇjı` ) Yu´shı` Wu´wa´ng naˇi yuˇ qı´ huıxia # ` zhua`ngshı` So king-name then with its command soldier shu` qia#n re´n ye` wa´ng qu`, . . . several thousand person night escape go Wu´wa´ng zhı # qı` qı´ ju#n wa´ng yeˇ, king-name POSS abandon its army flee PART ju#n suı` kuı`. army then collapsed ‘Therefore Wuwang and a few thousand of his soldiers fled during the night. . . When Wuwang abandoned his army and fled, his army collapsed.’ 2.2.7. Dynamic yeˇ 也 clauses introducing new information This study also finds some clauses in which dynamic yeˇ 也 introduces new information, although such examples are not common. For instance, the yeˇ 也 clauses in (32)–(34) narrate sequential historical events with ‘‘fled’’, ‘‘commit suicide’’ and ‘‘enter’’ as new information. Apparently this type of

38

Qianrui Chen

yeˇ 也 is quite di¤erent from the dynamic particle yıˇ 矣. Yıˇ 矣 can be used to indicate an event that happened a long time ago and then confirm its current relevance. Yeˇ 也 in (32)–(34), on the other hand, introduces new information, similar to sentence final le了 indicating a change of situation in Modern Chinese. Nonetheless, this is not a typical characteristic of the perfect marker, which usually appears in the latest stage of the development of a perfect marker and is closer to the function of a perfective marker.6 (32) 问其名居, 不告而退, 遂自亡也。(4th Century BC, Zuoˇzhua`n) We`n qı´ mı´ng ju#, bu´ ga`o e´r tuı`, Ask it name location not tell and leave suı` zı` wa´ng yeˇ. then self escape PART ‘(Zhaodun) asked for his name and his hometown, but he left without getting a reply and then fled.’ (33) 子恶闻之, 遂自杀也。(4th Century BC, Zuoˇzhua`n) Zıˇ ’e` we´n zhı,# suı` zı`sha# yeˇ. Name hear it so suicide PART ‘Zi’e heard it and then committed suicide.’ (34) 欲以激励应侯。应侯惧, 不知所出。蔡泽闻之, 往入秦也。 (1st Century BC, Shıˇjı` ) Yu` yıˇ jı #lı` Yı`ngho´u. Yı`ngho´u ju`, bu` zhı # suoˇ chu#. Want with stimulate name name scared not know where out Ca`ize´ we´n zhı,# waˇng ru` Qı´n yeˇ. Name hear it go enter country-name PART ‘(Zhaowang) wanted to motivate Yinghou, but Yinghou was scared and didn’t know what to do. Caize heard it, and then went into Qin.’

6. Givo´n (2001: 293–297) argues that anterior is the most complicated and subtle aspect. It has the following four characteristics: anteriority, perfectivity, lingering relevance, and sequentiality. Lingering relevance is equivalent to current relevance. Perfectivity includes the aforementioned properties of dynamic state and change. Anteriority refers to the function of providing a timeline and signifies that an event has happened before a certain point in time. Sequentiality refers to whether a description of an event follows the timeline of that event.

The development of the Chinese aspectual sentence-final marker yeˇ

39

2.2.8. Yeˇ 也 used with adverbs indicating future time Yeˇ 也 is also used with adverbs indicating future time like jia#ng 将 ‘be going to’ and da`i 殆 ‘almost’ as shown in (35) and (36) taken from Archaic Chinese sources. This use of yeˇ 也 is di¤erent from the typical dynamic and static uses and was later to independently develop a future function in Early Modern Chinese. Luo (1994) has pointed out that yeˇ 也 was used to refer to changes in the future in Early Modern Chinese, and Jiang (2004) also mentioned that the same use of yeˇ 也 appeared in Medieval Chinese. (35) 子贡闻之曰：‘‘. . . 夫子殆将病也。’’ (1st Century BC, Lıˇjı`i) Zıˇgo`ng we´n zhı # yue#: ‘‘. . . Fu#zi da`i jia#ng bı`ng yeˇ.’’ Name hear it say Confucius probably will be-sick PART ‘Zigong heard it and said: Confucius is going to fall ill.’ (36) 夫子曰：‘‘. . .予殆将死也。’’ (1st Century BC, Lıˇjı`i) Fu#zi yue#: ‘‘. . .Yu´ da`i jia#ng sıˇ yeˇ.’’ Confucius said I probably will die PART ‘Confucius said: . . . I am probably going to die.’ As mentioned above, it is reasonable to assume that dynamic yeˇ 也 engendered the perfect function because it underwent the same process of grammaticalization as the development of the perfect marker in English and other languages. 3. The development of dynamic yeˇ 也 in Medieval Chinese and Early Modern Chinese 3.1. The use of dynamic yeˇ 也 in Medieval Chinese (from 1st Century to 6th Century) In this section, the use of dynamic yeˇ 也 in three Medieval Chinese texts is explored. Wei (2002) found at least five sentences using dynamic yeˇ 也 in a Buddhist scripture entitled Fo´beˇn Xı´ngjı´jı #ng 佛本行集经 ‘The Collective Sutra of the Buddha’s Past Acts’ translated in the Sui Dynasty (557–581). The present author collected 11 sentences in So#ushe´njı` 搜神记 ‘Anecdotes about Spirits and Immortals’7 from the Eastern Jin Dynasty (317–420) and 7. Even though it has been mixed with many new words developed in later history, So#ushe´njı` has been widely used as a data source and thus still has a comparative value. In So#ushe´njı`, Long (2004: 165) found 12 sentences of the dynamic use of yeˇ 也, Jiang (2004) found 22 sentences.

40

Qianrui Chen

at least 15 sentences in Lu`nhe´ng 论衡 ‘Critical Discussions’ from the Eastern Han Dynasty (25–220). Compared to the use of dynamic yeˇ in Archaic Chinese, the frequency of dynamic yeˇ 也 in Medieval Chinese shows an increase but still remains sparse. In the text of Lu`nhe´ng, we have found some new characteristics of the usage of yeˇ 也 which may be of developmental significance. First, yeˇ 也 is used in yoˇu 有 ‘have’ sentences to express the absolute stative situation, as in (37) and (38). Second, yeˇ 也 is used with verbs like xı´ng 行 ‘move’ as in (39) to express an activity situation. Third, yeˇ 也 is frequently used with yıˇ 已 ‘already’ as in (38) and (39).8 (37) 陈仲子. . . 三日不食, 耳无闻, 目无见也。. . . 三咽然后 耳有闻, 目有见也。(1st Century, Lu`nhe´ng) Che´nzho`ngzıˇ . . . sa#n rı` bu` shı´, eˇr wu´ we´n, Name three day not eat ear not hear mu` wu´ jia`n yeˇ . . . Sa#n ya`n ra´nho`u eˇr yoˇu we´n, eye not see PART Three gulp then ear have hear mu` yoˇu jia`n yeˇ.’’ eye have see PART ‘Chenzhongzi . . . starved himself for three days, and then he could neither hear, nor see. . . . After eating three plums he got back to his senses of hearing and vision.’ (38) 《尚书》曰 ‘‘诗言志, 歌咏言’’, 此时已有诗也。 (1st Century, Lu`nhe´ng) Sha`ngshu# yue# ‘‘Shı # ya´n zhı`, ge# yoˇng ya´n’’, Book-name say poem express ideal song sing word cıˇ shı´ yıˇ yoˇu shı # yeˇ. that time already have poem PART ‘In Shangshu it said: ‘‘To compose a poem is to show ambition and [to compose] a song is to show feeling.’’ At that time, poems existed already.’ (39) 列星着天, 天已行也。(1st Century, Lu`nhe´ng) Lie` xıng # zhuo´ tia#n, tia#n yıˇ xı´ng yeˇ. Arranged star cling-to sky sky already move PART ‘When the stars are in the sky, the sky has been moving.’ 8. Jiang (2004) found seven sentences in the texts of Lu`nhe´ng where yeˇ 也 and yıˇ 已 ‘already’ co-exist.

The development of the Chinese aspectual sentence-final marker yeˇ

41

The examples of dynamic yeˇ 也 collected in So#ushe´njı` appear in absolute stative situations, like wu´fu` 无复 ‘no longer have’ in (40), and in stative situations, like keˇ zhı # 可知 ‘be able to know’ in (41), which includes both an auxiliary and a mental verb. The dynamic meaming of yeˇ 也 in those sentences is made evident in contrast to the antecedent clause. (40) 虽强语笑, 无复气味也。(3rd Century, So#ushe´njı` ) Suı # qiaˇng yuˇ xia`o, wu´fu` qı`we`i yeˇ. Although force speak smile no-longer-have taste PART ‘Even though she forced herself to smile, she had lost her style.’ (41) 或曰：‘‘投女则船为进。’’ 皆曰：‘‘神意已可知也。. . .’’ (3rd Century, So#ushe´njı`i) Huo` yue#: ‘‘To´u nuˇ¨ ze´ chua´n we´i jı`n.’’ Some say throw women so boat be go Jie# yue#: ‘‘She´n yı` yıˇ keˇ zhı # yeˇ. . .’’ all say god intention already can know PART ‘Somebody suggested: Let’s throw women in the river, so that the ship can get through. Others said: We already know God’s intention.’ In Fo´beˇn Xı´ngjı´jıng, # dynamic yeˇ 也 is found to be used with new types of verbs in di¤erent situations. For instance, qıˇng 请 ‘invite’ in (42) is used with yeˇ 也 in an activity situation and zhı # 知 ‘know’ in (43) in a stative situation. (42) 彼等问言：‘‘大德沙门, 是谁所请？’’ 佛即报言： ‘‘汝辈和上, 已请我也。’’ (6th Century, Fo´beˇn Xı´ngjı´jıng) # Bıˇdeˇng we`n ya´n: ‘‘Da` de´ sha#me´n, shı` she´i suoˇ qıˇng?’’ they ask speak big virtue monk is who PART invite Fo´ jı´ ba`o ya´n: Buddha at-once report say ‘‘Ruˇ be`i he´sha`ng, yıˇ qıˇng woˇ yeˇ.’’ you fellow monk already invite me PART ‘They asked: ‘‘Who invited the eminent monk?’’ Buddha answered: ‘‘The monks have already invited me.’’ ’ # (43) 世尊已知我名也。(6th Century, Fo´beˇn Xı´ngjı´jıng) Shı`zu#n yıˇ zhı # woˇ mı´ng yeˇ. honored-person already know my name PART ‘The honored person already knew my name.’

42

Qianrui Chen

It appears that dynamic yeˇ 也 had existed at least by the time of Medieval Chinese (from 1st Century to 6th Century). Although some slight changes occurred to the situation-type dynamic yeˇ 也, aspectual yeˇ 也 is still the same as it was in Archaic Chinese due to the lack of the new types of verbs mentioned above, and the low frequency of dynamic yeˇ 也. There is no remarkable di¤erence between dynamic yeˇ 也 in Chinese traditional secular documents and Chinese versions of Buddhist scriptures in Medieval Chinese, although the frequency of yeˇ 也 in the Chinese traditional secular documents is higher. 3.2. The use of dynamic yeˇ 也 in Early Modern Chinese. (from 7th Century to 18th Century) This section explores the use of dynamic yeˇ 也 in Zuˇta´ngjı´ 祖堂集 ‘Collection of Buddhist Texts’ in Early Modern Chinese. Several studies (including this paper) have counted the frequency of dynamic yeˇ 也 in Zuˇta´ngjı´. The results are displayed in Table 1. Table 1. Frequencies of dynamic yeˇ 也 in Zuˇta´ngjı´ Studies

Number of dynamic yeˇs 也

Wei (2002: 495)

255

Long (2004: 165) Jiang (2004)

234 312

This Paper

227

The di¤erent frequencies of yeˇ 也 collected from the same text, Zuˇta´ngjı´, reflect di¤erent understandings of yeˇ ’s 也 dynamic usage. For example, both Wei (2002: 510) and Long (2004: 165) deem sentences (44) and (45) to be expressing the dynamic usage of yeˇ 也, but this paper treats yeˇ 也 in these two sentences as static since they are direct answers to questions about age. Only the yeˇ 也 in (46) is considered dynamic, based on the temporal reference poin jınnia # ´ n 今年 ‘this year’ located in the result clause connecting the result from the past to the time of speaking. Hence, care should be taken when considering the use of dynamic yeˇ 也 in a context where stative yeˇ 也 plays a dominant role.9 9. Ohta (1988: 151) thinks that the most frequent use of yeˇ 也 is the stative yeˇ 也.

The development of the Chinese aspectual sentence-final marker yeˇ

43

(44) 师问僧：‘‘此水牯牛年多少？’’ 僧无对。师云：‘‘七十七也。’’ (9th Century, Zuˇta´ngjı´ ) Shı # we`n se#ng: ‘‘Cıˇ shuıˇguˇniu´ nia´n duo#shaˇo?’’ Master ask monk this water-bu¤alo year how-many Se#ng wu´ duı`. Shı # yu´n: ‘‘Qı #shı´qı # yeˇ.’’ monk not answer master say seventy-seven PART ‘The Master asked the monk: ‘‘How old is this water bu¤alo?’’ The monk could not tell. The Master told him: ‘‘Seventy-seven years old.’’ ’ (45) 师乃问毱多曰：‘‘汝年几岁耶？’’ 子曰：‘‘年十七岁也。’’ (9th Century, Zuˇta´ngjı´ ) Shı # naˇi we`n Ju´duo# yue#: ‘‘Ruˇ nia´n jıˇsuı` ye# ? ’’ Master so-then ask name say you year how-old PART Zıˇ yue#: ‘‘Nia´n shı´qı # suı` yeˇ.’’ He say year seventeen year PART ‘The Master then asked Juduo: ‘‘How old are you?’’ He answered: ‘‘I am seventeen.’’ ’ (46) 师云：‘‘那个师僧若在, 今年七十四也。’’ (9th Century, Zuˇta´ngjı´ ) Shı # yu´n: ‘‘Na` ge` shı #se#ng ruo` za`i, Master say that classifier monk if alive jı #n nia´n qı #shı´sı` yeˇ.’’ this year seventy-four PART ‘The Master said: ‘‘If that monk is still alive, he should be seventyfour years old this year.’’ ’ Inspired by the theories of prototype and grammaticalization, this paper proposes that the prototypical uses of dynamic and static yeˇ 也 are quite di¤erent, and the non-prototypical uses are hard to determine and may have di¤erent readings. The following part discusses the use of dynamic yeˇ 也 in Zuˇta´ngjı´ and its gramaticalization into a perfect marker.

44

Qianrui Chen

3.2.1. Dynamic yeˇ 也 used with di¤erent types of verbs In the Zuˇta´ngjı´, dynamic yeˇ 也 is used with many types of verbs including those expressing achievements, accomplishments or activities, like daˇ woˇ 打我 ‘hit me’ and daˇ zho#ng 打钟 ‘chime bell’ in (47), mental stative verbs, like shı´ 识 ‘know’, jieˇ 解 ‘understand’ and huı` 会 ‘can’ in (48), as well as non-absolute stative verbs, like da`nbo´ 淡泊 ‘low desire’ in (49) and absolute stative verbs, like yoˇu 有 ‘have’ in (50). It is also notable that yeˇ 也, when used at the end of a typical judgment sentence, conveys the meaning of a change of state, like shı` 是 ‘be’ in (51). (47) 院主报和尚：‘‘打钟也 , 请和尚上堂。’’ (9th Century, Zuˇta´ngjı´ ) Yua`nzhuˇ ba`o he´sha`ng: ‘‘Daˇ zho#ng yeˇ, Principle report monk hit bell PART qıˇng he´sha`ng sha`ng ta´ng.’’ please monk enter temple ‘The principal told the monk: ‘‘The bell has rung, will you please enter the temple?’’ ’ (48) 云：‘‘汝还知不？’’ 师云：‘‘我会也 , 我会也。’’ (9th Century, Zuˇta´ngjı´ ) Yu´n: ‘‘Ruˇ ha´i zhı # bu`?’’ Shı # yu´n: Say you still know not master say ‘‘Woˇ huı` yeˇ, woˇ huı` yeˇ.’’ I know PART I know PART ‘. . . asked: ‘‘Do you know it now?’’ The Master answered: ‘‘I know it now, I know it now.’’ ’ (49) 仰山云：‘‘. . . 仰山今时, 早已淡泊也。’’ (9th Century, Zuˇta´ngjı´ ) Yaˇngsha#n yu´n: ‘‘. . .Yaˇngsha#n jın# shı´, Name say name present time zaˇo yıˇ da`nbo´ yeˇ.’’ early already low-desire PART ‘Yangshan said: ‘‘. . . I do not have much desire now.’’ ’ (50) 座主曰：‘‘与摩则有弟二月也。’’ (9th Century, Zuˇta´ngjı´ ) Zuo`zhuˇ yue#: ‘‘Yuˇmo´ ze´ yoˇu dı` e`r yue` yeˇ.’’ Examiner say this so has younger-brother two month PART ‘The host said: ‘‘If so, you have had a younger brother for two months.’’ ’

The development of the Chinese aspectual sentence-final marker yeˇ

45

(51) 众生若是利根者, 忽闻真教, 顿晓性地, 便是智人也。 (9th Century, Zuˇta´ngjı´ ) Zo`ngshe#ng ruo` shı` lı`ge#n zheˇ, hu# we´n zhe#n jia`o, All-people if is smart PART sudden hear true religion du`n xiaˇo xı`ngdı`, bia`n shı` zhı` re´n yeˇ.’’ instantly understand truth so is wise person PART ‘If the commoners are smart and have a chance to accept the truth, they will understand it instantly. Those people have become wise people.’ 3.2.2. The current relevance of dynamic yeˇ 也 clause For the present study over 100 instances of dynamic yeˇ 也 were collected that occurred in conditional sentences from Zuˇta´ngjı´, which constitutes the main part among the types of current relevance expressed by the dynamic yeˇ 也 clause. Besides cause-result relationships in conditional sentences, two other types of currently relevant instances are found. One type is illustrated in (52), where yeˇ 也 is used to report new information. The other is demonstrated in (53), where yeˇ 也 indicates the persistent situation of the action of the verb. Note that yeˇ 也 in the pattern la´i . . . yeˇ 来 . . . 也 ‘have been . . . for’ expresses an aspectual meaning, similar to the function of the aspectual sentence-final le了. These two types of currently relevant cases are among the main types of the perfect aspect in typology (Schwenter 1994). (52) 停腾之间, 更有一人来报和尚云：‘‘适来迁化僧却来也。’’ (9th Century, Zuˇta´ngjı´ ) Tı´ngte´ng zhıjia# # n, ge`ng yoˇu yı` re´n la´i Break between another have one person come ba`o he´sha`ng yu´n: ‘‘Shı`la´i qia#nhua` se#ng que` la´i yeˇ.’’ report monk say just die monk again come PART ‘While they were waiting, someone came to tell the monk: ‘‘The monk who just died came back again.’’ ’ (53) 隔阙来多少时也 ? (9th Century, Zuˇta´ngjı´ ) Ge´que# la´i duo#shaˇo shı´ yeˇ ? Separate PART how-many time PART ‘How long has the separation lasted?’

46

Qianrui Chen

3.2.3. Yeˇ 也 indicating the immediate future Ohta ([1988] 1991: 150) points out that in Zuˇta´ngjı´, yeˇ 也 can express the meaning of beginning or be used with a future time phrase as in the result clause of a conditional sentence to express future perfect meaning. It should be noted that some of the instances of yeˇ 也 mentioned by Ohta actually indicate the immediate future, as in (54), the meaning of farewell as in (55), and the meaning of urgency or reminder as in (56). In none of these three examples did we find a future time phrase in the yeˇ 也 clause, yet the events expressed by the yeˇ 也 clauses are clearly about to happen. Therefore, we conclude that this is the use of yeˇ 也 indicating the immediate future, a function of the anterior marker at the later stage of its grammaticalization. (54) 三人不敢倾茶。女云：‘‘看老婆呈神通去也。 ’’ 拈起盏子, 便泄行茶。(9th Century, Zuˇta´ngjı´ ) Sa#n re´n bu` gaˇn qıng # cha´. Nuˇ¨ yu´n: Three person not dare pour tea woman say ‘‘Ka`n laˇopo´ che´ng she´nto#ng qu` yeˇ.’’ look woman show magic-power go PART Nia#nqıˇ zhaˇnzi, bia`n xie` xı´ng cha´. pick-up teacup then pour pass tea ‘The three dared not pour the tea. The woman said, ‘‘Let a woman show you the magic power.’’ So she picked up the teacup and poured the tea for the guests.’ ’’ (55) 师却云：‘‘. . . 归去也！归去也！呵呵珍重！ (9th Century, Zuˇta´ngjı´ ) Shı # que` yu´n: ‘‘. . . Guı # qu` yeˇ ! guı # qu` yeˇ ! Master but say return go PART return go PART He#he# zhe#nzho`ng!’’ laugh take-care ‘The Master said: ‘‘. . . I’m going to go. I’m going to go. Haha, Best regards!’’ ’ ’’ . . . 师遂告寂。 (56) 师 . . . 告众曰：‘‘法堂倒也, 法堂倒也。 (9th Century, Zuˇta´ngjı´ ) Shı .# . . ga`o zho`ng yue#: ‘‘. . . Faˇta´ng daˇo yeˇ, Master tell people say temple fall-down PART

The development of the Chinese aspectual sentence-final marker yeˇ

47

faˇta´ng daˇo yeˇ ’’. . . . Shı # suı` ga`ojı`. Temple fall-down PART master then die ‘The Master told the others: ‘‘. . . The temple is falling down, the temple is falling down.’’ . . . Then he died.’ 3.2.4. Yeˇ 也 co-occurring with ta`i 太 ‘too much’ Wei (2002) cited (57) and (58) as examples of yeˇ 也 co-occurring with an adverb of degree like ta`i 太 ‘too much’, but the present author regards these instances of yeˇ 也 as indicating mood rather than aspect. Apparently, the evolution of mood from aspect or tense is a general and subjective process. (cf. Lamarre 2003) (57) 师. . .云：‘‘太多也。’’ (9th Century, Zuˇta´ngjı´ ) Shı .# . . yu´n: ‘‘Ta`i duo# yeˇ.’’ Master say too much PART ‘The Master . . . said: ‘‘Too much.’’ ’ (58) 洞山云：‘‘太迟也。’’ (9th Century, Zuˇta´ngjı´ ) Do`ngsha#n yu´n: ‘‘Ta`i chı´ yeˇ.’’ Name say too late PART ‘Dongyun said: ‘‘It is too late.’’ ’ Bybee, Perkins, and Pagliuca (1994: 64) defined the ‘‘young’’ anterior as the gram that has only one type of anterior in the early stages of grammaticalization. They (1994: 80) listed other uses of the anterior such as future anterior, past anterior, immediate anterior, remote anterior, completive, state commence and evidentials and termed them ‘‘old’’ anteriors because they are older than the anterior that has only one type of semantic function. From the aforementioned analysis of dynamic yeˇ 也 in Zuˇta´ngjı´, we find that the uses of the present and future anterior, the immediate future and mood all exist in Zuˇta´ngjı´, which seems to suggest that dynamic yeˇ 也 in Zuˇta´ngjı´ was no longer a young anterior but rather had become an old anterior. 4. The typological significance of yeˇ 也 as an aspectual particle 4.1. The systematic meaning of static yeˇ 也 In this paper, the static use of yeˇ 也 in Archaic Chinese is categorized as stative or resultative in a broad sense, integrating Chinese aspectual

48

Qianrui Chen

research and research on aspectual typology. Some researchers, such as Smith (1991: 38), have also suggested the term ‘situation aspect’ for situation type, as contrasted to ‘‘viewpoint aspect’’ or ‘‘grammatical aspect’’. However, it is unusual to categorize a specific stative situation as stative. In fact, there are subtle di¤erences between ‘‘stative situation’’ and ‘‘stative’’, in that the former is a semantic type of predicate and its key arguments, while the latter is a grammatical device for marking this type of semantic property. Nedjalkov and Nedjalkov (1988: 241) define ‘‘stative’’ as ‘‘the aspect of state’’ when they introduced a stative form into the description of Evenki. Using a similar terminology, the term ‘‘stative’’ as describing ‘‘aspect of state’’ has been common in the Russian linguistic tradition. Chinese lacks a similar morphological variation. Therefore, the present study uses ‘‘stative’’ to refer to its aspectual meaning. Chen (2003) categorizes ingressive aspect, successive aspect, completive aspect, and resultative aspect as basic phrasal aspects. With this understanding, stative, as part of the resultative in a broad sense, belongs to the category of basic phrasal aspects indicating a continuing state. Using termination as a criterion, Chen (2003) also classifies basic phrasal aspects into two di¤erent types: the unbounded type (e.g. successive aspect) and the bounded type (e.g. ingressive aspect, completive, and resultative). We argue here that a specific analysis of the termination for the resultative is necessary. A resultative in the narrow sense with a previous action implied may be bounded while a stative may be unbounded. Sil’nickij (1988: 98) emphasizes that a stative form can be regarded as resultative in the broad sense in a particular language only if it is related to a corresponding lexical source. A stative resultative must be defined as a lexical derivative of its lexical source. Based on this analysis, static yeˇ 也 in Chinese is not a derivative, even though it can be traced back to its function as a judgment particle with the property of a predicate. This is because the extension and semantic shift of yeˇ 也 from ‘‘the mood of judgment’’ to ‘‘the mood of stressing a stative situation’’ removed the property of predicate from its judgmental function. The contrast between sentence-final static yeˇ 也 and dynamic yıˇ 矣 in Archaic Chinese reflects the one between the resultative and the anterior synchronically. This type of contrast did not receive a detailed description and explanation in the study of aspectual typology. Anterior is usually contrasted with progressive or perfective while resultative usually evolves into anterior. In a sense, the contrast between anterior and resultative is similar to the contrast between a ‘‘father’’ and a ‘‘son’’. Moreover, the

The development of the Chinese aspectual sentence-final marker yeˇ

49

‘‘son’’ even partly replaces the ‘‘father’’ in expressing perfect function during Medieval Chinese. According to Comrie (1976: 25), the continuous aspect can be divided into two types: progressive and non-progressive. From this point of view, yeˇ 也 should be categorized as non-progressive, in contrast to perfect or perfective. Comrie’s view had a considerable impact on the analysis of aspect. However, his view has been questioned by some researchers. For example, Bybee, Perkins & Pagliuca (1994: 139) have pointed out that Comrie does not clearly di¤erentiate the continuous marker from the non-progressive marker. In fact, Comrie’s theory does not stipulate that every type has its own corresponding gram-type, especially since the di¤erentiation between continuous and progressive is not found in crosslinguistic analyses. Therefore, Comrie’s theory has both pros and cons. On the one hand, it allows for categorization of static yeˇ 也 as a non-progressive marker. On the other hand, it does not provide an explanation for the evolution of yeˇ 也 from resultative to perfect, a drawback, that can be well explained by Nedjalkov and Jaxontov’s (1988) and Bybee, Perkins & Pagliuca’s (1994: 68) frameworks. In this case, it may be more appropriate to define yeˇ 也 as a stative or a resultative in a broad sense. In the study of the resultative, most research in Nedjalkov (1988) focuses on the resultative in a narrow sense. There is only limited discussion and description of the stative. In part because of this, the theoretical value of describing and studying yeˇ 也 in the Chinese language is especially important. 4.2. The typological importance of the development of dynamic yeˇ 也 The study of the development of yeˇ 也 from static use to dynamic use is important for at least two reasons. Firstly, the evolution of yeˇ 也 in general supports Bybee, Perkins & Pagliuca’s Source Determination Hypothesis (1994: 51), viz. that grams from stative verb sources evolve along di¤erent paths in contrast to those originating from dynamic verbs. In particular, auxiliaries from stative verbs go through a resultative stage while auxiliaries from dynamic verbs go through a stage of signaling completive or anterior aspects before becoming past and perfective. Secondly, yeˇ 也 has gone through a development from young anterior to old anterior in the course of its evolution. In Zuˇta´ngjı´, anteriors (both young and old) or perfect markers are found. The dynamic particle yıˇ 矣 appears in 165 cases, most of which are used in paragraphs describing

50

Qianrui Chen

Buddhists’ biographic experience; only 40 or so instances appear in conversations or citations (Wei 2002: 534)10. Therefore, yıˇ 矣 can be considered a first generation perfect. Dynamic yeˇ 也 appears 227 times in the same text, serving as the second generation of perfect. Among the 227 instances, 42 are the bisyllabic dynamic particle leyeˇ 了也 serving as the third generation. In Zuˇta´ngjı´, we also found 77 examples of le 了 used at the end of sentences, which serve as the fourth generation. This study considers le了 and leyeˇ 了也 in Zuˇta´ngjı´ young anterior. Among the four generations of the anterior or perfect marker, the latter three mostly appear in conversations and citations. Additionally, the second generation of yeˇ 也 has been used more often as a perfect marker than any of the other three generations. These results are useful for the investigation of the function and evolution of aspectual markers, and even for the study of language evolution in general.

5. Conclusion The present paper analyzes two hypotheses about the relationship between static and dynamic yeˇ 也. It has been argued that the Sound-record Hypothesis is not supported by su‰cient phonetic and dialectical evidence, while the Extension Hypothesis does not provide a complete explanation for the functional development of yeˇ 也, and does not relate yeˇ 也 to its ancient use when searching for dialectical evidence. Based on the grammaticalization path of the aspectual marker, this paper argues that dynamic yeˇ 也 is the result of the grammaticalization of yeˇ 也 as a stative element. In many respects, such as adaptation to situation types, the development of current relevance, temporal reference, and situation meaning, the use of dynamic yeˇ 也 conforms with the grammaticalization path of the perfect or anterior both in Chinese and other languages. Therefore, the evolution of yeˇ 也 supports the Extension Hypothesis from three perspectives: diachronic development, typology and function. However, when it comes to the Extension Hypothesis, there are still some important theoretical questions awaiting future research. For example, while both yeˇ 也 and yıˇ 矣 indicate the dynamic state, what are the di¤erences between the function of yeˇ 也 and yıˇ 矣 in Archaic Chinese? 10. In these sentences, not all yıˇ s 矣 are used as anteriors. Some are used as interjections.

The development of the Chinese aspectual sentence-final marker yeˇ

51

In some cases, such as in historical event narration, yeˇ 也 cannot be replaced by yıˇ 矣. When dynamic yeˇ 也 extends or increases its function, theoretically the use of yıˇ 矣 should be decreased. If this is the case, how do the two interact in the process of transition? Are the functional and evolutional di¤erences between yeˇ 也 and yıˇ 矣 related to their di¤erent sources? If so, how is this relationship related to the Source Determination Hypothesis?11 This study shows that typological studies of aspect markers may illuminate our analysis of the developmental relationship between the two uses of yeˇ 也. The result of our analysis provides evidence for the status and developmental path of yeˇ 也 and at the same time raises meaningful questions for future research into aspectual typology. Since Chinese is a language with relatively prominent aspectual expressions, abundant historical texts and various dialects, we hope that this study can shed some light on cross-linguistic research in the domain of aspectual typology.

References Bybee, Joan, Revere Perkins, and William Pagliuca 1994 The Evolution of Grammar: Tense, Aspect, and Modality in the Languages of the World. Chicago: The University of Chicago Press. Cao, Guangshun 1987 Yuˇqı`cı´ ‘‘le’’ yua´nliu´ qiaˇnshuo# [Analysis on the origin of the particle le]. Yuˇwe´n Ya´njiu# [Language Research] (2), 10–15. Cao, Guangshun 1995 Jı`nda`i Ha`nyuˇ Zhu`cı´ [The Particles in Early Modern Chinese Language]. Beijing: Yuwen Chubanshe. Carey, Kathleen 1995 Subjectification and the development of the English perfect. In Subjectivity and Subjectivisation: Linguistic Perspectives, Dieter Stein, and Susan Wright (eds.), 83–102. Cambridge/New York: Cambridge University Press. Chen, Qianrui 2003 Ha`nyuˇ tıˇma`o xı`toˇng ya´njiu# [A study of the Chinese aspectual system]. Doctoral dissertation, Huazhong Normal University.

11. These problems were pointed out by Professor Shaoyu Jiang when the author presented this paper at the Center for Chinese Linguistics at Peking University.

52

Qianrui Chen

Comrie, Bernard 1976 Aspect: An Introduction to the Study of Verbal Aspect and Related Problems. Cambridge: Cambridge University Press. Dai, Zhaoming 2006 Co´ng ‘‘yeˇ’’ da`o ‘‘a#’’ [From yeˇ to a# ]. Zho#ngguo´ Yuˇya´n Xue´ba`o [Chinese Linguistcs] 12, 140–152. Beijing: Shangwu Yingshuguan. ¨ sten Dahl, O 2000 Tense-aspect systems of European languages in a typological perspective. In Tense and Aspect in the Languages of Europe, ¨ sten Dahl (ed.), 3–25. Berlin/New York: Mouton de Gruyter. O Fang, Yixing 2004 Co´ng zho#ngguˇ cı´huı` de te`diaˇn ka`n ha`nyuˇshıˇ de fe#nqı # [Determining the stage development of Chinese language based on the characteristics of vocabulary in Medieval Chinese]. Ha`nyuˇshıˇ Xue´ba`o [Chinese Language History Journal ] 4, 178–184. ShangHai: Shanghai Jiaoyu Chubanshe. Givo´n, Talmy 2001 Syntax: An Introduction. Vol. 1. Amsterdam/Philadelphia: John Benjamins. He, Baozhang 1992 Situation types and aspectual classes of verbs in Mandarin. Doctoral dissertation, Ohio State University. He, Leshi 2004 Zuoˇzhua`n de yuˇqı`cı´ ‘‘yeˇ’’ [Modal particles yeˇ in Zuoˇzhua`n]. In Zuoˇzhua`n Xu#c´ı Ya´njiu# [Function Words in Zuoˇzhua`n] (revised edition), 412–446. Beijing: Shangwu Yinchuguan. Originally in Yuˇya´nxue´ Lu`nco´ng [Language Research Forum] 16, 1986. Jiang, Nan 2004 Guˇha`nyuˇ zho#ng biaˇo che´nshu` ju`weˇi yuˇqı`cı´ de yaˇnbia`n [The development of declarative particles for statement in Ancient Chinese]. Master thesis, Peking University. Lamarre, Christine 2003 Shı`lu`n we`icı´ de yuˇyı` te`zhe#ng he´ yuˇfaˇhua` de gua#nxı` [On semantic features of predicates and grammaticalization]. In Yuˇfaˇhua` yuˇ Yuˇfaˇ Yaˇnjıu# (1) [Grammaticalization and Grammar Research (Vol. 1)], Fuxiang Wu, and Bo Hong (eds.), 2–242. Beijing: Shangwu Yinshuguan. Li, Charles N., Sandra A. Thompson, and R. McMillan. Thompson 1982 The discourse motivation for the perfect aspect: The Mandarin Chinese particle LE. In Tense and Aspect: Between Semantics and Pragmatics, Paul Hopper (ed.), 19–44. Amsterdam: John Benjamins. Long, Guofu 2004 Ya´oqı´n Yı´jıng # Zhu`cı´ Ya´njiu# [Research on the Clitics in the Chinese Versions of Buddhist Scriptures in the Yaoqin Dynasty]. Changsha: Hunan Normal University Press.

The development of the Chinese aspectual sentence-final marker yeˇ Lu¨, Shuxiang 1982

Lu¨, Shuxiang 2002

Ma, Jianzhong 2005

Luo, Ji 1994

53

Zho#ngguo´ We´nfaˇ Ya`olu¨e` [Chinese Grammar Summary]. Lu¨ Shuu#ia#ng We´n jı´ [Lu¨ Shuxiang Selected Work] Vol. 1. Beijing: Shangwu Yinshuguan. Original edition, 1942. We´nya´n Xu#z`ı [Functional Words in Classical Chinese]. In Lu¨ Shu#xia#ng Qua´njı´ [Lu¨ Shuxiang Collected Work], Vol. 9. Shenyang: Liaoning Jiaoyu Chubanshe. Original edition, Shanghai: Kaiming Shudian, 1944. Maˇshı` We´nto#ng. In Maˇshı` We´nto#ng Du´beˇn [A Reading Book of Mashi Wentong] (2nd edition), Shuxiang Lu¨ & Haifen Wang (eds.). Shanghai: Shanghai Shiji Chuban Jituan. Original edition, ShangHai: Shangwu Yinshuguan, 1898.

Beˇiso`ng ju`weˇi yuˇqı`cı´ ‘‘yeˇ’’ ya´njiu# [Sentence-final particle yeˇ in the Northern Song Dynasty]. Guˇha`nyuˇ Ya´njiu# [Research in Ancient Chinese Language] (4), 29–32. Nedjalkov, Vladimir P. (ed.) 1988 Typology of Resultative Constructions. Amsterdam: John Benjamins. Translated from Russian, English translation edited by Bernard Comrie. Original edition, 1983. Nedjalkov, Vladimir P., and Sergej Je. Jaxontov 1988 The typology of resultative constructions. In Typology of Resultative Constructions, Vladimir P. Nedjalkov (ed.), 3–62. Amsterdam: John Benjamins. Nedjalkov, Igor’ V., and Vladimir P. Nedjalkov 1988 Stative, resultative, passive, and perfect in Evenki. In Typology of Resultative Constructions, Vladimir P. Nedjalkov (ed.), 241– 257. Amsterdam: John Benjamins. Ohta, Tatsuo 1987 Zho#ngguo´yuˇ Lı`shıˇ We´nfaˇ [Historical Grammar of Modern Chinese]. Beijing: Peking University Press. Translated from Japanese into Chinese by Shaoyu Jiang and Changhua Xu. Original edition, Tokyo: Konan Shoten, 1958. Ohta, Tatsuo 1991 Ha`nyuˇshıˇ To#ngkaˇo [Comprehensive Studies of the History of the Chinese Language]. Translated from Japanese into Chinese by Lansheng Jiang and Weiguo Bai. Chongqin: Chongqin Chubanshe. Original edition, Tokyo: Hakuteisha, 1988. Pulleyblank, Edwin G. 2006 Guˇ Ha`nyuˇ Yuˇfaˇ Ga#ngya`o [Outline of Classical Chinese Grammar]. Translated by Sun, Jingtao. Beijing: Yuwen Chubanshe. Original edition, Vancouver: University of British Columbia Press, 1995.

54

Qianrui Chen

Pulleyblank, Edwin G. 1995 Guˇ ha`nyuˇ tıˇta`i de ge` fa#ngmia`n [Aspects of aspect in Classical Chinese]. Guˇda`i Ha`nyuˇ Ya´njiu# [Research in Ancient Chinese Language] (2), 1–13. Sil’nickij, George G. 1988 The structure of verbal meaning and the resutative. In Typology of Resultative Constructions, Vladimir P. Nedjalkov (ed.), 87–100. Amsterdam: John Benjamins. Shimura, Ryoji 1995 Zho#ngguo´ Zho#ngshı` Yuˇfaˇshıˇ Ya´njiu# [History of Grammar of Medieval Chinese]. Translated from Japanese into Chinese by Lansheng Jiang and Weiguo Bai. Beijing: Zhonghua Shuju. Original edition, 1984. Smith, Carlota S. 1991 The Parameter of Aspect. Dordrecht: Kluwer Academic Publishers. Schwenter, Scotta A. 1994 Hot news and the grammaticalization of perfects. Linguistics 33(6), 995–1028. Sun, Xixin 1999 Jı`nda`i Ha`nyuˇ Yuˇqı`cı´ [Modality Particles in Early Modern Chinese]. Bejing: Yuwen Chubanshe. Wang, Li 1989 Ha`nyuˇ Yuˇfaˇshıˇ [History of Chinese Grammar]. Beijing: Shangwu Yinshuguan. Wang, Yinzhi ` n Shı`cı´ [Explanation of Functional Words in Chinese 2000 Jıngzhua # Classical Scripture]. Nanjing: Jiangsu Guji Chubanshe. Original edition, 1798. Wei, Peiquan 2002 Zuˇta´ngjı´ zho#ng de zhu`cı´ ‘‘yeˇ ’’ – Jia#nlu`n xia`nda`i ha`nyuˇ zhu`cı´ ‘‘le’’ de la´iyua´n [The Particle yeˇ in Zuˇta´ngjı´: The discussion of the origin of the particle le in Modern Chinese]. In Ha´nzha#ng Gua#nghua`: Da`i Lia´nzha#ng Xia#nsheng Qızhı # ` Zhe´da`n Lu`nwe´n Jı´ [Collected Work for Lianzhang Dai’s 70th Anniversary], Edition group (eds.), 491–544. Taibei: Liren Shuju. Yang, Shuda 1986 Cı´qua´n [Explanation of Words]. Shanghai: Shanghai Guji Chubanshe. Original edition, ShangHai: Shangwu Yinshuguan, 1928. Yang, Xiufang 1991 Co´ng lı`shıˇ yuˇfaˇ de gua#ndiaˇn lu`n mıˇnna´nyuˇ ‘‘le’’ de yo`ngfaˇ – Jia#nlu`n wa´nche´ngma`o zhu`cı´ ‘‘yıˇ ’’ (‘‘yeˇ ’’) [On the history grammar of le and the discussion of perfect particle yıˇ (yeˇ )]. Ta´ida` Zho#ngwe´n Xue´ba`o [Journal of Taibei University Chinese Department] (4), 214–282.

The emergence of a definite article in Beijing Mandarin: The evolution of the proximal demonstrative zhe`* Mei Fang

Traditionally, the description of demonstratives is based on the deictic center, relative to which demonstratives can be divided into distal, proximal and intermediate demonstratives. The spatial distinction between the referent and the deictic center is considered a main factor in the choice of demonstrative. In recent years, some studies on demonstratives in spoken Chinese suggest that the usage of demonstrative depends on the spatial relationship between the referent and the deictic center, and some other factors such as discourse modes (dialogue or narrative), the referential features (referential or non-referential), the degree of relevance between the referent and discourse context, the speaker’s attitude toward the referent, the speaker’s designs on expressions, and so on (cf. Tao 1999; Huang 1999). The current study, based on naturally occurring conversational data taken from Beijing Mandarin, suggests that the definite article, derived from the proximal demonstrative zhe` in a case of grammaticization,1 and the indefinite article are both emerging in contemporary Beijing Mandarin.

* This study was supported by U.S. Dept. of Education Grant #PO17A60031. I am grateful to Charles N. Li and Sandra A. Thompson for their munificent support and insightful discussions when I worked at UC Santa Barbara. This paper was presented at the 1st Kent Ridge International Roundtable on Chinese Linguistics, National University of Singapore, 2001. I would like to thank Professors Tao Hongyin, Liu Danqing and Zhang Bojiang for their helpful comments on the first draft. The Chinese version of this paper was published in Zhongguo Yuwen 4: 343–356, 2002. 1. The term ‘‘grammaticalization’’ stresses the historical perspective on grammatical forms, while the term ‘‘grammaticization’’ focuses on the implications of continually changing categories and meanings from a synchronic perspective. This paper uses the term ‘‘grammaticization’’ rather than ‘‘grammaticalization’’ to refer to a process whereby an item is entering the grammar of a language synchronically and may become fixed and constrained in distribution as described by Hopper and Traugott (1993: XVI).

56

Mei Fang

1. Discourse semantics of ‘‘Demonstrative B Noun’’ 1.1. Functions in discourse This paper will focus on the grammaticization of zhe` and na` in contemporary spoken Beijing Mandarin. In terms of on Himmelmann’s (1996) categorization, we find that the demonstratives zhe` and na` are used in the following ways.2 1) Situational use The demonstratives zhe` and na` are used to denote entities in situations of reported speech, as shown in (1) (1) A: 她动不动就大耳贴子贴我，跟拍苍蝇一样。 Ta# do`ngbu´do`ng jiu` da` eˇrtie#zi tie# woˇ, ge#n pa#i ca#ngying yı´ya`ng. 3SG often then big slap slap 1SG as swat fly same

我想不通，我想不通。 Woˇ xiaˇng bu` to#ng, woˇ xiaˇng bu` to#ng. 1SG think NEG clearly 1SG think NEG clearly ‘She often gives me big slaps like she was swatting a fly. I can’t get over it, I can’t get over it.’ B: 我不信，一定有演义，作家嘛! Woˇ bu´ xı`n, yı´dı`ng yoˇu yaˇnyı´, zuo`jia# ma! 1SG NEG believe must have fabrication writer PRT ‘I can’t believe it. You must have concocted some parts of the story. Anyway, you are a writer.’ A: 啊？我演义了？你瞅这脸上，你瞅， A? woˇ yaˇnyı` le? Nıˇ choˇu zhe` liaˇn sha`ng, nıˇ choˇu, INT 1SG fabricate PRT 2SG look Prox-Dem face on 2SG look

你还不信呀? 得，我也不怕寒碜。 Nıˇ ha´i bu´ xı`n ya? De´, woˇ yeˇ bu´ pa` ha´nchen. 2SG still NEG believe PRT INT 1SG also NEG afraid ridicule

你看这腿。 Nıˇ ka`n zhe` tuıˇ. 2SG look Prox-Dem leg 2. As for the discourse use of demonstratives, discussions on zhe` and na` in this paper refer to their monosyllabic forms only.

The emergence of a definite article in Beijing Mandarin

57

这就是跪搓板儿跪的！ Zhe` jiu` shı` guı` cuo#baˇnr guı` de! Prox-Dem just COP kneel washboard Kneel PRT ‘Ah? Me a liar? Look at my face, look . . . don’t you believe me? Well, put my face aside, look at my leg. I knelt on a washboard!’ The ‘‘demonstrative þ noun’’ construction in (1a) cannot be used as an answer to the question ‘Which one. . . ?’. Therefore, the demonstrative zhe` cannot be interpreted in the same way as a modifier. 2) Tracking use Demonstratives have an anaphoric use as shown in (2). (2)

它是棒子面儿，和得很瓷实， Ta# shı` ba`ngzi mia`nr, huo` de heˇn cı´shı´ 3SG COP corn meal mix RST very solid

弄一个大圆饼，就放到锅里煮。 no`ng yı´ ge da` yua´n bıˇng, jiu` fa`ng da`o guo# li zhuˇ. make one CL big roundcake just put into pan in boil.

煮出来就连汤带这大饼, zhuˇ chu#la´i jiu` lia´n ta#ng da`i zhe`i da` bıˇng, boil out then with soup and Prox-Dem big cake

就一块儿吃。

jiu` yı´kua`ir chı #. then together eat ‘Use cornmeal, knead it thoroughly by hand, then make it into a big round cake. Cook the big cake in water, and eat it with the soup.’ (3) A: 还有一东西是，叫疙瘩的。 Ha´i yoˇu yı´ do#ngxi shı` jia`o ge#da de. still have one thing COP name dough-drop PRT

就是，在山西农民，

jiu` shı`, za`i Sha#nxı # no´ngmı´n, just COP in Shanxi peasant

晋南一些人 Jı`nna´n yı`xie# re´n Southern-Shanxi some people

58

Mei Fang

就吃这种东西。

jiu` chı # zhe` zhoˇng do#ngxi then eat Prox-Dem kind thing

‘There is another food called dough-drop. The peasants in Shanxi Province or Southern Shanxi often eat it.’ B: 那疙瘩是， Ne`i ge#da shı`, Dist-Dem dough-drop COP

多半儿是白面做的是吧？ duo#ba`nr shı` ba´imia`n zuo` de shı` ba? mostly COP white-flour make REL COP PRT ‘The dough-drop is mostly made of wheat flour, isn’t it?’ (4)

对，乌贼，乌贼。他们常吃那东西。

Duı` wu#ze´i, wu#ze´i. Ta#men cha´ng chı # ne`i do#ngxi. right squid squid 3PL often eat Dist-Dem stu¤ ‘Yes, squid, squid. They often eat that stu¤.’ The ‘‘demonstrative þ noun’’ construction in tracking use referring to an established referent in the preceding discourse can be replaced with zhe` ge ‘this one’ or na` ge ‘that one’. 3) Discourse/Textual use The demonstrative here makes reference to a statement rather than a referent established in the preceding discourse. Examples (5) and (6) are examples of this use. (5) A: 听说你揍过他？ Tıngshuo# # nıˇ zo`u guo ta#? hear say 2SG beat PERF 3SG ‘It is said that you beat him.’ B: 揍。这你也听说啦？ Zo`u. Zhe` nıˇ yeˇ tı #ngshuo# la? beat Prox-Dem 2SG also hear PRT ‘Beat? You heard that too?’ (6)

要是闹出个人命来， Ya`oshı` na`o chu# ge re´nmı`ng la´i, if make out CL life come

The emergence of a definite article in Beijing Mandarin

59

那可不是闹着玩儿的。 na` keˇ bu´ shı` naˇozhewaˇnr de. Dist-Dem quit NEG COP make joke PRT ‘It’s not funny if someone gets killed.’ 4) Recognitional use In the recognitional use, the referent is introduced into the discourse with lower degree of identifiability, but it represents knowledge shared by the participants in the conversational situation, as shown in (7) and (8). (7) 以前我在北方的时候， Yıˇqia´n woˇ za`i beˇifa#ng de shı´hou, years ago 1SG in North REL time

有这小米面饼子。现在还有吗？ yoˇu zhe xiaˇomıˇmia`n bıˇngzi. Xia`nza`i ha´i yoˇu ma? have Prox-Dem millet cake now still have PRT ‘When I was in the North, there was a kind of cake made of millet. Is that still available now?’ (8) 还有, 还有那芝麻酱烧饼。 Ha´i yoˇu, ha´i yoˇu na zhı #majia`ng sha#obing. still have still have Dist-Dem sesame-paste baked-cake

我常常想起这个。想极了。 Woˇ cha´ngcha´ng xiaˇng qıˇ zhe` ge. Xiaˇng jı´le. 1SG often think up Prox-Dem CL miss very much ‘Also there were those baked sesame paste cakes which I often think about. I miss them a lot.’ Demonstratives in situational use, textual use and tracking use refer to entities which exist in the context, preceding discourse, or conversational situation. The situational use and the recognitional use are similar in so far as the referent of the noun following the demonstrative is mentioned for the first time. Unlike in the situational use, the textual use or the tracking use, a referent in the recognitional use is an entity that neither exists in the discourse context, nor is a topic that the speaker will talk about at a later point in the conversation. Furthermore, the choice between zhe` and na` made by a speaker does not depend on the spatial distance from the referent

60

Mei Fang

to the deictic center,3 but rather on the speaker’s realization of or attitude toward the referent. Generally speaking, zhe` and na` have two phonological variants: one with a single final vowel (zhe` /na`) and one with a final diphthong (zhe`i/ ne`i). The latter is much more colloquial than the former. According to our data, the diphthongal variants zhe`i and ne`i are much more likely to appear in anaphoric expressions. In Beijing Mandarin, however, the distribution of the two variants appears to show a di¤erent pattern, as seen in Table 1. Table 1. Distribution of the demonstratives in Beijing Mandarin

这 那

situational use

tracking use

textual use

recognitional use

zhe` /zhe`i

zhe`i

zhe`

zhe

ne`i/na`

ne`i

na`

na/ne

1.2. Referential features of the ‘‘Demonstrative þ Noun’’ combination It is interesting to note that the ‘‘Demonstrative þ Noun’’ phrase does not necessarily identify a referent, nor does it have to be co-referential or referential. Following are types of ‘‘Demonstrative þ Noun’’ usage found in our corpus:4 1) [þco-referential], [þreferential], [þidentifiable] (9) 那个吃辣的吃得邪啊。 Na` ge` chı # la` de chı # de xie´ a. Dis-Demo CL eat spicy MON eat RST extremely PRT

什么都没有，就是干辣椒， She´nme do#u me´i yoˇu, jiu` shı` ga#n la`jia#o, what all NEG have just COP dry pepper 3. Traditional treatment of demonstratives puts much emphasis on the deictic center, which is the starting point of reference to time, space and person. The division between distal demonstratives and proximal demonstratives is based on a traditional approach. Distal demonstratives are used when the referent is distant from the deictic center, while proximal demonstratives appear when the referent is close to the deictic center. 4. As for the theoretic concepts like co-referential, referential, identifiable and generic, all these terms refer to those used by Ping Chen (1987).

The emergence of a definite article in Beijing Mandarin

61

磨成粉呐， mo` che´ng feˇn na, pulverize into powder PRT

然后就倒到饭里和， ra´nho`u jiu` da`o da`o fa`n li huo`, and then pour into rice inside mix

和得那饭呐都成红颜色， huo` de ne`i fa`n do#u che´ng ho´ng ya´nse, mix RST Dist-Dem rice all become red color

然后就那么吃。

ra´nho`u jiu` na`me chı.# then just that way eat ‘Eating spicy foods becomes an extreme phenomenon. There is no vegetable, just dry pepper. It is pulverized, and mixed with the rice until everything turns red. They eat it just like that.’ 2) [co-referential], [þreferential], [þidentifiable] (10) 比方现在美国的市场里买点儿那些个， Bıˇfa#ng xia`nza`i Meˇiguo´ de shı`chaˇng li maˇi diaˇnr na` xie# ge, instance now U.S. POSS market in buy a-few Dist-Dem some CL

罐头里的菜，你打开罐头以后啊， gua`ntou li de ca`i, nıˇ daˇka#i gua`nto´u yıˇho`u a, can in NOM vegetable 2SG open can after PRT

那四季豆也黄不啦叽的，菠菜也黄不啦叽的，

na sı`jı`do`u yeˇ hua`ngbu`laji de, bo#ca`i yeˇ hua´ngbu`la#j ı # de, Dist-Dem beans also yellowish PRT spinach also yellowish PRT

什么都黄不啦叽的。

she´nme do#u hua´ngbu`la#j ı # de. everything all yellowish PRT ‘For instance, you buy some, some canned vegetables in a U.S. supermarket, and when you open those cans, (you will see) the beans are kind of yellowish, the spinach is also kind of yellowish. Everything in cans is yellowish.’

62

Mei Fang

4) [–co-referential], [þgeneric] (11) 你们什么时候儿听说过, Nıˇmen she´nme shı´ho`ur tıngshuo# # guo`, 2PL what time hear-say PERF

这文化人办文化上的事儿， zhe we´nhua`re´n ba`n we´nhua` sha`ng de shı`r, Prox-Dem intellectual run culture about NOM thing

还自个儿掏钱的？ ha´i zı`geˇr ta#o qia´n de? still oneself take out money PRT

都是要掏别人腰包。 Do#u shı` ya`o ta#o bie´ re´n ya#obo. all COP want take out other people pocket ‘When did you hear that intellectuals run cultural activities with their own money? In fact, they always want to take money from others.’ (12) A: 现在吃肉大概比以前稍微好点儿吧? Xia`nza`i chı # ro`u da`ga`i bıˇ yıˇqia´n sha#owe#i haˇo diaˇnr ba? now eat meat maybe than before a-little-bit well a-little PRT ‘Now, it is easier to have meat than before, isn’t it?’ B: 对。那城市里，很多人就是， Duı`. Na che´ngshı` li, heˇnduo# re´n jiu` shı`, right Dist-Dem city in many people just COP

每个星期都要去买点儿肉。

meˇi ge xı #ngqı # do#u ya`o qu` maˇi diaˇnr ro`u. every CL week all will go buy a-little meat ‘Right, many people in cities can buy meat every week.’ 5) [–co-referential], [–referential] (13) 有的人他是把白薯煮熟了以后， Yoˇu de re´n ta# shı` baˇ ba´ishuˇ zhuˇ shu´ le yıˇho`u, have NOM people 3SG COP BA sweetpotatoe boil well PERF after

风干，弄成白薯干儿。 fe#ngga#n, no`ng che´ng ba´ishuˇga#nr. dry make become dried potato

The emergence of a definite article in Beijing Mandarin

63

我很喜欢吃那个玩艺儿，

Woˇ heˇn xıˇhua#n chı # ne`i ge wa´nyı`r, 1SG very like eat Dist-Dem CL stu¤

一咬就跟那橡皮筋儿一样。

yı` yaˇo jiu` ge#n na xia`ngpı´jı #nr yı´ya`ng. one chew just as Dist-Dem rubber same ‘Some people like to boil sweet potatoes and hang them in the air to dry. I like that stu¤ very much. When you chew it, it feels like rubber.’ In general, the referential status of a bare noun is quite uncertainand the pronominal demonstrative contributes a great deal to the referential interpretation of the ‘‘demonstrative þ noun’’ phrase. Nonetheless, a pronominal demonstrative does not always lead to a referential or identifiable reading in discourse. 2. Functional extensions 2.1. Reduced predicate marker The ‘‘yı 5# ‘one’ þ verb’’ construction is a common grammatical construction in Mandarin Chinese. Examples (14) and (15) are from Xiandai Hanyu Babai Ci ‘Eight Hundred Modern Chinese Words’ by Lu¨ et al. (1982). (14) 医生一检查，是肺炎。 Yı #she#ng yı` jiaˇncha´, shı` fe`iya´n. doctor one check COP pneumonia ‘A doctor checked (him); it is pneumonia.’ (15) 我一想，他回去一趟也好。 Woˇ yı` xiaˇng, ta# huı´qu yı´ tang yeˇ haˇo. 1SG one think 3SG go-back one CL also good ‘I thought it might be a good idea for him to take a trip back.’ The ‘‘yı # ‘one’ þ verb’’ construction is syntactically dependent. It does not denote an event but a condition or a reason related to the following clause. 5. In spoken Chinese, the numeral word yı # ‘one’ occurs in its original tone in counting contexts. In other contexts, it changes to the second tone yı´ before a fourth-tone syllable and changes to the fourth tone yı` before syllables with other tones. We take the original form yı # ‘one’ here representing both yı´ and yı`.

64

Mei Fang

First of all, the verb in the construction cannot be followed by an aspectual marker such as le, zhe and guo, as shown in (16). (16) a. *她一检查了，是肺炎。 *Ta# yı` jiaˇncha´ le, shı` fe`iya´n. 3SG one check PERF COP pneumonia b. *我一想着/过，他回去一趟也好。 *Woˇ yı` xiaˇng zhe/guo, ta# huı´qu yı´ ta`ng yeˇ haˇo. 1SG one think PROG/PERF 3SG go-back one CL also good Secondly, the construction ‘‘yı # ‘one’ þ verb’’ is syntactically incomplete. It cannot be used as a declarative sentence; rather, it is used to introduce a clause expressing a condition or time, as shown in (17). Even with a sentence final particle, it is not grammatically acceptable. (17) a. *我一检查 *woˇ yı` jiaˇncha´ 1SG one check b. *她一检查了 *ta# yı` jiaˇncha´ le 3SG one check PERF c. *她一检查呢 *ta# yı` jiaˇncha´ ne 3SG one check PRT Thirdly, when the ‘‘yı # ‘one’ þ verb’’ combination follows another verb, it can only be interpreted as an object. This is shown in (18). (18) 想好了，决心一死。 Xiaˇng haˇo le, jue´xı #n yı` sıˇ. think well PERF decide one die ‘(I) thought it through well and decided to die.’ The verb used with yı` in (18) does not have the typical characteristic of a predicate and the whole clause with ‘‘yı # ‘one’ þ verb’’ displays no real participation in the event. However, if a demonstrative is added to the ‘‘yı # ‘one’ þ verb’’ construction, the whole phrase acts as a co-referential noun phrase tracking an action mentioned in the preceding discourse. In our corpus, all relevant examples are of this type.

The emergence of a definite article in Beijing Mandarin

65

(19) A: 我哭了，实在忍不住了。 Woˇ ku# le, shı´za`i reˇnbu´zhu` le. 1SG cry PERF really can’t-help PERF/PRT ‘I cried, I really couldn’t help crying.’ B: 这一哭，所有问题都解决了。 Zhe`i yı` ku#, suoˇyoˇu we`ntı´ do#u jieˇjue´ le. Prox-Dem one cry all problem all solve PERF/PRT ‘The crying solved all problems.’ It has been observed that a personal pronoun can be used in front of ‘‘dem þ yı #þ verb’’, as shown in (20). (20) A: 我哭了，实在忍不住了。 Woˇ ku# le, shı´za`i reˇnbu´zhu` le. 1SG cry PERF really can’t help PERF/PRT ‘I cried, I really couldn’t help crying.’ B: 你这一哭， 所有问题都解决了。 Nıˇ zhe`i yı` ku#, suoˇyoˇu we`ntı´ do#u jieˇjue´ le. 2SG Prox-Dem one cry all problem all solve PERF/PRT ‘Your crying solved all problems.’ Furthermore, a demonstrative can be placed before the verb in a ‘‘pron þ demonstrative þ verb’’ construction without yı,# as shown in (21). (21) A: 我哭了， 实在忍不住了。 Woˇ ku# le, shı´za`i reˇnbu´zhu` le. 1SG cry PERF really can’t help PERF/PRT ‘I cried, I really couldn’t help crying.’ B: 你这哭太管用了， Nıˇ zhe`i ku# ta`i guaˇnyo`ng le, 2SG Prox-Dem cry so e¤ective PRT

所有问题都解决了。 suoˇyoˇu we`ntı´ do#u jieˇjue´ le. all problem all solve PERF/PRT ‘Your crying was so e¤ective, it solved all problems.’ In (21), ‘‘pron þ demonstrative þ verb’’ phrase is used for tracking an action presented in the preceding discourse. The demonstrative 这 can be pronounced as zhe` or zhe`i. However, the use of the co-referential demon-

66

Mei Fang

strative is heavily restricted. An example is provided by ta# zhe`i chı # (‘his eating’) in (22). (22) A: 你跟他挺熟， Nıˇ ge#n ta# tıˇng shu´, 2SG with 3SG very familiar

你觉得他最大的特点是什么? nıˇ jue´de ta# zuı` da` de te`diaˇn shı` she´nme. 2SG think 3SG most big MOD character COP what ‘You are quite familiar with him. In your opinion, what is his most notable characteristic?’ B: 我就佩服他这吃。他可真能吃! Woˇ jiu` pee`fu ta# zhe`i chı #. Ta# keˇ zhe#n ne´ng chı #! 1SG only admire 3SG Prox-Dem eat 3SG quite really able eat ‘I admire his eating. He is able to eat quite a lot!’ In the case of (22), although ‘‘pron þ demonstrative þ verb’’ (e.g. ta# zhe`i chı # ‘his eating’) does not appear in the previous discourse, it either represents knowledge shared by the two speakers, or information identifiable from their common knowledge. In other words, the meaning ‘his eating’ is of higher accessibility. Additionally, the demonstrative 这 is pronounced zhe`i in the ‘‘pron þ demonstrative þ verb’’ construction, which appears to have developed from the parallel structure ‘‘pron þ demonstrative þ noun’’ in discourse. (23) 你要是不答应, Nıˇ ya`oshı` bu` da#y`ıng, 2SG if NEG agree

我就把我这头磕出脑浆子来！ woˇ jiu` baˇ woˇ zhe`i to´u ke# chu# naˇojia#ngzi lai! 1SG then BA 1SG Prox-Dem head knock out brains come ‘If you do not agree, I will knock my brains out!’ (24) 咱这社会主义比资本主义好在哪儿了？ Za´n zhe`i She`huı`zhuˇyı` bıˇ zıbe # ˇ nzhuˇyı` haˇo za`i naˇr le? 1PL Prox-Dem socialism than capitalism good in where PRT

就是有人情味儿。 jiu` shı` yoˇu re´nqı´ngwe`ir. just COP have human kindness ‘How is socialism better than capitalism? It is just that the people (in a socialist country) are full of human kindness.’

The emergence of a definite article in Beijing Mandarin

67

In this use, the noun refers to a unique entity in the real world or an entity identifiable in the context. Therefore, although the demonstrative takes a modifier slot, the two cannot be clearly di¤erentiated. In fact, ‘‘pron þ demonstrative þ verb’’ and ‘‘demonstrative þ verb/adjective’’ constructions appear to have developed from the ‘‘pron þ demonstrative þ noun’’ construction by analogy. The ‘‘pron þ demonstrative þ noun’’ construction appeared in Hong Lou Meng (cf. Lu¨ 1985) written in the Qing Dynasty (16th–19th Century). Furthermore, demonstratives can be used in front of a verb or an adjective in the ‘‘demonstrative þ verb/adjective’’ construction, such as zhe`i chı # jiaˇozi ‘the eating of dumplings’ in (25) and zhe`i sha`ng shu` ‘the climbing of trees’ in (26). (25) A: 不产面的地方, Bu` chaˇn mia`n de dı`fa#ng, NEG produce flour REL place

实在没有这个东西的地方，

shı´za`i me´i yoˇu zhe` ge do#ngxı # de dı`fa#ng, indeed NEG have Prox-Dem CL stu¤ REL place

就是，每人过年发二斤面。

jiu` shı`, meˇi re´n guo` nia´n fa# e`r jı #n mia`n. just COP every one celebrate New-Year distribute two CL flour

这样话能吃饺子。

zhe`ya`ng ne´ng chı # jiaˇo zi. this way can eat dumplings ‘For places that do not produce flour, everyone was only given two jins of wheat flour when New Year approached, so they could have dumplings.’ B: 过年这吃饺子这事儿得办到。 Guo`nia´n zhe`i chı # jiaˇozi zhe`i shı`r deˇi ba`n da`o. New-Year Prox-Dem eat dumpling Prox-Dem thing must do achieve lit. ‘The eating of dumplings must be done for New Year.’ (26) A: 还有一个菜，我觉得名字很有意思， Ha´i yoˇu yı´ ge ca`i, woˇ jue´de mı´ngzı` heˇn yoˇuyı`si. and have one CL dish 1SG think name very interesting

其实这菜并没有什么了不起， qı´shı´ zhe`i ca`i bı`ng me´i yoˇu she´nme liaˇobu`qıˇ, actually Prox-Dem dish but NEG have what splendiferous

68

Mei Fang

叫蚂蚁上树 啊。 jia`o maˇyıˇ sha`ng shu` a. called ant climb tree PRT ‘Besides there was a dish whose name is very interesting, I think. Actually this dish is not that high class. It is called Ants Climbing Trees.’ B: 这上树是怎么回事? Zhe`i sha`ng shu` shı` zeˇnme huı´ shı`? Prox-Dem climb tree COP what CL thing ‘What does the Climbing Trees mean?’ The ‘‘demonstrative þ adjective’’ construction is also found in the tracking use, and the demonstrative in it is realized as zhe`i, as in (27). (27) A: 您扔砖头哪? Nı´n re#ng zhua#nto´u na 2SG throw brick PRT ‘Was it you who threw a brick?’ B: 就听 ‘‘扑通’’。 Jiu` tıng # ‘‘pu#to#ng’’. just hear sound of splash ‘(I) Just heard the sound of a splash.’ A: 深。 She#n. deep ‘So deep.’ B: 就冲这深 . . . Jiu` cho`ng zhe`i she#n. . . just for Prox-Dem deep ‘Just because this river is so deep. . .’ A: 跳。 Tia`o. jump ‘Jump.’ B: 不跳! Bu´ tia`o. NEG jump ‘Don’t jump.’

The emergence of a definite article in Beijing Mandarin

69

In this case, she#n is an adjective, but zhe`i she#n ‘the deep’ functions as an object of the preposition cho`ng ‘to, for’, and the demonstrative induces a nominal reading of the adjective she#n (‘the depth’). This is a clear indication of how the demonstrative zhe`i helps to shift the function of the lexical item she#n from adjectival to nominal. 2.2. Topic marker The co-referential use is found not only in front of a noun but also in front of a verb. In Beijing Mandarin, a demonstrative can be used as a topic marker in a ‘‘demonstrative þ verb’’ construction placed at the beginning of a sentence. It represents an action described in previous utterances. Consider the following examples. (28) A: 不过，你要说到糖的话， Bu´guo`, nıˇ ya`o shuo#da`o ta´ng dehua`, however 2SG want say sugar as-for

南方做菜总是要搁糖的。 na´nfa#ng zuo`ca`i zoˇngshı` ya`o ge# ta´ng de. south cook always need put sugar PRT ‘However, as for sugar, the people in the south like to add sugar to dishes.’ B: 啊，可是那搁糖好像是调味儿性的。 A, Keˇshı` ne`i ge# ta´ng haˇoxia`ng shı` tia´owe`irxı`ng de. INT but Dist-Dem put sugar seem like COP seasoned PRT ‘Yes, but it seems that sugar is used for seasoning in cooking.’ (29) 您就拿我来说吧， 过去净想发财， Nı´n jiu` na´ woˇ la´i shuo# ba, guo`qu` jı`ng xiaˇng Fa#ca´i, 2SG ADV take 1SG come talk PRT years-ago only think become-rich

过去这个想法就不对， guo`qu` zhe`i ge xiaˇngfaˇ jiu` bu´ duı`, years-ago Prox-Dem CL thought actually NEG correct

净想发财， jı`ng xiaˇng fa#ca´i, only think become-rich

提起这发财来有个笑话。 Tı´qıˇ zhe`i fa#ca´i la´i yoˇu ge xia`ohua`. mention Prox-Dem become-rich come have CL joke

70

Mei Fang

‘As for me, yearning for money used to occupy my entire mind. My earlier desire for money was wrong. By the way, as for making a pile, there is a joke.’ In (28) and (29), ‘‘Prox-Dem þ verb’’ serves as a tracking device referring to an action or an event which was established by predicates in the discourse. It behaves just like a ‘‘Prox-Dem þ noun’’ construction serving as a noun phrase tracking device. As a reduced predicate phrase, ‘‘ProxDem þ verb’’ can be replaced by zhe`ige ‘this one’ or ne`ige ‘that one’ functioning as a NP. ‘‘Prox-Dem’’ co-occurring with a VP can be seen in (30) and (31) below. (30) 这过日子难免不铁勺碰锅沿儿。 Zhe guo` rı`zi na´nmiaˇn bu` tieˇ sha´o pe`ng guo# ya´nr. Prox-Dem livelife inevitable NEG iron spoon bump pan edge ‘In living the family life, friction is inevitable.’ (31) A: 宫里当差的, 人家谁要个乡下丫头？ Go#ng li da#ngcha#i de, re´njia# shuı´ ya`o ge xia#ngxia` ya#tou? palace inside worker REL person who want CL countryside girl ‘How can a man who works in the palace want such a country girl?’ B: 这不你女儿命好吗？ Zhe` bu` nıˇ nuˇ¨ ’e´r mı`ng haˇo ma? Prox-Dem NEG 2SG daughter fate good PRT ‘It shows how lucky your daughter is, doesn’t it?’ A: 谁呀？ Shuı´ ya who PRT ‘Who is it?’ B: 大太监，庞总管! Da` ta`ijia#n, Pa´ng zoˇngguaˇn! chief eunuch Pang director

你也听说过庞总管吧？

Nıˇ yeˇ tı #ngshuo# guo Pa´ng zoˇngguaˇn ba? 2SG also hear PERF Pang director PRT ‘Chief Eunuch, Director Pang! I bet you must have heard of this name!’

The emergence of a definite article in Beijing Mandarin

71

A: 这要孩子给太监做老婆， Zhe ya`o ha´izi geˇi ta`ijia#n zuo` laˇopo, Prox-Dem ask child for eunuch become wife

我怎么对得起女儿啊？

woˇ zeˇnme duı`deqıˇ nu¨ˇ ’e´r a? 1SG how owe daughter PRT ‘Asking the child to be a eunuch’s wife – how can I do such a thing to my daughter?’ In (30), the VP guo` rı`zi ‘to live the family life’ following the demonstrative is not mentioned in the previous discourse. Moreover, the whole sentence expresses a comment on an argument between a husband and his wife described in the previous conversation. In (31), ya`o ha´izi geˇi ta`ijia#n zuo` laˇopo ‘asking the child to be a eunuch’s wife’ is not mentioned but is accessible based on common knowledge shared by both speakers in the conversation. Since the ‘‘Dem þ VP’’ combination is not used for tracking, the proximal demonstrative 这 cannot be realized as zhe`i – as it is possible in the tracking use – but must be realized as zhe (without tone), because it instantiates the recognitional use. A demonstrative as a topic maker occurs when a new entity is introduced into the discourse. It marks the new information as given and identifiable. 2.3. Definite article The Chinese language has been classified as one that has neither definite nor indefinite articles. However, based on our data from Beijing Madarin, the definite article is an emerging grammatical category in discourse derived from the proximal demonstrative zhe` via its recognitional use. According to Himmelmann (1996), the grammaticization from a demonstrative to an article is identified by the following criteria: 1. Demonstratives cannot be used to refer to unique things, such as. *this/that sun or *this/that queen, while articles can be used as determiners. 2. Demonstratives cannot be used to refer to frame-based things. For example, if ‘‘tree’’ has appeared in the previous discourse, and one wants to choose an expression to refer to the branches of this tree, one would have to say ‘‘the branch’’ rather than ‘‘*this/that branch’’. Using these criteria in the context of Beijing Mandarin, we can see that the use of the proximal demonstrative zhe` can be summarized as follows.

72

Mei Fang

1. Zhe` co-occurs with a proper noun which is not tracking but accessible through shared knowledge. So the construction ‘‘demonstrative þ proper noun’’, such as zhe Le´iFe#ng, who is famous for ‘‘serving people heart and soul’’ in (32), is a generic expression referring to a type of person with the characteristics of Le´iFe#ng rather than an individual in the context. (32) 你以为呢！这雷锋可不是那么好当的。 Nıˇ yıˇwe´i ne! Zhe Le´iFe#ng keˇ bu´ shı` na`me haˇo da#ng de. 2SG think PRT DEF LeiFeng quit NEG COP so easy act PRT ‘What do you think? It’s not so easy to act as Lei Feng.’ 2. Zhe` occurs in front of a bare noun in the ‘‘demonstrative þ noun’’ construction, e.g. zhe wa`iguo´re´n ‘foreigner’ in (33) and zhe na´nde ‘males’ in (34), referring to a generic referent rather than a specific entity in discourse contexts or conversational situations. (33) 你知道吗，就这外国人呐 Nıˇ zhıda # ` o ma, jiu` zhe wa`iguo´re´n na, 2SG know PRT just DEF foreigner PRT

他们说话都跟感冒了似的，没四声。 ta#men shuo#hua` do#u ge#n gaˇnma`o le shı`de, me´i sı` she#ng. 3PL speak all as catch-cold PERF same NEG four tone ‘Do you know? The foreigners sound like they are catching a cold; they don’t use tones at all.’ (34) A: 你说我们这位吧， Nıˇ shuo# woˇmen zhe we`i ba, 2SG say 1PL Prox-Dem CL PRT,

过去挺好的，任劳任怨。 guo`qu` tıˇng haˇo de, re`nla´ore`nyua`n. past very good PRT, work-hard-no-complaining

现在倒好，成大爷了! Xia`nza`i da`o haˇo, che´ng da`ye´ le! now rather good become master PERF/PRT ‘As for my husband, he used to be very good, worked hard without complaint. But now he’s changed and become a master in the house!’

The emergence of a definite article in Beijing Mandarin

73

B: 没错。这男的呀，稍微长点本事， Me´icuo`. Zhe na´nde ya, sha#owe#i zhaˇng diaˇn beˇnshı`, right DEF menfolks PRT slight go-up a-little capability

就跟着长脾气。 jiu` ge#nzhe zhaˇng pı´qi. then follow grow temper ‘Yes. All men’s tempers rise when their capabilities rise.’ 3. Zhe` is used in front of a non-anaphoric noun whose referent is a frame-based specific entity rather than a specific one that appears in the previous discourse: (35) 在中国你要做炸酱面， Za`i Zho#ngguo´ nıˇ ya`o zuo` zha`jia`ngmia`n, in China 2SG want make noodles-with-soybean-paste

把这肉搁里面，噼里啪啦一爆，

baˇ zhe ro`u ge# lıˇmia`n, pılı# pa# # la# yı´ ba`o, BA DEF pork put inside sizzle one fry

把酱往里一搁，就行了。 baˇ jia`ng wa`ng lıˇ yı` ge#, jiu` xı´ngle. BA soybean-paste to inside one put just OK ‘In China, if you want to make soybean paste noodles, just put the minced pork into the pan and fry quickly, then take some soybean paste and mix it with the minced pork. Done!’ The function of the demonstrative in (35) is to make the indefinite referent noun ro`u ‘pork’ definite. Therefore, the demonstrative serves the function of a definite article. Such functional and phonological (with a neutral tone) characteristics appear to signal that the demonstrative has become grammaticalized into a definite article. From our data, only the proximal demonstrative can be used as a definite article while the distal demonstrative cannot. 3. Proximal vs. distal in asymmetry As for the asymmetry between the proximal demonstrative zhe` and the distal demonstrative na`, the demonstrative zhe` is the 10th most frequently used word among commonly used Chinese words, while na` is the 182nd (cf. Xu 1988). However, according to statistical surveys of zhe` and na` by

74

Mei Fang

Tao (1999), the frequencies of use of zhe` and na` are similar in situational uses but in textual uses, the proximal demonstrative zhe` is more frequent than the distal na`. In addition to the spatial di¤erence between the two demonstratives, Tao (1999) has identified five factors determining a speaker’s choice between the proximal and the distal usage in conversational discourse. 1. Discourse mode: Zhe`r ‘here’ in a dialogue refers to the immediate place where the conversation is taking place. But in a story or reported speech, it refers to a place in the conversation rather than the place of the current conversation. 2. Textuality: Textual uses of demonstratives, which refer to propositions or events as a whole rather than to specific entities, constitute one kind of anaphora. Only zhe` is likely to serve this textual function. 3. Hypotheticality of reference: It is found that the demonstrative zhe` tends to be used with a referent in realis clauses while a distal na` occurs with a referent in irrealis clauses. 4. Assumed familiarity: The proximal demonstrative zhe` is used when the referent is assumed to be non-identifiable by the speaker or new to the listener, whereas the distal na` is used for referents that are assumed to be new but identifiable to the addressee. 5. Social distance: This refers to the speaker’s attitude toward the referent being discussed. The distal na` is used for a referent that is not favored in the context. Tao’s observations are quite insightful. Our data suggest that zhe` and na` di¤er in two more ways: 1. With its function of coherence in discourse,6 the proximal zhe` is more likely to introduce a new topic than na`, which is the basis for an element functioning as a topic marker. 2. The proximal zhe` is preferred when referring to definite entities. 3.1. Aboutness The proximal zhe` has a strong influence on discourse coherence. This is supported by two factors: 1) it is more frequent in textual use (93%) than

6. Givo´n (1995) mentions an important finding. In spontaneous spoken English, when a new topic is introduced into the discourse, it has two referring expressions: one is the construction of ‘‘a þ N’’; the other one is ‘‘this þ N’’, which is especially used for introducing an important topic.

The emergence of a definite article in Beijing Mandarin

75

the distal na` (cf. Tao 1999); and 2) it is much more likely to be used to retain topic continuity than the distal na`. In discourse, when a referent is assumed to be new information to the listener, it always takes up the object position, especially the object position in existential sentences. When this referent is mentioned again as a topic, the proximal demonstrative zhe` is much more likely to be used than the distal na`, as shown in (36). So, the proximal demonstrative can introduce an entity as a topic or hold a referent on a topic chain in discourse. The function which tells hearers what is talked about can be called ‘‘aboutness’’. (36) [01] A: 还有窝头没有? Ha´i yoˇu wo#to´u me´i yoˇu? still have Wotou NEG have ‘Is there still Wotou?’ [02] B: 有窝头，对。那窝头就是黄金塔啊。 Yoˇu wo#to´u, duı`. Na` wo#to´u jiu` shı` hua´ngjı #ntaˇ a. have Wotou right Dist-Dem Wotou just COP golden tower PRT ‘Yes, it is. Wotou is just that golden tower!’ [03] A: 以前说起这1窝头来啊， Yıˇqia´n shuo#qˇı zhe1 wo#to´u la´i a, past mention DEF Wotou come PRT [04]

以前还有那么个笑话儿, yıˇqia´n ha´i yoˇu na`me ge xia`ohua`r, years ago still have that CL joke

[05]

就是讽刺慈禧太后的事儿。 jiu` shı` feˇngcı` Cı´Xıˇ ta`iho`u de shı´r. just COP satirize Cixi Empress-Dowager POSS thing

[06]

这1慈禧太后，清朝末年的时候儿，

´o Zhe1 Cı´Xıˇ ta`iho`u, Qıngcha # mo`nia´n de shı´ho`ur, DEF Cixi Empress-Dowager Qing-dynasty end year POSS time [07]

人民的生活也是越来越苦了 re´nmı´n de she#nghuo´ yeˇ shı` yue`la´iyue` kuˇ le people POSS life also COP more and more bitter PRT

是不是？ shı` bu´ shı`? PRT COP NEG COP

76

Mei Fang

[08]

老百姓啊 Laˇobaˇixı`ng a common people INT

这个生活真是苦极了， zhe` ge` she#nghuo´ zhe#nshı` kuˇ jı´le, Prox-Dem CL life really bitter extremely [09]

一天到晚啊就吃棒子面儿窝头，

yı` tia#n da`o waˇn a jiu` chı # ba`ngzimia`nr wo#to´u, one day to night INT only eat cornmeal Wotou [10]

没什么别的东西可以吃。

me´i she´nme bie´ de do#ngxi keˇyıˇ chı.# NEG what other thing can eat [11]

这1慈禧太后就说， Zhe1 Cı´Xıˇ ta`iho`u jiu` shuo#, DEF Cixi Empress-Dowager then say

[12]

‘‘呵，吃棒子面窝头， ‘‘he, chı # ba`ngzimia`n wo#to´u, INT eat cornmeal Wotou

棒子面儿窝头是什么东西啊? ba`ngzimia`nr wo#to´u shı` she´nme do#ngxi A? cornmeal Wotou COP what thing PRT [13]

给我吃一吃，我看一看,

geˇi woˇ chı # yı` chı,# woˇ ka`n yı´ ka`n, let 1SG eat one eat 1SG look one look [14]

这2老百姓生活到底是怎么样。’’ zhe2 laˇobaˇixı`ng she#nghuo´ da`odıˇ shı` zeˇnmeya`ng.’’ DEF commoner life on earth COP how

[15]

就跟这2御厨说啊，做窝头。 Jiu` ge#n zhe2 yu`chu´ shuo# a, zuo` wo#to´u. then to DEF executive chef say PRT make Wotou

[16]

这1御厨一想啊, Zhe1 yu`chu´ yı` xiaˇng a, DEF chef one think PRT

The emergence of a definite article in Beijing Mandarin

[17]

77

这可得小心点儿，不能乱做。

zhe` keˇ deˇi xiaˇoxın# diaˇnr, bu` ne´ng lua`nzuo`. Prox-Dem quit must careful little NEG can run wild [18]

这1慈禧太后来了， Zhe1 Cı´Xıˇ ta`iho`u la´i le, DEF Cixi Empress-Dowager come PERF/PRT

[19]

把这1窝头拿起来一吃。

baˇ zhe1 wo#to´u na´ qıˇlai yı` chı.# BA DEF Wotou take up one eat [20]

‘‘哟!’’ 说 ‘‘这好吃极了’’! Yo Shuo# ‘‘Zhe`i haˇochı # jı´le’’! INT say Prox-Dem delicious extremely

[21]

这1老百姓吃这个还一天到晚诉苦，

Zhe1 laˇobaˇixı`ng chı # zhe`i ge ha´i yı` tia#n da`o waˇn su`kuˇ, DEF people eat Prox-Dem CL still one day to night complain [22]

这还得了！ Zhe` ha´i de´liaˇo. Prox-Dem even how can ‘There is a story about the Empress Dowager Cixi and Wotou. Long ago, in the late Qing dynasty, the commoners’ life was getting worse and worse, right? They had no other foods but Wotou day after day. The Empress Dowager Cixi said: ‘‘What is Wotou? Let me taste. I will see what the common people’s life feels like.’’ Then she asked the executive chef to make Wotou for her. The chef realized that they had to be be careful in making it. . . . Then Cixi came. After eating one, she said: ‘‘It is so delicious. How can they have so many complaints when they have such food!’’ ’

In (36), the proximal demonstrative occurs much more often than the distal one. The distal demonstrative only occurs once in tracking use, while the proximal demonstrative occurs seven times and takes di¤erent roles in the discourse: zhe1 is used for tracking, as in lines [03], [06], [11], [16], [18], [19], [21]; and zhe2 is used to introduce a new entity for recognition, zhe laˇobaˇixı`ng she#nghuo´ 这老百姓生活 ‘the commoners’ life’ in [14] and zhe yu`chu 这御厨 ‘the executive chef ’ in [15].

78

Mei Fang

When a new entity is introduced by zhe, it is used as a topic marker which puts the new information into given information packaging. The proximal demonstrative is preferred to keep discourse coherent. In this case, zhe cannot be replaced by the distal na` /ne`i or na. On the other hand, when a distal demonstrative occurs in front of a topic, it establishes a comparison with the referent in the previous discourse. Consider the following examples. (37) 美国人就喝一杯咖啡，烤两块面包。 Meˇiguo´ re´n jiu` he# yı` be#i ka#fe#i, kaˇo liaˇng kua`i mia`nba#o. American people just drink one cup co¤ee bake two piece bread

那中国人主要讲究喝粥。 Na` Zho#ngguo´ re´n zhuˇya`o jiaˇngjiu# he# zho#u. Dist-Dem Chinese people mainly particular drink porridge ‘Americans just drink a cup of co¤ee and toast two pieces of bread, while Chinese people prefer to take porridge.’ (38) 西方人吃面条儿的时候很小心的， Xı #fa#ng re´n chı # mia`ntia´or de shı´ho`u heˇn xiaˇoxın# de, western people eat noodles REL time very careful PRT

弄个叉子在那儿转悠转悠，转悠半天， no`ng ge cha#zi za`i na`r zhua`nyou, zhua`nyou ba`ntia#n, use CL fork in there turn-around turn-around quite-a-while

然后再往嘴里搁。 ra´nho`u za`i wa`ng zuıˇ li ge#. after that then to mouth in put

那中国人， Na` Zho#ngguo´ re´n, Dist-Dem Chinese people

要是那样就不成其为吃面了。

ya`oshı` na`ya`ng jiu` bu` che´ng qı´ we´i chı # mia`n le. if that-way then NEG call its COP eat noodles PRT ‘People in the west are very careful when eating noodles. They wind the noodles around their forks again and again, then put them in their mouths. If a Chinese acts like that, you see, it’s just not eating noodles anymore.’ In both (37) and (38), the distal demonstrative na` expressing comparison cannot be replaced by the proximal demonstrative zhe`/zhe`i.

The emergence of a definite article in Beijing Mandarin

79

In Beijing Mandarin, the distal demonstrative in adnominal use is also found in recognitional use without an anaphoric function or the function of introducing a topic. In this use, ‘‘Dist-Dem þ noun’’ occurs in a grammatical position which never occurs with definite reference. (39) 有的人他是把白薯煮熟了 Yoˇu de re´n ta# shı` baˇ ba´ishuˇ zhuˇ shu´ le have NOM people 3SG COP BA sweet-potatoe boil well PERF

以后，风干，弄成白薯干儿。 yıˇho`u fe#ngga#n, no`ng che´ng ba´ishuˇga#nr. after dry make become dried potato

我很喜欢吃那个玩艺儿，

woˇ heˇn xıˇhua#n chı # ne`i ge wa´nyı`r, 1SG very like eat Dist-Dem CL stu¤

一咬就跟那橡皮筋儿一样。

yı` yaˇo jiu` ge#n na xia`ngpı´jı #nr yı´ya`ng. one chew just as Dist-Dem rubber same ‘Some people like to boil sweet potatoes and [hang] them in the air to dry. I like that stu¤ very much. . . . when you chew it, it feels like rubber.’ The phrase ge#n . . . yı´ya`ng ‘the same as. . .’ has two meanings. The first one is ‘same (in number, color, size or quality, etc)’; the second one is ‘like’ used with figurative expressions. In a metaphoric expression, such as ge#n na xia`ngpı´jınr # yı´ya`ng ‘like rubber’in (39), the noun with a figurative meaning only occurs with na, never with zhe. In summary, the proximal demonstrative is preferable to express ‘‘aboutness’’ in discourse with the following features: 1. The proximal demonstrative is preferred for keeping topic continuity in discourse. 2. The proximal demonstrative in adnominal use can introduce a new topic. 3. The distal demonstrative does not serve the function of introducing a new topic. 4. Systematicity of grammaticization and significance for linguistic typology The emergence of the definite article in Beijing Mandarin has taken place simultaneously with the emergence of the indefinite article. In synchronic

80

Mei Fang

data gathered from Modern Chinese,7 yı # (‘one/a’) is used as an indefinite article in its second tone morpheme ‘yı´ ’. 4.1. Indefinite article yı´ 一 ‘one’ Lu¨ (1944) suggests that yı´ge 一个 ‘one’ in Chinese has extended its function from the quantitative usage to several other usages, such as modifier of a non-countable entity, an action, or a situation. With such extension of function, the ‘‘numeral þ classifier’’ construction yı´ge is usually shortened to ge (i.e. without the numeral co-occurring) which acts as an indefinite article. In fact, the usage of yı´ge or ge has been common since the Yuan Dynasty (12th Century). The data of this study seem to suggest that an indefinite article yı´ 一 is emerging in Contemporary Beijing Mandarin as indicated by the following example. (40) A: 这女的是你妹妹？ Zhe`i nu¨ˇ de shı` nıˇ me`imei? Prox-Dem female NOM COP 2SG younger sister ‘Is this woman your younger sister?’ B: 不是。 Bu´ shı`. NEG COP ‘No.’ A: 你姐姐？ Nıˇ jieˇji? 2SG elder-sister ‘Your elder sister?’ B: 一亲戚。 Yı´ qı #nqi. a relative ‘A relative.’ 7. In Contemporary Beijing Mandarin, when the numeral yı # ‘one’ is used before a bare noun, it takes the second tone yı´ and has an indefinite function. However, in colloquial Beijing Mandarin if it is used in the ‘numeral þ classifier’ combination yıge # ` , it is pronounced as yı´ge. I pointed out this phenomenon at the 30th International Conference of Sino-Tibetan Languages and Linguistics, Beijing, 1997.

The emergence of a definite article in Beijing Mandarin

81

A: 什么亲戚? She´nme qı #nqi? what relative ‘What relative?’ B: 八杆子打不着的亲戚。 Ba#ga#nzidaˇbu`zha´ode qınqi. # distantly-related relative ‘A distant relative.’ Here, the recurrence of qı #nqi ‘relative’ indicates that the referent of this noun is assumed by the speaker to be non-identifiable to the hearer. This use of yı´ 一 comes from yı´ge 一个 without the classifier ge 个. An indefinite article yı´ 一 di¤ers from its numeral status in the following ways: 1. In Mandarin Chinese, when yı´ 一 is followed by a first-tone syllable, a second-tone syllable or third-tone syllable, it changes into a fourth tone, such as, yı # she#ng ! yı` she#ng 一生 ‘one life’ yı # pı´ng ! yı` pı´ng 一瓶 ‘one bottle’ 一起 ‘together’ yı # qıˇ ! yı` qıˇ And when yı´ 一 is followed by a fourth-tone syllable, it has to be changed into a second tone, for instance, yı # cı` ! yı´ cı` 一次 ‘one time’. But the indefinite use yı´ 一 always retains its lexical second tone, regardless of the tones of its adjacent syllables, such as, yı´ shızi # 一狮子 ‘a lion’ yı´ pe´ngyou 一朋友 ‘a friend’ yı´ laˇowa`i 一老外 ‘a foreigner’ yı´ ha`ozi 一耗子 ‘a rat’ The behavior of yı´ 一 in keeping its lexical second tone suggests that it is a reduced form consisting of the ‘‘yı´ þ ge’’ construction without the classifier ge 个, which is a fourth-tone word. 2. The numeral in ‘‘yı´ þ noun’’ construction has its stress on the noun rather than yı´. 3. The ‘‘yı´ þ noun’’ construction cannot establish a comparison with other numeral expressions. For example,

82

Mei Fang

(41) a. 我就带了一个帮手儿，可是他领了仨。 Woˇ jiu` da`i le yı´ ge ba#ngshoˇur, keˇshi ta# lıˇng le sa#. 1SG only take PERF one CL assistant but 3SG take PERF three ‘I had only one assistant with me, but he took three.’ b. 我就带了一帮手儿，可是他领了仨。 Woˇ jiu` da`i le yı´ ba#ngshoˇur, keˇshi ta# lıˇng le sa#. 1SG only take PERF INDEF assistant but 3SG take PERF three ‘I had one assistant with me only, but he took three.’ 4. The combination ‘‘yı´ þ noun’’ is more likely to be used in the object position and never serves as tracking in discourse. (42) 只当是东三省被占了， Zhıˇ da#ng shı` do#ng sa#n sheˇng be`i zha`n le, only deem COP east three provinces by occupy PERF/PRT

我是一你们用得着的少帅行吗？ woˇ shı` yı´ nıˇmen yo`ngdezha´o de sha`oshua`i xı´ngma? 1SG COP INDEF 2PL useful MOD Young-Marshall all-right ‘It was only deemed that the three provinces in the east of China were to be occupied, I was one of the Young Marshalls you needed, all right?’ (43) 我替我姐说吧。 Woˇ tı` woˇ jieˇ shuo# ba. 1SG replace 1SG sister say PRT

你还不能算一坏人。 Nıˇ ha´i bu` ne´ng sua`n yı´ hua`i re´n. 2SG still NEG can be considered INDEF bad person ‘I will say it for my sister that you are not a bad person.’ (44) 我在一小单位里工作。 Woˇ za`i yı´ xiaˇo da#nwe`i li go#ngzuo`. 1SG in INDEF small company in work ‘I work for a small company.’ (45) 有一朋友倒是愿意帮这个忙儿。 Yoˇu yı´ pe´ngyoˇu da`oshı` yua`nyı` ba#ng zhe` ge ma´ngr. have INDEF friend indeed will help Prox-Dem CL issue ‘A friend of mine is willing to help (you).’

The emergence of a definite article in Beijing Mandarin

83

(46) 他给我一官儿当。 Ta# geˇi woˇ yı´ gua#nr da#ng. 3SG give 1SG INDEF o‰cer serve ‘He o¤ered me a position as o‰cer.’ When speakers in a conversation cannot identify the referent based on their common knowledge, they usually use the combination ‘‘yı´ þ noun’’. The referent in this use has not been established in previous discourse, and it is assumed by the speaker to be unfamiliar to the hearer. Relatively speaking, the emergence of the definite article zhe came one generation earlier than that of the indefinite article yı´. Beijing Mandarin data from the early 1980s reveals that the definite article zhe existed in the corpus of the older generation’s conversations while the indefinite article yı´ is not attested in the corpus until the 1960s.8 We have shown that the emergence of the definite article in Beijing Mandarin is not isolated. The definite article zhe and the indefinite article yı´ are interrelated. So the functional extension of the demonstrative zhe is part of a systemic change. 5. Summary In this paper, the proximal demonstrative zhe` and the distal demonstrative na` in Contemporary Beijing Mandarin have been investigated. The results show that the two demonstratives serve di¤erent discourse and pragmatic functions and have undergone functional shifts and grammaticization. We have seen that the definite article was derived from the demonstrative zhe` in its recognitional use, accompanied by the emergence of the use of yı´ ‘one’ as an indefinite article. This functional shift from a demonstrative to a definite article and from a numeral to an indefinite article is a clear case of grammaticization. As a result, a new grammatical category, definiteness, has emerged in Contemporary Beijing Mandarin: zhe`ge ‘this; this one’ ! zhe definite article yı´ge ‘one’ ! yı´ indefinite article 8. In the early 1980s, Prof. Lin Tao from the Department of Chinese Language and Literature, Peking University, did fieldwork on Beijing Mandarin in both urban and rural areas of Beijing, as well as some parts of Hebei Province. At that time, I took part in this investigation in the urban area of Beijing. At present, these corpora are preserved in the Department of Chinese Language and Literature, Peking University.

84

Mei Fang

Although these patterns are not observed in written Mandarin Chinese (i.e. Putonghua), they are preferred by native Beijing Mandarin speakers in oral discourse.

References Ariel, Mira 1991 Chen, Ping 1987

The function of accessibility in a theory of grammar. Journal of Pragmatics 16: 443–463. On referentiality vs. non-referentiality, definiteness vs. indefiniteness, specificity vs. nonspecificity, and genericity vs. individuality in Chinese. Zhongguo Yuwen 2: 81–92.

Diessel, Holger 1999 Demonstratives: Form, Function, and Grammaticization. Philadelphia/Armsterdam: John Benjamins. Du Bois, John W. 1980 Beyond definiteness: the trace of identity in discourse. In The Pear Stories: Cognitive, Cultural and Linguistic Aspects of Narrative Production, Wallace L Chafe (ed.), 203–274. Norwood: Ablex Publishing Co. Du Bois, John W. and Sandra A. Thompson 1991 Dimensions of a theory of information flow. Manuscript, UC Santa Barbara. Fang, Mei 1997 Xiandai Beijinghua de yufa tedian [Characteristics of the Modern Beijing dialect.] Paper presented at the 30th International Conference of Sino-Tibetan Languages and Linguistics. Beijing. Givo´n, Talmy 1995 Coherence in text vs. coherence in mind, In Coherence in Spontaneous Text, Gernsbacher, Morton Ann and T. Givo´n (eds.), 59– 115. Philadelphia/Armsterdam: John Benjamins. Heine, Bernd, Ulrike Claudi and Friederike Hunnemeyer 1991 Grammaticalization: A Conceptual Framework. Chicago: The University of Chicago Press. Himmelmann, Nikolaus P. 1996 Demonstratives in narrative discourse: a taxonomy of universal uses. In Study in Anaphora, Barbara A. Fox (ed.), 205–254. Philadelphia/Armsterdam: John Benjamins. Hopper, Paul J. and Elizabeth Closs Traugott 1993 Grammaticalization. Cambridge: Cambridge University Press. Huang, Shuanfan 1999 The emergence of a grammatical category definite article in spoken Chinese. Journal of Pragmatics 31–l: 77–94.

The emergence of a definite article in Beijing Mandarin Liang, Jingmei 2002

85

Zhe-, na- de yuyong yu huayu gongneng yanjiu [A Study on the Pragmatic and Discourse Function of Demonstratives zheand na-], Ph.D. dissertation, Institute of Linguistics in Chinese Academy of Social Sciences. Liu, Jian, Lansheng Jiang, Weiguo Bai, Guangshun Cao 1992 Jindai Hanyu Xuci Yanjiu [Study on the Empty Words in Early Modern Chinese.] Beijing: Yuwen Press. Lu¨, Shuxiang 1944 Gezi de yingyong fanwei, fulun danwei ci qian yi zi de tuoluo [On the range of use of ge and the loss of yi before measure words.] In Hanyu Yufa Lunwenji [Chinese Grammar Essays], 145–175 [1984]. Beijing: Commercial Press. Lu¨, Shuxiang 1982 Xiangdai Hanyu Babai Ci [Mandarin Eight Hundred Words.] Beijing: Commercial Press. Lu¨, Shuxiang 1985 Jindai Hanyu Zhidaici [Demonstratives in Early Modern Chinese.] Shanghai: Xuelin Press. Ota, Tatsuo 1987 Zhongguoyu Lishi Wenfa [A Historical Grammar of Modern Chinese], translated by Jiang Shaoyu & Xu Changhua. Beijing: Peking University Press. Ota, Tatsuo 1991 Hanyushi Tongkao [A Historical Study of Chinese language], translated by Jiang Lansheng & Bai Weiguo. Chongqing: Chongqing Press. Payne, Thomas E. 1997 Describing Morphosyntax: A Guide for Field Linguistics. Cambridge: Cambridge University Press. Quirk, Randolph et al. If there are less than 3 authors, you have to list them 1985 A Comprehensive Grammar of the English Language. London: Longman. Shen, Jiaxuan 1999 Bu Duichen he Biaojilun [Asymmetrical Phenomena in Chinese Grammar and the Theory of Markedness.] Nanchang: Jiangxi Jiaoyu Press. Tao, Hongyi 1999 The grammar of demonstratives in Mandarin conversational discourse: A case study. Journal of Chinese Linguistics 27: 69–103. Teng, Shou-hsin 1981 Deixis, anaphora, and demonstratives in Chinese. In Cahiers de Linguistique-Asie Oriental 10: 5–18. Xu, Dan 1988 Qiantan zhe/na de buduichengxing [A preliminary study of the asymmetries between zhe and na.] Zhongguo Yuwen 2: 128–130.

86

Mei Fang

Xu, Yulong 1987

A study of referential function of demonstratives in Chinese discourse. Journal of Chinese Linguistics 15: 132–151. Zhang, Bojiang and Mei Fang 1996 Hanyu Gongneng Yufa Yanjiu [Functional Studies on Chinese Grammar.] Nanchang: Jiangxi Jiaoyu Press. Zhu, Dexi 1982 Yufa Jiangyi [Lectures on Grammar.] Beijing: Commercial Press.

Abbreviation of grammatical terms used in this paper: 1SG 1PL 2SG 2PL 3SG 3PL ADV ASP BA CL COP DEF DEG Dist-Dem INDEF INT MOD NEG NOM POSS PRON Prox-Dem PERF PROG PRT REL RST

first person singular first person plural second person singular second person plural third person singular third person plural adverb aspect marker disposal marker classifier copula definite degree complement marker distal demonstrative indefinite interjection modifier negative nominalizer possessive pronoun proximal demonstrative perfective progressive particle relativizer resultative complement marker

The grammaticalization of the directional verb ‘la´i’: A construction grammar approach1 Cheng-hui Liu 1. Introduction In recent years, grammaticalization has been widely applied to the study of historical Chinese grammar since it coincides closely with the traditional philological scholars’ insight into the distinction between content words (实字 shı´-zı` ) and function words (虚字 xu#-zı` ), yet supplements the traditional view with the processes that a content word may undergo when becoming a function word, and thus leads us toward a better understanding of the history of the language. Nevertheless, ways to apply methods of grammaticalization to the rebuilding of historical Chinese grammar still need to be scrutinized. As an overview of recent works reveals, studies of the same grammatical marker can come to drastically di¤erent conclusions about the processes of its change. This confusing phenomenon has to do with the properties of the language, its lack of morphology and the nature of its written system. Ancient Chinese, as preserved in written documents, seems to have 1. This paper focuses on the necessity of adopting a framework to reconstruct the changes 来 (la´i) has undergone, a highly complicated case in the history of Chinese. To make our point as explicit as possible, we leave out detailed reviews of former studies as well as part of the supporting evidence in favor of the present resolution, all of which have been discussed in a Chinese paper entitled as ‘Shanggu dao Zhonggu ‘Lai’ de Yanbian – Goushi Guandian de Yufa Fenxi’ [The Changes of ‘Lai’ from Archaic to Medieval Chinese: An analysis from the Perspective of Constructions], by this author. For readers interested in such details, please refer to the aforementioned paper. We herein use a corpus set up by the Institute of Linguistics, Academia Sinica, to search data from historical documents. To avoid redundancy, the data that repeat that of former studies of 来 (la´i) have been, to the extent of our knowledge, noted in the Chinese paper mentioned above, and will not be referred to here. My gratitude goes to anonymous reviewers of the paper and especially to Prof. Janet Xing, who not only made useful comments on this paper, but also kindly provided assistance for making the English version more readable.

88

Cheng-hui Liu

been ‘morphology-less,’ even though its proto-form may have exhibited some morphology. To add to the confusion, the Chinese characters do not comprise a phonetic alphabet and are, therefore, incapable of reflecting the phonological changes that might occur in their oral counterparts through di¤erent historical stages. The formal evidence for tracing linguistic history used in general by linguists is simply missing in the written documents of Chinese. Such confusion is exemplified by the case of 来 (la´i). The character represents several grammatical items, including a verb and various functional markers, the latter of which are derived from the former. When and how they split constitutes one of the core interests of grammaticalization studies. Quite a few studies have probed into historical details to find an answer, but the answers are in striking disagreement, which points to the necessity of a) establishing formal criteria for distinguishing divergent items or markers represented by the same character, and b) establishing reliable criteria to track the chronological order of linguistic changes. Let’s first compare the following examples: (1) 日之夕矣，羊牛下来。(诗经 Shı #jı #ng, 6th c. BC) rı` zhı # xı # yıˇ ya´ng niu´ xia` la´i sun ZHI sunset MOD, sheep cattle descend come2 ‘At sunset, the sheep and cattle descended and came here.’ (2) 乃有四大天王下来，诣于其边。 (佛本行集经 Fo´beˇnxı´ngjı´jı #ng, 6th c.) naˇi yoˇu sı`-da`-tia#n-wa´ng xia` la´i yı` yu´ qı´ bia#n then exist Four-Devarajas descend come, call-on to his side ‘Then Four Devarajas descended and came to call on him.’ (1) and (2) appear in documents more than one thousand years apart. In both examples, 下来 (xia` la´i) are written with the same characters. There is no formal evidence, on the surface at least, signaling whether they are identical linguistic items or not, and this points to a di‰culty in the observation of linguistic changes in Chinese. 2. MOD stands for epistemic modality markers. 之 (zhı )# in 日之夕矣 (rı` zhı # xı # yıˇ ) ‘it was sunset’ is a structural marker of a special kind; since it is irrelevant to the issue of 来 (la´i), we use ZHI to refer to its function to avoid unnecessary discussion of its grammatical property. The same goes for SUO in (37). In addition, NEG refers to negative adverbs, CONJ to conjunctions, TOP to topic markers, and PLU to plural markers.

A construction grammar approach

89

By the ninth century AD, quite a few grammatical functions of 来 (la´i) had developed, mostly related to the semantic notion of aspect. Since they are notionally close to one another, classifying them is a di‰cult undertaking indeed. In Cao (1995), all the functions relevant to aspect are grouped into one single category, namely, ‘auxiliary words of situational aspect’ (事态助词 shı`-ta`i zhu`-cı´ ). Wu (1996) distinguishes ‘auxiliary words of sentential aspect’ (事态助词 shı`-ta`i zhu`-cı´ ),3 from ‘auxiliary words of verbal aspect’ (动态助词 do`ng-ta`i zhu`-cı´ ). Wu (2002: 27) further divides verbal aspect markers in the form of 来 (la´i) into ‘phase complements’ (动相补语 do`ng-xia`ng buˇ-yuˇ ), ‘auxiliary words of perfective aspect’ (完成 体助词 wa´n-che´ng tıˇ zhu`-cı´ ), ‘markers of continuative aspect’ (持续体标 记 chı´-xu` tıˇ bia#o-jı` ), and ‘markers of stative complement’ (状态补语标记 zhua`ng-ta`i buˇ-yuˇ bia#o-jı` ), without referring to the term of ‘auxiliary words of verbal aspect’. Liang (2007) recognizes ‘auxiliary words of sentential aspect’ (事态助词 shı`-ta`i zhu`-cı´ ) and ‘phase complements’ (动相补语 do`ng-xia`ng buˇ-yuˇ ), the former of which coincides with the corresponding class of Wu (1996), while the latter roughly includes Wu (2002)’s ‘phase complements,’ ‘auxiliary words of perfective aspect,’ and ‘markers of continuative aspect.’ Categorizations for individual instances vary accordingly. Let’s take 生计 抛来诗是业 (she#ng-jı` pa#o-la´i shı # shı` ye`) ‘writing poetry has become my career since I gave up the job I earned a living at’ as an example: 来 (la´i) of 抛来 ( pa#o-la´i) is grouped into the category of perfective aspect marker according to Wu (2002: 27), and into the category of phase complement according to Liang’s (2007: 160) analysis. Moreover, there is disagreement about when and how certain changes occured: Cao (1995: 98), for example, suggests that the auxiliary word 来 (la´i) emerged at the beginning of the seventh century; but Liang (2007: 129–130) moves the date up to the fifth century. Wu (2002: 27) proposes that the phase complement 来 (la´i) came from the directional complement (趋向补语 qu#-xia`ng buˇ-yuˇ ) and split further into auxiliary words of perfective aspect and markers of continuative aspect categories. Liang (2007: 160–166), on the other hand, argues against Wu’s proposal that the phase complement descended directly from the directional complement. He suggests, instead, that the phase complement arose from the phrasal pattern ‘PþVþ 以来 ( yıˇ la´i).’4 Since all the studies rely heavily on notional criteria of their own, there is hardly common ground for comparison and evaluation. It is, therefore, not unreasonable to doubt the overall usefulness of Chinese historical 3. Wu uses the same Chinese term as Cao’s, but with a di¤erent meaning. 4. See Liang (2007: 162); also see Note 14.

90

Cheng-hui Liu

documents for research on grammaticalization. The present paper is concerned with this issue, and will, by displaying examples of 来 (la´i) in documents from two consecutive historical stages, argue that there exists formal evidence of a di¤erent kind, namely, from constructions, which has, unfortunately, been ignored in the past. Historical changes in Chinese grammar are hidden behind the numerous constructions that have arisen as the language moved through di¤erent historical stages. It is through evidence of the way lexical items or grammatical markers diverge in distinct constructions that Chinese historical documents, and the findings from them, will make solid contributions to the area of grammaticalization. The term ‘construction’ is used as Goldberg (1995: 4) uses it, who states that ‘‘C is a CONSTRUCTION i¤ def C is a form-meaning pair 3Fi, Si4 such that some aspect of Fi or some aspect of Si is not strictly predictable from C’s component parts or from other previously established constructions.’’ The historical stages involved in our discussion are as follows: Archaic Chinese (上古汉语 sha`ng-guˇ ha`n-yuˇ ): 5th century BC–AD 1st century Medieval Chinese (中古汉语 zho#ng-guˇ ha`n-yuˇ ): AD 2nd century–6th century The discussion will also be extended to include the early period of PreModern Chinese (近代汉语 jı`n-da`i ha`n-yuˇ ), around the seventh to ninth centuries, in order to complete the discussion of the splitting of several verbal aspect markers from a common origin. We would like to suggest that i) grammatical markers in the form of 来 (la´i) became diversified in distinguishable constructions, ii) only based on constructions do certain markers continue to change through time, and iii) owing to this, the verb 来 (la´i) has outlived some of its descendants. We argue that, because grammatical changes occur within the framework of constructions, the verb form was able to remain active for more than two thousand years.

2. Historical changes within the framework of constructions The case of 来 (la´i) is among the most complicated in the annals of grammaticalization and in the history of the language per se. To begin with, it concerns a verb turning into a component of a directional verb compound. Secondly, it also concerns a verb turning into a component of a resultative verb compound, which is then grammaticalized further into a verb su‰x of the perfective aspect. Moreover, the same verb undergoes grammatical-

A construction grammar approach

91

ization into a sentence-final marker of the perfect aspect, into a hortative marker, and into even more forms that will not be discussed due to space limitations.5 One may ask if the above-mentioned varieties are derived from homonymous words rather than one single word. As far as this issue is concerned, it is agreed that all of them originate from the verb 来 (la´i). The disagreement mainly concerns the question of how and when these varieties come into existence. Up to this point, all studies of grammaticalization in historical Chinese grammar trace changes through both combinatory and functional properties. The combinatory property is formal in nature. The problem is its insu‰ciency, as shown in (1)–(2). It is incapable of revealing if 下来 (xia` la´i) in Medieval Chinese is a preserved form of that same phrase in Archaic Chinese. Consequently, it is hard to determine whether the function of 来 (la´i) in this phase underwent changes during that period. If, on the other hand, we apply the framework of constructions as we argue, issues of the above sort can be revisited and reanalyzed with perhaps more success and more enlightening results. We would like to make it clear that the present argument is not completely unknown in the field. Some other Chinese historical linguists have also seen the innovation of functional markers as being relevant to their constructions. As Xie (1987: 213) puts it: That a content word becomes a function word relies primarily on its meaning, and its grammatical position. That is to say, a word shifting from content word to function word status is generally due to its frequently occurring in the position where a certain grammatical relationship is implied, which reduces its lexical sense and denotation progressively, and leads it to be conventionalized as a function word.6

5. Most of the functions that have developed from 来 (la´i) have been discussed in Wang (2004), Li (2004), and Liang (2007), the latter one of which takes conventionalized phrasal patterns into account, and thus is the closest to the present study. Nevertheless, it does not trace every phrasal pattern to its very origin, and, more importantly, it appeals to notional criteria from time to time, and puts aside constructional evidence in such occasions. Subsequently, it comes to di¤erent conclusions about the changes of 来 (la´i) than this study does. See Notes 10, 14 and 15 for details. 6. The following is the quote in Chinese: 实词的虚化，要以意义为依据，以句

法地位为途径。也就是说，一个词由实词转化为虚词，一般是由于它经 常出现在一些适于表现某种语法关系的位置上，从而引起词义的逐渐 虚化，并进而实现句法地位的固定，转化为虚词。(Xie 1987: 13).

92

Cheng-hui Liu

‘Position’ and ‘grammatical relationship’ in the above quote can be interpreted in terms of construction, ‘position’ corresponds to ‘form’ and ‘relationship’ to ‘functional meaning’, and the conventionalization of a position-relationship pair can thus be interpreted as corresponding to that of form-meaning pair. Liu, Cao and Wu (1995: 161), in a parallel manner, point out that the second position of the serial verb construction induces verbs to change into aspect markers: Take the perfective aspect markers as an example. The perfective markers in Pre-Modern Chinese, such as 将 ( jia#ng), 着 (zha´o), 取 (quˇ ) and 得 (de´ ), are derived from verbs. Their shift towards perfective marker function begins with their stepping into the serial verb construction.7

The semantic interaction of elements in a syntagmatic sequence might trigger the element in it to be grammaticalized, as pointed out above. But what happens after grammaticalization? The grammatical relationship of elements in the sequence is unlikely to remain unchanged afterwards. Rather, the process of grammaticalization is accompanied by an alteration in the relationship of the elements. In our opinion, at the moment when the verbs 将 ( jia#ng), 着 (zha´o), 取 (quˇ ) and 得 (de´) were grammaticalized in the serial verb construction, a construction novel to the language of the previous historical stage was simultaneously innovated. Whereas Liu, Cao and Wu (1995) stress the changes of individual elements, treating them like free morphemes, we will, in this paper, pay equal attention to both the elements and their grammatical contexts, taking the elements as components within the contexts of constructions. Constructions are common to the Chinese language in all historical stages, and, therefore, the history of grammar should be reconstructable by examining the changes in the overall constructions as well as their components. 3. The verb 来 (la´i) in Archaic and Medieval Chinese The verb 来 (la´i) is the root of quite a few functional markers of identical form. To maintain clarity, it will be written as 来 (la´i)VERB below. 来 (la´i)VERB can be traced back to Archaic Chinese and even earlier. It is 7. The following is the Chinese quote: 以动态助词为例。近代汉语动态助词 ‘将’ ‘着’ ‘取’ ‘得’ 等均由动词演变而来。它们从动词向动态助词演变的 过程是从进入连动式开始的。(Liu, Cao and Wu 1995: 161)

A construction grammar approach

93

usually translated into English as ‘to come’; but, to be more accurate, it denotes ‘to arrive at the location the speaker occupies or where the narrator sets as a reference point in a narrative discourse.’ Moreover, it is usually subcategorized as a directional verb (趋向动词 qu#-xia`ng do`ng-cı´ ), although it behaves like an achievement verb (达成动词 da´-che´ng do`ng-cı´ ). The property of 来 (la´i)VERB falls squarely into the category of achievement verbs at the Archaic Chinese stage. For example: (3) 十二月，祭伯来，非王命也。(左传 Zuoˇzhua`n, 4th c. BC) shı´-e`r-yue` Zha`i-bo´ la´i fei# wa´ng mı`ng yeˇ December NAME be-here be-not emperor order MOD ‘In December, Count Zhai was here – it was not ordered by the emperor.’ (4) 有朋自远方来，不亦乐乎？(论语 Lu´nyuˇ, 5th c. BC) yoˇu pe´ng zı` yuaˇn fa#ng la´i bu´ yı` le` hu# exist friend from distant-place come, NEG also be-happy MOD ‘When friends come from afar, don’t you also feel happy?’ (5) 故远人不服，则修文德以来之。(论语 Lu´nyuˇ, 5th c. BC) gu` yuaˇn-re´n bu` fu´ so people-far-away NEG obey, ze´ xiu# we´n de´ yıˇ la`i zhı 8# then cultivate civility morality so-as-to cause-come them ‘So if people far away are not obedient, then improve civility and morality to smooth their way.’ Unlike directional verbs in Mandarin, 来 (la´i)VERB in Archaic Chinese barely allows a locative element to follow it, as illustrated in (3)–(4). It permits a shift to a causative verb, as shown in 来之 (la`i zhı )# ‘to cause them to come’ in (5), which is typical of achievement verbs of that time. It is closely associated with the concept of directionality, however, when modified by a locative prepositional phrase, as shown in (4) 自远方来 (zı` yuaˇn-fan# g la´i) ‘to come from afar’. To our knowledge, there is only one case in the documents of Archaic Chinese in which 来 (la´i)VERB precedes a locative element, as shown in (6):

8. The character 之 (zhı )# here represents the third person pronoun, which di¤ers from the structural marker 之 (zhı )# in (1).

94

Cheng-hui Liu

(6) 日，吾来此也，非以狄为荣 . . . (国语 Guo´yuˇ, 4th c. BC) rı` wu´ la´i cıˇ yeˇ the other day, I come here MOD, fei# yıˇ dı´ we´i ro´ng be-not think-of the-Di-tribe be honor ‘The other day when I came here, it was not for the honor of being a member of the Di tribe . . .’ The appearance of 此 (cıˇ ) here after 来 (la´i)VERB is redundant given the denotation of the verb and thus it might be regarded as a sign of change. At the point of transition between Archaic and Medieval Chinese, 来 (la´i)VERB begins a shift towards becoming a directional verb, one that takes a locative object. For example: (7) 即令他夫人饰，从御者数十人，为邢夫人来前 (史记 Shıˇjı`, 1st c. BC) jı´ lı`ng ta# fu#-re´n shı` right-away order other lady play-the-part, co´ng-yu`-zheˇ shu` shı´ re´n we`i xı´ng-fu#-re´n la´i qia´n follower several ten person, make-up Lady-Xing come front ‘Then (the emperor) ordered another lady to pretend to be Lady Xing, and, in the company of tens of followers, to come before (Lady Yin).’ (8) 田单因宣言曰：‘‘神来下教我。 ’’ (史记 Shıˇjı`, 1st c. BC) Tia´n-da#n yı #n xua#n ya´n yue# NAME thus declare words say she´n la´i xia` jia#o woˇ divine-being come underside teach me ‘Tian-dan thus claimed: ‘‘A divine being came below to teach me.’’’ (9) 又顷梦青蝇数十来鼻头上，驱之不去。 (世说新语 Shı`shuo#xı #nyuˇ, 5th c.) ´ng shu` yo`u qıng # me`ng qıng-yı # shı´ la´i bı´-to´u sha`ng, again short-while dream green-fly several ten come nose upside qu# zhı # bu´ qu` drive them NEG go ‘Again, for a while, he dreamed that some green flies landed on his nose and there was no way to drive them o¤.’

A construction grammar approach

95

来 (la´i)VERB turns into a directional verb when it starts to take directional nouns as its object. In (7)–(8), the objects of 来 (la´i) are simple directional nouns 前 (qia´n) ‘before (in front)’ and 下 (xia`) ‘below’ respectively; in (9), it is a noun phrase headed by a directional noun 上 (sha`ng) ‘above.’ When 来 (la´i)VERB first takes locative objects, it is restricted to directional nouns. The restriction holds during the transition time, but is removed afterwards. 4. The hortative marker 来 (la´i) In Archaic Chinese, the form of 来 (la´i) refers to an expression of ‘calling others to be in the speaker’s location,’ as shown in (10), as well as a marker that indicates ‘inciting to action,’ as shown in (11). (10) 孔某曰：‘‘来！吾语女. . .’’ (墨子 Mo`zıˇ, 4th c. BC) Koˇng-moˇu yue# la´i wu´ yu` ruˇ NAME-certain say-that come I tell you ‘A certain Mr. Kong said: ‘‘Come here! Let me tell you. . .’’’ (11) 虽然，若必有以也，尝以语我来！(庄子 Zhua#ngzıˇ, 3rd c. BC) suı # ra´n ruo` bı` yoˇu yıˇ yeˇ even-if so you must have reason MOD cha´ng yıˇ yu` woˇ la´i try by-it tell me Hortative ‘Even so, you must have a reason for this. Would you try to tell me about it, please!’ The forms of 来 (la´i) in (10)–(11) occur in imperative sentences. When used in the imperative form, a verb denoting ‘to be physically where the speaker is’ would metaphorically extend to denote ‘to be figuratively at the point the speaker occupies,’ and that is an example of 来 (la´i) being used as a calling expression, as shown in (10).9 来 (la´i) as the marker of inciting to action, as shown in (11), occurs in the sentence-final position, just as epistemic modality markers (语气词 yuˇ-qı`-cı´ ) do, only the latter are used in all sentence types, while the 9. The calling expression 来 (la´i), as shown in (10), may be considered as a hortative marker. However, it is, by convention, treated as a verb used in an imperative form, and we follow this convention here.

96

Cheng-hui Liu

former is restricted to the imperative sentence. It will be written as 来 (la´i)HORTATIVE.10 It is possible that 来 (la´i)HORTATIVE was derived from the calling expression and further diversified via fixation of its grammatical position; it is also possible that it was derived from the verb directly. Whatever the case may be, the change to 来 (la´i)HORTATIVE relates directly to its sentence-final position in imperative sentences. It can be used in circumstances in which one tries to persuade oneself, or where one talks to an object mimicking a second person, as shown in (12)–(13). It has a grammatical status parallel to 乎 (hu# ), and thus may occur freely before or after 乎 (hu#). (12) 孔子在陈曰：‘‘盍归乎来！. . .’’ (孟子 Me`ngzıˇ, 3rd c. BC) Koˇng-zıˇ za`i Che´n yue# Confucius in Chen-state say-that he´ guı # hu# la´i why-not return-home MOD Hortative ‘Confucius was in the state of Chen when he said (to himself ): ‘‘Why not return home! . . .’’’ (13) 歌曰：‘‘长铗归来乎！食无鱼。 ’’ (战国策 Zha`nguo´ce`, 1st c. BC) ge# yue# cha´ng-jia´ guı # la´i hu# shı´ wu´ yu´ sing say-that long-sword return Hortative MOD eat not-exist fish ‘(He) sang out: ‘‘Long sword, would you return home! There is no fish for our meals.’’’

10. Liang (2007: 209–218) observes that there is a variety of 来 (la´i) in Medieval Chinese, to wit: a ‘modality word of the imperative form’ (祈使语气词 qı´shıˇ yuˇ-qı`-cı´ ), which is grammaticalized from the phrasal pattern ‘VþNP þ来 (la´i)’ in Archaic Chinese, on the condition that V is featured with [–carrying (携持 xı-chı # ´ )] and that the phrase occurs in the imperative sentence. The imperative sentence type is crucial to the change, indeed and yet we do not think this feature of the verb has much to do with the change. Moreover, Liang (p. 214–215) claims that i) there is a modality marker 来 (la´i) which indicates exclamation in Archaic Chinese, but became extinct in Medieval Chinese, and ii) the ‘modality word of imperative form’ 来 (la´i) in Medieval Chinese was innovated separately. 来 (la´i) in (13) is thus treated as a marker of exclamation. However, there is insu‰cient evidence to support such claims, especially in trying to prove that the modality 来 (la´i) in Archaic Chinese di¤ers from that in Medieval Chinese. In our opinion, there is only one single modality marker 来 (la´i), which is 来 (la´i) HORTATIVE. For more on this, see Liu (2009).

A construction grammar approach

97

Accordingly, both the calling expression 来 (la´i) and 来 (la´i)HORTATIVE were developed through specific constructions; the former through the imperative sentence types, and the latter, through the sentence-final position imperative. They do not end with Archaic Chinese but remain active in Medieval Chinese, as shown in (14)–(15): (14) 呼何共坐曰：‘‘来！来！此是君坐。 ’’ (世说新语 Shı`shuo#xı #nyuˇ, 5th c.) hu# He´ go`ng zuo` yue# la´i la´i cıˇ shı` ju#n zuo` call NAME together sit say-that be-here be-here this is you seat ‘(He) called He to sit with him by saying: ‘‘Come here! Come here! This is your seat.’’’ (15) 若比丘向女人作隐覆语言，‘‘姐妹沐浴来！噉果来！出毒来！ ’’ (摩诃僧祇律 Mo´he#se#ngzhıˇlu¨`, 5th c.) ruo` bıˇ-qiu# xia`ng nuˇ¨ -re´n zuo` yıˇn-fu` yuˇ-ya´n suppose Buddhist-monk towards woman say equivocal expression jieˇ-me`i mu`-yu` la´i da`n-guoˇ la´i chu#-du´ la´i sister bath Hortative eat-fruit Hortative drive-out-sin Hortative ‘Suppose that a Buddhist monk used such equivocal expressions with women as ‘‘Hi, sis! Would you take a bath! Would you take some fruits! Would you drive your sins out!’’’ 5. The perfective aspect marker 来 (la´i) As displayed in (1), 下来 (xia` la´i) can be traced back as far as Archaic Chinese. Back then, however, it was exceptional, since 来 (la´i) rarely served as the second component of the serial verb construction [VþV]. This very form remained in use through the next historical stage, Medieval Chinese, as shown in (2) and (16): (16) 复有住在虚空不下，复有下来自然碎末。 (佛本行集经 Fo´beˇnxı´ngjı´jı #ng, 6th c.) fu` yoˇu zhu`-za`i xu#-ko#ng bu´ xia`, fu` yoˇu also exist stay be-in-location air NEG descend also exist xia`-la´i zı`-ra´n suı`-mo` descend-PERFECTIVE naturally small-pieces ‘There were also some of them which stayed in the air without descending, and also some others which broke into small pieces at the moment they descended.’

98

Cheng-hui Liu

下来 (xia`-la´i) in (16) di¤ers from that in (2). Let us look into the context: 复有下来自然碎末 ( fu` yoˇu xia`-la´i zı`-ra´n suı`-mo`) ‘there were also some others which broke into small pieces at the moment they descended’ relates to one single situation, in which the breaking of the objects and their descending occur simultaneously; 来 (la´i) indicates the final phase of the descent, that is, the instantaneous completion of it, and thus is a perfective aspect marker. On the other hand, 来 (la´i) in (2) preserves its verbal content, just as in (1), although it is uncertain whether its content is unchanged or whether it has shifted to a half-way mark somewhere between a full verb and a pure functional marker. The instance in (16) reveals that [V þ 来 (la´i) PERFECTIVE] emerged in late Medieval Chinese. But what triggered 来 (la´i) PERFECTIVE? In accordance with Liu, Cao and Wu (1995)’s comments on the verbs 将 ( jia#ng), 着 (zha´o), 取 (quˇ ), 得 (de´) turning into perfective aspect markers in the serial verb construction, it should be rooted in a general trend of linguistic change in Chinese. As observed by Chinese linguists, the split of [VþR] (R ¼ resultative component) from [VþV] got its start in early Medieval Chinese,11 through metaphorical extension from the conceptual domain of temporality to that of causality. According to several previous studies,12 dozens of verbs in the second position of [VþV] underwent change and became part of the resultative component of [VþR] in Medieval Chinese, and they were mostly achievement verbs and stative verbs. 来 (la´i) was originally an achievement verb, and thus, according to the observed pattern, was a candidate for the change. The following examples confirm that speakers of Medieval Chinese indeed regarded it as a resultative component: (17) 王告大臣，汝等方便引水令来。(十诵律 Shı´so`nglu`¨ , 5th c.) wa´ng ga`o da`-che´n ruˇ-deˇng fa#ng-bia`n yıˇn-shuıˇ-lı`ng-la´i king tell minister you-PL be-convenient induce-water-cause-come ‘The king told his ministers that you, in the ways that are convenient to you, induce water to be here.’ (18) 即以神力，感魔使来。(贤愚经 Xia´nyu´jı #ng, 5th c.) jı´ yıˇ she´n-lı` gaˇn-mo´-shıˇ-la´i right-away by spiritual-power move-demon-cause-come ‘(He) brought the demon here right away by spiritual power.’ 11. For a review on the rise of [VþR], see Liu (2002). 12. See Liu (1999), Wei (2000), and Zhao (2000).

A construction grammar approach

99

(19) 皆是他方剎土风吹使来。(增壹阿含经 Ze#ngyı #a#ha´njı #ng, 4th c.) jie# shı` ta#-fa#ng cha` tuˇ fen# g chuı-shı # ˇ-la´i all be elsewhere country sand wind blow-cause-come ‘All are sands from other countries blown here by the wind.’ [V þ 令 (lı`ng)/使 (shıˇ ) þ R] is a marked variation of [VþR] in Medieval Chinese, with 令 (ling) or 使 (shi) marking the component after it to denote the resultative state of the event indicated by the preceding verb. 来 (la´i) in 引水令来 ( yıˇn-shuıˇ-lı`ng-la´i) ‘to induce water to be here,’ 感魔 使来 (gaˇn-mo´-shıˇ-la´i) ‘to cause the demon to be here,’ and 吹使来 (chuı-# shıˇ-la´i) ‘to blow (the sands) (to be) here’ testify to 来 (la´i)’s being the R of [VþR]. It also occurs in the [VþR] compounds as R, as shown in (20)–(21): (20) 问：‘‘冬何得此葵？’’ 云 ‘‘从日南买来。 ’’ (齐民要术 Qı´mı´nya`oshu`, 6th c.) we`n do#ng he´ de´ cıˇ kuı´ yu´n co´ng rı`-na´n maˇi-la´i ask winter how obtain this sunflower answer from NAME buy-come ‘(One) asked: ‘‘How can sunflowers of this kind be obtained in the winter?’’ The other answered: ‘‘They were bought and brought here from Ri-nan.’’’ (21) 遣人捕来，勅使杀之。(贤愚经 Xia´nyu´jı #ng, 5th c.) qiaˇn re´n buˇ-la´i chı` shıˇ sha# zhı # dispatch person catch-come order cause kill him ‘(The king) dispatched someone to arrest him and bring him here and gave the order to kill him.’ What 来 (la´i) denotes in 买来 (maˇi-la´i) ‘to buy sth. to be here’ and 捕来 (buˇ-la´i) ‘to arrest sb. to be here’ is the resultative state of directionality, the resultative part being derived from the construction meaning of [VþR] in general, and the directionality part, from the meaning of the component 来 (la´i) in particular. Let us compare (20)–(21) with (16). 来 (la´i) in 下来 (xia` la´i) of (16) indicates an instantaneous completion of the downward movement denoted by the preceding verb 下 (xia`), which is more grammaticalized than the resultative component. That is, it shifts further towards grammatical function than the 来 (la´i) in 买来 (maˇi-la´i) and 捕来 (buˇ-la´i) of (20)–(21) does. The di¤erence is attributable to the properties of the preceding verbs. The verb in (16) is a directional verb, and the sense of directionality is included in the verb’s denotation. When 来 (la´i) occurs in [VDIRECTIONAL þ 来 (la´i)RESULTATIVE], this very sense, also inclusive in 来 (la´i)VERB, becomes

100

Cheng-hui Liu

redundant and fades out, which results in the eventual emergence of [VDIRECTIONAL þ 来 (la´i)PERFECTIVE]. On the other hand, the verbs in (20)–(21) are non-directional verbs 买 (maˇi) and 捕 (buˇ ) and do not cause a similar e¤ect. In these cases, 来 (la´i) retains the sense of directionality as a resultative state of the preceding event. In principle, all the instances of [VþV] where the second component has a resultative implication are under the influence of the above-mentioned trend in Medieval Chinese, i.e. to evolve into [VþR]. But the instances of [VþV] in circulation before the trend can resist change, and, therefore, the long-existing forms and the new ones end up being mixed together. This explains why 下来 (xia` la´i) in (2) (which may still belong to [VþV]), and 下来 (xia-lai) in (16), an instance of [VDIRECTIONAL þ来 (la´i)PERFECTIVE]) which is further developed from [VþR], co-exist in Medieval documents.13 6. The perfect aspect marker 来 (la´i)

来 (la´i) is also a marker of the perfect aspect, written as 来 (la´i) PERFECT below. Both 来 (la´i) PERFECTIVE and 来 (la´i) PERFECT are related to the concept of completion, yet from di¤erent perspectives. The former has to do with the completion of an event, and the latter, the situation at a temporal reference point, usually the time of speaking, which results from a prior event or situation. The former is attached to verbs; the latter, to predicates. 来 (la´i) PERFECT is derived from 来 (la´i) VERB via [PþN/VþC þ来 (la´i)], where P stands for preposition, and C, for conjunction.14 This construction 13. 来 (la´i) is the second component of certain disyllabic directional markers such as 上来 (sha`ng-la´i) ‘up; upwards,’ 下来 (xia`-la´i) ‘down; downwards,’ 出来 (chu#-la´i) ‘out; outwards,’ 进来 ( jı`n-la´i) ‘in; inwards,’ and the second component of the disyllabic inchoative marker of 起来 (qıˇ-la´i), in Mandarin. Their common origin should be [VDIRECTIONAL þ来 (la´i) PERFECTIVE]. 14. Liang (2007: 125–126) relates 来 (la´i) with 以来 ( yıˇ-la´i), and suggests that the ‘auxiliary word of sentential aspect’ (事态助词 shı`-ta`i zhu`-cı´ ), an equivalent of the perfect aspect marker, 来 (la´i) is derived from the phrase ‘V(þNP)þ 来 (la´i),’ with 来 (la´i) (>以来 ( yıˇ-la´i)) as a ‘temporal directional word’ (时 间方位词 shı´-jia#n fa#ng-we`i-cı´ ) in early Medieval Chinese. We trace its origin even further back to Archaic Chinese, and propose that 来 (la´i) began its shift to a temporal marker before Medieval Chinese. Furthermore, Liang (2007: 133–134) thinks it possible that the phase complement (动相补语 do`ng-xia`ng buˇ-yuˇ ) 来 (la´i) (which emerges shortly after Medieval Chinese) may shift into a perfect aspect marker in certain contexts, for the two functions are notionally close to each other. Because 来 (la´i) in a di¤erent function undergoes change in a di¤erent construction, this possibility should be negated.

A construction grammar approach

101

was originally used to express directional movement, as exemplified in (22): 自外而来 (zı` wa`i e´r la´i) ‘to come from the outside’ consists of 自 (zı` ) ‘from,’ a preposition indicating the point of departure of a spatial movement, 外 (wa`i) ‘outside,’ referring to the place of departure, 而 (e´r) ‘and,’ a conjunction that links the place of departure and the place of arrival ‘here,’ which is implied by 来 (la´i). 来 (la´i) denotes directional movement, and thus is a directional verb. (22) 其夫早自外而来，士适出。(韩非子 Ha´nfei# zıˇ, 3rd c. BC) qı´ fu# zaˇo zı` wa`i e´r la´i shı` shı` chu# her husband early from outside CONJ come gentleman just go-out ‘Her husband came in from outside early, just as the gentleman went out.’ (23) 由周而来，七百有余岁矣。(孟子 Me`ngzıˇ, 3rd c. BC) yo´u Zho#u e´r la´i from the-Zhou-dynasty CONJ come qı # baˇi yo`u yu´ suı` yıˇ seven hundred more year MOD ‘It has been more than seven hundred years since the Zhou dynasty.’ The spatial concept of the construction is further mapped onto the temporal concept, as displayed in 由周而来 ( yo´u Zho#u e´r la´i) ‘since the Zhou dynasty’ in (23), where 由 ( yo´u) ‘from’ is the preposition, 周 (Zho#u), the point of departure of the temporal movement, 而 (e´r), the conjunction, and 来 (la´i), the directional verb which implies the end point ‘now.’ The mapping is accompanied by a subtle semantic shift in 来 (la´i), as can be seen in (24). 由孔子而来 ( yo´u Koˇng-zıˇ e´r la´i) ‘since the time of Confucius’ falls under the construction, except that 来 (la´i) shifts and focuses on the sense of directionality, and the end point in this case is indicated by the following phrase 至于今 (zhı` yu´ jın) # ‘arriving at the present.’ It signifies an increase of prominence of the directionality of 来 (la´i) in the temporal construction, which is consistent with the tendency of the semantic shift of 来 (la´i) VERB from an achievement verb to a directional one. (24) 由孔子而来，至于今百有余岁。(孟子 Me`ngzıˇ, 3rd c. BC) yo´u Koˇng-zıˇ e´r la´i zhı` yu´ jın# baˇi yo`u yu´ suı` from Confucius CONJ come arrive to now hundred and plus year ‘It has been a hundred-plus years since the time of Confucius.’

102

Cheng-hui Liu

It is worth noting that the object of the preposition is allowed to be a nominalized verb phrase, as can be seen in (25): the verb phrase 有生民 ( yoˇu she#ng-mı´n) ‘there exists mankind’ serves as the object of the preposition 自 (zi). (25) 自有生民以来，未有孔子也。(孟子 Me`ngzıˇ, 3rd c. BC) zı` yoˇu she#ng-mı´n yıˇ la´i we`i yoˇu Koˇng-zıˇ yeˇ From exist mankind CONJ come never exist Confucius MOD ‘Since the dawn of man, there has never been a person the equal of Confucius.’ In the early instances of [PþN/VþC þ 来 (la´i)], P can be 由 (yo´u), 自 (zı`), or 从 (co´ng) alternatively, and C can be 而 (e´r) or 以 ( yıˇ ). Although the actual morphemes are alterable, P and C are both obligatory. In Medieval Chinese, P and C became optional, and 来 (la´i) became a more functional component. When the construction was first formed, 来 (la´i) was a verb; after further conventionalization, 来 (la´i) as a temporal marker became more prominent, and began to lose its original verbal content. It is at this point in the process that P and C turn redundant, and, subsequently, optional. In Medieval Chinese, the construction split into several variations, among which [PþV þ 来 (la´i)] and [V þ 来 (la´i)], as exemplified by 从生来 (co´ngshe#ng-la´i) ‘from birth till now’ in (26) and 生来 (she#ng-la´i) ‘birth till now’ in (27), respectively, are closely related to the innovation of 来 (la´i) PERFECT. (26) 我从生来无有大过。(贤愚经 Xia´nyu´jı #ng, 5th c.) woˇ co´ng-she#ng-la´i wu´ yoˇu da`-guo` I from-be-born-TEMPORAL NEG exist serious-mistake ‘I have made no serious mistakes since I was born.’ (27) 但念生来未曾妄语。(贤愚经 Xia´nyu´jı #ng, 5th c.) da`n nia`n she#ng-la´i we`i ce´ng wa`ng yuˇ only think-of be-born-TEMPORAL never ever unfounded words ‘I believe myself never to have talked nonsense since I was born.’ When C is obligatory in the construction, V and 来 (la´i) are separated by C, and there is no room for semantic interaction between them. When C is omitted after becoming optional, the temporal component 来 (la´i) gets positioned next to V, where its verbal property will be further reduced. Consequently, 从生来 (co´ng-she#ng-la´i) in (26) degrades from a verb phrase to a prepositional phrase.

A construction grammar approach

103

Further change occurs when P in [PþV þ 来 (la´i)] is omitted. In principle, there is more than one way to change. Let’s take (26) as an example and examine what would happen after the omission of P. One of the possibilities is that 来 (la´i) will remain a temporal marker, and the impact of the change will fall upon the phrasal relationships. That is, 从生来 (co´ngshe#ng-la´i) is an adverbial adjunct of the following predicate 无有大过 (wu´ yoˇu da`-guo`) ‘to make no serious mistakes,’ and after the omission of 从 (co´ng), 生来 (she#ng-la´i) turns out to be its topic, as shown in Figure I:

Figure I. Change of [V þ来 (la´i)] – Option 1

Another possible way for it to change is for 生来 (she#ng-la´i) to be associated with the topic 我 (woˇ ). Subsequently, 我生来 (woˇ she#ng la´i) serves as topic/protasis, and 无有大过 (wu´ yoˇu da`-guo`), as predicate/apodosis, as shown in Figure II:

Figure II. Change of [V þ来(la´i)] – Option 2

In practice, both of the possibilities are realized in Medieval Chinese, as can be seen in the following examples, with (28) corresponding to Figure I, and (29) to Figure II. (28) 彼二人者，亡来七日。(中本起经 Zho#ngbeˇnqıˇjı #ng, 3rd c.) bıˇ e`r re´n zheˇ wa´ng-la´i qı # rı` that two person TOP die-TEMPORAL seven day ‘Those two persons, it has been seven days since they died.’

104

Cheng-hui Liu

(29) 是等诸幻我见来，以是意中不贪乐。 (佛本行集经 Fo´beˇnxı´ngjı´jı #ng, 5th c.) shı`-deˇng zhu# hua`n woˇ jia`n la´i this-PL all-kinds-of mirage I see PERFECT yıˇ shı` yı`-zho#ng bu` ta#n le` by this in-heart NEG be-vehement-for happiness ‘Ever since I have seen these kinds of mirages, I have had no vehement desire for happiness.’ In (28), 彼二人者 (bıˇ e`r re´n zheˇ ) ‘those two persons’ serves as the topic of 亡来七日 (wa´ng-la´i qı # rı` ) ‘their being dead is seven days,’ in which 亡来 (wa´ng-la´i) ‘their being dead’ is the topic of the nominal predicate 七日 (qı # r`ı ) ‘seven days.’ In (29), 是等诸幻我见来 (shı`-deˇng zhu# hua`n woˇ jia`n la´i) ‘I have ever seen these kinds of mirages,’ ending with 来 (la´i) PERFECT, is the protasis, and 以是意中不贪乐 ( yıˇ shı` yı`-zho#ng bu` ta#n le`) ‘because of this, in my heart, there is no vehement desire for happiness’ is the apodosis of the conditional sentence. The protasis-apodosis relationship is merely a random outgrowth of the change of [PþV þ 来 (la´i)] > [V þ 来 (la´i)]. 来 (la´i) PERFECT, however, is not limited to the relationship, as shown in (30)–(32): (30) 吾已食来，不须复办。(法句譬喻经 Faˇju`pı`yu`jı #ng, 4th c.)15 wu´ yıˇ shı´ la´i bu` xu# fu` ba`n I already eat PERFECT NEG need again prepare ‘I have already taken a meal; there is no need to prepare one again.’ 15. This example is cited from Liang (2007: 132), yet his opinion about the function of 来 (la´i) di¤ers from ours. Liang (p. 133) believes it to be a perfective aspect marker as it indicates completion of the action that the verb denotes. But, as mentioned in the beginning of the section, the perfect marker 来 (la´i) also connects with the notion of complement. Liang admits, however, that (30) is the one and only instance that he was able to find out of a large number of Medieval documents wherein 来 (la´i) indicates the perfective aspect marker. As far as we are concerned, the perfective aspect marker 来 (la´i) in Medieval Chinese is restricted to [VDIRECTIONAL þ来 (la´i) PERFECTIVE]. It seems to us that Liang treats 来 (la´i) in (30) di¤erently from that of (31) because of the verbs: 来 (la´i) in (30) collocates with 食 (shı´ ), a verb of action, and in (31), with 亡 (wa´ng), a verb of resultative state. In our opinion, the perfect aspect marker in Chinese may go with all kinds of predicates, including predicates headed by verbs of action, and 来 (la´i) in (30) is the same as that in (31). For more on the properties of perfect aspect markers in Archaic and Medieval Chinese, see Liu (2007, 2010).

A construction grammar approach

105

(31) 家人皆怪之曰：‘‘玄石亡来，服以阕矣。 ’’ (搜神记 So#ushe´njı`, 4th c.) jia#-re´n jie# gua`i zhı # yue# Xua´n-shı´ wa´ng la´i family-member all be-strange it say-that NAME die PERFECT fu´ yıˇ que` yıˇ mourning-clothes already be-left-o¤ MOD ‘The family members all felt strange, saying: ‘‘Xuan-shi died, and the mourning clothes have already been taken o¤.’’’ (32) 问：‘‘东吴有长柄壶卢，卿得种来不？’’ (世说新语 Shı`shuo#xı #nyuˇ, 5th c.) we´i we`n Do#ng-wu´ yoˇu cha´ng-bıˇng hu´-lu´ only ask the-Dong-wu-state exist long-stem gourd qı #ng de´ zhoˇng la´i foˇu you cause-grow PERFECT NEG ‘(He) merely asked: ‘There are long-stem gourds in Dong-wu; have you planted any or not?’’’ Both (30) and (31) consist of two coordinating clauses in temporal order, where 吾已食来 (wu´ yıˇ shı´ la´i) ‘I have already taken meal’ and 玄石亡来 (Xua´n-shı´ wa´ng la´i) ‘Xuan-shi died’ are independent of the subsequent clauses. In (32), the clause 卿得种来不 (qıng # de´ zhoˇng la´i foˇu) ‘have you planted any or not’ is in juxtaposition with the clause before it. 来 (la´i) PERFECT in (30)–(32) indicates a ‘perfect of result,’ in Comrie’s (1976) terms, and it can indicate an ‘experiential perfect’ and a ‘perfect of the recent past,’ too, as exemplified in (33)–(34) respectively: (33) 我昔曾闻是事来，现在我身亲自见。 (佛本行集经 Fo´beˇnxı´ngjı´jı #ng, 6th c.) woˇ xı # ce´ng we´n shı` shı` la´i I past ever hear this thing PERFECT xia`n-za`i woˇ-she#n qın-zı # ` jia`n now my-body in-person see ‘I have heard of this kind of thing in the past; now I see it in person.’ (34) 翠篾初织来，或如古鱼器。(酒趋 Jiuˇqu#, 皮日休 Pı´, Rı`xiu#, 834–883) cuı` mie` chu# zhı # la´i green thin-bamboo-strip just weave PERFECT huo` ru´ guˇ yu´-qı` slightly like ancient fishing-gear ‘The green thin bamboo strips were just woven; it is a bit like ancient fishing gear.’

106

Cheng-hui Liu

曾 (ce´ng) ‘ever’ in (33) is an experiential marker. 初 (chu#) ‘just’ in (34) indicates that the event happened recently. Examples like these support the analysis that 来 (la´i) serves as marker of the perfect aspect in the sentencefinal position of non-imperative sentences. 7. The post-Medieval Chinese duration/continuation marker 来 (la´i) As mentioned in the previous section, the change of [PþV þ 来 (la´i)] > [V þ 来 (la´i)] gives rise to 来 (la´i) PERFECT in Medieval Chinese. Out of this particular change also comes a duration marker: 来 (la´i) DURATION. Let’s return to (26) and examine a third possibility of change, as illustrated in Figure III:

Figure III. Change of [V þ来 (la´i)] – Option 3

The prepositional phrase 从生来 (co´ng-she#ng-la´i) is an adverbial adjunct of the predicate 无有大过 (wu´ yoˇu da`-guo`) with 从 (co´ng) indicating the initial point in time,16 and 来 (la´i), the ending point in time ‘now.’ After changing, the adverbial function is likely to be retained in 生来 (she#ng-la´i) for two reasons. The first reason is that [PþV þ 来 (la´i)] and [V þ 来 (la´i)] co-existed in Medieval Chinese, and the pairs with or without P, such as 从生来 (co´ng-she#ng-la´i) and 生来 (she#ng-la´i), were used interchangeably for hundreds of years. It is, therefore, a reasonable inference that 生来 (she#ng-la´i) took over the function of 从生来 (co´ng-she#ng-la´i) after the latter ceased to exist. Secondly, 来 (la´i), no matter whether in [PþV þ 来 (la´i)] or [V þ 来 (la´i)], is a temporal marker, and the notion of temporality is usually associated with the function of adjunct. The replacement of [PþV þ 来 (la´i)] by [V þ 来 (la´i)] might have begun shortly after Medieval Chinese, as shown in (35)–(36), which are drawn from the documents of the seventh to ninth centuries.

16. See Meisterernst (2004: 222).

A construction grammar approach

107

(35) 孟子死来江树老，烟霞犹在玉门山。 (襄阳过孟浩然故居 Xia#ngya´ng guo` Me`ng-Ha`ora´n gu`ju#, 陈羽 Che´n, Yuˇ, 8th–9th c.) Me`ng-zıˇ sıˇ-la´i jia#ng shu` laˇo NAME die-DURATION river tree be-old ya#n xia´ yo´u za`i Yu`-me´n-sha#n mist rosy-cloud still be-in-location Yu-men-Mountain ‘The trees along the river have aged since the death of Meng-zi, yet mist and rosy clouds still surround the Yu-men Mountain.’ (36) 洛下闲来久，明朝是十年。 (岁除夜对酒 Suı`chu´ye` duı` jiuˇ, 白居易 Ba´i, Ju#y`ı, 772–846) Luo`-xia` xia´n-la´i jiuˇ the-Lo-area become-unoccupied-DURATION long mı´ng-zha#o shı` shı´ nia´n tomorrow-morning be ten year ‘I have long been unoccupied in Lo; it will be ten years tomorrow morning.’17

来 (la´i) in 死来 (sıˇ-la´i) ‘since becoming dead’ and 闲来 (xia´n-la´i) ‘since becoming unoccupied’ of (35)–(36) marks o¤ the ‘duration’ between the event indicated by the verb and the time of utterance. It further extends to mark o¤ ‘continuation’ of the state indicated by the verb,18 as in 闲来 (xia´n-la´i) ‘being at leisure’ and 深来 (she#n-la´i) ‘being far-reaching’ in (37)– (38):

17. According to the convention of Tang poetry, the first person subject is implied in the verse. 18. 来 (la´i) DURATION and 来 (la´i) CONTINUATION are especially used in poetry due to the rhetorical motivations of the poetic genre. However, we do not exclude the possibility that both functions can be more easily triggered in poetry, rather than in daily language, and thus be stylistic variations. Even so, the variations have to arise from the fact that [PþV þ来 (la´i)] is on the verge of change, and, more importantly, their shift into markers of duration and continuation must not violate the contemporary linguistic trend. In addition, Jiang (1980: 111–112) points out that 来 (la´i) frequently acts as a marker of continuation or perfective aspect in the poems of the Tang dynasty. Due to the limitations of space in the present paper, we only discuss how 来 (la´i) turns into a marker of continuation.

108

Cheng-hui Liu

(37) 闲来对镜自思量，年貌衰残分所当。 (对镜偶吟 Duı` jı`ng oˇu yı´n, 白居易 Ba´i, Ju#y`ı, 772–846) xia´n-la´i duı` jı`ng zı` sı #-lia´ng at-leisure-CONTINUATION at mirror myself think nia´n ma`o shua#i ca´n fe`n suoˇ da#ng age appearance frail broken propriety SUO fit ‘I assessed myself while facing the mirror at leisure, recognizing that the frail and broken appearance was fitting for my age.’ (38) 嫖姚何日破重围，秋草深来战马肥。 (闻官军既至未睹凯旋 We´n gua#nju#n jı` zhı` we`i duˇ kaˇixua´n, 韦庄 We´i, Zhua#ng, 836–910) pia´o-ya´o he´ rı` po` cho´ng-we´i commander what day break-through multi-layered-blockade qiu# caˇo she#n-la´i zha`n-maˇ fe´i autumn weed far-stretching-CONTINUATION war-horse stout ‘What date is the commander going to break through the multilayered blockade of the enemy, while the weeds are far-reaching in late autumn, and the war-horses grow stout?’ The shift from 来 (la´i) DURATION to 来 (la´i) CONTINUATION should be initiated by instances like 闲来 (xia´n-la´i), where the verb relates to both the notions of event and state. In short, when the verb of [V þ 来 (la´i)] denotes an event (such as 死 (sıˇ) ‘to die’ in (36)), 来 (la´i) indicates duration; when it denotes a state (such as 深 (she#n) ‘to be far-stretching’ in (38)), 来 (la´i) indicates continuation. Whenever it allows the denotations of either event or state, such as with 闲 (xia´n) ‘to become unoccupied; to be at leisure’ in (36)–(37), 来 (la´i) carries out a function with respect to the property of the verb in its context. In the original construction, the temporal marker 来 (la´i) goes with verbs that denote events only; after the shift, 来 (la´i) outgrows its original limit of collocation by way of its co-occurrence with dual function verbs such as 闲 (xia´n).

8. Conclusion According to the discussion in Sections 4–7, the historical changes of 来 (la´i) can be diagrammed as follows:

A construction grammar approach

109

Diagram I. The splitting of the verb 来 (la´i)

As shown in Diagram I, 来 (la´i) VERB shifts to 来 (la´i) HORTATIVE when occurring at the end of an imperative sentence. It shifts to 来 (la´i) PERFECTIVE when it comes immediately after another directional verb in the frame of [VþR]. The construction [PþN/VþC þ来 (la´i)] is crucial to the development of 来 (la´i), and one of its descendants [PþV þ 来 (la´i) TEMPORAL] brings forth some more changes. 来 (la´i) PERFECT comes into being, when the omission of P induces a structural reanalysis wherein the verb in [V þ 来 (la´i) TEMPORAL] is treated as an immediate constituent to the topic of the sentence. Moreover, 来 (la´i) DURATION and 来 (la´i) CONTINUATION are innovated to be the adjunct of the succeeding predicate after [V þ 来 (la´i) TEMPORAL] replaces [PþV þ 来 (la´i) TEMPORAL]; 来 (la´i) DURATION collocates with verbs that denote events, and 来 (la´i) CONTINUATION, with those that denote states. Both [VEVENT þ VDURATION] and [VSTATE þ VCONTINUATION] are adjuncts, not able to serve as independent predicates. What if the frame of constructions were to be ignored? The routes of changes might be realized as shown in Diagram II. These two diagrams reflect di¤erent views on how a lexical item becomes grammaticalized into a functional marker. The latter understands the item under change to be a free morpheme, while the former treats it as an inseparable part of its construction. Diagram II provides an over-simplified conclusion, and no actual studies on the grammaticalization of 来 (la´i) look like this. On the other hand, no other studies on 来 (la´i) come to the same conclusion we have, as shown in Diagram I, either.

110

Cheng-hui Liu

Diagram II. The routes of changes of 来 (la´i)

Most of the previous studies place particular emphasis on the serial verb construction [VþV]. But, in reality, the imperative sentence type and [PþN/VþC þ 来 (la´i)] are of equal importance in setting up frames for changes in 来 (la´i). Over-emphasis on [VþV] and ignorance about the existence of other constructions may lead to all kinds of mistaken conclusions. Let’s first compare (27), where 生来 (she#ng-la´i) makes no di¤erence in instances of [VþV] on the surface, with (16), where 下来 (xia`-la´i) has the same surface form as 下来 (xia` la´i), an instance of [VþV]. Suppose we overlook the fact that 生来 (she#ng-la´i) comes from [V þ 来 (la´i)] (< [PþV þ 来 (la´i)] < [PþN/VþC þ 来 (la´i)]) and that 来 (la´i) PERFECTIVE is restricted to [VDIRECTIONAL þ 来 (la´i) PERFECTIVE] in Medieval Chinese. In this scenario, [V þ 来 (la´i) PERFECTIVE] and [V þ 来 (la´i) DURATTION] cannot help but be attributed to a common origin of [VþV], and such questions as how they split from [VþV] or which is derived from the other, might arise by mistake. Moving on to 来 (la´i) HORTATIVE and 来 (la´i) PERFECT, both of them appear in the sentence-final position, with only the former appearing in imperative sentences, and the latter, in sentences of other types. Without information about the construction in which 来 (la´i) PERFECT is formed, it could possibly be inferred from the idea of complementary distribution, that one is derived from the other, and, following this inference, a misleading question about the motivation of a hortative marker turning into a perfect marker might arise, since 来 (la´i) HORTATIVE first appears in Archaic Chinese, and 来 (la´i) PERFECT emerges in Medieval Chinese. Problems of this sort are inevitable at least until all of the constructions relevant to the changes of 来 (la´i) have been made explicit. Returning to 生计抛来诗是业 (she#ng-jı` pa#o-la´i shı # shı` ye`), the confusing case given in Section 1, we note a few things in the original document: It is paired with the couplet 家国忘却酒为乡 ( jia#-guo´ wa`ng-que` jiuˇ we´i xia#ng) ‘the country has been forgotten, and wine is my homeland’ in a

A construction grammar approach

111

poem, where 抛来 ( pa#o-la´i) and 忘却 (wa`ng-que`) are a parallelism. Since 却 (que`) is generally recognized as a perfective aspect marker, 来 (la´i) tends to be taken as performing the same function. But the question is whether 来 (la´i) must perform an identical function with 却 (que`) just because of the parallelism in the poem. The parallelism rather ascertains that 来 (la´i) and 却 (que`) occupy a similar grammatical status but the actual function of 来 (la´i) depends on the construction it belongs to. According to the discussion in Section 7, 抛来 ( pa#o-la´i) belongs to [VEVENTþVDURATION], and 来 (la´i) here is a marker of duration. On top of providing resolutions to confusing cases as well as preventing erroneous questions over grammaticalization, the constructionist view seems the only available perspective to explain why the 来 (la´i) VERB has remained vital over an extraordinarily long period of time, without obvious influence from all the changes relating to it. The reason for its long life may be that the form of the verb remains at all times a free morpheme and change takes place only after the form has stepped into specific constructions and deviated from its original form.

References Cao, Guangshun 1995 Jindai Hanyu Zhuci [Auxiliary Words in Pre-Modern Chinese.] Beijing: Yuwen Chubanshe. Comrie, Bernard 1976 Aspect. Cambridge: Cambridge University Press. Goldberg, Adele 1995 A Construction Grammar Approach to Argument Structure. Chicago: The University of Chicago Press. Jiang, Shaoyu 1980 Dushi ciyu zhaji [Some remarks on the poems of dufu] Yuyanxue Luncong [Linguistic Review] 6: 94–127. Beijing: Shangwu Yinshuguan. Li, Ming 2004 Quxiang dongci ‘lai / qu’ de yongfa ji qi yufahua [The usages of directional verbs ‘lai / qu’ and their grammaticalization.] Yuyanxue Luncong [Linguistic Review] 29: 291–313. Beijing: Shangwu Yinshuguan. Liang, Yinfeng 2007 Hanyu Quxiang Dongci de Yufahua [Grammaticalization of Directional Verbs in Chinese.] Shanghai: Xuelin Chubaishe.

112

Cheng-hui Liu

Liu, Cheng-hui 1999 Shilun shichengshi de laiyuan ji qi chengyin [On the origins of the verb-resultative construction and the motivations of Change.] Guoxue Yanjiu [Studies of Chinese Culture] 6: 349–386. Liu, Cheng-hui 2002 Hanyu Dongbu Jiegou Lishi Fazhan [Historical Development of the Verb-Complement Construction in Chinese.] Taipei: Hanlu Tushu. Liu, Cheng-hui 2007 Xianqin ‘yi’ de gongneng ji qi fenhua [The functions of ‘yi’ and its semantic extensions in Pre-Qin Chinese.] Language and Linguistics 8(3): 743–766. Liu, Cheng-hui 2009 Shanggu dao zhonggu ‘lai’ de yanbian – Goushi guandian de yufa fenxi [The changes of ‘lai’ from Archaic to Medieval Chinese: An analysis from the perspective of constructions]. Paper presented at Workshop on Chinese Directionals: History and Dialectal Variation, August 26–27, 2009, Academia Sinica. Liu, Cheng-hui 2010 Zhonggu yijing ‘yi’ dui jindai ‘le’ de yingxing – Yuyan jiechu ruhe qiandong yufa yanbian [The influence of ‘yi’ in the translated Buddhist texts on the rise of the Chinese verb su‰x ‘le’: How language contact has impact on grammatical changes.] The Bulletin of the Institute of History and Philology 81(3): 467–512. Taipei: Academic Sinica. Liu, Jian, Kuangshun Cao, and Fuxiang Wu 1995 Lun youfa hanyu cihui yufahua de ruogan yinsu [On certain motivations of grammaticalization of lexical items in Chinese.] Zhonghuo Yuwen [Chinese Language] 1995(3): 161–169. Meisterernst, Barbara 2004 Temporal phrases in the Shiji: adverbials indicating a point of time. Cahiers de Linguistique Asie Orientale 33(2): 203–250. Wang, Jin-hui 2004 ‘Wang’ ‘Lai’ ‘Qu’ Lishi Yanbian Zonglun [On the Diachronic Changes of ‘Wang,’ ‘Lai ’ and ‘Qu.’] Taipei: Liren Suju. Wei, Pei-chuan 2000 Shuo Zhonggu Hanyu de Shichengshi [On the Causative Constructions in Middle Chinese.] The Bulletin of the Institute of History and Philology 71(4): 807–947. Wu, Fuxiang 1996 Dunhuang Bianwen Yufa Yanjiu [A Study on the Grammar of Dunhuang Bianwen.] Changsha: Yuelu Shushe. Wu, Fuxiang 2002 Nanfang fangyan jige zhuantai buyu biaoji de laiyuan (II) [Origin of some markers of stative complement in Chinese southern dialects (Part II).] Fangyan [Dialect] 2002(1): 24–34.

A construction grammar approach Xie, Huiquan 1987

113

Tan shici xuhua [On how content words become function words.] Yuyan Yanjiu Luncong [Language Research Review] 4: 208–227. Tianjin: Nankai Daxue Chubanshe.

Zhao, Changcai 2000 Hanyu Shubu Jiegou de Lishi Yanjiu [A Diachronic Study of the Verb-complement Construction in Chinese.] Ph.D. diss., Zhongguo Shehui Kexueyuan.

The degree-evaluative construction: Grammaticalization in constructionalization Mei-Chun Liu and Chun Chang 1. Introduction One of the important issues in recent linguistic research is the interaction between syntax and lexical semantics. The focus in such an approach is normally on verbs and nouns. The information structure of adjectives and adverbs has recived much less attention. To restore the missing piece in lexical semantic research, we take on the task of re-examining the nearly obligatory use of a degree adverb with a predicative adjective in Mandarin Chinese (hereafter MC). It is a known fact that, in MC, predicative adjectives are treated as stative predicates, which constantly collocate with the preceding degree adverb heˇn 很 ‘very, quite’: (1) 他 很高/很帥/很用功。 ta# heˇn ga#o/heˇn shua`i /heˇn yo`nggo#ng he DEG tall/DEG handsome /DEG hard-working ‘He is quite tall/handsome/hard-working.’ Traditionally, the above examples have been treated essentially as an adjectival phrase with the adjectival predicate as the structural head modified by the degree adverb heˇn 很. Previous studies have mostly centered on the lexical semantic constraints of the adjectival element: What types of adjectives or stative predicates might optionally or obligatorily occur in the pattern and what are the semantic or pragmatic features involved (see Zhu 1956; Lu¨ 1980; Chui 2000; Tang 2000a; Zhang 2002, among many others)?1 Moving in a totally di¤erent direction, this paper adopts a constructional approach (Kay 1995; Kay and Fillmore 1999; Goldberg 1995, 2003) and regards the sequence [Degree þ X] as a form-meaning pairing, 1. The literature on degree adverbs in MC, especially on the high-degree adverb heˇn 很 ‘very’, is plentiful. For detailed descriptive discussion, see Zhu (1956, 1982); Lu¨ (1980); Li and Thompson (1981); Tang (1989); Cheng (2000); Chui (2000); Tang (2000a); Tang (2000b); Zhang (2000), and the references cited therein.

116

Mei-Chun Liu and Chun Chang

taking the degree modifier as a constructional operator which requires and coerces a quality-implicating element in the following slot, signaling an attributive evaluation of the predicated subject. The radically di¤erent analysis proposed in this study is based on observations of a large amount of naturally-occurring data, consisting of contemporary corpora (Sinica Corpus, Google and udndata.com) and conversational data (139 minutes). The data reveal that a surprisingly diverse range of grammatical categories can occur in the X slot of the [Degree þ X] pattern, among them proper nouns, non-stative verbs, and demonstrative pronouns, which were previously considered to be incompatible with degree modification. The elements found to co-occur with a degree modifier cover a full range of grammatical categories: NP, AP, VP, ADVP, DP and even the interrogative category,2 as briefly illustrated below: (2) with nouns:

很陳水扁。 ta# heˇn che´n-shuıˇ-biaˇn he DEG Name ‘He is very much like Shuibian Chen (the former President).’

(3) with verbs:

他很殺。

ta# heˇn sha# he DEG kill ‘He is quite killing (attractive or forceful).’ (4) with determiners:

他很那個。

ta# heˇn na`ge he DEG that ‘He is very that (inferrable from context).’ Diverging from studies based on lexical properties, we find that a construction-based approach renders a unified, non-compositional account of the seemingly heterogeneous data, as quite successfully proved in other 2. Abbreviations used in this paper are as follows: X ¼ variable; NP ¼ noun phrase; AP ¼ adjective phrase; VP ¼ verb phrase; ADVP ¼ adverb phrase; DP ¼ determiner phrase; ASP/DUR ¼ aspect marker; BEI ¼ passive marker; DE ¼ genitive or relative marker; LIAN ¼ focus particle; CL ¼ classifier; SFP ¼ sentence-final particle; and SIP ¼ sentence-initial particle.

Grammaticalization in constructionalization

117

studies (cf. Fillmore et al. 1988; Goldberg 2006; Jackendo¤ 1997; Su 2002b; Lai 2003; Biq 2004; Liu 2005, Lien 2005, 2006; Boas 2007, 2008, and the references cited therein).3 The predicational pattern under study is thus redefined as the Degree-Evaluative Construction (henceforth DEC), associating the syntactic sequence of a degree modifier plus an attributeimplicating constituent [Degree þ X] with the semantic property of qualitative evaluation. The construction is partially filled with an overtly marked and lexically specified degree adverb, typically heˇn 很 ‘very’ as the default choice. The degree modifier is thus construed as the constructional operator that coerces the constructional meaning of degree evaluation over an attributive quality implicated by the co-occurring element. To further account for the ‘quality-implicating’ mechanism of the post-head element, principles of the Qualia Structure, proposed in Pustejovsky (1995), are utilized to derive four related senses, namely, formal, constitutive, telic and agentive properties, as the basis for meaning coercion. This constructional account has a deeper theoretical implication for research on grammaticalization. In deriving the Degree Evaluative Construction, the degree adverbial is grammaticalized as the constructional operator in the process of constructionalization. Its grammatical status changes from a lexical adverbial expressing a scalar degree to a grammatical marker signaling the DEC constructional function. What we see here is a process of grammaticalization realized in the process of constructionalization, an interesting phenomenon that has not been fully addressed in previous literature. This paper is structured as follows. Section 2 reviews some relevant observations on the relationship between the degree modifier and the postadverbial element as commonly cited in the literature. Section 3 introduces the sources of the data and the distribution of the target pattern in the collected data. Section 4 discusses the findings and accounts for the diverse 3. In the sense of Construction Grammar (Goldberg 1995, Jackendo¤ 1999?), a construction is considered to be a form-meaning pair that exists independently of the individual words (components). As a basic form-meaning bearing unit, the syntactic configuration of a construction contributes semantic content above and beyond what is contained in the constituent lexical items. In the past two decades, scholarly works tackling linguistic issues based on Construction Grammar have been legion. For studies in English, see Fillmore et al. 1988; Goldberg 1995, 2006; Kay 1995; Michaelis and Lambrecht 1996; Jackendo¤ 1997, 2007, 2008; Kay and Fillmore 1999; Goldberg and Jackendo¤ 2004; Boas 2007, 2008. For those in Mandarin, see Biq 2004; Su 2002b; Liu 2005; Chang and Lien 2009. For those in Taiwanese, see Lien 2005, 2006. For those in Hakka, see Lai 2003, 2006.

118

Mei-Chun Liu and Chun Chang

range of categories occurring in the X slot in [Degree þ X], from the perspectives of Construction Grammar (Kay 1995; Kay and Fillmore 1999; Goldberg 1995, 2003; Jackendo¤ 1997, 2008, etc.) and Qualia Structure a` la Pustejovsky (1995). Finally, in Section 5 theoretical implications are discussed before conclusions are drawn. 2. The range of co-occurring post-adverbial categories Scholarly works of the past five decades on the relation between degree adverbs (e.g. in particular, the adverb heˇn 很 ‘very’) and the post-adverbial categories in MC can be divided in two ways. One trend is descriptionoriented, with the relevant studies aiming to exhaust the distributional features for the co-occurrence of degree adverbs with stative and/or psychological predicates. The consensus reached in these studies is that the categories allowed to follow a degree adverbial are grammatically and semantically restricted and that the degree adverb may render a scalar property pertaining to semantic polarity of the modified element. For example, Tang (1989) claims that, as a modifying adjunct, heˇn 很 is used to specify a relatively stronger degree with positive polarity within the modificational scope. Similar claims can also be found in later studies (see Tang 2000a, 2000b; Zhang 2000; Zhang 2002).4 The other trend is function-oriented and characterized by studies that aim to explicate the actual uses of the degree adverb in daily conversations or compiled corpora (e.g. Chui 2000; Cheng 2000). For example, Cheng (2000), by analyzing the Sinica Corpus, points out that degree adverbs such as heˇn 很 ‘very’ can collocate with auxiliary verbs denoting ability or epistemic meaning (e.g., ne´ng 能 ‘can’ or keˇne´ng 可能 ‘possibly’). Chui (2000) examines the semantic and morphosyntactic changes of the degree adverb heˇn 很 as observed in spoken discourse: heˇn 很 undergoes a process of semantic shift from a free content word (with a degree sense) to a bound constituent (via cliticization and compounding). As a full-fledged degree adverb, heˇn 很 is often used for communicative purposes and semantically indicates the extent of gradation as being higher than the expected average. 4. Tang (1989) di¤erentiates the syntactic behaviors of heˇn 很 using a semantic explanation. The main concern is that heˇn 很 ‘very’ can render a standard of comparison to its modified element (adjective) to license the adjective in the a‰rmative sentence (e.g. ta# *(heˇn) ga#o 他很高 ‘he is *(very) tall’). For more detailed descriptive accounts, see Tang (2000a); Tang (2000b); Zhang (2000) and Zhang (2002).

Grammaticalization in constructionalization

119

However, the data we collected show a quite di¤erent picture. The degree modifier heˇn 很 is able to coexist with a surprisingly wide range of grammatical as well as semantic categories. Its distribution with a postadverbial element appears to be much less restricted than previously thought. To account for the flexible use of the degree marker in the pattern, a systematic account is needed to provide an encompassing explanation for the observed phenomena and, at the same time, make sensible predictions as to possible novel uses in the future. This paper attempts to provide such an account by using a constructional approach to redefining the co-occurrence as a partially-filled construction, headed by the degree marker in a construction of the form [Degree þ X], mapped to the essential function of evaluative predication of an attributive quality, coerced from the following element X. The construction is termed the ‘DegreeEvaluative Construction’ (DEC), a unique form-meaning mapping construct that plays an important role in expressing evaluative judgment.

3. The database: corpora and spoken conversation5 In order to e¤ectively support our arguments, naturally-occurring data are used in this study, which have been drawn from three types of online resources and recorded conversations for a balance of data types. The corpus data include: 1) the 5-million-word Sinica Corpus, largely composed of written and spoken texts of Taiwan Mandarin (http://www.sinica.edu.tw/ SinicaCorpus);6 2) the daily-updated online archive, Google (http://www. google.com), containing the latest information that includes various internetbased texts; 3) the Udndata news database (http://udndata.com/library/), which provides the latest news from five local newspapers in Taiwan. As for the conversational database, it consists of 139 minutes of naturally occurring spoken conversation, with a variety of registers such as compliments, 5. In cases in which the examples demonstrated in this work may be controversial, we checked more than ten native speakers to see whether they would accept the questionable collocations. Meanwhile, the methods of data collection and analysis are along the lines of Leech’s (1991: 74) statement that it is a question of corpus plus intuition, rather than of corpus or intuition. 6. Chui and Lai (2008: 122) indicate that ‘‘Academia Sinica Balanced Corpus of Modern Chinese (Sinica Corpus) includes texts which have been collected since 1990 from di¤erent areas. Spoken data are just a small part of the corpus which consists of speech drafts, play scripts, actors’ lines, conversations, and written records of speeches and meetings’’.

120

Mei-Chun Liu and Chun Chang

interviews and storytelling. Among the data sources, the Sinica Corpus is taken as the main resource; Google and udndata are used as supporting references to (re)confirm our interpretation of special or marginal cases. Given that a number of lexical entries may be used as degree adverbs, we first counted the total number of occurrences of the target pattern with all possible degree adverbs in Sinica Corpus (SinC) and the spoken conversation (SpoC).7 Then, we selected two high-frequency degree adverbs to be examined in greater detail: the default degree modifier heˇn 很 ‘very’ (12988 tokens in SinC and 277 tokens in SpoC), and the minimal degree adverb yoˇudiaˇn 有點 ‘a little’ (469 tokens in SinC and 13 in SpoC). These two adverbs happen to mark two di¤erent polarities of degree modification. The analysis proposed in this paper is mainly based on the selected examples of these two commonly used degree modifiers. 4. Constructional account with Qualia Structure In this section, we will show that the correlation between the degree adverb and the post-adverbial element in the [Degree þ X] sequence is of a constructional nature. As Liu (2005) demonstrates, the application of Construction Grammar (Goldberg 1995, 2003; Jackendo¤ 1997) to the analysis of a formally uniform but semantically heterogeneous pattern will render a unified and consistent account of the grammatical and semantic properties allowed in the pattern. As will be detailed below, the default degree marker heˇn 很 may co-occur with almost any kind of grammatical category, including NP, AP, VP, ADVP, DP and even the interrogative category. Selected examples are given in (5) for illustration: (5) Diverse range of post-adverbial elements: a. common N:

或許他們很草莓，光鮮不耐壓。

huo`xuˇ ta#me#n heˇn caˇome´i, gua#ngxia#n bu` na`iya# maybe they DEG strawberry bright not durable ‘Maybe they are very much like strawberries, shining but not durable.’ 7. The term ‘frequency’ refers to the number of occurrences in the Sinica Corpus and conversation data. Due to the limited space, we focus on two high-frequency degree adverbs, heˇn 很 ‘very’ and yoˇudıˇan 有點 ‘a little’, since it is impossible to give a full compendium in the present paper. More analyses of collocations between other degree adverbs and their post-adverbial categories are left for future study.

Grammaticalization in constructionalization

b.

121

proper N:

感覺聲音很蔡依琳。

gaˇnju´e she#ngyı #n heˇn Ca`i Yı #lı´n feeling voice DEG name ‘Her voice is very Yilin Cai-like (sweet and touching).’8

c. active V:

他是律師 . . . 玩起小白球卻很殺。

ta# shı` lu`shı.# . . waˇnqıˇ xiaˇoba´iqı´u qu`e heˇn sha# he is lawyer play golf but DEG kill ‘He is a lawyer, yet he is quite killing (forceful) on the golf course.’

d.

ADVP:

他很有點害怕。 ta# heˇn yoˇudiaˇn ha`ipa` he DEG a little afraid ‘He is quite sort of afraid.’

e. Demonstrative pronoun:

場面很那個。

chaˇngmia`n heˇn na`ge scene DEG that ‘The scene is quite that (erotic).’ f.

Interrogative:

這首歌的歌詞很什麼。 zhe` shoˇu ge# de# ge#c´ı heˇn she´nme this CL song DE lyrics DEG what ‘The lyrics of this song were quite what (people say).’

Note that in all the above examples, the element following heˇn 很 contributes a ‘degree-modifiable attribute’ that may or may not be explicitly specified. In (5e) and (5f ), the demonstrative na`ge` 那個 ‘that’ and the interrogative word she´nme# 什麼 ‘what’ (not signaling a question here) are both pronominal in nature, implying an underspecified quality that must be contextually inferred. The distributional flexibility is taken as deriving from the constructional alignment with meaning coercion. It also 8. Ca`i Yı #lı´n is a famous female pop singer in Taiwan, whose voice is sweet and touching.

122

Mei-Chun Liu and Chun Chang

suggests that the degree of heˇn’s grammaticalization has increased, similar to the grammaticalization of lia´n 連 ‘even’, for example, which became more grammaticalized when its scalar function spread from modifying Ns to Vs and then to adverbial phrases (Xing 2004). To derive the implied semantic facet from the post-heˇn element, the qualia roles typically associated with a nominal in qualia structure (Pustejovsky 1995) are drawn upon to further explain the possible range of meaning coercion crucial in the realization of the construction. In the following, we will first define the proposed construction with a specification of its function and constraints, and then illustrate and discuss its distribution with di¤erent grammatical categories in detail. 4.1. Properties of the Degree-Evaluative Construction The recurrent sequence [Degree þ X], traditionally thought of as a degreemodified adjectival phrase, is re-analyzed as a semi-fixed, degree-headed construction termed the ‘Degree-Evaluative Construction’ (abbreviated as DEC). The construction maps a surface form [Degree þ X], with a construction-specific meaning pertaining to evaluative predication. Drawing on the insights of a Constructional Approach (mainly Goldberg 1995, 2003), the construction may be defined as shown in (6): (6) The definition of the Degree-Evaluative Construction (DEC): NP þ [ stative Degree ADV þ X] Degree evaluation of a scalar attribute or quality, encoded by X Function: The degree adverb is taken as the constructional operator that signals and coerces the degree evaluation of a given attributive quality implied by the X element. Example: (a) Possible Degree ADV: heˇn 很 ‘quite, very’, yoˇudiaˇn 有點 ‘a little’, fei# cha´ng 非常 ‘very’, shı´fe#n 十分 ‘pretty’, etc. (b) Possible X: N, V, Adj, Adv, Demonstrative or Q-word Constraints: (a) syntactic: Ns and Vs for X may not be a fully instantiated phrasal category (*full NP or VP). (b) semantic: Ns and Vs for X are non-referring and nondenotational: Ns do not refer to the existence of an entity; Vs do not refer to the actual occurrence of an event. (c) discourse: Ns and Vs are non-prototypical Ns and Vs in that they do not perform their prototypical discourse functions to introduce a manipulable entity and to report the happening of an event, respectively (c.f. Hopper and Thompson 1984). Form: Meaning:

Grammaticalization in constructionalization

123

To avoid unconstrained, excessive application of the construction, the grammatical, semantic and discourse constraints as specified above are added to rule out the unacceptable cases containing a determiner-headed full NP (*heˇn yı´ ge# keˇa`i de Ca`i Yı #lı´n 很一個可愛的蔡依琳 ‘DEG þ a cute singer Yilin Cai’) or a tense-aspect marked full VP (*heˇn sha# le# re´n 很殺了人 ‘DEG þ killed a person’), as further illustrated below: (7) a.

b.

Incompatibility of a full NP and VP with DEC *聲音很 [ NP 一個可愛的蔡依琳]。 she#ngyı #n heˇn yı´ ge keˇa`i de Ca`i Yı #lı´n voice DEG one CL cute DE Name ‘*The voice is of the quality of a cute singer Yilin Cai.’ *他打起球來很 [VP 殺了一個人 ]。 ta# daˇqıˇ qiu´ la´i heˇn sha# le yı´ ge re´n he play golf ASP DEG kill ASP one CL person ‘*He is quite forceful in killing a person when playing golf.’ Or ‘He will kill a person in golf, quite forcefully.’

For adjectives and adverbs, fewer restrictions may apply since they are typically and lexically quality-denoting. For demonstratives and interrogatives, there seems to be a tendency for certain words to be preferred over others. For example, na`ge 那個 ‘that’ is more frequently used than zhe`ge 這個 ‘this’; she´nme 什麼 ‘what’ tends to occur in the DEC more frequently than other question words such as she´i 誰 ‘who’ or naˇ lıˇ 哪裡 ‘where’. 4.2. Grammatical categories in the X slot In general, the grammatical constituent fitting the X slot in the DEC is responsible for providing a scalar quality, be it explicitly or implicitly encoded in the expression. When following a degree head, this X-element, may trigger, through constructional coercion, a variety of scalar evaluative properties in a category-specific way: A common noun denotes a type-referring property that is culturally or contextually salient (e.g. heˇn caˇome´i 很草莓 ‘very strawberry,’ understood as being fragile and nonpressure resistant); a proper name denotes a characteristic feature of the named entity which is perceived to be contextually relevant (e.g. heˇn Ca`i Yı #lı´n 很蔡依琳 ‘very Yilin Cai’, singing like the female pop singer Yilin Cai,); the most commonly occurring adjective encodes a lexically specified scalar value (e.g. heˇn kua`ile` 很快樂 ‘very happy’); the rarely occurring dynamic verb may encode an attributive property inferable from the

124

Mei-Chun Liu and Chun Chang

manner or e¤ect associated with the action (e.g. heˇn sha# 很殺 ‘kill’, referring to the forceful manner of killing); an additional adverbial may also follow the head degree modifier encoding an adverbially modified value (e.g. heˇn yoˇudiaˇn chı #jıng # 很有點吃驚 ‘quite sort of surprised’); a demonstrative pronoun may anaphorically or cataphorically refer to a quality that is contextually evoked or implied (e.g. heˇn na`ge` 很那個 ‘very that’); an interrogative Q is potentially a nominal with a semantically underspecified value (e.g. heˇn she´nme# 很什麼 ‘very what’). No matter what follows the degree head in the DEC, some kind of evaluative quality, lexically or contextually inferred, will be borne out through constructional coercion. That is the main reason why the DEC is defined as a unique construction, mapping the constructional form with the constructionallytriggered meaning. In what follows, the grammatical categories in the X slot are further explored respectively. 4.2.1. Nominal X Let us begin with the grammatical category N. In the database, there are two kinds of nouns that co-occur with a degree operator: common nouns, such as huoˇ 火 ‘fire,’ (see example (8a) below) and proper nouns, such as Che´ng-pıˇn 誠品 ‘Eslite (name of a famous bookstore)’ (see (8b) below). As a nominal, neither kind is used to refer to a specific participant or entity manipulable in the subsequent discourse; they are not loaded with referential intent. Rather, they are used to predicate a perceivable quality, characteristic of the person or type of entity denoted by the noun. They manifest a conceptual transfer from person to quality or object to entity. Such a transfer – from a nominal entity to its characteristic quality – is made possible via the basic cognitive mechanism of metaphorization (Lako¤ and Johnson 1980, Glucksberg 2003).9 The implicational extension, from person to quality or from object to quality, complies nicely with the direction and principles of metaphorical extension, as explicitly suggested in Claudi and Heine (1986: 301)’s implicational relation10: PERSON > 9. According to Glucksberg (2003: 96), people apprehend metaphorical meanings as quickly and as automatically as they apprehend literal meanings. With continued use, once novel metaphors become conventionalized, their metaphorical senses enter into people’s mental dictionaries. 10. The implicational relation is unidirectional (see Claudi & Heine 1986: 330). The direction ‘QUALITY PROCESS SPACE OBJECT PERSON’ is possible, while the possibility of the opposite direction is ruled out (*‘QUALITY ! PROCESS ! SPACE ! OBJECT ! PERSON’), since it is hard for humans to use abstract properties to describe concrete entities (see also Lako¤ and Johnson 1980; Su 2002a).

Grammaticalization in constructionalization

125

OBJECT > SPACE > PROCESS > QUALITY. The concrete PERSON or OBJECT may serve as the cognitive reference for conceptualizing the abstract QUALITY associated with the source entity as a result of metaphorical transfer.11 Moreover, the specific QUALITY or scalar property extracted from a concrete nominal may be further identified through four possible meaning facets, following the conventions of the qualia structure. As Pustejovsky (1995: 8586) puts it, qualia structure is a structured representation that encodes the relational force of a lexical item. It allows multiple meaning facets to be associated with the same word. For nominal entities, a qualia structure specifies four essential dimensions of the noun, termed qualia role: QC (constitutive) ‘the internal constitution of an entity’, QF (formal) ‘the perceptual properties of an entity’, QT (telic) ‘what an entity is for’, and QA (agentive) ‘the life history of the entity’. The qualia roles may help specify the evaluative quality implied in a particular DEC utterance. For example, huoˇ 火 ‘fire’ in (8a) would fit both the constitutive role (QC) and the formal role (QF); the same appears to hold in (8b). While denoting a qualia role, the noun following a degree is restricted to type-reference, rather than token-reference, as evidenced in the quantifier test presented below: adding the numeral-classifier yı´-ge 一個 ‘one’ to the quality-inferring noun makes it referentially specific and thus unacceptable in the pattern: (8) Incompatibility of full NPs with the DEC a. 王力宏的人氣很 (*一個) 火。 Wa´ng Lı`ho´ng de# re´nqı` heˇn (*yı´-ge) huoˇ NAME DE publicity DEG (one-CL) fire ‘The singer Lihong Wang is very popular.’

11. Tang (2000a: 298) also uses metaphorical extension to account for collocations between the degree adverb heˇn 很 and a noun, as in heˇn qio´ngya´o/heˇn baˇobe`i 很瓊瑤/很寶貝 ‘very romantic/valuable’. This view of metaphorical extension is reminiscent of many others in previous literature (Lako¤ and Johnson 1980; Lako¤ 1987; Tang 2000a; Chui 2000; Su 2002a; Ko¨vecses 2003). However, our analysis di¤ers slightly from those in the literature: we further take the implicational relation and qualia role as applying to the metaphorical extension of the collocation between degree adverb and noun category.

126

Mei-Chun Liu and Chun Chang

b. 二手書店很 (*一個) 誠品。 e`rshoˇu shu#dia`n heˇn (*yı´-ge) Che´ng-pıˇn second-hand bookstore DEG (one-CL) NAME ‘The second-hand bookstore is very Eslite-like.’ The evaluative property implied by the common noun huoˇ ‘fire’ or proper noun Che´ng-pıˇn ‘bookstore name’ is triggered through the constructionbased coercive mechanism as mentioned above and schematically represented in (9). Through the combined operations of metaphorical extension and qualia role assigning in the DEC, the highlighted qualities can be derived from an OBJECT, as in (8a): the blazing flames of fire provide the source for understanding the flaming popularity of the predicated target (a singer) via similarity associations, relying on the constitutive as well as formal qualia roles of fire ([QC ¼ characteristics of the physical nature of fire]) ([QF ¼ blazing]). In (8b), the appearance/decorative design of the secondhand bookstore is compared to that of a well-known, elegantly-furnished bookstore, Eslite, via metaphorically-transferrable similarity associations, based on constitutive and formal properties of the reference bookstore ([QC ¼ characteristics of the salient features of Eslite] and [QF ¼ spatial, bright, and stylish]). The meaning shift in using a proper N to refer to its characteristic features is a case of type coercion in qualia reasoning. (9) Figure 1: The associative relation of contextually or culturally defined quality in the DEC

The metaphorical operation and the qualia role association allow the Ns in the X-slot to be used as nominal predicates implying quality-like properties. Such a meaning extension is constructionally triggered simply by the co-occurrence of a N with the degree adverb heˇn 很. The co-occurring pattern renders a constructionally-derived scope of modification for the degree-following N to saturate (cf. Tang 1989). The very existence of the

Grammaticalization in constructionalization

127

degree marker heˇn 很 presupposes a comparative background for scalar evaluation of the inferred quality. Without heˇn 很, the sentences are incomprehensible due to the lack of su‰cient frame-evoked background information (c.f. Chafe 1987). This is why the degree marker is considered to be the constructional operator: it functions as a triggering device and, as such, is crucial and mandatory in the construction. 4.2.2. Adjectival X Adjectival modifiers are, presumably, the most common category that follows a degree adverb. Previous works focus mainly on the semantic correlation between the degree adverb and various kinds of adjectives denoting stative values (see Zhu 1982; Tang 2000a; Tang 2000b; and the references cited therein). Most works conclude that non-gradable adjectives with a negative polarity, such as cuo` 錯 ‘wrong,’ are unacceptable with degree modification. However, based on the corpus examples as shown below, this commonly held restriction needs to be reconsidered. Besides inherently gradable and evaluative adjectives, the possible range of adjectivals occurring in the X slot encompasses three finely-distinguished types in terms of gradability (c.f. Kennedy and McNally 2005)12: (i) the relative gradable adjectives, featuring color terms (e.g. la´n 藍 ‘blue’ or lu¨` 綠 ‘green’) as in (10); (ii) the absolute gradable adjectives denoting physical disability, such as xia# 瞎 ‘blind’ and lo´ng 聾 ‘deaf ’ in (11), and (iii) the non-gradable adjectives entailing logical and epistemic judgments, such as zhe#n 真 ‘true’, jiaˇ 假 ‘false’, duı` 對 ‘right’ and cuo` 錯 ‘wrong’ as in (12). Consider the following examples of the three types of adjectives: (10) Relative gradable adjectives: a. 茂伯現在很紅。 Ma`o-bo´ xia`nza`i heˇn ho´ng Name-uncle now DEG red ‘Uncle Mao is currently very popular.’ 12. According to Kennedy and McNally (2005), there are three types of adjectives that comply with the distinct criterion of ‘whether they can co-occur with degree adverbs (e.g. very) or proportional adverbs (e.g. completely)’: relative gradable adjectives, absolute gradable adjectives and non-gradable adjectives. Following their classification of adjectives, we argue that these three types of adjectives (gradable adjectives, absolute gradable adjectives and non-gradable adjectives) interact with heˇn 很 ‘very’ through the mechanism of metaphorical extension triggered separately by the construction.

128

Mei-Chun Liu and Chun Chang

b. 在計程車司機中，我相信他很藍/綠。 za`i jı`che´ngche#-sı #jı # zho#ng, woˇ xia#ngxı`n ta# heˇn la´n/lu`¨ . at taxi-driver among, I believe he DEG blue/green ‘Among all the taxi drivers, I believe he is one who is quite pro-KMT/pro-DPP.’ (11) Absolute gradable adjectives: a. 周說蔡打電話說他被拍到很瞎。 Zho#u shuo# Ca`i daˇdia`nhua` shuo# ta# be`i pa#ida`o heˇn xia#. Zhou say Cai call say he BEI shoot DEG blind/silly ‘Zhou said that Cai called him to tell him that being captured in a picture is quite silly.’ b. 大嫂耳朵很聾。 Da`saˇo eˇrduo# heˇn lo´ng. sister-in-law ear DEG deaf ‘My sister-in-law is quite deaf.’ While the above adjectives are recognized as denoting a gradable value to some extent, the following adjectives are generally considered to be nongradable: (12) Non-gradable adjectives: a. 圖中每一樣都很真實，樹很真。 tu´-zho#ng meˇi-yı´-ya`ng do#u heˇn zhe#nshı´, shu` heˇn zhe#n. picture-in everything all DEG true, tree DEG true ‘Everything in the picture looks real; the trees are quite vivid.’ b. 負面新聞多後，大眾批她笑容很假。 fu`mı`an xınwe # ´ n duo# ho`u, negative news more after, da`zho`ng pı # ta# xia`oro´ng heˇn jiaˇ. people criticize her smile DEG fake ‘After a number of negative news reports, people criticized her smile as being quite insincere.’ c. 打是愛，罵是關心，這乍聽似乎很對。 daˇ shı` a`i, ma` shı` gua#nxın, # zhe` zha` tıng # sı`hu# heˇn duı`. hit is love, scold is care, this first-hear seem DEG right ‘Hitting is love and scolding is caring; this at first sounds quite reasonable.’

Grammaticalization in constructionalization

129

d. 法律不外乎人情，這句話實在很錯。 faˇlu` bu´-wa`i-hu# re´nqı´ng, zhe`-ju` hu`a shı´za`i heˇn cuo`. law no-more-than relation, this-CL words truly DEG wrong ‘The sentence ‘‘Law is nothing but human relations’’ is indeed quite wrong.’ The range of adjectivals found in the above examples is quite unrestricted and productive. The lexically-specified senses of the adjectives may be relatively discrete, far from being bona fide gradable, but they all render a gradable sense once they occur in the [Degree þ X] construction. Again, a constructionally-derived evaluative meaning is triggered for these adjectives to embrace scalar values, by means of cognitively or contextually available inferences. As Paradis (2001, 2003) suggests, cognitive operations may a¤ect the boundaries of adjectives with degree modification, as an integral manifestation of cognitive functions.13 As a result of cognitive manipulations, lexically non-gradable adjectives may be perceived as having gradable and scalar properties, via constructional coercion. The first type of adjectival in (10) consists of color terms that are culturally and socially associated with a gradable sense, e.g. degree of popularity in (10a) or degree of comradeship with political parties (blue for KMT, green for DPP) in (10b). When occurring in the DEC, the color term is coerced by the Degree operator into acquiring a scalar value that is socio-culturally linked to the degree of political commitment, representing the symbolic constitutive properties of the color ([QC ¼ characteristics of political party]). The second type of AP in (11) has an absolute gradable sense that is shifted to the relative gradable sense under constructional coercion. The adjectives denoting the physical impairment of eyes or ears are viewed as evaluable in terms of degrees of seriousness and degrees of disability. A further metaphorical extension operates in the use of heˇn xia# 很瞎 ‘very blind’ in (11a) that equates the degree of blindness (physical disability) with the degree of stupidity (mental disability): THE QUALITY OF INTELLIGENCE IS THE QUALITY OF EYESIGHT. The conceptual 13. Paradis (2001: 58) discusses the issue of English non-gradable meaning. Even non-gradable adjectives may be coerced into a gradable reading. It is possible to perceive the non-gradable adjective Swedish as gradable and scalar in expressions such as ‘That woman is very Swedish’. This example clearly shows that the content part and the schematic domain collaborate in an intimate and sophisticated manner in contributing to our conceptualization of the various adjectives.

130

Mei-Chun Liu and Chun Chang

metaphor in operation here is Eyesight as Intelligence. The physiological attribute (poor eyesight) stands metaphorically for a mental attribute (poor reasoning strength). The third type of adjectival is the non-gradable, logical operator of truth conditions, such as zhe#n 真 ‘true’, jiaˇ 假 ‘false’, duı` 對 ‘right’ and cuo` 錯 ‘wrong’, as exemplified in (12) above.14 Contrary to previous findings, they can also be used in the [Degree þ X] construction with a coerced gradable meaning. As logical operators of truth value, they are non-gradable in nature; but when occurring in the DEC, they render an evaluative sense as relational opposites of a scalar domain, i.e., degree of truthfulness vs. degree of falsehood, or degree of correctness vs. degree of wrongness. Thus, they can be viewed as denoting opposite end-portions of an evaluative continuum. Through conceptual manipulation, the originally discretely complementary ends of the truth domain (correct vs. wrong) are extended to two relatively comparable portions, scalarly opposing each other (portion of correctness vs. portion of wrongness). This magic shift or emergence in meaning is a result of the construction-triggered sense alignment, a mechanism well-motivated and theoretically imbedded in a constructional approach and also interestingly echoing the theory of pragmatic halos as described in Lasersohn (1999).15 4.2.3. Verbal X Traditionally, only verbs of state, such as gradable psych-verbs, have been considered to be compatible with a degree modifier, as in (13), while eventive predicates, in general, may not be degree-modified, as in (14) below:

14. Chui (2002) suggests that duı` 對 develops a pragmatic function of agreement as a result of conventionalizing the conversational implication that commitment to truthfulness entails agreement with the content. Take the adjective duı` 對 in zhe` zha` tıng # sı`hu# heˇn duı` 這乍聽似乎很對 ‘this seems to be very right’ for example, duı` 對 here is forced to mean ‘reasonable or in agreement’ instead of ‘truth-value’. It is taken as a pragmatic marker functioning as ‘agreement’ denoting a state of a¤airs, which refers to the speaker’s agreement with someone’s utterances, words, or attitudes. What the evaluator tries to do is strengthen or emphasize the possibility of the utterances. 15. The pragmatic halos in Lasersohn (1999) provide a schema for determining how much deviation from what is actually true still counts as ‘close enough to’ to the truth in any context to be an acceptable range of semantic deviation.

Grammaticalization in constructionalization

131

(13) Stative-psych verbs 他很高興/生氣。 ta# heˇn ga#oxı`ng/she#ngqı`. he DEG happy/angry ‘He is quite happy/angry.’ (14) Eventive-process verbs *他很吃/完成/消失。 ta# heˇn chı #/wa´nche´ng/xia#oshı #. he DEG eat/finish/disappear *‘He eats/finishes/disappears very much.’ However, our data show that less-typical stative predicates and eventive predicates can also occur in the target construction. As illustrated below, less-prototypical stative predicates are commonly found with a degree adverb, including cognition verbs (15), causative co-verbs (16), and modal auxiliaries (17). More surprisingly, a degree modifier is also found with atelic activity verbs (e.g., tia`o 跳 ‘jump’), telic achievement-type verbs (e.g., chu#xia`n 出現 ‘show up’, po`hua`i 破壞 ‘damage’), and even accomplishment verbs preceded by an evaluative operator (e.g., na´n wa´nche´ng 難完成 ‘di‰cult to accomplish’), turning the entire predicate into a state predicate. The examples found cover three di¤erent types of predicates: quasi-stative predicates (cognitive verb, causative co-verb, or modal auxiliary) as in (15)–(17), derived stative predicates (telic verbs) with an incorporated stative operator such as na´n ‘di‰cult’ as in (18), and telic or atelic eventive predicates as in (19)–(20). In terms of frequency, the majority of cases have a stative predicate following the degree adverb, as states may readily encode an evaluative quality compatible with degree modification (c.f. Smith 1991). On the other hand, eventive predicates are less susceptible to degree modification, though they are still found in the DEC.16 Let’s first look at the three types of stative predicates: (15) Cognitive verb

我很了解/認識他。 woˇ heˇn liaˇojieˇ/re`nshı` ta#. I DEG understand/know him ‘I understand/know him quite well.’

16. There are no such expressions shown in the Sinica corpus and spoken conversation. The data demonstrated here are extracted from the online Google archives: these sorts of usages are rare when compared with semantically stative events.

132

Mei-Chun Liu and Chun Chang

(16) Causative co-verb

他很令/使/讓/叫媽媽擔心。

ta# heˇn lı`ng/shıˇ/ra`ng/jia`o ma#ma# da#nxın. # he DEG make/make/make/make mother worry ‘He made his mom worry a lot.’

(17) Modal aux

你很會照顧你父親。

nıˇ heˇn huı` zha`ogu` nıˇ fu`qın# you DEG able take care your father ‘You are very capable of taking care of your father.’

The above verbs can be viewed as denoting some stative quality that is readily compatible with degree modification. Besides the psych-verbs that are stative and evaluative in nature, the cognition verbs liaˇojieˇ 了解 ‘understand’ and re`nshı` 認識 ‘know’ are used statively in (15) as referring to the degree of understanding/knowing; causative co-verbs such as lı`ng/ shıˇ/ra`ng/jia`o 令/使/讓/叫, all meaning ‘make, let or cause’ in (16), may be seen as reflecting a gradable degree of causation. According to Chang (2005), verb phrases anchored with causative markers are usually modified by a degree adverb such as heˇn 很. As for the modal auxiliary huı` 會 ‘can’ in (17), it implies an evaluative judgment of one’s capability, a stative and scalar value consistent with the constructional meaning. In contrast, eventive verbs were previously thought to be incompatible with the degree adverb heˇn 很 ‘very’. However, the data retrieved from Google, as exemplified in (18)–(20) below, show that it is possible, though less commonly found, for eventive verbs pertaining to physical action to occur in the X-slot in the DEC. They may be used in two kinds of environment: in (18), the post-degree predicate contains a lexically incorporated evaluative operator, haˇo ‘good’ or na´n ‘di‰cult’, which serves to convert the whole predicate to a stative meaning, while in (19) and (20) the dynamic verbs are used alone in the X-slot without such a state-converting evaluative operator: (18) a. 鉛筆很好用 qia#nbıˇ heˇn haˇo yo`ng. pencil DEG good use ‘The pencil is good to use (easy to use).’

Grammaticalization in constructionalization

133

b. 論文很難完成。 lu`nwe´n heˇn na´n wa´nche´ng thesis DEG hard finish ‘The thesis is di‰cult to complete.’ c. 仇恨很難消失。 cho´uhe`n heˇn na´n xia#oshı # hatred DEG di‰cult disappear ‘Hatred is hard to expunge.’ (19) a. 顏色很跳。 ya´nse` heˇn tia`o color DEG jump ‘The color is quite jumping (eye-catching).’ b. 個人形象很破壞。 ge`re´n xı´ngxia`ng heˇn po`hua`i individual image DEG destroy ‘Personal image is quite destroyed.’ c. 打擊很出現 daˇjı´ heˇn chu#xia`n batting DEG appear ‘(Speaking of baseball) The batting is quite evident.’ (20) 這種說話方式 很殺/很搞笑。 zhe#-zho#ng shu#ohua` fa#ngshı` heˇn sha#/heˇn gaˇoxia`o. this-kind talking way DEG kill/DEG make-fun ‘This way of talking is quite ‘‘killing’’/funny.’ As mentioned above, though dynamic eventive verbs are thought to be incompatible with degree modification, they may still be used in the DEC. Incorporating an evaluative operator (e.g. haˇo 好 ‘good’ and na´n 難 ‘di‰cult’) in the predicate makes them readily acceptable in the DEC, as shown in (18). The evaluative operator helps convert a process verb to a gradable, stative predicate with scalar properties (c.f. Lien 2009). More interestingly, the degree head heˇn 很 in (19) and (20) is found to directly take an action verb, such as tia`o 跳 ‘jump’, po`hua`i 破壞 ‘damage’, chu#xia`n 出現 ‘show up’ and sha# ‘kill’, imbuing it with the dynamic e¤ect associated with the type of event it denotes. In line with the history of the Chinese language, telic verbs may appear as stative predicates without any

134

Mei-Chun Liu and Chun Chang

morphological change, such as sıˇ 死 ‘to die/dead’ and huo´ 活 ‘to live/ alive’17. These verbs do not refer to any specific occurrence of an event, but to a generalized dynamic property implied by the verb. In (19a), the verb tia`o 跳 ‘jump’ implies the quality of jumping out to the eyes (i.e., eye-catching) as a direct e¤ect of the action; the verb po`hua`i 破壞 ‘damage’ in (19b) implies the state of being damaged; the verb chu#xia`n 出現 ‘appear, show up’ in (19c) refers to the end result of showing up. Since the X-slot in the DEC is required to have a stative quality for evaluation, the process predicates mentioned above undergo a type-shift and behave like a stative predicate through type coercion triggered by the very use of the construction. 4.2.4. Adverbial X The use of the slightly misleading term ‘adverbial X’ here is intended to emphasize the presence of another degree adverb in the X-slot, as the post-degree element incorporates an additional adverbial modifier. The juxtaposition of more than one degree adverb in the surface form was previously thought to be unacceptable (see Zhu 1982; Tang 2000a; Tang 2000b), but it occurs quite often in the database, as illustrated below: (21) 李文秀很有點害怕/妒忌。 Lıˇ-We´nxiu` heˇn yoˇudiaˇn ha`ipa`/du`jı`. Name DEG slightly afraid/jealous ‘Wenxiu Li is quite slightly afraid/jealous.’ As the example shows, heˇn 很 ‘very, quite’ and yoˇudiaˇn 有點 ‘a little, slightly’ are juxtaposed in a nested structure of degree modification of a stative predicate: [Degree þ [Degree þ Pred]]. Though the two degree adverbs form what seems to be an adverbial compound, they actually perform di¤erent functions in the construction. The first or outer degree marker, heˇn 很 ‘very, quite’, being the constructional operator, exerts its modificational force over the second, embedded degree marker yoˇudiaˇn 有點 ‘a little, slightly’ in the X-position. The embedded X [ x yoˇudiaˇn þ X] is itself a degree expression referring to a sub-scale towards the weak end [slightly þ X]. This value then feeds back into the constructional operator heˇn ‘very, quite’ towards the strong end. The combined value points to some point in the middle of the scale, as a result of a positive degree

17. This historical fact is pointed out by one of the reviewers.

Grammaticalization in constructionalization

135

modification of a somewhat negative degree. Such a chaining operation in scalar evaluation can be represented schematically as shown in Figure 2. As separate degree adverbs, heˇn 很 ‘very’ and yoˇudiaˇn 有點 ‘slightly’ happen to occupy opposite polarities (strong vs. weak) on the same scale of X. The weak-degree modifier yoˇudiaˇn 有點 ‘a little, slightly’ ranges over a sub-scale towards the weak end of minimal degree, but then the weaker value is modified by the strong degree adverb heˇn 很 ‘very’ to render a positive assertion of the weaker portion. The example demonstrates a nested, double-layered, relation of degree modification. The constructional degree operator on the outer layer cannot completely override the value of the degree-embedded X. That is why the expressions in the above example (heˇn yoˇudiaˇn ha`ipa` 很有點害怕 ‘quite sort of afraid’ and heˇn yoˇudiaˇn du`jı` 很有點妒忌 ‘quite sort of jealous’) tend to be skewed to the relatively weaker end, designated by the inner, embedded degree marker in the Xslot. Pragmatically speaking, the competing interaction of the two degree adverbs results in a compromise of degree strength and manifests a form of reserved speech, with the pragmatic e¤ect of signaling the speaker’s reservation or indirectness. It seems that the less desirable attribute, e.g. ha`ipa` 害怕 ‘afraid’ or du`jı` 妒忌 ‘jealous,’ should not be asserted too explicitly or emphatically as it may run against the convention of courtesy (Levinson 1983, Zhang 2002).18 (22) Figure 2: polarity scale for heˇn [ yoˇudiaˇn þ x]

4.2.5. X as demonstrative Demonstratives, or demonstrative pronouns, are also found in the X slot in the DEC, such as na`ge# 那個 ‘that one’ and zhe`ge# 這個 ‘this one’, 18. Zhang (2002) states that certain degree adverbs can co-occur in certain contexts for pragmatic reasons. For example, sha#owe´i yoˇudıˇan qı`chuˇan 稍微有 點氣喘 ‘a little asthmatic’ and yoˇudıˇan ta`i yı`ng le 有點太硬了 ‘a little too hard’, etc. However, this juxtaposition of heˇn 很 and yoˇudıˇan 有點 is a new finding that is not discussed in Zhang (2002).

136

Mei-Chun Liu and Chun Chang

as exemplified below.19 Without lexically-specified semantic contents, demonstratives are used in the DEC as deictic pointers for anaphoric or cataphoric inferences, similar to a pro-form whose semantic reference or content is deictically defined and contextually evoked: (23) Demonstratives referring to previous established quality a. 電影中有段感情戲，場面很那個. . . dia`nyıˇng zho#ng yoˇu dua`n gaˇnqı´ng-xı`, movie midst have section romantic-episode, chaˇngmia`n heˇn na`ge#. . . scene DEG that ‘There’s a romantic episode in the movie; the scene is quite that (erotic).’ b. 感情戲它很這個 . . . 那個 . . . 該怎麼說呢? gaˇnqı´ng-xı` ta# heˇn zhe`ge#. . . na`ge# . . . ga#i zeˇnme# shuo# ne#? romantic episode it DEG this that should how say SFP ‘How can I describe the romantic scene? It is quite this. . . quite that. . .’ The demonstrative pronouns are used deictically in the above examples to refer to a referentially marked but semantically underspecified stative quality that is only traceable and retrievable in a discourse exchange anchored in the particular context. When the explicit mention of a valueloaded X is less preferred and the speaker purposely uses a semanticallyvoid demonstrative na`ge# 那個 ‘that one’ and zhe`ge# 這個 ‘this one’ in the DEC, it is expected that certain contextual information will be drawn upon for the hearer to fill in the relevant attribute, deictically recoverable in the given situation. The specific quality implied by na`ge# 那個 ‘that one’ and zhe`ge# 這個 ‘this one’ is semantically undetermined but contextually inferrable.20 In the examples, the prior mention of gaˇnqı´ngxı` 感情戲 19. Quirk et al. (1985: 265–268) claim that the definite article ‘the’ is used to mark the phrase that it introduces as definite ‘referring to something which can be identified uniquely in the contextual or general knowledge shared by speaker and hearer’. 20. Huang (1999: 88) classifies two distinct characteristics of na`ge# 那個 ‘that one’ and zhe`ge# 這個 ‘this one’: a more local (or less global) retrieval di‰culty is marked by na`ge# 那個 if the target of retrieval is a more familiar lexical item. On the other hand, zhe`ge# 這個 functions as a pause marker for local syntactic planning or for a general lexical search.

Grammaticalization in constructionalization

137

‘romantic drama’ entails a qualia property ([QC ¼ characteristics of romantic drama]) which seems to be su‰cient to imply a mutually understood attribute (something erotic or pornographic). The trick here is that the intended meaning is not explicitly expressed, but implicitly via a deictic pointer (the demonstrative) which leads to the intended meaning. 4.2.6. X as Interrogative Interrogatives or question words, such as zeˇnya`ng 怎樣 ‘how’ and she´nme# 什麼 ‘what,’ are also found as the X-element DEC. As question words, they are referentially and semantically underspecified: (24) Interrogatives referring to an underspecified quality a. 她聊著那個歌手唱的怎樣，這首歌的歌詞很什麼. . . ta# lia´o zhe na`ge# ge#shoˇu cha`ng de# zeˇnya`ng, she chat DUR that singer sing DE how zhe`-shoˇu ge# de# ge#c´ı heˇn she´nme. . . this-CL song DE lyrics DEG what ‘She’s chatting about how the singer sings, and the lyrics of the song were pretty something (it’s like). . .’ b. 我覺得並沒有說代表很怎樣 . . .當然你也不是說很無情. woˇ jue´de# bı`ng me´iyoˇu shuo# da`ibiaˇo heˇn zeˇnya`ng. . . I feel also NEG say representative DEG how da#ngra´n nıˇ yeˇ bu´shı` shuo# heˇn wu´qı´ng. certainly you also NEG say DEG heartless ‘I think it’s not to say that the representative is very whatever. Certainly you cannot say he’s quite heartless, either.’ In contrast to deictic pronouns that are situationally anchored and referentially framed, interrogatives zeˇnya`ng 怎樣 ‘how’ and she´nme 什麼 ‘what’ are semantically open-ended without a deictic component. They are used to signal a non-instantiated target quality that the interlocutor cannot point out at the moment of utterance. Although zeˇnya`ng 怎樣 ‘how’ and she´nme 什麼 ‘what’ are semantically unspecified, they are intended to refer to some kind of a commonly assumed quality that can be contextually recovered in the preceding or upcoming context. That is to say, the quality phrased by a question word is intended to be open to interpretation, on the basis of what is reasonably assumed in the context. For the possible range of interpretations intended by the interrogative X, relevant information is supposed to be present in the immediate context.

138

Mei-Chun Liu and Chun Chang

Normally, the unfilled quality is some prototypical property commonly associated with the predicated NP-subject. For instance, in (24), the predicated subject ge#c´ı ‘lyrics’ ([QC ¼ characteristics of lyrics]) and da`ibiaˇo 代表 ‘the representative’ ([QC ¼ personalities]) provide the basic clues for the non-instantiated value. Subsequent discourse may also be helpful. In example (24b), the subsequent mention of the attribute wu´qı´ng 無情 ‘heartless’ may be taken as semantically relevant to a sensible inference for the intended quality. It is interesting to note that the use of an underspecified demonstrative na`ge# 那個 ‘that’ or an open-ended interrogative she´nme 什麼 ‘what’ tends to imply a negatively evaluated or less preferred quality, as illustrated in the above examples. The use of the ‘‘unspecified’’ pronominal form for a socially taboo sense in (23) and the use of an unfilled question word for an intended criticism in (24) may both be taken as a pragmatic strategy to tone down or avoid directly asserted negative comments. The intended negative evaluation is thus expressed with a lexically-void substitute, in an understated, to-be-filled-in-by-the hearer manner. In sum, the interpretation of the X-slot in the DEC discussed so far can be briefly captured as follows, taking into consideration its syntactic behavior and semantic properties: (25) The Degree-Evaluative Construction (DEC), formally realized as [Degree þ X], can signal and coerce a degree evaluation of a given attributive quality implied through type-coercion or semantic extension, by the post-degree element X. When the evaluative quality in the X slot is underspecified, contextual information must necessarily be relied upon to instantiate the semantic content (a stative quality) required in the construction. 4.3. A constructional variant: the Minimal Degree Construction The proposed DEC may have di¤erent constructional variants when taking di¤erent degree modifiers. This section turns to the discussion of the contrast between the positive degree adverb heˇn 很 ‘very, quite’ and the minimal degree adverb yoˇudiaˇn 有點 (shorter form for yoˇuyı`diaˇn) ‘a bit, a little, slightly’. Corpus observation reveals that the minimal degree adverb yoˇudiaˇn 有點 is distinct in its distributional pattern as illustrated below. It is found that the occurrence of yoˇudiaˇn 有點 seems to collocate more frequently with relatively negative attributes. It serves as a euphemistic device to tone down the assertion of a disfavored quality expressed in the X-slot:

Grammaticalization in constructionalization

139

(26) Minimal degree with pejorative attribute a. 嗯，就是有點樣板啦！ en, jiu`shı` yoˇudiaˇn ya`ngbaˇn la# um that’s DEG exemplar SFP ‘Um, it’s just a little old-fashioned!’ b. 啊！真有點 嚇人哪！ a#! zhe#n yoˇudiaˇn xia`re´n na# Ah really DEG frightening SFP ‘Ah! It is truly somewhat scary.’ The minimal degree marker yoˇudiaˇn 有點 tends to take a negatively valued quality in the X-slot in the DEC, signaling a pragmatically more marked utterance, together with a sentence-final particle (SFP), e.g. la 啦 or na 哪. As yoˇudiaˇn 有點 encodes the minimal strength of the implied value, it is often used to minimize the assertional force when addressing a socially or contextually less-preferred or to-be-avoided attribute. The use of the minimal degree marker yoˇudiaˇn 有點 with a quality-denoting X can be viewed as a constructional variant of the DEC, associated with a specific interpretation of the speaker’s stance toward the evaluated quality. In contrast to the assertion of a stronger degree, the minimal degree variant is used pragmatically to tone down the assertion of a less-favored, less acceptable attribute. In discourse, the contrastive strength between yoˇudiaˇn 有點 and heˇn 很 may also be utilized by the speaker to indicate a weak vs. strong assertion of an evaluated quality. In the conversation below, with hen 很 interlocutor A makes a negative comment on a student, and interlocutor B uses yoˇudiaˇn 有點 to downplay his compliance with the negative evaluation, focusing on the student’s state of laziness, due perhaps to the demands of etiquette (Levinson 1983): (27) polite compliance of criticism with yoˇudiaˇn 有點: A: 這個學生我知道他的行為不是很好。 zhe`ge# xue´she#ng woˇ zhıda # ` o ta#de# xı´ngwe´i bu#shı` heˇn haˇo. this student I know his behavior NEG DEG good B: 有點懶散， . . .什麼事情都迷迷糊糊喔。 yoˇudiaˇn laˇnsa`n, she´nme` shı`qı´ng do#u mı´mı´hu´hu´ o# DEG lazy, whatever thing all dazed SFP

140

Mei-Chun Liu and Chun Chang

A: 你看那個題目都很好考. . .他也考不過。 nıˇ ka`n na`ge# tı´mu` do#u heˇn haˇokaˇo . . . ta# yeˇ kaˇo bu´guo` you see that question all very easy . . . he still test NEG A:

‘I know this student. . . (He) does not behave well.’

B:

‘(He’s) a little lazy. . . absent-minded about everything.’

A:

‘You see, the questions in the test were all very easy. . . yet he still couldn’t pass it. . .’

Below are more examples to illustrate the use of the minimal degree yoˇudiaˇn 有點 with predicates that lexically encode a negatively-valued, undesirable attribute, such as jie´jie´ba#ba# ‘stammering’, hua#nghua#ngzha#ngzha#ng 慌慌張張 ‘hurry-scurry’: (28) Minimal degree with reduplicative adjectives a. 他甚至說話都有點結結巴巴. . . ta# she`nzhı` shuo#hua` do#u yoˇudiaˇn jie´jie´ba#ba# he even talk all DEG stammering ‘Even when he carries on a normal conversation, he stammers a little.’ b. 他的酒意消退了，甚至還有點慌慌張張. . . ta#de jiuˇyı` xia#otuı` le#, his drunkenness disappear ASP, she`nzhı` ha´i yoˇudiaˇn hua#nghua#ngzha#ngzha#ng even still DEG hurry-scurry ‘When his drunkenness is gone, he is still a little panicky.’ Interestingly, reduplicative predicates (represented as AABB), usually considered unacceptable with degree modification, are found to be abundant in the variant pattern [ yoˇudiaˇn þ X], as shown above. While reduplication itself is taken as a way of marking intensive degree, it can be counterbalanced by the minimal degree marker yoˇudiaˇn 有點 ‘a bit’ to reduce the morphologically-derived assertional strength. Pragmatically, yoˇudiaˇn 有點 may be used as a hedge of uncertainty while making a weak assertion of a socially undesirable quality, due to concerns of politeness (c.f. Levinson 1983, Lasersohn 1999). Reduplicated predicates frequently collocate with the minimal degree marker yoˇudiaˇn 有點, but rarely occur with its counterpart, the positive degree marker heˇn. The reason might be that it is semantically redundant as both heˇn and reduplication indicate

Grammaticalization in constructionalization

141

a stronger degree in assertional strength. In addition, when taking reduplicative predicates, the minimal degree adverb tends to co-occur with perspective-taking markers, such as she`nzhı` 甚至 ‘even’ or jiaˇnzhı´ 簡直 ‘simply,’ as a means to convey strong subjectivity along with the highly expressive utterance (see Traugott 1989, 2003; Paradis 2003). Contrary to Tang’s claim (2000b: 214–215) that the co-occurrence is unacceptable, the use of a degree adverb with reduplicates is actually a discourse strategy with a designated purpose and should be treated as a pragmatic tendency rather than an absolute grammatical constraint. 4.4. Interim summary: Degree marker as grammaticalized constructional operator Whenever the [Degree þ X] construction is used, an evaluative scale is implied. This constructional coercion is triggered by the constructional operator, the degree marker, which imposes a degree modificational scope over the following element, converting it into an evaluable quality. Among the choices of lexically-denoted degree, the monosyllabic heˇn 很 occurs most frequently in text corpora (12,988 times) and spoken conversations (1.99 times/minute). Compared with other counterparts such as yoˇudiaˇn 有點 ‘a bit, a little’, heˇn 很 serves as the unmarked, default degree marker, widely used with all possible types of quality-denoting elements, as discussed above. Given its status as the invariant constructional trigger, the unmarked degree adverb heˇn 很 in [Degree þ X] can be viewed as undergoing a process of grammaticalization from a lexically-specified degree adverbial to a constructionally-defined operator, grammatically and semantically associated with the proposed Degree-Evaluative Construction (DEC). It is grammaticalized to function as a constructional operator that triggers constructional coercion in deriving the appropriate meaning designated for the construction. 5. Conclusion To understand the collocational nature of combining a degree adverb with a post-adverbial element, we explored the [Degree þ X] pattern in a large text corpora and conversation recordings. Surprisingly, an unexpectedly diverse range of grammatical categories occur in the X slot, preceded by the degree marker heˇn 很 ‘very, quite’, which include nominal, adjectival, verbal and adverbial elements, as well as demonstratives and interrogatives. To accommodate all the distinct categories following the degree marker

142

Mei-Chun Liu and Chun Chang

heˇn 很, a constructional approach is applied to redefine the [Degree þ X] sequence as a form-meaning pairing construct, signaled by the degree marker in implying a scalar value for evaluative predication. The DegreeEvaluative Construction (DEC) is thus proposed to account for the unexpected variety of X-elements found in the database. The construction can be clearly defined with specifications of its form, meaning, function and constraints. The proposal attempts to shift the focus and challenge the traditional thinking that the degree modifier is a secondary adjunct with no significant grammatical e¤ect. It departs from the traditional view in taking the degree modifier as the semantic and constructional operator for triggering the crucial interpretation of a scalar attribute. The exact attribute may be derived lexically or metaphorically through type coercion of qualia properties. The default degree marker heˇn 很 is further contrasted with the minimal degree marker yoˇudiaˇn 有點 ‘a bit,’ which may be viewed as marking a constructional variant of the DEC with a specific pragmatic connotation. The proposal is of particular interest to the study of grammaticalization (c.f. Traugott and Heine 1991, Hopper and Traugott 1993, Traugott 2003), as the adverb heˇn may be taken as undergoing a process of grammaticalization to serve as a constructional operator or trigger that would constructionally coerce an evaluative meaning out of the following element. When necessary, the constructional coercion is accomplished through cognitive manipulations of metaphorical extensions and/or contextual inferences. Four implications of this detailed work on the evaluative construction can be drawn as follows: 1. A reconsideration of the degree modification construction in Mandarin is needed. With plausible manipulation of conceptual schema, the cooccurrence of a degree adverb with a flexible range of quality-implying elements in daily conversation is more productively used than has been traditionally thought. An adaptable and predictive model is proposed in this study to account for the emerging novel uses in daily interchanges. 2. The contextually-evoked diverse range of coerced interpretations can be readily explained by the proposed solution without postulating extra lexical rules or additional senses in the lexicon. The postulation of the Degree Evaluative Construction (DEC) has a number of advantages over the lexical-rule approach: (i) it may avoid the unnecessary addition of attributive senses to nominal predicates; (ii) it may allow sense extensions of seemingly non-gradable attributes to derive constructionbased gradable senses; (iii) it may account for the creative sense

Grammaticalization in constructionalization

143

extension of verbal predicates through type-coercion; (iv) it may explicate the juxtaposition of more than one degree adverb (e.g. [Degree [Degree þ X]]); and (v) it may account for the fact that demonstratives and interrogatives are also used with a degree modifier to refer to a given attributive quality extractable from the relevant context. 3. Language works as the mind works. The correlation between the degree adverb and the post-adverbial element depends largely on how they are perceived as functionally compatible, with the flexible application of cognitive mechanisms in a given context. This indicates that language use is closely related to our cognitive capacity. And the interaction between syntax, semantics, and discourse pragmatics is constantly in play to determine what is uttered at a given point of discourse. The form-meaning pairing constructions such as the DEC emerge as routinized strategies in language use. Looking at naturally-occurring data may provide a more complete picture for further work in this direction. 4. In view of distributional evidence, the most commonly used degree adverb heˇn 很 ‘very, quite’ is taken to be the default, prototypical candidate as the grammaticalized operator in instantiating the proposed construction. The grammaticalization of heˇn 很 takes place alongside the constructionalization of the Degree Evaluative Construction. What we see here is a process of grammaticalization manifested in the process of constructionalization, similar to the processes discussed in Traugott (2003). The grammatical status of heˇn 很 thus highlights a new dimension of grammaticalization as it may interact with the process of constructionalization. By examining the unique properties of the Degree Evaluative Construction (DEC), this study demonstrates how grammaticalization can go hand in hand with constructionalization in shaping the grammar of Mandarin Chinese.

References Biq, Yung-O. 2004 Boas, Hans C. 2007

Construction, reanalysis, and stance ‘V yi ge N’ and variation in Mandarin Chinese. Journal of Pragmatics 36: 1655–1672. Construction Grammar in the twenty-first century. English Language and Linguistics 11 (3): 569–585.

144

Mei-Chun Liu and Chun Chang

Boas, Hans C. 2008

Resolving form-meaning discrepancies in Construction Grammar. In Constructional Reorganization, Leino, J. (ed.), 11–36. Amsterdam: John Benjamins. Brinton, Laurel J. and Elisabeth C. Traugott 2005 Lexicalization and language change. Cambridge: Cambridge University Press. Chafe, Wallace 1987 Cognitive constraints on information flow. In Coherence and Grounding in Discourse, Russell S. Tomlin (ed.), Amsterdam: John Benjamins. Chang, Chun and Chinfa Lien 2009 Interpretations and Constraints of Implicit Objects in Chinese. Hua´yuˇwe´n jia#oxue´ ya´njiu` [Journal of Chinese Language Teaching] 6 (1): 39–58. Chang, Lili 2005 The causativisation of Chinese Sheˇ-yı` Construction. Ta´ida´ We´nsheˇzhe´ Xue´ba`o [Humanitas Taiwanica] 62: 119–152. Cheng, Ying. 2000 Co´ng Yuˇlia`oku` Ka`n Ha`nyuˇ Zhu`do`ngcı´ De Yuˇfaˇ Te`dia`n [A Discussion of Syntactic Behaviors of Mandarin Modality via Corpus]. Proceedings of Research on Computational Linguistics Conference XIII (ROCLING XIII ): 157–170. Claudi, Ulrike and Bernd Heine 1986 On the metaphorical base of grammar. Studies in Language 10 (2): 297–338. Chui, Kawai 2000 Morphologization of the degree adverb Hen. Language and Linguistics 1 (1): 45–59. Chui, Kawai 2002 Ritualization in evolving pragmatic functions: a case study of DUI. Language and Linguistics 3 (4): 645–663. Chui, Kawai, and Hueiling Lai 2008 The NCCU corpus of spoken Chinese: Mandarin, Hakka, and Southern Min. Taiwan Journal of Linguistics 6 (2): 119–144. Fillmore, Charles J., Paul Kay, and Mary C. O’Connor 1988 Regularity and idiomaticity in grammatical construction: The case of let alone. Language 64: 501–538. Givo´n, Talmy 1993 English grammar: A function-based introduction. Amsterdam: John Benjamins. Glucksberg, Sam 2003 The Psychlinguistics of metaphor. Trends in Cognitive Sciences 7 (2): 92–96.

Grammaticalization in constructionalization

145

Goldberg, Adele E. 1995 Constructions: A Construction Grammar Approach to Argument Structure. Chicago: The University of Chicago Press. Goldberg, Adele E. 2003 Constructions: a new theoretical approach to language. Trends in Cognitive Sciences 7 (5): 219–224. Goldberg, Adele E. 2006 Constructions at Work: The nature of generalization in Language. Oxford: Oxford University Press. Goldberg, Adele E. and Ray Jackendo¤ 2004 The English resultative as a family of constructions. Language 80: 532–568. Hopper, Paul J. and Sandra A. Thompson 1984 The Discourse Basis for Lexical Categories in Universal Grammar. Language (60): 703–753. Hopper, Paul J. & Elisabeth Closs Traugott 1993 Grammaticalization. Cambridge: Cambridge University Press. Huang, Shuanfan 1999 The emergence of a grammatical category definite article in spoken Chinese. Journal of Pragmatics 31: 77–94. Jackendo¤, Ray 1997 Twistin’ the night away. Language 73 (3): 534–559. Jackendo¤, Ray 2007 Language, Consciousness, Culture: Essays on Mental Structure. Cambridge: The MIT Press. Jackendo¤, Ray 2008 Construction after Construction and its theoretical Challenges. Language 84 (1): 1–28. Kay, Paul ¨ stman and 1995 Construction grammar. In Jef Verschueren, Jan-Ola O Jan Blommaert (eds.), Handbook of Pragmatics: Manual, 171– 177. Amsterdam: John Benjamins. Kay, Paul and Charles J. Fillmore 1999 Grammatical constructions and linguistic generalizations: the What’s X doing Y? construction. Language 75 (1): 1–33. Kennedy, Christopher and Louise McNally 2005 Scale Structure, Degree Modification and the Semantics of Gradable Predicates. Language 81 (2): 345–381. Ko¨vecses, Zolta´n 2003 Emotion, metaphor and narrative. Trends in Cognitive Sciences 8.4: 154–156. Lai, Hueiling 2003 Hakka LAU constructions: A constructional approach. Language and Linguistics 4 (2): 357–378.

146

Mei-Chun Liu and Chun Chang

Lai, Hueiling 2006

Iconic coding of conceptualization: Hakka reduplicative constructions. Language and Linguistics 7 (2): 483–499. Lako¤, George and Mark Johnson 1980 Metaphors we Live by. Chicago: The University of Chicago Press. Lako¤, George 1987 Women, Fire and Dangerous Things. Chicago: The University of Chicago Press. Lasersohn, Peter 1999 Pragmatic halos. Language 75: 522–551. Leech, G 1991 The state of the art in corpus linguistics. In English corpus linguistics: Studies in honour of Jan Svartik, K. Aijmer and B. Altenberg (ed.), 8–29. London: Longman. Levinson, Stephen C. 1983 Pragmatics. Cambridge: Cambridge University Press. Li, Charles N. and Sandra A. Thompson 1981 Mandarin Chinese: A Functional Reference Grammar. Berkeley: University of California Press. Lien, Chinfa 2005 Families of ditransitive constructions in Li Jing Ji: Language and Linguistics 6 (4): 707–737. Lien, Chinfa 2006 Bong2 in Taiwanese Southern Min: a lexical and constructional perspective. Journal of Chinese Linguistics 34: 328–354. Lien, Chinfa 2009 Middles in Taiwanese Southern Min: The interface of lexical meaning and event structure. Lingua 1618–1632. Liu, Meichun 2005 Lexical information and beyond: meaning coercion and constructional inference of the Mandarin verb GAN. Journal of Chinese Linguistics 33 (2): 310–332. Lu¨, Shuxiang 1980 Xı`anda`i Ha`nyuˇ Ba#baˇi Cı´ [800 Words in Modern Chinese]. Beijing: Shangwu Yingshuguan. Michaelis, Laura and Knud Lambrecht 1996 Toward a construction-based theory of language function: The case of nominal extraposition. Language 72: 215–247. Paradis, Carita 2001 Adjectives and boundedness. Cognitive Linguistics 12: 47–65. Paradis, Carita 2003 Between epistemic modality and degree: the case of really. In R. Facchinetti, F. Palmer & M. Krug. (ed.), Modality in contemporary English, 197–220. Berlin: Mouton de Gruyter.

Grammaticalization in constructionalization

147

Pustejovsky, J. 1995 The Generative Lexicon. London: The MIT Press. Quirk, Randolph, Sidney Greenbaum, Geo¤rey Leech, and Jan Svartvik 1985 A Comprehensive Grammar of the English Language. London: Longman. Smith, Carlota 1991 The Parameter of Aspect. Dordrecht: Kluwer. Su, Iwen 2000 Mapping in thought and language as evidenced in Chinese. Chinese Studies 18: 395–424. Su, Iwen 2002a What can metaphors tell us about culture? Language and Linguistics 3 (3): 589–614. Su, Iwen 2002b Why a construction – that is the question! Concentric: Studies in English Literature and Linguistics 28 (2): 27–42. Tang, Tingchi 1989 Guˇoyuˇ Yuˇfaˇ Ya´njı`u Lu`njı´ [On the Study of Mandarin Syntax]. Taipei: Student Bookstore. Tang, Tingchi 2000a Ha`nyuˇ Yuˇfaˇ Lu`njı´ [Papers on Chinese Syntax.] Taipei: Pyramid Press. Tang, Tingchi 2000b Ha`nyuˇ Cı´faˇ Lu`njı´ [Papers on Chinese Morphology.] Taipei: Pyramid Press. Traugott, Elizabeth C. 1989 On the rise of epistemic meanings in English: An example of subjectification in semantic change. Language 65: 31–55. Traugott, Elizabeth C. 2003 Constructions in grammaticalization. In A Handbook of Historical Linguistics, Brian D. Joseph and Richard D. Janda. (eds.), 624–647. Oxford: Blackwell. Traugott, Elizabeth C. and Bernd Heine (eds.) 1991 Approaches to Grammaticalization. Vols. I and II. Amsterdam: John Benjamins. Xing, Janet Z. 2004 Grammaticalization of the focus particle lian in Mandarin Chinese. Journal of Historical Pragmatics 5 (1): 81–106. Zhang, Yajun 2002 Fu`cı´ Yuˇ Xia`ndı`ng Mia´ozhua`ng Go#ngne´ng [Adverbs and Descriptive Functions]. Anhui: Jiaoyu Chubanshe. Zhang, Yisheng 2000 Xia`nda`i Ha`nyuˇ Fu`cı´ Ya´njiu` [Study of Modern Adverbs]. Shanghai: Xuelin Chubanshe.

148

Mei-Chun Liu and Chun Chang

Zhu, Dexi 1956 Zhu, Dexi 1982

Xia`nda`i Ha`nyuˇ Xı´ngro´ngcı´ Ya´njiu` [On adjectives in Modern Mandarin]. Yuˇya´n Ya´njiu` [Language Studies] 1: 83–112. Yuˇfaˇ Jiaˇngyı` [Lectures on Chinese Syntax]. Beijing: Shangwu Yingshuguan.

Website Resources Academia Sinica Balanced Corpus of Modern Chinese http://www.sinica.edu.tw/SinicaCorpus/index.html Google www.google.com Udndata com (Lı´anhe´ zhı #shı` ku`) http://udndata.com/library/

The semantic historical development of modal verbs of volition in Chinese Alain Peyraube* and Li Ming** 1. Introduction The history of the modal auxiliaries, sometimes also called ‘‘can-wish’’ verbs (ne´ngyua`n do`ngcı´ 能愿动词) after Wang Li (1947), is very complex, and any brief discussion of the subject inevitably involves a gross oversimplification. It is all the more di‰cult to determine the main lines of their evolution since the use of the modal auxiliaries in Standard Classical Chinese (Late Archaic Chinese of the pre-Qin period, 5th–3rd centuries BC) is itself far from clear. Several years ago, Peyraube (1999 on the auxiliaries of possibility, 2001 on the modal auxiliaries of volition) and Li Ming (2001, 2002) began a study of these verbs in Classical and Medieval Chinese in order to better understand their evolution. We would like to take up again the subject of modal auxiliaries, as well as their evolution from Archaic Chinese to Modern Chinese via the intermediate stage of Medieval Chinese. Based on an investigation into representative texts in each period of Chinese history, we will examine how volition is expressed throughout its entire history, and try to find out the pathways of recurrent semantic change. The following definition for modal auxiliaries was put forward in Peyraube (1999): a modal auxiliary is an auxiliary verb which allows only predicative objects and which expresses the modality of the following VP. This modality (ability, possibility, probability, certainty, obligation, volition, etc.) can be characterized as epistemic, deontic, or dynamic, following Palmer (1986). Four verbs of volition in Classical Chinese were identified in Peyraube (2001), all meaning ‘‘dare, wish, intend, be willing, consent’’ and they are: gaˇn 敢, keˇn 肯, yu` 欲, and yua`n 愿. Before resuming our investigation of these verbs of volition and examining them in more detail along with others including jia#ng 将, ya`o 要, xiaˇng 想, etc., we will quickly summarize some of the hypotheses and find* CRLAO (Centre de recherches linguistiques sur l’Asie Orientale), EHESS & CNRS, Paris, France. ** Institute of Linguistics, Chinese Academy of Social Sciences, Beijing, China.

150

Alain Peyraube and Li Ming

ings which were elaborated upon in Peyraube (1999, 2001) in the following section of this paper (section 2).

2. Previous hypotheses and findings 2.1. The modal verbs were all originally single full verbs, allowing nominal objects, which were later grammaticalized and became auxiliary verbs, allowing only predicative objects. Such an evolution is cross-linguistically robust (see Hopper and Traugott 1993: 45–48; Lightfoot 1979: 81; Harris and Campbell 1955: 172; Traugott and Dasher 2002: 105–151.). We do not know, however, for Chinese, when these evolutions took place precisely, nor, above all, when the new category of auxiliary verbs was definitely established. Some of these verbs, particularly those of volition, while retaining full verb status, could also serve as auxiliary verbs when they took an object VP. 2.2. The notion of modality is vague and leaves room for a number of possible definitions. Descriptions that have been put forward so far include but are not limited to ‘attitudes or opinions’, ‘speech acts’, ‘subjectivity’, ‘nonfactivity’, and ‘possibility and necessity’. We agree with Alleton’s view that modality is ‘‘a set of processes by which the speaker takes his/her distance concerning the information in his/her utterance’’ (Alleton 1984: 1). What modality signals is the speaker’s attitude toward the proposition. The distinction made between subjective and objective modality, and the one which most Chinese linguists adopt, is not very easy to apply. The distinction between epistemic and deontic modalities as prototypes is more useful and clearer for the analysis of modal auxiliaries in Classical Chinese than is objective versus subjective modality (Peyraube 1999). Epistemic modality is concerned with matters of knowledge, belief or opinion rather than fact (Lyons 1977: 681–2, 793). It involves ‘‘the degree of commitment by the speaker to what he says’’ (Palmer 1986: 51). Deontic modality is concerned with the necessity or possibility of actions performed by morally responsible agents (Lyons 1977: 823). This distinction does not mean, of course, that a modal verb has a single interpretation (epistemic or deontic). Most of them have both epistemic and deontic interpretations. Palmer’s model, which argues that there are di¤erent degrees in which the speaker may be involved, and then proposes to recognize not two, but three types of modality: epistemic, deontic, and dynamic was also adopted. Dynamic modality is said to be concerned with ‘ability and disposition’ and

The semantic historical development of modal verbs of volition in Chinese

151

may also be involved with intent and desirability. It is generally agentoriented, and mostly is not concerned with subjectivity. Espistemic modality is subjective, dynamic modality is objective, and deontic modality may be subjective, or objective, or somewhere in between the two. 2.3. The following two hypotheses concerning the historical evolution of modal auxiliaries of possibility in Classical Chinese were formulated: (i) the first is that the auxiliary modals have undergone a shift from nonepistemic to epistemic: non-epistemic > epistemic; (ii) the second is that there was also a shift from weak subjective epistemicity to more strongly subjective epistemicity: weak epistemicity > strong epistemicity. These hypotheses support the claim that meanings tend to move toward greater subjectivity and are in accordance with the analysis of Traugott (1989) on epistemic meanings in English, confirming the general tendencies in semantic change that she has stated. The latter tendency is especially confirmed, in the sense that meanings tend to become increasingly based on the speaker’s subjective beliefs or attitudes toward the proposition. 2.4. As for the modal auxiliaries of volition that express will, hoping and wishing, their status as auxiliary verbs is probably more debatable than that of the modals of possibility. Certain works, such as Ma Jianzhong (1898), clearly treat gaˇn, yu` and yua`n as full verbs. Li and Thompson (1981: 174) do the same for ya`o ‘want, wish’ in contemporary Chinese. Given the definition of auxiliary verbs provided above, it is nevertheless argued that these verbs are also modal auxiliaries, even if this means we have to recognize volition as a peripheral type of modality. It is also di‰cult to assign to them a feature of either epistemic modality or deontic modality. They obviously are not strictly deontic. Yet, equally, they are not epistemic, since they do not express the degree of the speakers’ commitment to what they are saying. As they are more like deontic modals in the sense that they are concerned more with possible action than with truth or knowledge, Palmer’s (1986: 115) proposal to use the term ‘deontic’, for the sake of convenience was adopted. This is also Givon’s view (1995: 111–115), though his preference, however, is to talk in terms of ‘valuativedeontic’. Among ‘valuative-deontic’ attitudes, he listed the following notions: desirability, preference, intent, ability, obligation and manipulation. Another approach, equally valuable, would be to consider the volitives as dynamic verbs. Let’s now look at the evolution and historical development of these modal verbs of volition in more detail. We will distinguish three categories and not four, as in Peyraube & Li (2009): the first category expresses the

152

Alain Peyraube and Li Ming

meaning of yua`nyı` 愿意 ‘be willing to, be ready, wish, like, want’, and includes A gaˇn 敢, keˇn 肯, and yua`n 愿; the second expresses the meaning of xıwa # ` ng 希望, yua`nwa`ng 愿望 ‘hope, wish, expect, desire, aspire’ and includes yu` 欲, jia#ng 将, and yua`n 愿; while the third expresses the meaning of yı`zhı` 意志 ‘intention, want, will’, and includes yu` 欲 and ya`o 要. This last category codes the strongest volition. 3. Development of the volitives in the first category This first category of volition verbs expresses the weakest volition: volition that is neither active nor voluntary. The agent is not conceptualized as eager to carry out the action. Gaˇn means ‘dare to do something’; keˇn means ‘consent to another’s request’; yua`n means ‘be willing to do something (at the request of another, or due to the will of the agent per se).’ 3.1. Gaˇn 敢 Gaˇn occurred for the first time in the Bronze Inscriptions of the Western Zhou period (11th–6th c. BC), and exhibited two functions: one, a modal verb whose meaning is ‘dare to’; the other, a special kind of adverb, usually called biaˇojı`ng fu`cı´ 表敬副词 ‘respect-signaling adverb’ in the literature, since, in this use, it has almost lost its lexical meaning and indicates the speaker’s respect towards the relevant person. As claimed earlier in Peyraube (2001), gaˇn is found mostly in the forms of negation. In a few cases, gaˇn also occurs in a‰rmative sentences, but judging from the context, it is apparently accompanied by an interrogative or exclamatory tone. Many examples are given in Peyraube (2001). We can hypothesize that the adverbial use of gaˇn was derived from the modal use. At the beginning, gaˇn had the pragmatic meaning of qıˇ gaˇn 岂敢 ‘how dare’ due to its frequent occurrence in rhetorical questions. When this meaning was gradually conventionalized into the lexical meaning of gaˇn, it evolved into an adverb. The developmental pathway gaˇn followed is laid out below: (1) volitional modal verb > adverb signaling the speaker’s respect. The adverbial use of gaˇn declined after the pre-Qin period (before 221 BC), while the meaning of modal gaˇn has been maintained up to the present time, with a remarkable stability that has not undergone any important changes. It is of little use then to expand our discussion further on this first volitive gaˇn.

The semantic historical development of modal verbs of volition in Chinese

153

3.2. Keˇn 肯 Keˇn means ‘accept a request’ in contemporary Chinese. In pre-Qin Chinese, keˇn has to be interpreted as ‘be willing to, be ready to’ [ yua`nyı` 愿意], rather than ‘accept a request’ [keˇn 肯]. As mentioned above, keˇn codes a special kind of willingness; it is the willingness to accept the requests of others. Keˇn has thus undergone the following semantic change: (2) ‘be willing to’ > ‘consent to’, ‘accept a request’, which is a sort of semantic specialization or narrowing of meaning. Similar to gaˇn, keˇn is first documented in Early Archaic Chinese (11th– 6th c. BC), especially in the Sha`ng Shu# 尚书 ‘Book of Documents’ and in the Shı # Jıng # 诗经 ‘Book of Odes’, with the meaning of ‘be willing to.’ Example: (3) 厥子乃弗肯播矧肯获? (尚书: 大诰, ca. 9th–8th c. BC) Jue´ zıˇ naˇi fu´ keˇn bo# sheˇn keˇn huo` 3SG child even NEG willing sow:seed let:alone willing harvest ‘His child is not even willing to sow seed, let alone willing to harvest.’ (4) 此邦之人不我肯穀 (诗经: 黄鸟, ca. 10th–7th c. BC) cıˇ ba#ng zhı # re´n bu` woˇ keˇn guˇ this country GEN people NEG 1SG willing nourish ‘People from this country are not willing to nourish me.’ In the majority of cases in the pre-Qin period, there is no such context in which keˇn can be definitely interpreted as ‘accept a request’ [keˇn 肯], while there is no di‰culty in interpreting keˇn as ‘be willing to’ [ yua`nyı` 愿意], just as the two examples above show. Another example follows: (5) 先生不肯视，行不辍，何邪? (庄子: 人间世, 4th c. BC) xia#nshe#ng bu` keˇn shı`, xı´ng bu` chuo`, he´ ye# master NEG willing look act NEG stop why PRT ‘[A carpenter left for the state of Qi. When he arrived at Quyuan, he saw an oak tree in a shrine for the God of the Earth. . . Many people were watching it, and the place was like a bazaar. The carpenter did not watch it at all and he just kept going. His apprentice said:] ‘‘. . . Master, you were not willing to look at it and kept going. Why is this so?’’’ In this example, nobody asked the carpenter to look at the oak tree, and it was the carpenter himself who was not willing to look at it. Clearly,

154

Alain Peyraube and Li Ming

keˇn in this case cannot be understood as ‘accept a request’ and should be interpreted as ‘be willing to’. Nevertheless, we have to recognize the appearance of a few cases of this second meaning of [keˇn 肯] as ‘accept a request, agree’ during preQin times, as in: (6) 请盟，齐侯不肯 (左传: 文公十六年, 5th c. BC) qıˇng me´ng Qı´ ho´u bu` keˇn ask alliance Qi marquis NEG agree ‘[Jiwenzi] asked the [Marquis of Qi] to make an alliance, but the Marquis of Qi did not agree.’ (7) 吾已召之矣，丙怒甚不肯来 (韩非子: 内储说上, 3rd c. BC) Wu´ yıˇ zha#o zhı # yıˇ, 1SG already call:upon 3SG PRT Bıˇng nu` she`n bu` keˇn la´i Bing angry very NEG consent come ‘I have already called upon Bing, but he was very angry and did not consent to come.’ In (6), keˇn is a main verb in the clause while in (7), keˇn is a modal auxiliary followed by another verb. By the time of the Han and Weijin periods (1st–4th c. AD), most of the cases of keˇn could already be interpreted as ‘consent to, accept the request’. We estimate that it was not until the Southern and Northern Dynasties period (5th–6th c. AD) that keˇn fully completed its semantic change from ‘be willing to’ > ‘consent to, accept a request’. Since then, keˇn has been in continual use with the meaning ‘accept a request.’ From the Eastern Han to the Tang (1st–10th c.), one can also find some cases in which keˇn expresses root possibility, which probably belongs to dynamic modality. Example: (8) 百年谁肯保 (王梵志诗: 卷五, 8th c.) Baˇi nia´n she´i keˇn baˇo 100 year who can ensure ‘Who can guarantee living for 100 years?’ The meaning of this sentence is not ‘Who is willing to guarantee living for 100 years?’ but ‘who is able to guarantee it?’ During the Tang Dynasty (7th– 10th c.), keˇn was also sometimes used to express an epistemic possibility.

The semantic historical development of modal verbs of volition in Chinese

155

(9) 名位岂肯卑微休? (杜甫: 徐卿二子歌, 8th c.) mı´ng we`i qıˇ keˇn be#iwe´i xiu# fame position how can humble PRT ‘How can someone with his fame and position be humble?’ Keˇn does not indicate volition here or any enabling condition and can be only interpreted as expressing an epistemic possibility. To sum up, the historical development of keˇn followed this path: (10) ‘be willing to’ > ‘accept the request’ > root possibility > epistemic possibility 3.3. Yua`n 愿 Among the modal verbs carrying the meaning of ‘be willing to’, the most common is undoubtedly yua`n, which first occurred in the Lu´n yuˇ 论语 ‘Analects’. For example: (11) 非曰能之，愿学焉! (论语: 先进 , 5th c. BC) Fe#i yue# ne´ng zhı,# yua`n xue´ ya#n NEG say capable 3SG willing learn PRT ‘[I] don’t say that I am capable of it. [Just that I] am willing to learn.’ (12) 回不愿仕 (庄子: 让王, 4th c. BC) Huı´ bu` yua`n shı` Hui NEG willing be:an:o‰cial ‘I (Hui) am not willing to be an o‰cial.’ In the pre-Qin period, yua`n and keˇn were semantically nearly synonymous whereas their syntactic distributions were almost complementary: the subject of yua`n is essentially the speaker, and this is exemplified by (11) and (12) above, while the subject of keˇn is never the speaker. On the other hand, keˇn was mostly confined to the negative form (rhetorical questions can be equally treated as negative forms) while yua`n is free of this restriction. As the meaning of keˇn gradually shifted from ‘be willing to’ to ‘accept a request’ during the Han-Wei-Jin times, the semantic di¤erence between yua`n and keˇn became more and more striking. The di¤erence, just before the meaning of keˇn shifted from ‘be willing to’ to ‘accept a request,’ was precisely this: keˇn denoted the agent’s acceptance of another person’s request and necessarily involved inter-subjective communication, while

156

Alain Peyraube and Li Ming

yua`n only emphasized the agent’s willingness, no matter whether this volition was due to his own initiative, such as in (11), or came at the request of other people, such as in (12). The meaning of yua`n ‘be willing to’ has been maintained right up to its present use in Contemporary Chinese. Hence we will give no further examples. During Tang-Song times (7th–13th c.), a new verb appeared expressing the meaning of ‘be willing to’: this was qı´ngyua`n 情愿: (13) 情愿将身作夫妻 (变文: 破魔变 , 9th–10th c.) qı´ngyua`n jia#ng she#n zuo` fu#qı # willing take 1SG make husband:and:wife ‘[I] am willing to marry you.’ In the Yuan dynasty (13th–14th c.), the common modal verb yua`nyı` 愿意, widely used today in contemporary Chinese, appeared, but without any object: (14) 你既愿意，就配与元帅为夫人者 (元曲选: 玉箫女, 14th c.) Nıˇ jı` yua`nyı`, jiu` pe`i yuˇ yua´nshua`i we´i fu#re´n zheˇ 2SG since willing then match to Marshal as wife PRT ‘Since you are willing, [I will] marry [you] to the Marshal.’ There is no evidence before the Qing period (1644–1911), however, of the use of yua`nyı` with another VP as its object (i.e. as a modal auxiliary): (15) 谁愿意穿这些 (红楼梦, 31回, 18th c.) She´i yua`nyı` chua#n zhe`ixie# who willing wear these ‘Who is willing to wear these [clothes]?’ 3.4. Summary As shown above, the first type of volitive denotes the weakest volition, consisting of three subsets in terms of semantic meaning: ‘dare to’, ‘be willing to’ and ‘consent to, accept the request’. The first one is expressed by gaˇn 敢, the third one by keˇn 肯, and the second one by both keˇn 肯 and yua`n 愿. During Tang-Song times (Late Medieval, 7th–13th c. AD), the verb qı´ngyua`n 情愿 appeared and from the 17th or 18th century onwards, yua`nyı` 愿意 began to act as a modal verb. Both these verbs express the notion ‘be willing to.’

The semantic historical development of modal verbs of volition in Chinese

157

4. Development of volitives of the second category The volitives of this second category express wishes and hopes ( yua`nwa`ng 愿望). The degree of volition is stronger than the one expressed by the volitives of the first category. It represents a kind of voluntary and active volition cherished by people. Three main verbs are involved in this second category: yu` 欲, jia#ng 将 and yua`n 愿. 4.1. Yu` 欲 The earliest examples of yu` are found in the bronze inscriptions of the Early Archaic Chinese period, as in: (16) 俗 (欲) 我弗作先王忧 (毛公鼎, ca. 10th c. BC) su´ ( yu`) woˇ fu´ zuo` xia#n wa´ng yo#u wish 1SG NEG make former king adversity ‘(I) wish that I might not get former kings into trouble.’ In the Sha`ng Shu# (also Early Archaic Chinese), there are several examples of yu` with this meaning of ‘wish’ ( yua`nwa`ng): (17) 欲王以小民受天永命 (尚书: 召诏, ca. 9–8th c. BC) yu` wa´ng yıˇ xiaˇo mı´n sho`u tia#n yoˇng mı`ng wish king with small people receive Heaven perpetual mandate ‘(We) wish that the king, with his people, might receive Heaven’s mandate in perpetuity.’ In these examples, yu` expresses the speaker’s wish or desire and not his intention. Later, in the Late Archaic period (5th–2nd c. BC), this meaning of ‘wish, desire’ will be conveyed by yua`n 愿 (see section 4.3 below) while yu` will come to express an intention, rather than a wish or desire. In Early Archaic Chinese there was no yua`n. We can therefore conclude that yu` originally expressed a ‘wish’ or a ‘hope’ of the speaker, but yua`n, which appeared later, replaced yu` in this use in the Spring and Autumn period, while at the same time, yu` evolved a new sense, that of ‘intend to’. The semantic shift of yu` might be portrayed thus: (18) wish, hope > intention 4.2. Jia#ng 将 Jia#ng followed the same developmental path as yu` during the pre-Qin times, i.e. a shift from ‘wish’, ‘hope’ to ‘intention’. The first meaning was already attested in Early Archaic Chinese, as exemplified in (19 below):

158

Alain Peyraube and Li Ming

(19) 将子无怒。 (诗经: 氓 , ca. 10th–7th c. BC) Jia#ng zıˇ wu´ nu` wish 2SG NEG angry ‘Wishing that you do not become angry.’ (20) 古之为关也，将以御暴。 (孟子: 尽心下, 4th c. BC) Guˇ zhı # we´i gua#n yeˇ jia#ng yıˇ yu` ba`o ancient PRT make fort PRT intend with resist violence ‘In ancient times, people made the fort because they wanted to resist violence.’ However, apart from the two usages shown above, jia#ng acted mostly as an adverb of futurity, with the meaning of ‘will, be going to’ in pre-Qin texts. In fact, the modal verb uses – one with the meaning ‘wish, hope’ and the other with the meaning ‘intend, want’ – are especially rare and the only previous study that referred to this was Mei (2004) who makes a clear distinction between the modal senses of jia#ng and its future meaning. We can hypothesize that jia#ng underwent a grammaticalization process that changed it from a modal verb into an adverb by the Spring and Autumn period (8th–5th c. BC) and caused it to function almost exclusively as an adverb since that time. Thus, the modal uses only show residues of this development and are not well documented in the extant texts. In contrast to the development of yu`, the semantic shift of jia#ng can be characterized as follows: (21) wish, hope > intention > will (future marker) 4.3. Yua`n 愿 As mentioned above, yua`n first appeared in the Lu´n Yuˇ with the meaning of ‘willing to’, but it also has the meaning of ‘wish, hope’, as in: (22) 愿闻子之志 (论语: 公冶长, 5th c. BC) yua`n we´n zıˇ zhı # zhı` wish hear 2SG GEN ideal ‘(I) would like to hear your ideal.’ (23) 札虽不才，愿附于子臧 (左传: 襄公十四年, 5th c. BC) Zha# suı # bu` ca´i yua`n fu` yu´ Zıˇza`ng Zha though NEG ability wish attach to Zizang ‘Though I (Zha) have no ability, I wish to follow Zizang’s example [and would not like to be the king of Wu].’

The semantic historical development of modal verbs of volition in Chinese

159

This meaning of ‘wish’ is recorded right up to the 18th century: (24) 你愿他死了，有什么好处? (红楼梦, 25回, 18th c.) Nıˇ yua`n ta# sıˇ le yoˇu she´nme haˇochu` 2SG wish 3SG die PFV have what benefit ‘You wish him to die, and what’s the good of this?’ The preceding examples, however, are not representative of the essential usages of yua`n. During a very long period, up to the Yuan dynasty (14th c.), in fact, a great majority of yua`n’s usages were found in situations wherein the speaker was making a polite request to the hearer and therefore were performative in use, i.e. indicating that the speaker was performing an act by requesting the hearer to do something, as in (22). Performative uses of yua`n can also be divided into two types: one wherein the speaker hopes to do something himself which more or less concerns the hearer, and, in this case, yua`n takes a VP as its complement, as shown in (22); the other when the speaker asks the hearer to carry out an action, and, in this case, yua`n takes a clausal complement, as shown in (25): (25) 抑臣愿君安其乐而思其终也 (左传: 襄公十一年, 5th c. BC) yı` che´n yua`n ju#n a#n qı´ le` e´r sı # qı´ zho#ng yeˇ however 1SG wish 2SG enjoy that joy and think that end PRT ‘However, I would like you to enjoy that joy and think about [keeping it to] the end.’ This last example is the earliest one showing yua`n with a clausal complement. From the 8th or 9th century on, another particular performative use of yua`n can be found that expresses the speaker’s blessing. This usage is retained today. For example: (26) 愿圣人万岁, 万万岁! (变文: 韩擒虎话本 , 9th–10th c.) yua`n she`ngre´n wa`n suı` wa`n wa`n suı` wish emperor 10,000 year 100,000,000 years ‘May your majesty live for thousands of years!’ As mentioned above, the non-performative uses of yua`n were uncommon and disappeared at the end of the Qing dynasty, around the 19th century. In contrast, performative uses of polite request were quite popular only until the 15th century, and the performatives of blessing were also frequently used from 8th–9th centuries on. We can conclude from this that yua`n is a highly subjective modal verb, i.e. it is usually related to the speaker’s wish.

160

Alain Peyraube and Li Ming

A last remark: In Contemporary Mandarin Chinese, in addition to the yua`n which expresses the speaker’s blessing, another verb, the disyllabic, da`nyua`n 但愿 also expresses the speaker’s wish. This verb was already in use in the 8th–9th centuries, but it was not until the Yuan dynasty (13th– 14th c.) that it became a compound and was specialized to express the speaker’s wish. 4.4. Other verbs expressing wishes and hopes Other verbs have been used to express wishes and hopes, and some of them are still used today with the same meaning. The first one, jı` 冀, was already being used during the Warring States period (5th–3rd c. BC), and is quite frequently attested during the Han (206 BC–220 AD) as well. For example: (27) 冀复得兔 (韩非子: 五蠹, 3rd century BC) jı` fu` de´ tu` hope again obtain hare ‘(He) hopes to get another hare.’ The second one, wa`ng 望, underwent a semantic change in Late Archaic Chinese, evolving from the meaning ‘look far away’ to that of ‘hope’, and was quite common in Late Medieval Chinese, from the 7th c. to the 13th c. Some examples include: (28) 王如知此, 则无望民之多于邻国也 (孟子: 梁惠王上, 4th c. BC) Wa´ng ru´ zhı # cıˇ King if know thus ze´ wu´ wa`ng mı´n zhı # duo# yu´ lı´ng guo´ yeˇ then NEG hope people PRT many than neighbor state PRT ‘If you (the king) know this, you should not hope that your people are more numerous than in neighboring states.’ (29) 伏望大圣慈悲 (变文: 维摩诘经讲经文 (五), 9th c.) fu´ wa`ng da`she`ng cı´be#i prostrate hope mahatma show:mercy ‘I sincerely hope that you (mahatma) will show mercy to me.’ The third one, xı # 希, underwent the same semantic change as wa`ng, from a verb of physical perception ‘look,’ to a mental verb meaning ‘look forward to’, but it was not until the Tang and Five Dynasties period (8th– 10th c.) that xı # as a modal verb really came into use, as in:

The semantic historical development of modal verbs of volition in Chinese

161

(30) 心心希听受 (变文: 维摩诘经讲经文 (一), 9th c.) xı #n xın# xı # tıng # sho`u mind mind hope hear receive ‘Everyone hopes to hear [Buddha preach].’ The fourth one, pa`n 盼, was originally an adjective describing the beauty of female eyes. It first evolved along metonymical lines into a verb expressing the meaning ‘gaze at’ after which the physical verb developed into a mental verb ‘look forward’. Example: (31) 盼梦里他来到 (元曲选: 梧桐雨，第四折, 14th c.) pa`n me`ng lıˇ ta# la´i da`o hope dream in 3SG come arrive ‘I look forward to his coming to my dream.’ Finally, as wa`ng 望 has been undoubtedly the most popular volitive verb to express wishes and hopes, since the Tang dynasty (7th–10th c.), it is not surprising that it became a component of several disyllabic volitive verbs, including che´ngwa`ng 承望 (attested during the Tang), zhıˇwa`ng 指望 (under the Song), pa`nwa`ng 盼望 (under the Yuan) and xı #wa`ng 希望 (only documented in Contemporary Chinese).

5. Development of volitives of the third category In this third and last category, we actually group two types of volitive verbs expressing intention. In Contemporary Chinese, intention can be clearly divided into two subsets: one is xiaˇng 想 ‘think, want’, and the other is ya`o 要 ‘want’. The latter now expresses a stronger volition than the former. However, this is not true for most of the periods of Ancient Chinese. As xiaˇng was not established as a separate semantic category before the 17th century, the modal verbs indicating intention cannot be divided into two subsets before that time. Moreover, even after the 17th century, the verb ya`o 要 could still convey both xiaˇng and ya`o meanings. Thus, we think that it is more appropriate to regard these two types as one when dealing with the historical evolution of the modal verbs of volition. 5.1. Yu` 欲 We have seen above that yu` underwent a semantic change from ‘wish’ to ‘intend, want’ in the Late Archaic period [see (18)]. For example:

162

Alain Peyraube and Li Ming

(32) 夫子疾病，不欲见人 (左传: 昭公四年, 5th c. BC) fu#zˇı jı´bı`ng bu` yu` jia`n re´n master ill NEG intend see people ‘The Master is ill and does not intend to receive guests.’ In the majority of cases, it is not easy to distinguish whether yu` expresses only an intention (the meaning of xiaˇng) or a stronger volition (the meaning of ya`o). However, we think that it is more appropriate for yu` to be glossed in the same way as xiaˇng when the volitive is preceded or followed by a negation, as in (32) above. As early as the 1st c. BC (the Former Han period), yu` developed a future meaning ‘be going to’: (33) 殿欲坏者三 (史记: 李斯列传, 1st c. BC) dia`n yu` hua`i zheˇ sa#n palace be:going:to broken NOM three ‘The palace is going to be ruined for the third time.’ (34) 臣朔饥欲死 (汉书: 东方朔传, 1st c. AD) che´n shuo` jı # yu` sıˇ 1SG Shuo hungry be: going: to die ‘I am hungry (to the extent that) I am going to die.’ In (33), the subject is inanimate and yu` cannot express an intention. In (34), although the subject is [þHuman], the event expressed by the verb (sıˇ ‘to die’) is certainly not what people intend to do. We thus have the following semantic change: (35) intention > future There were other verbs expressing an intention that co-existed with yu` during short periods of time, but they never really challenged the strong status of yu` or ya`o. These verbs are: guı # 规 ‘intend’ (mainly used during the Six Dynasties period, 4th–6th c.), nı # 拟 ‘intend’ (quite popular during the Tang period, 7th–10th c.), and da`i 待 ‘want’ (10th–14th c.). 5.2. Xiaˇng 想 Xiaˇng appeared during the Yuan dynasty (13th–14th c.), but was not frequently used, and probably kept its original meaning of ‘think of ’, up to the 16th–17th centuries, as can be seen by comparing examples (36) and (37):

The semantic historical development of modal verbs of volition in Chinese

163

(36) 不想回家 (金瓶梅, 12回, 16th c.) bu` xiaˇng huı´ jia# NEG think:about/want return home ‘(He) does not think of returning home / He does not want to return home.’ (37) 不思想来家 (金瓶梅, 98回, 16th c.) Bu` sıxia # ˇ ng la´i jia# NEG think:about come home ‘(He) does not think of coming back home.’ After the 17th century, xiaˇng’s only competitor was ya`o (to be discussed in the next section). It has survived up to the present day by taking over one part of the meaning of ya`o (the meaning ‘think of ’), while ya`o underwent a semantic change from ‘think of ’ to ‘want’. 5.3. Ya`o 要 The earliest examples of ya`o as a modal verb meaning ‘to want’ are found in texts of the Tang dynasty (7th–10th c.): (38) 若要相知者 (拾得诗, 7th c.) ruo` ya`o xia#ng zhı # zheˇ if want each:other know PRT ‘If one wants to know me . . .’ (39) 汝只要我道不得 (祖堂集: 卷六，投子和尚, 952) Ru´ zhıˇ ya`o woˇ da`o bu` de´ 2SG only want 1SG say NEG able ‘You only want me to be unable to answer.’ Ya`o developed from an original meaning ‘to request, to invite, to ask’ that was in common use during the Three Kingdoms period (3rd century), e.g.: (40) 公瑾昔要子敬来东 (三国志: 吴志, 吕蒙传, 3rd c.) Go#ngjıˇn xı´ ya`o Zıˇjıˇng la´i do#ng Gongjin in: the :past invite Zijing come east ‘In the past, Gongjin invited Zijing to come to the East.’ The semantic change can be illustrated thus: (41) ‘to request, to ask’ > ‘to think of, to want’.

164

Alain Peyraube and Li Ming

It is important to keep in mind that from the 7th century to the 17th century, ya`o could express the meanings of both ‘think of ’ (xiaˇng) and ‘want’ ( ya`o). It was not until the Contemporary Chinese period that xiaˇng eventually became the exclusive modal indicating ‘to think of ’, while ya`o retained only the ‘want’ meaning and abandoned the meaning ‘think of ’. A clear dichotomy then arose in the semantic domain of intention. The following examples, taken from texts before the time of Contemporary Chinese, show that ya`o was used with the xiaˇng meaning and would be replaced by xiaˇng in today’s Chinese: (42) 我只要拿一拿大姐的俊手儿 (团圆梦, 16th c.) woˇ zhıˇ ya`o na´ yı # na´ da`jieˇ de ju`n shoˇur 1SG only want take CL take elder : sister SUB pretty hand ‘I only want to take your pretty hand for a while.’ (43) 不要吃了 (金瓶梅, 56回, 16th c.) bu` ya`o chı # le NEG want eat PRT ‘(I) don’t want to eat any more.’ We also have to mention, finally, that ya`o, like yu` before (see above) developed a new meaning expressing futurity around the 11th–12th centuries. Example: (44) 到工夫要断绝处 (朱子语类: 卷 8, 12th c.) da`o go#ngfu# ya`o dua`njue´ chu` reach work be:going:to break:o¤ place ‘When the work is going to break o¤. . .’ The semantic change looks like this: (45) ‘to think of, to want’ > future marker. Several further usages developed out of this future meaning, but they will not be discussed here.

6. Conclusion 6.1. The historical development of the modal verbs of volition in Chinese involved several important semantic changes. 1. Intention > Future

The semantic historical development of modal verbs of volition in Chinese

165

This path of grammaticalization, which recurs in the history of Chinese, is typologically well attested in general linguistics (cf. Bybee, Perkins, and Pagliuca 1994). It concerns the following verbs: yu` 欲, da`i 待 and ya`o 要. An interesting issue can be raised here, viz: why is it that only modals having the meaning of ‘intention’ and belonging to the third category have developed into future markers, and not the other volitives? And why did xiaˇng never become a future marker? We would like to put forward the following reason: this semantic change is only possible for modals implying that the agent is eager to carry out the action with insistence, and thus the modal predicts the action. 2. Weak volition > Strong volition This semantic change relates to jia#ng 将 and yu` 欲. They both originally only expressed ‘wish, hope’, and belonged to the second category, but later, they came to indicate intention, and switched into category III. 3. Physical domain > Mental domain The semantic change from physical domain to mental domain is well illustrated by the development of xı # 希, wa`ng 望 and pa`n 盼 in category II. They were originally verbs of vision, but developed later into mental volitional verbs meaning ‘hope, look forward’. 6.2. As already mentioned in Section 2, volition is assuredly not a core category in the domain of modality. This claim is confirmed by the analysis undertaken in this study from di¤erent aspects: (i) Syntactically speaking, the forms in category II, such as xı #wa`ng 希望, very often take a clausal complement, though it is also normal for them to take a VP complement. This characteristic is obviously not in accordance with the syntactic property of ordinary modal verbs, since modal verbs should be followed by VPs. The forms in category III, such as yu`, xiaˇng and ya`o can also take clausal complements. The verbs in category I alone cannot take clausal complements. (ii) Semantically speaking, except for yua`n 愿, volitional verbs show little subjectivity involving a speaker’s attitude or opinion. Although some verbs in category III might develop a meaning of futurity (epistemic uncertainty), the other verbs, (with the possible exception of keˇn 肯), have no connection at all with the prime constituents of modality, i.e., epistemicity or deonticity.

166

Alain Peyraube and Li Ming

References Alleton, Viviane 1984 Les Auxiliaires de Mode en Chinois Contemporain. Paris: Maison des Sciences de l’Homme. Bybee, Joan, Robert Perkins & William Pagliuca 1994 The Evolution of Grammar – Tense, Aspect, and Modality in the Languages of the World. Chicago: University of Chicago Press. Givo´n, Talmy 1995 Functionalism and Grammar. Amsterdam: John Benjamins. Harris, Alice C. & Lyle Campbell 1995 Historical Syntax in Cross-linguistic Perspective. Cambridge: Cambridge University Press. Hopper, Paul. J. & and Elizabeth C. Traugott 1993 Grammaticalization. Cambridge: Cambridge University Press. Li, Charles N. & and Sandra A. Thompson 1981 Mandarin Chinese. A Functional Reference Grammar. Berkeley and Los Angeles: University of California Press. Li, Ming 2001 Hanyu zhudongci de lishi yanbian yanjiu [Research on the historical development of Chinese auxiliary verbs]. Ph. D. diss., Beijing University. Li, Ming 2002 Liang Han shiqi de zhudongci xitong [The system of the auxiliary verbs under the Han], Yuyanxue luncong, vol. 25, 257–283. Lightfoot, David 1979 Principles of Diachronic Syntax. Cambridge: Cambridge University Press. Lyons, John 1977 Semantics. Cambridge: Cambridge University Press. Ma, Jianzhong 1898 Ma Shi Wen Tong [Grammar by Master Ma]. Beijing: Zhonghua shuju, 1988 reed. Ma shi wen tong xiao zhu. Mei, Kuang 2004 Functional categories in Classical Chinese. Ms. Meisterernst, Barbara 2008a Modal verbs in Han period Chinese, Part I: the syntax and semantics of ke and keyi. Cahiers de Linguistique Asie Orientale 37–1, 85–120. Meisterernst, Barbara 2008b Modal verbs in Han period Chinese, Part II: Negative markers in combination with the modal auxiliary verbs ke and keyi. Cahiers de Linguistique Asie Orientale 37–2, 197–222. Palmer, Frank R. 1979 Modality and the English Modals. London: Longman.

The semantic historical development of modal verbs of volition in Chinese

167

Palmer, Frank R. 1986 Mood and Modality. Cambridge: Cambridge University Press. Peyraube, Alain 1999 On the modal auxiliaries of possibility in Classical Chinese. In Selected Papers from the 5th International Conference on Chinese Linguistics, H.S. Wang, F. Tsao & C. Lien (eds.), 27–52. Taipei: Crane Publishing Company. Peyraube, Alain 2001 On the modal auxiliaries of volition in Classical Chinese. In Chinese Grammar: Synchronic and Diachronic Perspectives, Chappell H. (ed.),172–187. Oxford: Oxford University Press. Peyraube, Alain & Ming Li 2009 Hanyu yiyuan dongci de lishi yanbian [Historical evolution of the volitive verbs in Chinese]. Hanyu shi, vol. 8, 17–35. Traugott, Elizabeth C. 1989 On the rise of epistemic meanings in English: An example of subjectification in semantic change. Language, vol. 65 (1), 31–55. Traugott, Elizabeth C. & Richard B. Dasher 2002 Regularity in Semantic Change. Cambridge: Cambridge University Press. Wang, Li 1947 Zhongguo xiandai yufa [Contemporary Chinese Grammar] 2 volumes Beijing: Shangwu Yinshuguan.

Semantic change in the grammaticalization of classifiers in Mandarin Chinese Janet Zhiqun Xing 1. Introduction The classifier has attracted many scholars in the past due to its unique grammatical function. Western linguists have devoted much energy to trying to shed light on why some languages developed a classifier system while others did not. Greenberg (1990: 177) categorizes the Chinese numeral classifier as ‘‘an individualizer which performs the same function as a singulative derivational a‰x in languages with the collective/singulative distinction.’’ Based on Greenberg’s study, Bisang (1998, 1999) further elaborates on the function of the Chinese numeral classifier saying that it provides a way to individualize and classify its referent. Grammarians of Chinese (e.g. Ohta 1958, Huang 1964, Peyraube 1998, Zhang 2008), on the other hand, have focused more on the morpho-syntactic development of numeral classifiers. The most recent development on this front is presented by Wu et al. (2006) who, going against the traditional analysis that states that the ‘numeral þ classifier þ noun’ combination was developed from the ‘noun þ numeral þ classifier’ construction, suggest that, due to the clear functional (syntactic, semantic and discourse pragmatic) di¤erences between the two aforementioned structures, it is impossible for one to be derived from the other. They claim that the ‘numeral þ classifier þ noun’ construction was derived from ‘numeral þ classifier þ zhı # 之 (possessive marker) þ noun’, in which the possessive marker was fused and lost over time. However, Zhang (this volume) provides quantitative and semantic evidence from Late Archaic Chinese (200 B.C.–200 A.D.) to Medieval Chinese (220 A.D.P618 A.D.) showing that numeral classifiers were derived from the ‘noun1 þ numeral þ noun1’ construction, in which the second noun was grammaticalized into a classifier. Another recent study (Tao 2006) takes into consideration both Western and Chinese researchers’ points of view and examines the Chinese numeral classifier from its developmental end point. Tao proposes a phono-syntactic account for the loss of classifiers in spoken Beijing Mandarin and claims that the loss of the general classifier

170

Janet Zhiqun Xing

ge` was caused by phonetic erosion through everyday use. All these studies investigate Chinese numeral classifiers from di¤erent angles and provide morpho-syntactic, phono-syntactic, historical and typological accounts of this linguistic element. However, what appears to be lacking in the literature, except for Zhang’s study in this volume, is a fine-grained study of the semantic change involved in the development of Chinese classifiers. In the literature of diachronic change over the last two decades (e.g. Heine et al. 1991, Hopper & Traugott 2003 [1993], Bybee et al. 1994, Traugott & Dasher 2002), two phenomena have been prominently discussed: semantic change and grammaticalization. This is probably due to the consensus regarding the inter-relationship between the two, namely that grammaticalization is a diachronic process in which a lexical item changes to a grammatical item and its meaning changes from more concrete/specific to more abstract/general (cf. Bybee et al. 1994). Clearly, semantic change is one of the major components in the process of grammaticalization. Without a good understanding of how meaning shifts from one semantic domain to another, it is di‰cult, if not impossible, to form a complete picture of the process of grammaticalization. For this reason, and because of the lack of systematic analyses of semantic change in the Chinese language mentioned above, the current study investigates the development of sixteen numeral classifiers in Mandarin Chinese (henceforth Chinese) and seeks answers to two major questions: 1) what factors, be they semantic, pragmatic, or constructional, trigger the emergence and disappearance of numeral classifiers? and 2) what factors lead to semantic change in the grammaticalization of numeral classifiers? In accordance with Zhang (this volume), this study hypothesizes that the emergence of numeral classifiers in Chinese was prompted by the use of a countable common concrete noun (e.g. tree, plant, branch)1 with a numeral, namely the [Num þ NP] construction. This constructional form, commonly used for counting in Archaic Chinese, signals that the NP in it has entered the process of grammaticalization 1. There has been a debate with regard to the original semantic functions of classifiers. Many grammarians of Chinese believe that the earliest classifiers were developed from repeaters (Wang 1958, Huang 1964, Peyraube 1998, Li 2000). Zhang in this volume, however, points out that the semantic properties of the few repeaters that existed in Oracle Bone Inscriptions (i.e. Early Archaic Chinese 1700 B.C.–1100B.C.) are not compatible with those of early classifiers in Late Archaic Chinese (200 B.C.–200 A.D.) and concludes that it is impossible for classifiers to have developed from repeaters. This finding leads us to question whether classifiers in Oracle Bone Inscription actually existed, let alone to use the characteristic of repeaters as criteria for classifiers developed later.

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

171

from noun to classifier. Once a NP is used in the [num þ np/cl þ np] construction, its semantic function gradually changes from a countable concrete noun to an abstract general classifier. Eventually, when the classifier loses its semantic function completely and serves purely as a filler in the [Num þ CL þ NP] construction, it can be omitted altogether, thus marking the end of the life cycle of this classifier. The sixteen most frequently used numeral classifiers in Modern Chinese (with over 10,000 occurrences each) were selected out of one hundred seventy eight instances from A Chinese-English Dictionary of Measure Words [Classifiers]2 (Jiao 2001). The frequency counts, shown in Table 1,3 are based on their occurrences with the numeral (Num) yı # 一 ‘‘one’’, demonstratives (Dem) zhe` /na` 这/那 ‘‘this/that’’, and interrogative words (Q) jıˇ 几 ‘‘how many’’ and naˇ 哪 ‘‘which’’, three environments where Modern Chinese numeral classifiers are used in the online corpus compiled by the Center for Chinese Linguistics, Peking University (2005). The following section (2) presents the development of numeral classifiers from their source meanings to their classifier meanings. Section 3 discusses the mechanisms identified in the semantic change of the sixteen cases in the course of their grammaticalization. Section 4 explores factors contributing to the disappearance of the general classifier ge` 个 in Modern Chinese and Section 5 focuses on the interface between semantic change and grammaticalization. The last section (6) summarizes the findings of this study and attempts to draw a conclusion.

2. The word ‘classifiers’ was added because the dictionary uses ‘‘measure words’’ to refer to both measure words and classifiers. 3. This article excludes measure words such as pı´ng 瓶 ‘bottle’ and be #i 杯 ‘cup’, as well as mass nouns such as qu´n 群 ‘group’. It should be noted that, apart from their classifying function, some of the classifiers included in this study also have another meaning/function (e.g. bu` 部 can be used as a noun meaning ‘‘bureau’’; baˇ can be used as an object marker; fe #ng can be used as a verb meaning ‘‘to seal’’ etc.). Their frequency counts were calculated by searching the classifiers used in the [numeral (一 yı # ‘‘one’’)/demonstrative (这/那 zhe`/na` ‘‘this/that’’)/interrogative word (几 jıˇ ‘‘how many’’ and 哪 naˇ ‘‘which’’) þ classifier þ NP] structure so that those non-classifying usages of the same character would be excluded. Although this process also excludes the classifier’s usage with other numerals (e.g. liaˇng ‘‘two’’, sa #n ‘‘three’’, shı´ ‘‘ten’’, yı #baˇi ‘‘one hundred’’ etc.), the results are relative and replicable because the same method has been applied to all sixteen classifiers for their frequency counts.

172

Janet Zhiqun Xing

Table 1. Frequency counts of the 16 classifiers in Modern Chinese Classifier

Semantic function

w/Num

w/Dem

w/Q

Total

ge` 个 we`i 位

generic

502,039

268,650

52,299

822,988

58,105

41,243

7,224

106,572

tia´o 条 jia`n 件

‘long object/concept’

48,097

13,423

2,244

63,746

‘divisible object’

25,685

17,548

1761

44,994

ju` 句 xia`ng 项

‘sentence’ ‘event/project/policy’

25,241 24,440

9,009 12,576

7,489 1,295

41,739 38,311

zhı # 只

‘object/animal (part of a pair)’

20,860

6,596

1,286

28,742

zha#ng 张 zuo` 座

‘flat object’

19,321

7,189

1,394

27,904

‘object with a base’ ‘book/magazine’

15,974 14,725

8,003 11,208

530 1,374

24,507 27,307

zhı # 支 fe`n 份

‘branchable concept’

14,117

4,414

426

18,957

‘document/sharable concept’

14,002

6,352

319

20,673

baˇ 把4 bu` 部

‘hands-on object’

11,440

1,763

328

13,531

‘collectable book’

13,343

11,506

958

25,807

pia#n 篇

‘essay’

10,342

6,099

901

17,342

fe #ng 封

‘letter’

7,327

3,779

407

11,513

beˇn 本

‘honorable person’

2. From lexical to grammatical Among grammarians of Chinese (e.g. Wang 1980 [1958]: 235, Ohta 1987 [1958]: 148–153, Liu 1965: 1, Peyraube 1998, 1991: 106), it is generally agreed that numeral classifiers in Chinese first emerged during the PreQin period (approx. 1100–221 BC), developed into a grammatical category (i.e. the classifier class) during the Han Dynasty (206 BC–220 AD), and expanded significantly in use during the Wei-Jin-Nan-Bei Dynasties (221–589 AD). Recently, based on quantitative evidence, Zhang (2008) 4. In Modern Chinese, baˇ 把 is predominantly used as a classifier for hands-on objects such as umbrellas, knives, chairs, fans, keys, etc. Only a small number of baˇ 把 is used in a measure word function. The frequency count in Table 1 excludes the usage as a measure word and its verbal and prepositional functions. This is also true of the statistics for baˇ 把’s distribution in historical texts given in Table 2 (also see Note #5).

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

173

indicates that many di¤erent types of classifiers (e.g. general classifiers, shape/size classifiers) indeed developed in Middle Chinese (i.e. the Han and Wei-Jin-Nan-Bei Dynasties), but the classifier class was not fully established until the Tang Dynasty (618–907 AD). These studies, conducted by Chinese grammarians, have clearly contributed to our understanding of what lexical forms emerged as classifiers and when. But as far as the motivational factors contributing to the development of the classifier class in Chinese are concerned, there is still little clarity. However, comparing related studies conducted by Western typologists, we discover that some of their findings are actually complementary to those found by Chinese grammarians. Erbaugh (1986: 399), for instance, suggests that ‘‘the historical development of classifiers was triggered by the need for explicit surface distinctions among nouns as the language became increasingly homophonous because of massive phonemic mergers.’’ Aikhenvald (2000: 373–91), on the other hand, provides evidence to show that the development of the classifier is often motivated not only by internal phonological and morphological factors (e.g. some north Australian languages eliminate vowel-initial words; as a result, free classifier forms have become noun-class prefixes), but also by external factors such as areal di¤usion (i.e. less prestigious languages borrow from more prestigious ones, cf. the spreading of numeral classifiers from the Indo-Aryan languages to the Dravidian languages), creolization and pidginization (i.e. most creoles and pidgins, such as the Russian-Chinese pidgin spoken in Harbin, China, have no numeral classifiers), and language obsolescence (e.g. obsolescence of numeral classifiers in Warekena in Arawak). Although the e¤ect of these external factors mentioned by Aikhenvald may be di‰cult to verify through written historical Chinese texts, the internal factors appear to coincide with the findings of Feng’s (2005) study. Bisang (1999), though not directly discussing the motivational factors for the emergence of the classifier class, points out that the classifier primarily serves three pragmatic purposes: classification, identification and individualization, which indirectly supports Erbaugh’s position. Based on these studies and the evidence from the studies done by Chinese grammarians, it is likely that the emergence of the Chinese classifier class is a result of both external – the communicative need to classify, identify or individualize noun class and internal factors – and the constructional (phonological, morphological and syntactic) need to correspond with other constructional developments (e.g. verb þ complement, serial verb, prepositional phrase þ verb) – occurring during the same time period.

174

Janet Zhiqun Xing

On the other hand, although many Chinese classifiers emerged in Middle Chinese (220 A.D.–618 A.D.) as noted earlier, some developed earlier and more rapidly than others. For instance, ge` 个 was used to classify ‘bamboo’ or ‘bamboo-related objects’ as early as the Pre-Qin period (1100–221 BC) and, by the 4th–5th Centuries, extended its function to classify not only a variety of ‘objects’, but also ‘humans’. In contrast, xia`ng 项 was used as a noun to refer to ‘the back part of the neck’ during the 3rd Century; it did not become a common classifier for events/projects until the Ming Dynasty (the 14th Century). More importantly, of the 16 Chinese numeral classifiers investigated in this study, some have developed from full-fledged nouns, similar to the origin of numeral classifiers in many other languages (cf. Langacker 1991: 165, Aikhenvald 2000: 353, Croft 2002: 27–38), while others have evolved from full-fledged verbs, such as fe#ng 封 ‘to bestow land’, jia`n 件 ‘to divide’ and zha#ng 张 ‘to draw a bow’. Although there are languages (e.g. Imonda a Papuan language, see Aikhenvald 2000: 362– 365) in which numeral classifiers have developed from verbs, it is much less common cross-linguistically for numeral classifiers to have developed from verbs than from nouns. I argue that the reason why some of the Chinese numeral classifiers have developed from verbs is the isolating characteristic of the Chinese language, in which nouns and verbs are not specified for category (e.g. no case, gender, agreement markers) (cf. Bisang 2008). I will come back to this point later. The distribution of the sixteen classifiers5 in historical texts is given in Table 2. They are listed alphabetically. The five historical periods were chosen due to their importance in the development of numeral classifiers in Chinese. As mentioned earlier, Chinese grammarians generally agree that Chinese numeral classifiers emerged sometime during the Pre-Qin period (in Old Chinese). They developed into a grammatical category in the Han and Wei-Jin-Nan-Bei (WJNB) Dynasties (in Middle Chinese). By the Tang Dynasty (in early Modern Chinese), the numeral classifier had become a fully established class in Chinese grammar. In Table 2, we see that 14 out of the 16 lexemes obtained the classifying function by the time of the WJNB period. During the Tang Dynasty, the 5. The distribution of the sixteen classifiers is the result of a search for the classifiers used in the [num þ cl þ (np)] construction in the Corpus compiled by the Center for Chinese Linguistics, Peking University (2005). The same corpus was used for the frequency of the classifiers in Modern Chinese given in Table 1. Please note that the statistics in Table 2 do not include any tokens of the classifiers used with demonstrative or interrogative words, usages developed since the Tang Dynasty.

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

175

Table 2. Distribution of the 16 classifiers in five historical texts6 Classifier baˇ 把7 beˇn 本 bu` 部 fe #n 分/ fe`n 份8 fe #ng 封 ge` 箇/個/个* jia`n 件 ju` 句 pia#n 篇 tia´o 條/条* we`i 位 xia`ng 項/项* zha#ng 張/张* zhı # 支 zhı # 隻/只* zuo` 座

Pre-Qin

0 4 2

13 0 36 0 0 6 1 11 0 0 0 0 0

Han

WJNB

Tang

Ming-Qing

Total

1 10 6 17 6 2 0 0 12 2 7 0 1 9 0 1

0 5 11 15 6 8 0 8 16 6 5 0 8 14 6 4

0 478 40 74 124 1863 136 135 91 376 154 1 93 75 157 166

942 588 521 923 1372 40690 3757 2968 445 3364 3307 101 2556 595 2436 2178

943 1085 580 1042 1508 42599 3893 3111 568 3749 3484 102 2658 693 2599 2349

* These are di¤erent versions of the logographs for the given lexeme: one traditional and the other simplified.

distribution of all classifiers (except for two, xia`ng 项 and baˇ 把) increased dramatically, and ge` 个 clearly emerged as the most frequently used (1,863) of all classifiers. Text analysis indicates that the relatively sudden increase of the classifier ge` 个 during the Tang period can be attributed to its general semantic function and its replacement by me´i 枚, another dominant general classifier in Middle Chinese (cf. Liu 1965, Wang S. 1992, Zhang 2008). Another noticeable fact in Table 2 is that the frequency of baˇ 把 in classifying function is rather low prior to the Ming-Qing Dynasties. Text analysis 6. Pre-Qin: 1100–221 BC; Han: 206 BC–220 AD; Wei-Jin-Nan-Bei (WJNB): 221– 589 AD; Tang: 618–907 AD; Ming-Qing: 1368–1911 AD. 7. Baˇ 把 began to be used as a measure word meaning ‘a handful of ’ and a classifier for ‘hands-on objects’ during the Han Dynasty. This study considers baˇ 把 in yi baˇ huoˇ 一把火 ‘a fire’ as a measure word, but the one in yi baˇ huoˇju` 一把火炬 ‘a torch’ as a classifier. The distribution of baˇ 把 in Table 2 only includes its classifier usage. 8. In Old Chinese, fe #n 分 began to be used as a classifier for ‘dividable/sharable objects’. The same logograph and function continued to be in use until the Ming Dynasty (14th Century) when it was gradually replaced by fe`n 份.

176

Janet Zhiqun Xing

shows that this is probably due to its frequent occurrence as a verb and measure word in Middle Chinese and during the Tang Dynasty. To investigate how the 16 classifiers have developed from nouns and verbs, their origins, or rather their source meanings, have been traced and their paths of semantic change have been analyzed on the basis of historical texts9 written from the Pre-Qin to modern times. Table 3 summarizes the source meanings and the current classifier meanings of the sixteen classifiers included in this study. Source meaning refers to the etymological meaning of the lexical item from which the classifier meaning has developed. All the source meanings provided in Table 3 are either deciphered from an earlier usage in Old Chinese texts or referred to in the first etymological dictionary, Shuo# We´n Jieˇ Zı` 说文解字 ‘Definition and Explanation of Words and Characters’, written by Xu Shen in the 2nd Century. Table 3 shows that among the sixteen classifiers, five of the source meanings (beˇn 本 ‘root of plant’, ge` 个 ‘bamboo pole’, pia#n 篇 ‘bamboo stalk’, tia´o 条 ‘tree branch’, and zhı # 支 ‘tree branch’) relate to plants, three relate to animate beings, humans or animals, (bu` 部 ‘tribe’, xia`ng 项 ‘neck’, and zhı # 只 ‘single bird’), five describe some kind of common/handling action (baˇ 把 ‘to hold with a hand’, fe#n 分 ‘to divide’, fe#ng 封 ‘to bestow land’, jia`n 件 ‘to divide’, and zha#ng 张 ‘to draw a bow’), one relates a part to a whole ( ju` 句 ‘string of words’), and two illustrate position or location (we`i 位 ‘location’, zuo` 座 ‘seat’). From a cognitive perspective, it is not surprising that all the source However, what makes the study of fe #n 分 more complicated than that of any other classifier included in this study is that, in addition to its verbal and classifying functions, fe #n 分 was also used as a measure word in Old Chinese to refer to a unit of area equal to approximately 66.67 m2. In Table 2, I tried to include only the classifying function of fe#n 分 as it appears in the five periods of historical texts. I have to admit though that, in some cases, it is very di‰cult to tell whether it is used as a classifier or a measure word as demonstrated in Note 5. 9. Most of the examples used in this study are quoted from texts of five periods: 1) Pre-Qin: Shı´sa #n Jıng # 十三经 ‘Thirteen Classics’, 2) Han Dynasty: Shıˇjı` 史记 ‘Record of History’; 3) Wei-Jin-Nan-Bei Dynasties: Shı`shuo# Xınyu # ˇ 世说 新语 ‘Records and Anecdotes’, We´nxuaˇn 文选 ‘Selected Essays’; 4) TangSong Dynasties: Bia`nwe´n 变文 ‘Say-sing Literature’, Zuˇ Ta´ng Jı´ 祖堂集 ‘Collection of Buddhist Texts’; and 5) Ming-Qing Dynasties: Xıyo # ´ u Jı` 西游记 ‘Journey to the West’, Sa #nguo´ Yaˇnyı` 三国演义 ‘Romance of the Three Kingdoms’, Shuıˇhuˇ Zhua`n 水浒传 ‘Outlaws of the Marsh’.

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

177

Table 3. Source meaning and classifier meaning Classifier baˇ 把 beˇn 本 bu` 部 fe #n 分/ fe`n 份 fe #ng 封 ge` 個/个 jia`n 件 ju` 句 pia#n 篇 tia´o 條/条 we`i 位 xia`ng 項/项 zha#ng 張/张 zhı # 支 zhı # 隻/只 zuo` 座

Source meaning ‘to hold with a hand’ ‘root of plant’ ‘tribe/army’ ‘to divide’ ‘to bestow land/title on subjects’ ‘bamboo stalk’ ‘to divide’ ‘sentence/string of words’ ‘bamboo stalk’ ‘tree branch’ ‘position/location’ ‘neck (back part)’ ‘to draw a bow’ ‘tree branch’ ‘single bird’ ‘seat’

Classifier meaning for for for for for

hands-on objects books and magazines collectable books documents/sharable concepts letters using an envelope

for for for for for for for for for for for

general usage divisible objects words/sentences articles/writing long objects honorable people projects/events/policies flat objects branchable concepts objects/animals (part of a pair) objects/statues with a base

meanings of the classifiers are related to objects, actions or events commonly seen in the ancient world, because plants, animals, language and humans were the most common and important entities in ancient society. They are countable, concrete, visible, and easy to identify and classify. This path of development is also evident in the evolution of classifiers in other languages (cf. Aikhenvald 2000: 354–359, 362–363). What makes the comparison between the source meaning and the classifier meaning interesting, though, is that classifiers with the same or a similar source meaning may not share common characteristics among their classifier meanings. For instance, both tia´o 条 and zhı # 支 derive from the source meaning ‘tree branch’, yet as a classifier, tia´o 条 is used for ‘long objects’ or ‘traceable concepts’ (as ‘fish’, ‘line’, ‘road’, ‘news’), whereas zhı # 支 is used for ‘branchable concepts’ (such as ‘troop’, ‘river’). The same is true of ge` 个 vs. pia#n 篇 and fe#n 分 vs. jia`n 件, whose similar original meanings are not transparent in their classifier meanings. On the other hand, if we compare the source meaning of beˇn 本 ‘root of plant’ with the source meaning of bu` 部 ‘tribe/army’, we find that they are quite di¤erent, yet their classifier

178

Janet Zhiqun Xing

meanings are somewhat similar (i.e. for ‘bound books’). An even more intriguing case is that the logograph baˇ 把 with its source meaning ‘to hold with a hand’ has developed two completely di¤erent grammatical functions: it is used as a classifier modifying ‘hands-on objects’ and as a preposition or object marker placed after a subject and before a verb. These examples provide evidence that the source meaning of a lexical item does not necessarily have a definable relation to its grammatical meaning (e.g. classifier, object marker), and this pattern is similar to that demonstrated by the semantic change and grammaticalization of verbs in Chinese (see Xing 2003: 118–119). Bybee et al. (1994: 12) clarify their position on the relation between the source meaning and the grammatical meaning as follows: Our claim is not that the source meaning gives a unique grammatical meaning, but rather that the source meaning uniquely determines the grammaticalization path that the gram will travel in its semantic development. When we take into consideration the meaningful units that comprise a grammatical construction, we find that the meaning present in the source construction bears a definable relation to the grammatical meanings that later arise and that these earlier meanings prefigure the grammatical meaning.

Apparently, Bybee and her associates focused their attention on the relationship between the source meaning of a construction (i.e. meaningful unit) and the corresponding grammatical meaning developed later. The evidence from this study (see Section 4) generally supports Bybee and her associates’ view that the meaning of the source construction, and not of the lexical source, foreshadows the later-developed grammatical meaning. In addition, this study, particularly the example of baˇ 把, shows that the source meaning does not necessarily determine the grammaticalizaton path that the gram has traveled in its semantic development. Now the central concern of this study is to shed light on how the sixteen classifiers developed from their lexical source meanings to their grammatical classifier meanings. Based on the database used for this study, historical polysemies10 involved in the development of the 16 classifiers are summarized below, and glossed examples are given whenever necessary throughout 10. Please note that the summary only includes polysemies relevant to the development of each lexeme’s classifier meaning. Some of the 16 lexemes have developed other polysemies diachronically, but they are not included in the summary. For instance, baˇ 把, in addition to its nominal polysemies, developed several verbal polysemies (e.g. ‘hold with a hand’ > ‘take/bring’) before it became grammaticalized into an object marker.

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

179

the paper. A more comprehensive list of examples showing the polysemies of the 16 classifiers is provided in the Appendix.11 baˇ 把:

‘to hold with a hand’ > ‘hands-on’ > mw ‘handful’/cl for hands-on objects beˇn 本: ‘root of plant’ > ‘foundation/body/origin’ > ‘document’ > cl for books bu` 部: ‘tribal/army’ > cl for organized group of people > cl for volume of written documents > cl for collectable books fe#n 分/ fe`n 份: ‘to divide’ > ‘one part of a whole/share’ > cl for sharable objects/documents fen# g 封: ‘to mark land with dirt as a border’ > ‘to bestow land/title on subjects’/ ‘to cover with dirt/seal’ > cl for letters (with sealable envelopes) ge` 個/个: ‘bamboo stalk’ > cl for bamboo/bamboo objects > cl for any objects/people jia`n 件: ‘to divide’ > division of/piece of an object (wood/ tools) > piece of a concept (things/crime) > cl for a piece of clothing/utensils/matter ju` 句 ‘string of words/sentence’ > cl for sentence ‘bamboo stalk’ > ‘essay written on bamboo stalk’ > cl pia#n 篇 for essays tia´o 條/条: ‘tree branch’ > long objects/abstract concepts > cl for long objects/abstract concepts we`i 位: ‘position/location’ > ‘(honorable) status (emperor/the dead)’ > cl for those of honorable status xia`ng 項/项 ‘back part of the neck’ > ‘kind/type’ > cl for projects/ events/policies zha#ng 張/张: ‘to draw a bow’ > ‘to open (flat objects)’ > cl for flat objects zhı # 支: ‘branch of plant’ > body limb/branch of family/river > cl for branchable concepts zhı # 隻/只: ‘single bird’ > ‘one of a pair’ > cl for objects/animals (part of a pair) zuo` 座: ‘seat’ > ‘base’ > cl for objects/statues with a base

11. Following are the abbreviations used for this paper: cl – classifier, poss – possessive marker, sg – singular, asp – aspect marker, part – particle, nom – nominalizer, pl – plural, mw – measure word, and pass – passive marker.

180

Janet Zhiqun Xing

Each of the polysemies to the left of the arrow (>) refers to the earlier meaning and that to the right refers to the later-developed meaning. Take beˇn 本 as an example: its source meaning is ‘root of plant’ as illustrated in (1a). From this source meaning, the semantic development appears to have branched into di¤erent directions. One branch developed by metaphorically extending the source meaning to the meaning ‘foundation’ or ‘origin’ (i.e root > foundation/origin), as in (1b)–(1c). The other branch represents an extension of the source meaning through metonymical shift to ‘body’ (i.e root > body), as in (1d). Then, again through a metonymical shift, the meaning, most likely ‘origin’, is shifted to refer to ‘document’ or ‘copy of document’, as in (1e). Finally, through reanalysis, this leads to the classifier use for ‘bound written materials’ such as books, magazines, dictionaries, etc., as in (1f ), which is the meaning that has been used in Chinese ever since. (1) a. 枝葉未有害，本實先撥。(Approx. 11th–6th C. BC, Shijing) zhı #ye` we`i yoˇu ha`i, beˇn shı´ xia#n bo´ branch-leaf no have hurt, root indeed first break ‘The root, in fact, shatters first before the branches and leaves are damaged.’ b. 君子務本。(approx. 3rd C. BC, Lunyu) ju#nzıˇ wu` beˇn gentleman work-on fundamental-issue ‘Gentlemen work on important issues.’ c. 樂者，. . . 其本在人心之感於物也。 (2nd C. BC, Liyueji) yue` zheˇ, . . . qı´ beˇn za`i re´n xın# zhı # gaˇn yu´ music person its origin reside person heart poss feel about wu` yeˇ things part ‘A musician’s profession (develops from sound). The essence (of this profession) is people’s feelings toward things.’ d.

. . .其大本臃腫，而不中繩墨。(3rd C. BC, Zhuangzi) qı´ da` beˇn yo#ngzhoˇng, e´r bu´ zho`ng she´nmo` its big body pu¤-up, so not middle measure-tool ‘Its big body is pu¤ed-up, so it’s not straight when measured by a rope ruler.’

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

181

e. 河北此書家藏一本。 (4th C., Rules of Yan’s Family) he´ beˇi cıˇ shu# jia# ca´ng yi beˇn river north this book home hide one cl ‘One copy of this book is hidden at home to the north of the river.’ f. 古人據個甚摩事去卻四十二本經論 (10th C., Zutang Ji) guˇ re´n ju` ge` she´nme shı` ancient people based cl what thing qu` que` sı`shı´ e`r beˇn jınglu # `n go purge forty two cl scripture-study ‘Why did ancient people purge forty-two volumes of scripture?’ From the summary of the sixteen classifiers and the illustrative examples given in (1), we see that the critical step in the development of a classifier meaning is when its semantic function is reanalyzed as a modifier classifying ‘people’, ‘objects’ or ‘concepts.’ Thus, we call this step a ‘categorical change’, i.e. verbal/nominal meaning > classifier meaning. Other semantic changes provided in the summary may be considered preparatory (i.e. those changes that occur prior to the categorical change) or extended (i.e. those that occur after the categorical change). According to this analysis, it appears that two stages of meaning are critical in the development of the classifier meaning: 1) the lexical meaning before a lexeme enters into the grammatical form, ‘num þ np/cl’, and 2) the classifier meaning in the grammatical form ‘num þ cl þ np’. From the summary, we can also see that the sixteen classifiers have undergone a di¤erent number of steps in their semantic change: some have undergone three, some two and a few only one step. Furthermore, some of those semantic changes are easier to detect than others. For instance, the two steps of semantic change of pia#n 篇 from its source meaning ‘bamboo stalk’ to its classifying function for ‘essays’ are more straightforward than the two steps in fe#ng’s 封 development from its source meaning ‘to bestow land’ to its classifier meaning used for ‘letters’. How these semantic changes (from nouns/verbs to classifiers) occurred and what mechanisms are involved in those changes is the primary focus of the following section. 3. Semantic change in grammaticalization Traugott & Dasher (2002: 24–40) developed a systematic account for the nature of semantic change in grammaticalization and suggest that ‘‘the chief driving force in the process of regular semantic change is pragmatic:

182

Janet Zhiqun Xing

the context-dependency of abstract structural meaning allows for change in situations of use, most particularly the speaker’s role in strategizing this dynamic use.’’ Furthermore, they recognize two mechanisms involved in semantic change: metaphoricalization, ‘‘primarily an analogical principle [that] involves conceptualizing one element of a conceptual structure Ca in terms of an element of another conceptual structure Cb’’, and metonymization, a contiguous or logical relation between two conceptual structures/ domains (e.g. part > whole and cause > result). The evidence from this study shows that metaphoricalization and metonymization are indeed two major mechanisms in the process of semantic change and grammaticalization of classifiers in Chinese, as exemplified in (2)–(3). (2) a. b. c.

beˇn 本: zhı # 支: tia´o 条:

(3) a. b. c.

zha#ng 张: ‘to draw a bow’ > ‘to open’ pia#n 篇: ‘bamboo stalk’ > ‘essay on bamboo stalk’ xia`ng 项: ‘neck (back part)’ > ‘type’

‘root of plant’ > ‘foundation/origin’ ‘branch of plant’ > ‘branch of family’ ‘tree branch’ > ‘long (object)’

All three examples in (2) are instances of metaphoricalization where two concepts, ‘root of plant’ vs. ‘foundation/origin’, ‘branch of tree’ vs. ‘branch of family’, and ‘tree branch’ vs. ‘long (object)’, share similar semantic properties in one way or another. In comparison, the three examples in (3) show that metonymization is at work because all three pairs of concepts, ‘to draw a bow’ vs. ‘open’, ‘bamboo stalk’ vs. ‘essay written on a bamboo stalk’, and ‘back part of the neck’ vs. ‘type’, demonstrate either a contiguous, part-whole relationship or a deductive relationship (i.e. since it is the ‘back part of the neck’and not the whole neck, it is, therefore, a ‘type/category of something’.) In addition to the two mechanisms discussed so far, this study has also found that semantic reanalysis is a crucial factor in the development of classifier meaning. Although reanalysis has been widely recognized as a mechanism in morphosyntactic and phonological change (Meillet 1912, Langacker 1977, Lightfoot 1991, Harris & Campbell 1995), not until recently was it introduced into semantic change in grammaticalization as semantic reanalysis (see Eckardt 2006). Based on the morphosyntactic definition that reanalysis refers to a ‘‘change in the structure of an expression or class of expressions that does not involve any immediate or intrinsic modification of its surface manifestation’’ (Langacker 1977: 58), this study

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

183

uses semantic reanalysis to refer to a change triggered by morphosyntactic reanalysis whereby a source lexical item that has an inherent meaning is interpreted as having a di¤erent syntagmatic function. Let us examine the semantic change of baˇ 把 and fe#ng 封 in (4)–(5). (4) a. 周公旦把大鉞，畢公把小鉞， (1st C., Shiji) Zho#ugo#ngda`n baˇ da` yue`, Bı`go#ng baˇ xiaˇo yue` Zhougongdan hold big weapon Bigong hold small weapon ‘Zhougongdan held a big weapon. Bigong held a small weapon.’ b. 於此水底得一把毛。 (4th Century, Baiyujin) yu´ cıˇ shuıˇ dıˇ de´ yi baˇ ma´o at this water bottom get one handful/mw hair ‘(Someone) got a handful of hair from the bottom of this water.’ (5) a. 秦將詐稱二世使人遣李良書，不封 (1st C., Shiji) Qı´n jia`ng zha` che#ng e`rshı` shıˇ re´n qiaˇn Qin chief cheat claim junior-emperor ask man dispatch Lıˇlia´ng shu#, bu` fen# g Liliang letter not seal ‘The Qin commander pretended to be the Emperor’s son to send a letter to Li Liang. (The letter) was not sealed.’ b. 子淵附書一封 (3rd C., Luoyang Jialian Ji) Zıˇyua#n fu` shu# yi fe#ng Ziyuan attach letter one cl ‘Ziyuan attached a letter’ In comparing the two usages of baˇ 把 in (4), we see that the sentence construction of (4a) clearly determines the meaning of baˇ 把 as ‘to hold’, whereas the construction in (4b) is di¤erent, with baˇ 把 being in between a numeral yi 一 ‘one’ and a noun ma´o 毛 ’hair’, leading the reader/listener to deduce the meaning as some kind of nominal modifier ‘a handful of ’. In other words, the morphosyntactic change from [Noun þ baˇ 把 þ Noun] to [Num þ baˇ 把 þ Noun] triggers the semantic reanalysis of baˇ from ‘to hold’ to ‘a handful’ and changes its syntactic category from verb to measure word. Similarly, the two fe#ngs 封 in (5a) and (5b) cannot be interpreted as having the same meaning. Since fen# g 封 in (6a) follows the negative marker bu` 不, it is most appropriately interpreted as a verb meaning ‘to seal (an envelope).’ Fe#ng 封 in (5b), on the other hand, follows the numeral yi 一 ‘one’. Therefore, it is best interpreted as a classifier for

184

Janet Zhiqun Xing

Table 4. Four Mechanisms of semantic change12 MTPH baˇ 把 beˇn 本

Z

bu` 部

Z

MTNM

SR

Z

Z

Z

Z

Z

Z

Z

Z

Z

Z Z

jia`n 件 ju` 句

Z

Z

Z

Z

pia#n 篇 tia´o 條/条

Z

Z

fe #n 分/ fe`n 份 fe #ng 封 ge` 個/个

Z

Z

Z

we`i 位 xia`ng 項/项

Z

Z

Z

Z

zha#ng 張/张

Z

Z

Z

zhı # 支 zhı # 隻/只 zuo` 座

Z Z

Z

Z Z

Z

Z

Z

the identification of the noun referent shu# 書 ‘letter’. This kind of interpretation of the two baˇ s 把 and two fe#ngs 封 reflects a process of semantic reanalysis based on their position in a sentence, which leads to the development of their polysemies. It should be pointed out that the process of semantic reanalysis (SR) di¤ers from metaphoricalization (MTPH) and metonymization (MTNM) in that semantic reanalysis is triggered by morphosyntactic reanalysis whereby a source lexical item that has an inherent meaning can be interpreted as having a di¤erent syntagmatic function. In comparison, the other two mechanisms arise out of a certain cognitive construal – metaphoricalization, which is triggered by analogy and metynomization through contiguous or logical relations. Table 4 provides a 12. The decision about which mechanism is at work in the development of the 16 classifiers is made primarily based on the written texts (the Corpus compiled by Peking University 2005 and those listed in Note 7) and on the author’s best judgment. In a few cases, it is di‰cult to tell if a particular change was the result of metaphoricalization (analogical relation) or metonymization (associative relation), especially when the time frame in which the change occurred is long and data sparse.

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

185

summary of the mechanisms involved in the development of the sixteen classifiers from nouns and verbs. The data in Table 4 yield three patterns of semantic change involved in the development of the sixteen classifiers. First, all 16 cases have undergone either metaphoricalization or metonymization or both in their evolution. This suggests that these two mechanisms are indispensable for semantic change. Furthermore, text analysis of the data used for this study shows that those two mechanisms may take place at any stage of the semantic change, as shown in (6). (6) a. 終南何有？有條有梅。 (approx. 7th C. BC, Shijing) Zho#ngna´n he´ yoˇu? yoˇu tia´o yoˇu me´i place-name what have? have tree have flower ‘What does Zhongnan have? (It) has trees and flowers.’ b. 群木蕃滋數大，條直以長。(3rd C. BC, Guanzi) qu´n mu` fa´nzı # shu`da`, tia´o zhı´ yıˇ cha´ng group plant grow big, branch straight and long ‘A lot of plants grew to a tremendous size, their branches straight and long.’ c. 有蕊一條，長於花葉 (3rd C., Liu 1965: 101) yoˇu ruıˇ yi tia´o, cha´ng yu´ hua# ye` have stamen one cl, long compare-to flower leave ‘(The flower) has a stamen longer than the flower’s leaves.’ d. 條屬者取出一條繩 (3rd C., Liu 1965: 101) tia´o shuˇzheˇ quˇ chu# yi tia´o she´ng long-rope owner fetch out one cl rope ‘The owner of the rope took out a rope.’ e. 萬邑王畿曠，三條綺陌平。(4th C., Liu 1965: 102) wa`n yı` wa´ng jı # kua`ng, ten-thousands mile emperor land far-wide sa#n tia´o qıˇmo` pı´ng three cl wide-road flat ‘The emperor’s land is thousands of miles long with three roads, wide and flat.’ Metonymization that occurred before tia´o 條/条 became grammaticalized into a classifier as shown in (6a)–(6b) (meaning ‘tree’ > ‘branch’ – associa-

186

Janet Zhiqun Xing

tive relation) and metaphoricalization occurred after it became a classifier as shown in (6c)–(6e) (classifying ‘rope’ > ‘road’ – analogical relation). Second, all sixteen cases have undergone semantic reanalysis, which makes this trait mandatory among the three mechanisms in the development of Chinese numeral classifiers from lexical nouns or verbs. Another issue closely related to the mechanisms of semantic change is their e¤ect on the semantic properties of the polysemies. The data analysed for this study show that in the course of the development of the sixteen classifiers, some changed their meaning from concrete to abstract (i.e. abstractification or AB) and some from specific to general (i.e. generalization or GN). Many classifiers restricted their semantic function to nouns that share a certain unique semantic property, such as ‘long’, ‘flat’, ‘single’ (i.e. specialization or SP, cf. Hopper 1991: 22), while some others transferred their function from one semantic domain to another, such as beˇn 本, whose meaning shifted from ‘root of plant’ to ‘body/foundation’ and xia`ng 项, which changed from ‘neck of the back part’ to ‘type’ (i.e. transference or TSF). Some changed from verbal to nominal (i.e. nominalization or NM), and still others did not change much at all (i.e. persistence or PST, cf. Hopper 1991: 22), despite the fact that they all became grammaticalized as classifiers, as summarized in Table 5. Since both AB and GN involve changes from a more concrete/substantial concept to a less concrete/substantial concept, they are grouped into one category. In Table 3, we can identify patterns that are also present in the development of other class types (i.e. prepositions, conjunctions, etc.), both in Chinese (cf. Xing 2003) and in other languages (Bybee et al. 1994) that indicate that almost all the lexemes’ semantic functions have become abstracted or generalized in their evolution from nouns/verbs to function words, a typical characteristic of metaphoricalization and metonymization. More importantly, the results in Table 5 indicate that the semantic functions of nouns and verbs tend to become specialized (SP) before they become grammaticalized (12 out of 16), a characteristic completely absent from the evolution of verbs to function words such as prepositions and conjunctions, where meanings only become generalized before they become grammaticalized (cf. Xing 2003). Notice that among the three classifiers that did not become specialized (i.e. jia`n 件, zhı # 只, ge` 个), all have vague semantic functions. Jia`n 件, for instance, appears to have inherited its etymological meaning ‘divide’ to classify ‘divided/ piece of ’ objects or concepts such as ‘tools’, ‘documents’, ‘clothes’, ‘crimes’, and ‘things/matters’. However, it is not obvious to ordinary language users why these objects and concepts are categorized as a jia`n 件 type,

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

187

Table 5. Outcomes led by four mechanisms of semantic change AB/GN baˇ 把 beˇn 本

Z

bu` 部 fe #n 分/ fe`n 份

Z

fe #ng 封 ge` 個/个 jia`n 件

Z

SP

TSF

NM

Z

Z

Z

Z

Z

Z

Z

Z Z

Z? Z Z

Z

Z Z

ju` 句 pia#n 篇 tia´o 條/条

PST

Z? Z

Z

Z

Z

Z

Z

we`i 位 xia`ng 項/项

Z

Z

Z

Z

Z

Z

zha#ng 張/张

Z

Z

Z

zhı # 支 zhı # 隻/只 zuo` 座

Z Z

Z

Z Z

Z

Z

Z

while others that share the divisible characteristic are not. One may argue that the di‰culty in pinpointing jia`n’s 件 semantic function as a classifier may be due to the possible evolution of jia`n 件 into a general classifier like ge` 个. While this study cannot confirm or disconfirm this argument, it is clear that the function of jia`n 件 as a classifier is not as specified or transparent as that of the majority of other classifiers. Another classifier that appears to share a similar vagueness of semantic function is zhı # 只. Historical data show that zhı # 只 cannot only be used to classify objects and humans/animals that have a set of two parts (such as ‘hand’, ‘ear’, ‘eye’, ‘sock’, ‘earring’, ‘chicken’, ‘sheep’, ‘tiger’, ‘butterfly’ etc. – the last four examples either have pairs of legs or a pair of wings), but also for objects that do not form part of a pair, such as ‘boat’, ‘vase’, ‘bowl’, ‘basket’, and ‘suitcase’. In this case, we cannot say that the semantic function of zhı # 只 has become specialized. In addition to jia`n 件 and zhı # 只, ge` 个 has completely lost its semantic function in Modern Chinese, although it did become specialized when it first became a classifier.

188

Janet Zhiqun Xing

Transference (TSF) and Persistence (PST) are two outcomes of semantic change that are in complementary distribution in the development of those classifiers investigated in this study. Text analysis indicates that transference tends to arise from metonymization and semantic reanalysis, whereas persistence does not involve any mechanism of semantic change; it simply retains the classifier’s source meaning. Strictly speaking, if a lexeme does not transfer its semantic function in the process of its grammaticalization, it must have retained its source meaning. However, the data used for this study do not provide such a black-and-white picture. Take fen# 分/fe`n 份, for example, we see that it has neither completely retained its source meaning (‘to divide’) when used as a classifier for ‘divisible’ objects/concepts (e.g. gifts, documents, work, contributions, etc.), nor has it completely transferred its classifier meaning from its source meaning. All the objects/concepts classified by fe#n 分/fe`n 份 still share at least one feature of the source meaning ‘divisible/sharable’. This is why a question mark is added to the semantic persistence of fe#n 分/fe`n 份 in Table 5. The last significant result of semantic change in the process of grammaticalization of classifiers is nominalization as noted in Table 5. Among the sixteen classifiers, five ( fe#n 分/fe`n 份, baˇ 把, fe#ng 封, jia`n 件, zha#ng 张) have developed from verbs. More specifically, these five verbs became nominalized before they obtained a classifier meaning. This is a natural step in the course of the grammaticalization of classifiers, because classifiers, after all, can be considered a subclass of nouns due to their similar semantic property (i.e. they, too, are used to refer to a person, place, thing, event, substance, quality, or idea) and grammatical function (i.e. modifying nouns). Notice that when the five verbs obtained the nominal function, their meanings changed as well: fe#n 分 changed from a common action ‘to divide’ to a more general meaning ‘part of a whole’; baˇ 把 changed from a specific action ‘to hold something with the hand’ to a nominal modifying ‘handson objects’; fe#ng 封 changed from a specific action ‘to seal’ to a general nominal meaning ‘a type of sealable object’; jia`n 件 changed from a specific action ‘to divide’ to a general nominal meaning ‘divisible objects’; and zha#ng 张 changed from a specific action ‘to open’ to a nominal meaning modifying ‘a type of flat object’. All these semantic changes, from more specific to more general, accompany the process of nominalization. However, what makes the process of nominalization unique among the characteristics of semantic change discussed so far is that Chinese verbs can be easily reanalyzed and reinterpreted as having nominal meanings [as illustrated in (4) and (5)], and from there, develop a classifier meaning.

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

189

This is rare among languages with numeral classifiers. For example, in the Austro-Asiatic languages (cf. Aikhenvald 2000: 366) verbs and nouns are often marked for number, case, animacy, etc, to make it less likely for a verb to be interpreted as a noun. In comparison, Chinese does not have any conjugation for agreement, so verbs can be used or interpreted freely as nouns, or vice versa, in discourse. So far we have seen that di¤erent mechanisms of semantic change correlate with di¤erent outcomes of semantic change. Metaphoricalization and metonymization often lead to abstraction and generalization of semantic function; semantic reanalysis leads to nominalization; and metonymization and semantic reanalysis can both be responsible for transference.

4. Disappearance of classifier meaning Among the 16 classifiers investigated in this study, only ge` 个 has completely lost its semantic function in Modern Chinese and is used now as a general classifier (cf. Erbaugh 1986, Liu 1965). The data analysed in this study suggest that any countable noun without a clear semantic characteristic of categorization, or without an established numeral classifier to classify it can be modified by the general classifier ge` 个. Take the loan word dia`nzıˇ yo´ujia`n 电子邮件 ‘electronic mail’ (probably calqued from English) as an example. Yo´ujia`n 邮件 ‘mail’ has been traditionally classified by jia`n 件 for ‘divisible object’ or by fe#ng 封 for regular snail mail which comes in a sealed envelope. However, since the character jia`n 件 is also part of the modified noun yo´ujia`n 邮件, it is not linguistically favorable to also use jia`n 件 as a classifier for that noun. As a result, some people use fe#ng 封 to classify dia`nzıˇ yo´ujia`n 电子邮件, even though it does not need to be physically put in an envelope in the way that regular snail mail does. Others simply use ge` 个 to classify dia`nzıˇ yo´ujia`n 电子邮件 ‘electronic mail’, as evidenced by Google’s frequency counts of the three classifiers categorizing dia`nzıˇ yo´ujia`n 电子邮件 ‘electronic mail’ given in Table 6. These statistics seem to suggest that although fe#ng 封 is still the most widespread classifier for ‘electronic mail’ (282,000 instances), ge` 个 is the most convenient classifier to use in place of fe#ng 封 without losing any of the linguistic (i.e. syntactic, semantic and pragmatic) functions in the numeral-classifier construction. Such an interaction between the general classifier ge` 个 and semantically distinctive classifiers is evident throughout our investigation of the sixteen classifiers’ modern usage. Table 6 provides the results of a Google search for the occurrences of four nouns used with di¤erent competing classifiers on the World Wide Web.

190

Janet Zhiqun Xing

Table 6. Frequency of classifiers used in internet websites13 noun

classifier

frequency

dia`nzıˇ yo´ujia`n

ge` 个 – general fe #ng 封 – for mail jia`n 件 – for dividable objects

132,000 282,000 346

chıˇzi 尺子 ‘ruler’

ge` 个 – general tia´o 条 – for long objects ba˘ 把 – for hands-on objects

4,250 421 155,000

la˘ohu 老虎 ‘tiger’

ge` 个 – general to´u 头* – for animals zhı # 只 – for objects/animals (part of a pair)

30,700 23,300 354,000

zı`dia˘n 字典 ‘dictionary’

ge` 个 – general bu` 部 – for collectable books beˇn 本 – for books

25,600 24,300 122,000

电子邮件

‘electronic mail’

* Classifiers not investigated in this study

In comparing the occurrences of the general classifier ge` 个 with those of semantically distinctive classifiers ( fen# g 封 for ‘mail in an envelope’, baˇ 把 for ‘hands-on objects’, zhı # 只 for ‘animals’, and beˇn 本 for ‘bound books’) categorizing the four di¤erent nouns in Table 6 (dia`nzıˇ yo´ujia`n 电子邮件 ‘electronic mail’, chıˇzi 尺子 ‘ruler’, laˇohu 老虎 ‘tiger’, and zı`diaˇn 字典 ‘dictionary’), we see that the four nouns are still predominantly classified by their semantically distinctive classifiers. Among the three competing classifiers, the general classifier ge` 个 is the second most frequently used one for those nouns. Clearly, the high frequency of ge` 个 13. The statistics given in Table 6 are taken from a Google search done by the author on January 9, 2009, for the targeted classifiers used in [num ( yi 一 ‘one’) þ cl þ np] construction. The frequency count may vary when the date of the search changes because online information changes every minute. One reason to search the World Wide Web (WWW) for frequency counts of those classifiers is that the WWW is an excellent source, (almost animate, if you will) for discovering how ordinary Chinese choose among those competing classifiers. Since the WWW embraces communication among people from all walks of life, it is an ideal place to find the most common, colloquial, and informal language forms in ordinary people’s communication.

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

191

in Modern Chinese is derived from its lack of distinctive semantic function. In other words, if ge` 个 still retained its semantic function to classify ‘bamboo-made objects’ as it did in Archaic Chinese (1700 B.C.–200 B.C.), it would be impossible to use it also to categorize animals, electronic mail, tigers or rulers. More importantly, I would argue that the two characteristics, lack of distinctive semantic function and high frequency, are the main factors responsible for the disappearance of the general classifier ge` 个 in Modern Chinese. Consider the following examples: (7) a. 爸爸，我带你看一东西。(Tao, 2006: 107) ba`ba woˇ da`i nıˇ ka`n yi do#ngxi dad I take you see one thing ‘Dad, I want to take you to see something.’ b. 我们报了一专利。(Tao, 2006: 107) woˇmen ba`o le yi zhua#nlı` 1pl. apply asp one patent ‘We have applied for a patent.’ c. 吃一桃吧。(Tao, 2006: 114) chı # yi ta´o ba chi one peach part ‘Eat a peach.’ Or ‘Please eat a peach.’ In all three examples in (7), the general classifier ge` 个 is omitted, leaving the numeral yi 一 ‘one’ to be followed directly by a noun. An explanation for this is o¤ered by Tao (2006) who proposes a phono-syntactic account. According to her analysis, phonetic erosion in language leads to the omission of the highly frequent classifier ge` 个 while retaining the numeral yi 一 ‘one’ in spoken Beijing Mandarin. While acknowledging Tao’s contribution to our understanding of the loss of the general classifier from a phono-syntactic point of view, I would add that the condition for the omission of the general classifier intrinsically depends on the classifier’s semantic and pragmatic function. That is, when a classifier has a distinct semantic function such as fe#ng 封 classifying ‘letter in an envelope’, it cannot be omitted (*yi xı`n 一信 ‘one letter’) no matter how frequently it is used in discourse, or whether it involves any phonetic erosion or tone sandhi. If it is omitted, language users would be unclear about the meaning of the phrase. In comparison, when the general classifier is omitted, it does not a¤ect language users’ understanding of the constructions given in (7). From a pragmatic point of view, then, the omission could be triggered

192

Janet Zhiqun Xing

by one of the principles of conversational implicature, specifically the Maxim of Quantity (Grice 1989: 26) ‘‘Do not make your contribution more informative than is required,’’ or the Second Heuristic (Levinson 2000: 32) ‘‘What is expressed simply is stereotypically exemplified.’’ Both of these generalizations essentially refer to the same situation, where ‘‘one need not say what can be taken for granted’’ (Levinson 2000: 37). The omission of the general classifier ge` 个 in spoken Beijing Mandarin seems to be a perfect example of this conversational implicature (cf. Fang’s analysis of the omission of yi in the classifier construction of Beijing Mandarin in this volume). From a historical point of view, we actually see a mirror image of the same construction [num þ np] without a classifier used in Middle Chinese, as shown in (8): (8) a. 獨見一女子，狀貌非常 (4th C., Shishuo Xinyu) du´ jia`n yi nuˇ¨ zıˇ, zhua`ngma`o fe#icha´ng only see one girl figure extraordinary ‘(Someone) only saw one girl with an extraordinary figure.’ b. 經一石橋，甚狹而峻。(3rd C., Soushenji) jıng # yi shı´qia´o, she`n xia´ e´r ju`n pass one stone-bridge, quite narrow and dangerous ‘(They) passed a quite narrow and dangerous stone bridge.’ c. 忽有一客來看之，乘一白馬。(3rd C., Soushenji) hu# yoˇu yi ke` la´i ka`n zhı #, che`ng yi ba´imaˇ suddenly have one guest come see it, ride one white-horse ‘Suddenly, a guest, riding a white horse, came to look at it.’ It is interesting to see the nil classifier used in classifier constructions both in Middle Chinese (220 A.D.P618 A.D.), when the classifier class itself was not yet fully developed (cf. Zhang 2008), and in Modern Chinese, where the general classifier ge` 个 can be omitted. We have just discussed some reasons why the general classifier ge` 个 can be omitted in Modern Chinese and we can only speculate why a classifier was not used in the examples in (8) taken from Middle Chinese. Based on the database used in this study, we know that me´i 枚, not ge` 个, was used as a general classifier in Middle Chinese. However, me´i 枚, in contrast to ge` 个, could not classify ‘humans’ (cf. Zhang 2008). Yet the distribution of ge` 个 was quite restricted at that time, as shown in Table 2. In other words, it was still uncommon for people in Middle Chinese to use ge` 个 to classify nouns, including human nouns. This is probably the reason why there is

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

193

no classifier in (8a). Text analysis indicates that the nil classifier in (8b) was probably the result of similar reasoning as observed in (8a), namely, the rarity of the semantically distinct classifier zuo` 座 for ‘bridge’ in Middle Chinese. The situation for (8c) is somewhat di¤erent from (8a) and (8b). We find that the noun, maˇ 馬 ‘horse’, is classified by pı´ 匹 in the same text where we find the example in (8c), which seems to suggest that the classifier pı´ 匹 was not semantically conventionalized for ‘horses’ in Middle Chinese. Of course another possible reason for the usage of the nil classifier in all three examples in (8) is that the classifying constructions had not yet been fully established in Middle Chinese, so that some nouns were modified by a classifier and used in a [num þ cl þ np] construction, whereas others were not. Nonetheless, the usage of the nil classifier in classifying constructions in Middle Chinese and Modern Chinese seems to be triggered by similar linguistic factors – both semantic and pragmatic. That is to say that in Middle Chinese, speakers/writers did not use certain classifiers because they did not know which one was semantically and/or pragmatically appropriate to modify certain nominal entities. In Modern Chinese, on the other hand, speakers/writers omit the general classifier ge` 个 because it does not have any distinctive semantic or pragmatic function in discourse. So far, we have discussed the development of classifiers and the loss of their semantic functions. The process can be said to follow a pattern of four developmental stages: specialization > generalization > loss of semantic function > loss of classifier. It should be noted that this schema is not absolute, but rather a tendency. In other words, to become a classifier, a noun is likely to first develop a distinctive semantic function, which then enables it to categorize or identify a certain type of entity (e.g. objects, things, animals, human activities, etc.). Once it becomes a classifier and is commonly used in discourse, it gradually loses its distinctive semantic property and becomes, instead, a dedicated classifier used to classify di¤erent types of entities. In the end, its ever increasing frequency in discourse along with the disappearance of its semantic function leads to the omission of the classifier altogether in the numeral-classifier construction.

5. Construction grammar, semantic change, and grammaticalization Construction Grammar (CG), developed in the last decade, is considered a ‘holistic’ framework which emphasizes the uniform representation of form and function and states that no one level of grammar, be it syntactic,

194

Janet Zhiqun Xing

semantic or phonological, is autonomous (cf. Croft 2001, Langacker 2005, ¨ stman 2004). Such a Croft & Cruse 2004, Goldberg 1995, 2006, Fried & O view of grammar is clearly echoed in the diachronic evidence of this study. The results of this study indicate that, in the transition of the sixteen lexemes from nouns and verbs to classifiers, in each case the construction (or the form termed by some researchers) in which they are used is the key to the accomplishment of that transition, which seems to coincide with Croft’s (2001: 261) suggestion that ‘‘in the grammaticalization process, construction as a whole changes meaning.’’ We have seen that when the nouns or verbs are used as a head of a phrase (i.e. a noun phrase or verb phrase) without a numeral, they convey a nominal or a verbal meaning, but when they are used with a numeral, they can express a classifier meaning, as shown in (9)–(10). Among the 16 investigated classifiers, there is no single exception to this generalization. (9)

a. 一人持本，一人讀析 (7th C., Taiping Yulian) yi re´n chı # beˇn, yi re´n du´xı # one person carry book, one person read-analyze ‘One person carries the book, the other person analyzes it.’ b. 公德數抄寫一本 (7th C., Bianwen) go#ngde´ shu` cha#oxieˇ yi beˇn achievement account copy-write one cl ‘An account of the accomplishments was recorded in one book.’

(10) a. 既張我弓，即挾我矢。 (approx. 7th C. BC, Shijing) jı` zha#ng woˇ go#ng, jı` jia# woˇ shıˇ if open 1sg bow, then carry 1sg arrow ‘If (you) open my bow, you should carry my arrow.’ b. 子產以帷幕九張行 (1st C. BC, Zuo Zhuan) Zıˇchaˇn yıˇ we´imu` jiuˇ zha#ng xı´ng Name use screen nine cl walk ‘Zichan used nine screens when walking.’ Here, we see that semantic change, from lexical as a noun in (9a) and a verb in (10a) to grammatical as a classifier in (9b)–(10b), does not occur independently at the semantic level (i.e. the function in Croft’s CG model 2001), but rather interacts with and relies on morpho-syntax (i.e. the form in Croft’s CG model). Without the ‘numeral þ noun’ construction/form

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

195

and its corresponding meaning/function, especially the numeral meaning, those nouns and verbs would not be interpreted as having a classifier meaning. Notice that such analysis provides a somewhat di¤erent picture from Langacker’s view (Langacker 1991) on CG, which states that form and function are paired as equal, and that meaning/function is not interpreted from syntax/form. However, this analysis is very much in line with the result of some other studies (cf. Bergs & Diewald 2008), which indicate that the development from lexical meaning to grammatical meaning rests on the construction and the relationship among the elements in the construction. The evidence found in this study clearly suggests that the meaning of the numeral classifier is derived from the ‘numeral þ NP/classifier’ construction and not the other way around. Furthermore, I would argue that the semantic change involved in the development of the classifier meaning in Chinese provides evidence for the interaction between CG and grammaticalization. Early studies (Peyraube 1999, Xing 2003) show that before Chinese verbs became grammaticalized into prepositions (e.g. baˇ 把 ‘patient marker’ and be`i 被 ‘agent marker’), adverbs (e.g. ha´i 还 ‘also’) and conjunctions (e.g. jiu` 就 ‘then’), they were all used in serial verb constructions, i.e., V (NP) V (NP). However, it is very di‰cult for researchers to verify the role of the serial verb construction in the process of the grammaticalization of those verbs, even though there is no lack of historical texts for investigation. The main reason for this di‰culty is that we do not know whether it is the serial verb construction or semantic change that triggered verbs to become grammaticalized. In other words, although we can observe some kind of relationship between the change of construction and meaning from the grammaticalization of verbs to function words, we are nonetheless unsure of exactly how they interact with each other. This uncertainty, however, disappears in the case of semantic change and the grammaticalization of classifiers investigated in this study. We know that the ‘numeral þ NP’ construction triggers the development of classifier meaning, a conclusion also supported by Bisang’s (2008: 586) study of Archaic Chinese, which shows that ‘‘a lexical item in a given position is coerced into a particular semantic interpretation associated with that position.’’ Although the term ‘grammaticalization’ was first introduced into linguistics in 1912 by Meillet, who considered the lexical item the primary input to grammmaticalization (i.e. lexical > grammatical), Western linguists did not extend the scope of grammaticalization sources to constructions (i.e. phrases or sentences) until Givo´n (e.g. 1979). Since then, many researchers

196

Janet Zhiqun Xing

(e.g. Hopper and Traugott 2003 [1993], Bybee et al. 1994, Lehmann 1992), as well as those working in CG (e.g. Kay and Fillmore 1999, Croft 2001), have considered both constructions and lexical items as the two main sources in the development of grammatical meaning. Hopper and Traugott (2003: 15) define grammaticalization as ‘‘the change whereby lexical items and constructions come in certain linguistic contexts to serve grammatical functions and, once grammaticalized, continue to develop new grammatical functions.’’ It is interesting to note that Hopper and Traugott altered their definition of grammaticalization from their earlier (in 1993) notion as ‘a process’ to ‘a change’ in 2003. In the development of Chinese classifiers, we have seen two types of change: 1) nominalization from verbs and 2) grammaticalization from nouns to classifiers. Both types appear to comply with Hopper and Traugott’s definition of grammaticalization. That is, in the first type of change, grammaticalization occurs when a verb (i.e. a lexical item) is used in the context of serving a nominal (i.e. grammatical) function and its meaning thereby changes from more specific to more general, as discussed in Section 3. This observation coincides with Givo´n’s (1990: 498–513) suggestion, who, after surveying di¤erent types of nominalization [e.g. (ir)realis nominalization and adverb/adjective nominalization] in Indo-European languages, points out that nominalization is ‘a grammatical process’, rather than a purely lexical one. As to the second type of change, it is a clear case of grammaticalization when a nominal phrase [num þ np] becomes a grammatical [num þ cl] construction. These two types of change only di¤er in terms of their input forms: one arises out of a lexical item and the other out of a construction. Furthermore, nominalization always occurs before the target construction becomes grammaticalized into a classifier construction.

6. Conclusion Through the investigation of semantic change and grammaticalization of sixteen classifiers in Chinese, this study has identified some general tendencies concerning the emergence, development, and disappearance of the classifier and classifier meaning. As for the emergence of the classifier and classifier meaning, it has been found that there is no direct relation between the source meaning and the classifier meaning. Any countable common noun or any common activity verb (handling daily activities) with a concrete specific semantic function can develop into a classifier once it has been

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

197

used in a construction with a numeral. In the development of classifier meaning, three mechanisms (metaphoricalization, metonymization, and semantic reanalysis), have been discussed, all of which are found to be indispensable in the grammaticalization of nouns and verbs to classifiers. In addition, metaphoricalization and metonymization may occur at any stage. The results of those semantic changes indicate that metaphoricalization and metonymization lead to abstraction and generalization, while metonymization and semantic reanalysis tend to lead to a more substantial or dramatic change, one that transfers the semantic function of a lexeme to a categorically di¤erent semantic domain. Concerning the disappearance of the classifier, two factors have been found to be responsible: high frequency in discourse and loss of semantic function. The complete process of emergence, development, and disappearance of a classifier may be diagrammed as shown in (11): (11) nouns/verbs (concrete meaning) > nouns/verbs (abstract meaning) > classifier (distinctive meaning) > classifier (loss of distinctive meaning) > loss of classifier One may consider the diagram in (11) as reflecting a process of change in both form (outside the parentheses) and meaning (inside the parentheses). Notice that the omission of the classifier at the end of the schema given in (11) leaves the conventional ‘numeral þ classifier þ noun’ construction as a fused ‘numeral þ noun’ unit. This is the end of a classifier’s life span, but the fused construction, which is identical to the input form or source construction of classifiers, provides a new opportunity for the noun after the numeral to emerge as another classifier, especially if the noun meets all the requirements (i.e. countable, conveying meaning related to human’s daily activities or characteristics of common objects) necessary for becoming a classifier. In other words, the output of grammaticalization of a classifier may be recycled as input of grammaticalization of an emerging classifier as diagrammed in (12): (12) Num þ (verb>) noun1 > num þ cl1 þ noun2 > num þ noun2 > num þ cl2 þ noun3 The cline in (12) indicates that noun1 can be changed into cl1 and noun2 into cl2. Based on this pattern and that discussed earlier, we may conclude that the process of semantic change and grammaticalization of classifiers in Chinese provides a sound example for the interface between construction unit/form and grammaticalization.

198

Janet Zhiqun Xing

Appendix: Illustrative examples in the development of 16 classifiers in Chinese 1.

baˇ 把 a. 周公旦把大鉞，畢公把小鉞，(1st C., Shiji) Zho#ugo#ngda`n baˇ da` yue`, Bı`go#ng baˇ xiaˇo yue` Zhougongdan hold big weapon Bigong hold small weapon ‘Zhougongdan held a big weapon. Bigong held a small weapon.’ b. 於此水底得一把毛。(4th C., Baiyujing) yu´ cıˇ shuıˇ dıˇ de´ yi baˇ ma´o at this water bottom get one handful/mw hair ‘(Someone) got a handful of hair from the bottom of this water.’ c. 只見兩扇門關著，一把鎖鎖著 (10th C., Bei Song Huaben Xuanji) zhıˇ jia`n liaˇng sha`n me´n gua#n zhe, yi baˇ suoˇ suoˇ zhe only see two cl door close asp, one cl lock lock asp ‘(They) only saw the two-leaved door was closed and there was a lock on it.’

2.

beˇn 本 a. 枝葉未有害，本實先撥。(Approx. 11th–6th C. BC, Shijing) zhıye #` we`i yoˇu ha`i, beˇn shı´ xia#n bo´ branch-leaf no have hurt, root indeed first break ‘The root, in fact, shatters first before the branches and leaves are damaged.’ b. 君子務本。(approx. 3rd C. BC, Lunyu) ju#nzıˇ wu` beˇn gentleman work-on fundamental-issue ‘Gentlemen work on important issues.’ c. 樂者，. . . 其本在人心之感於物也。 (2nd C. BC, Liyueji) yue` zheˇ, . . . qı´ beˇn za`i re´n xın# zhı # gaˇn yu´ music person its origin reside person heart poss feel about wu` yeˇ things part ‘A musician’s profession (develops from sound). The essence (of this profession) is people’s feelings toward things.’

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

199

d. . . . 其大本臃腫，而不中繩墨。(3rd C. BC, Zhuangzi) qı´ da` beˇn yo#ngzhoˇng, e´r bu´ zho`ng she´nmo` its big body pu¤-up, so not middle measure-tool ‘Its big body is pu¤ed-up, so it’s not straight when measured by a rope ruler.’ e. 河北此書家藏一本。 (4th C., Rules of Yan’s Family) he´ beˇi cıˇ shu# jia# ca´ng yi beˇn river north this book home hide one cl ‘One copy of this book is hidden at home to the north of the river.’ f. 古人據個甚摩事去卻四十二本經論 (10th C., Zutang Ji) guˇ re´n ju` ge` she´nme shı` qu` que` ancient people based cl what thing go purge sı`shı´ e`r beˇn jınglu # `n forty-two cl scripture-study ‘Why did ancient people purge forty-two volumes of scripture?’ 3.

bu` 部 a. 皆有分部也 (3rd C., Weiliaozi) jie´ yoˇu fe`n bu` yeˇ all have divide group part ‘All can be divided into groups.’ b. 王大為吏部郎， (4th C., Shishuo Xinyu) Wa´ngda` we´i lı` bu` la´ng Wangda serve-as personnel-a¤air unit/ministry o‰cial ‘Wang Da served as the minister of the Personnel A¤airs Department.’ c. 八部龍神千萬眾， (7th C., Bianwen) ba# bu` lo´ng she´n qia#n wa`n zho`ng eight cl dragon spirit thousand ten-thousand mass ‘Eight types of dragons and thousands of people.’ d. 皆是十二部尊經，(7th C., Bianwen) jie´ shı` shı´’e`r bu` zo#ng jıng # together be twelve cl worshipful scripture ‘Altogether there are twelve volumes of worshipful scripture.’

200

Janet Zhiqun Xing

e. 一部論語何嘗只說知仁！(10th C., Zhuzi Yulei) yı # bu` Lu´nyuˇ he´-cha´ng zhıˇ shuo# zhı # re´n one cl Confucius-Analects how-come only say know humanity ‘How come that one (collectable) book – The Analects of Confucius – only discusses humanity?’ 4.

fen# 分/ fe`n 份 a. 方以類聚，物以群分 (approx. 7th C. BC, Zhouyi) fa#ng yıˇ lei# ju`, wu` yıˇ qu´n fe#n shape according-to type gather, thing according-to type divide ‘Shapes are gathered according to their type; things are divided based on their type.’ b. 五分而得其一。(5th C. BC, Mozi) wuˇ fen# e´r de´ qı´ yı # five cl but receive its one ‘(One) only received one share out of five.’ c. 然則國財之一分在賈人。(3rd C. BC, Guanzi) ra´n ze´ guo´ca´i zhı # yi fen# za`i Jiaˇ re´n then however national-treasure poss one cl belong-to Jia clan ‘However, the Jia family has a share of the national treasure.’ d. 彼有牢信不狐疑, 集此諸法為一分。(2nd C., Donghan Fojing) bıˇ yoˇu la´o xı`n bu` hu´yı´, 3pl have solid belief not suspect jı´ cıˇ zhu# faˇ we´i yi fe#n collect this various law as one cl ‘They truly believe and do not suspect anything, so they have gathered all these laws (and made it) as one document.’ e. 劫來銀子，你拿二份，我受一份。(14th C., Baogong An) jie´ la´i yı´nzi, nıˇ na´ liaˇng fe`n, woˇ sho`u yi fe`n. rob come silver, 2sg take two cl, 1sg get one cl ‘Of the robbed silver, you take two shares, I take one.’

5.

fe#ng 封 a. 制其畿疆而沟封之 (approx. 7th C. BC, Zhouli) zhı` qı´ jı # jia#ng e´r go#u fe#ng zhı # legalize poss thousand-mile land and ditch mark-as-border it ‘(Emperor) legalized thousand miles of land and dug a ditch to mark it as a border.’

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

201

b. 諸公之地，封疆方五百里，(approx. 7th C. BC, Zhouli) zhu#go#ng zhı # dı`, fen# g jia#ng fa#ng wuˇ baˇi lıˇ every-duke poss land, bestow border extend five hundred miles ‘The land bestowed on every duke extended five hundred miles.’ c. 丘也貧，無蓋，于其封也，亦予之席 (approx. 7th C. BC, Zhali) qiu# yeˇ pı´n wu´ ga`i, yu´ qı´ fe#ng yeˇ, tomb also barren no cover, go it cover part, yı` yu` zhı # xı # also give its straw-mat ‘The tomb was barren without a cover. Go cover it with earth and mat made of straw.’ d. 籍吏民，封府庫，而待將軍。(1st C., Shiji) jı´ lı` mı´n, fe#ng fuˇ ku`, list o‰cial mass, seal government repository, e´r da`i jia#ngju#n then wait-for commander ‘Enlist ordinary people, seal government repositories, and then wait for the commander-in-chief.’ e. 秦將詐稱二世使人遣李良書，不封 (1st C., Shiji) Qı´n jia`ng zha` che#ng e`rshı` shıˇ re´n qiaˇn Qin chief cheat claim 2nd-emperor ask man dispatch Lıˇlia´ng shu#, bu` fe#ng Liliang letter not seal ‘The Qin’s chief commander pretended to be the Emperor’s son to send Liliang a letter that was not sealed.’ f. 吾聞子三封而三不成 (2nd C. BC, Kongzi Jiaoyu) wuˇ we´n zıˇ sa#n fe#ng e´r sa#n bu` che´ng 1sg hear sir three seal/cl but three NEG succeed ‘I heard you tried to seal it three times but failed three times.’ g. 子淵附書一封 (3rd C., Luoyang Jialanji) Zıˇyua#n fu` shu# yı # fe#ng Ziyuan attach letter one cl ‘Ziyuan sent a letter (along with it).’

202 6.

Janet Zhiqun Xing

ge` 箇/個/个 a. 楊僅入竹三萬箇。(1st C., Shiji) Ya´ngjıˇn ru` zhu´ sa#n-wa`n ge` Yangjin pay bamboo thirty-thousand cl ‘Yang Jin bought thirty thousand bamboo stalks.’ b. 縷綦以為奉，鹿皮四個 (1st C., Guoyu) lu¨ˇ qı´ yıˇwe´i fe`ng, lu`pı´ sı` ge` thread/clothing best serve-as o¤ering deer-skin four cl ‘The o¤erings include the best thread/clothing and four deer skins.’ c. 一個嫡女，執箕以晐姓于王宮 (1st C., from Zhang 2008) yi ge` dı´ nuˇ¨ , one cl close-relative girl zhı´ jı # yıˇ ge#i xı`ng yu´ wa´nggo#ng carry groom to prepare surname to palace ‘A relative’s girl carries a groom to prepare a marriage for the emperor.’ d. 誰論洛水，一個河神? (3rd C., from Liu 1965) she´i lu`n luo` shuıˇ, yi ge` he´ she#n who discuss Luo water, one cl river god ‘Who discusses the Luo water, a river god? e. 爾一生只作這個見解，(7th C., Foyulu) eˇr yishe#ng zhıˇ zuo` zhe` ge` jia`njieˇ 2sg whole-life only use-as det cl explanation ‘You’ve used this explanation your whole life.’

7.

jia`n 件 a. 凡為傢俱者，前件木，皆所宜種。(6th C., Qiminyaoshu) fa´n we`i jia# ju` zheˇ, qia´n jia`n mu`, anything for house tool part head divide wood, jie´ suoˇ yı´ zho`ng all pass convenient plant ‘Any household tools, if its head piece is wood, would be useful for planting.’

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

203

b. 具件階級數，令本曹尚書以朱印印之。(6th C., Weishu) ju` jia`n jie#j´ı shu`, all piece classify number lı`ng beˇn Ca´o sha`ngshu# yıˇ zhu# yı`n yı`n zhı # order this Cao high-o‰cial use red seal seal it ‘All pieces were categorized and sealed with a red seal by Cao the High O‰cial.’ c. 又求晉、趙起居注諸雜書數十件 (5th C., Songshu) yo`u qiu´ Jı`n again plead Jin Zha`o qıˇju# zhu` zhu# za´-shu# shu` shı´ jia`n Zhao o‰cial annotate various mix-book number ten cl ‘. . . pleaded with the Jin and Zhao o‰cials again for their annotations of over ten various books.’ d. 一狀所犯十人以上，所斷罪二十件以上，為大。 (7th C., Jiu Tangshu) yı # zhua`ng suoˇ fa`n shı´ re´n yıˇsha`ng, one accusation pass commit-crime ten person over suoˇ dua`n zuı` e`rshı´ jia`n yıˇsha`ng, we´i da` pass sentence crime twenty cl over consider big ‘One accusation about more than ten people who committed over twenty crimes is considered a big case.’ e. 喜有四件，優有四般 (7th C., Bianwen) xıˇ yoˇu sı` jia`n yo#u yoˇu sı` ba#n happiness have four cl excellence have four kind ‘There are four happy events and four excellent things.’ f. 一件袈裟掛在身 (7th C., Bianwen) yı # jia`n jia#sha# gua` za`i she#ng one cl robe wear on body ‘A robe is worn on his body.’ 8.

ju` 句 a. 經下篇旁行句讀。(5th C. BC, Mozi) jıng # xia`pia#n pa`ng ha`ng ju` do`u classic next-essay adjacent line sentence period ‘The adjacent line to the second essay of the classics is a full stop.’

204

Janet Zhiqun Xing

b. 四句，則唐時寫本已然矣，(3rd C. BC, Guanzi) sı` ju`, ze´ Ta´ng shı´ xieˇ beˇn yıˇre´n yıˇ four cl then Tang time write copy/book already-use part ‘The four-sentence (style) was already in use during the Tang period. (note: this refers to the Emperor Yao’s time, approx. 20th Century BC)’ 9.

pia#n 篇 a. 二篇之策，. . .，當萬物之數也。(approx. 7th C. BC, Zhouyi) e`r pia#n zhı # ce`, . . . , two writing-on-bamboo-stalk/cl poss book . . . , da#ng wa`n wu` zhı # shu` yeˇ as ten-thousand things poss number part ‘The book of the two essays. . . , serves as the number for ten thousand things.’ b. 昔者周公旦朝讀書百篇。(5th C. BC, Mozi) xı # zheˇ Zho#ugo#ngda`n zha#o du´ shu# baˇi pia#n past person name day read writing hundred cl ‘In the past, the person Zhougongdan read one hundred essays every day.’

10.

tia´o 條/条 a. 終南何有？有條有梅。 (approx. 7th C. BC, Shijing) Zho#ngna´n he´ yoˇu? yoˇu tia´o yoˇu me´i place-name what have? have tree have flower ‘What does Zhongnan have? (It) has trees and flowers.’ b. 群木蕃滋數大，條直以長。(3rd C. BC, Guanzi) qu´n mu` fa´nzı # shu`da`, tia´o zhı´ yıˇ cha´ng group plant grow big, branch straight and long ‘A lot of plants grew to a tremendous size, their branches straight and long.’ c. 有蕊一條，長於花葉 (3rd C., Liu 1965: 101) yoˇu ruıˇ yi tia´o, cha´ng yu´ hua# ye` have stamen one cl, long compare-to flower leave ‘(The flower) has a stamen longer than the flower’s leaves.’

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

205

d. 條屬者取出一條繩 (3rd C., Liu 1965: 101) tia´o shuˇzheˇ quˇ chu# yi tia´o she´ng long-rope owner fetch out one cl rope ‘The rope owner took out a rope.’ e. 萬邑王畿曠，三條綺陌平。(4th C., Liu 1965: 102) wa`n yı` wa´ng jı # kua`ng, ten-thousands mile emperor land far-wide sa#n tia´o qıˇmo` pı´ng three cl wide-road flat ‘The emperor’s land is thousands of miles long with three roads, wide and flat.’ 11.

we`i 位 a. 君子以思不出其位。 (approx. 7th C. BC, Zhouyi) ju#nzıˇ yıˇ sı # bu` chu# qı´ we`i gentlemen have idea not surpass its position ‘Gentlemen have thoughts but cannot surpass their social positions.’ b. 天子一位，公一位，侯一位，. . . 凡五等 (2nd C. BC, Mengzi) tia#nzıˇ yı # we`i, go#ng yı # we`i, ho´u yı # we`i, . . . emperor one cl, duke one cl, marquis one cl, . . . fa´n wuˇ deˇng together five level ‘There is one emperor, one duke, one marquis. . .all together five levels of positions.’ c. 忌日祭，只祭一位。 (1st C., Zhuzi) jı` rı` jı`, zhıˇ jı` yı # we`i memorial day pay-tribute only pay-tribute one cl ‘On the memorial day, only pay tribute to one person.’ d. 祭酒者，一位之元長也。 (2nd C., Yiwen Leiju) jı` jiuˇ zheˇ, yı # we`i zhı # yua´n-zhaˇng yeˇ pay-tribute liquor person one cl poss old-o‰cial part ‘The one who paid tribute with liquor was an old o‰cial.’

206 12.

Janet Zhiqun Xing

xia`ng 項/项 a. 行而俯項，非擊戾也；(1st C. BC, Xunzi) xı´ng e´r fuˇ xia`ng, fe#i jı # lı` yeˇ walk and lower neck/head, not hit hard part ‘(If someone) was walking with his head lowered, (this person) would not be hit seriously.’ b. 千頭萬項，然一透都透。(10th C., Zhuzi Yulei) qia#n to´u wa`n xia`ng, thousand head ten-thousand cl ra´n yı # to`u do#u to`u yet one penetrate all penetrate ‘It does not matter if it is one thousand heads or ten thousand; when one is penetrated, all are penetrated.’ c. 州府亦得此一項稅錢。(10th C., Zhuzi Yulei) zho#ufuˇ yıˇ de´ cıˇ yı # xia`ng shuı` qia´n prefecture also receive this one cl tax money ‘The Prefecture also received this tax money.’ d. 只一項事是一箇道理。(10th C., Zhuzi Yulei) zhıˇ yı # xia`ng shı` shı` yı # ge` da`olıˇ only one cl matter be one cl reason ‘Only one reason explains one thing.’

13.

zha#ng 張/张 a. 既張我弓，即挾我矢。(approx. 7th C. BC, Shijing) jı` zha#ng woˇ go#ng, jı` jia# woˇ shıˇ if open 1sg bow, then carry 1sg arrow ‘If (you) open my bow, you should carry my arrow.’ b. 琴瑟張而不均。(1st C. BC, Xunzi) qı´nse` zha#ng e´r bu` ju#n music-zither open/play but not even ‘(He) plays the zither unevenly.’ c. 勞不坐乘，暑不張蓋。(1st C., Shiji) la´o bu` zuo` che`ng, shuˇ bu` zha#ng ga`i work not sit carriage, hot not open cover ‘(One goes) to work without a carriage. (One does) not open a screen in hot weather.’

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

207

d. 子產以帷幕九張行 (1st C. BC, Zuo Zhuan) Zıˇchaˇn yıˇ we´imu` jiuˇ zha#ng xı´ng Zichan use screen nine cl walk ‘Zichan used nine screens (for protection from the sun or wind) when walking.’ e. 細切長縷，束為把，大如十張紙卷。(6th C., Qiminyaoshu) xı` qie # cha#ng lu¨ˇ shu` we´i baˇ, fine slice long thread, tie into bunch da` ru´ shı´ zha#ng zhıˇ jua`n big as ten cl paper roll ‘Slice them long and thin and tie them into a bunch as big as ten paper rolls.’ 14.

zhı # 支 a. 芄嵐之支，童子佩觿。(approx. 7th C. BC, Shijing) wa´nla´n zhı # zhı #, to´ngzıˇ pe`i xı # plant poss branch young-boy wear decorate ‘Young boys wore branches of the plant as decoration.’ b. 從腰以上五十一骨，四支百四骨。(2nd C., Donghan Fojing) co´ng ya#o yıˇsha`ng wuˇshı´ yı # guˇ, sı` zhı # baˇi sı` guˇ from waist above fifty one bone, four limb hundred four bone ‘There are fifty one bones above the waist. The four limbs have hundreds of bones.’ c. . . . 有江，支流三百六十。(7th C., Xin Tang Shu) yoˇu jia#ng, zhı # liu´ sa#n baˇi liu`shı´ have river, branch flow three hundred sixty ‘. . . there were rivers that had three hundred sixty branches.’ d. 有一個小女人逐水而來，中有七支蓮花 (7th C., Bianwen) yoˇu yı # ge` xiaˇo nuˇre´n zhuo´ shuıˇ e´r la´i, have one cl small female along-with water and come, zho#ng yoˇu qı # zhı # lia´nhua# middle have seven cl lotus-flower ‘A small lady came down with the flow of the water. There were seven lotus flowers in the water.’

208 15.

Janet Zhiqun Xing

zhı # 隻/只 a. 老而無子曰獨獨隻獨也言無所依也。 (2nd C., Shiming) laˇo e´r wu´ zıˇ yue# du´ du´ old yet no son consider single single zhı # du´ yeˇ ya´n wu´ suoˇ yı # yeˇ obey one of two single part word no pass part ‘An old man without a son is considered a single lonely person whose words are not heeded by anyone.’ b. 呂布令門侯于營門中舉一隻戟。(5th C., Sanguozhi) Luˇ¨ bu` lı`ng me´n ho´u yuˇ yı´ng me´n zho#ng Lu¨bu order close o‰cial arrive camp door middle juˇ yı # zhı # jıˇ hold one cl weapon ‘Lu Bu asked close o‰cials to come to the middle of the camp to hold up a weapon.’ c. 便买黄雌鸡十隻 (6th C., Qiminyaoshu) bia`n maˇi hua´ng cı´jı # shı´ zhı # then buy yellow hen ten cl ‘Then (he) bought ten yellow hens.’ d. 唯須小船一隻 (7th C., Bianwen) we´i xu# xiaˇo chua´n yı # zhı # only allow small boat one cl ‘Only a small boat is allowed.’ e. 憶師兄，哭太煞，失卻一隻眼 (10th C., Zutangji) yı` shı # xio#ng, ku# ta`i sha#, shı # que` yı # zhı # yaˇn miss master brother, cry too much, lose then one cl eye ‘(He) missed his master brother and cried so much he lost one of his eyes.’

16.

zuo` 座 a. 共坐有若孔子之座。(1st C., Lunheng) go`ng zuo` yoˇuruo` Koˇngzıˇ zhı # zuo` all sit as-if Kongzi poss seat ‘All sit as if this were Confucius’s seat.’

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

209

b. 通香虎皮檀象一座，(1st C., Zhaofeiyan Waizhuan) to#ng xia#ng huˇpı´ ta´n xia`ng yi zuo` generous fragrant tiger-skin light-red statue one cl ‘(It’s) a light red fragrant tiger-skin statue.’ c. 八座尚書，即古六卿之任也。(4th C., Wenxuan) ba# zuo` sha`ngshu#, eight cl o‰cial-position jı` guˇ liu` qıng # zhı # re`n yeˇ equivalent ancient six marquis poss post part ‘The eight o‰cial positions are equivalent to the six marquises of ancient times.’ d. 受業於九座山。(10th C., Zutangji) sho`uye` yu´ jiuˇ zuo` sha#n receive-profession at nine cl mountain ‘(Someone) works at the nine mountains.’

References Aikhenvald, Alexandra 2000 Classifiers: A Typology of Noun Categorization Devices. Oxford, UK: Oxford University Press. Bergs, Alexander and Gabriele Diewald (eds.) 2008 Construction and Language Change. Berlin/New York: Mouton de Gruyter. Bisang, Walter 1998 Grammaticalization and language contact, constructions, and positions. In The Limits of Grammaticalization, Ramat & Hopper (eds.), 13–58. Amsterdam/Philadelphia: Benjamins. Bisang, Walter 1999 Classifiers in East and Southeast Asian Languages. In Numeral Types and Changes Worldwide, Gvozdanovic J. (ed.), 112–185. Berlin/NewYork: Mouton de Gruyter. Bisang, Walter 2008 Precategoriality and syntax-based parts of speech – the case of Late Archaic Chinese. Studies in Language 32: 568–589. Bybee, Joan, Revere Perkins, and William Pagliuca 1994 The Evolution of Grammar: Tense, Aspect, and Modality in the Language of the World. Chicago: Chicago University Press.

210

Janet Zhiqun Xing

Croft, William 2001

Radical Construction Grammar: Syntactic Theory in Typological Perspective. Oxford: Oxford University Press. Croft, William and D. Alan Cruse 2004 Cognitive Linguistics. Cambridge: Cambridge University Press. Eckardt, Regine 2006 Meaning Change in Grammaticalization. New York: Oxford University Press. Erbaugh, Mary S. 1986 Taking stock: The development of Chinese noun classifiers historically and in young children. In Noun Classes and Categorization, Colette Craig (ed.), 399–436. Amsterdam/Philadelphia: Benjamins. Feng, Shengli 2005 Ha`nyuˇ Yu`nlu¨` Yuˇfaˇ Ya´njiu# [Study of Chinese Prosodic Structure.] Beijing: Beijing University Press. ¨ stman (eds.) Fried, Mirjam and Jan-Ola O 2004 Construction Grammar in a Cross-linguistic Perspective. Amsterdam/Philadelphia: Benjamins. Givo´n, Talmy 1979 On Understanding Grammar. New York: Academic Press. Givo´n, Talmy 1990 Syntax: A Functional-Typological Introduction, Vol. II. Amsterdam/Philadelphia: Benjamins. Goldberg, Adele E. 1995 Constructions: A Construction Grammar Approach to Argument Structure. Chicago: University of Chicago Press. Goldberg, Adele E. 2006 Construction at Work: The Nature of Generalization in Language. Oxford: Oxford University Press. Greenberg, Joseph 1990 Numeral classifiers and substantial number: Problems in the genesis of a linguistic type. In On Language: Selected Writings of Joseph H. Greenberg, Keith Denning and Suzanne Kemmer (eds.), 166–193. Stanford: Stanford University Press. Grice, Paul H. 1989 Studies in the Way of Words. Cambridge, MA: Harvard University Press. Harris, Alice C. and Lyle Campbell 1995 Historical Syntax in Cross-Linguistics Perspective. Cambridge: Cambridge University Press. Hopper, Paul 1991 On some principles of grammaticization. In Approaches to Grammaticalization, Elizabeth C. Traugott & Bernd Heine (eds.), vol. I, 17–36. Amsterdam/Philadelphia: Benjamins.

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

211

Hopper, Paul and Elizabeth C. Traugott 2003 Grammmaticalization. 2nd ed. Cambridge: Cambridge University Press. Huang, Daijun 1964 Co´ng jiaˇguˇwe´n, jı #nwe´n lia`ngcı´ de yı #ngyo`ng, kaˇocha´ ha`nyuˇ lia`ngcı´ de qıˇyua´n he´ fa #zhaˇn [Study of the origin and development of Chinese classifiers in oracle bone and bronze scripture.] Zhongguo Yuwen [Chinese Language] 6. Jiao, Fan 2001 A Chinese-English Dictionary of Measure Words [& Classifiers]. Beijing: Sinolingua. Joseph, Brian D. and Richard D. Janda (eds.) 2003 The Handbook of Historical Linguistics. Malden, MA and Oxford: Blackwell. Kay, Paul and Charles J. Fillmore 1999 Grammatical constructions and linguistic generalizations: The What’s X doing Y construction. Language 75, 1–33. Langacker, Ronald 1977 Syntactic reanalysis. In Mechanisms of Syntactic Change, Li Charles (ed.), 57–139. Austin: University of Texas Press. Langacker, Ronald 1991 Foundation of Cognitive Grammar. Stanford: Stanford University Press. Langacker, Ronald 2005 Construction grammar: Cognitive, radical, and less so. In Cognitive Linguistics: Internal Dynamics and Interdisciplinary Interaction, Francisco Jose´ Ruiz de Mendoza Iba´n˜ez, M. Sandra Pen˜a Cervel (eds.), 101–159. Berlin: Mouton de Gruyter. Lehmann, Christian 1992 Word order change by grammaticalization. In Internal and External Factors in Syntactic Change, Marinel Gerritsen & Dieter Stein (eds.), 395–416. Berlin/New York: Mouton de Gruyter. Levinson, Stephen C. 2000 Presumptive Meanings: The Theory of Generalized Conversational Implicature. Cambridge, MA: MIT Press. Lightfoot, David 1991 How to Set Parameters: Arguments from Language Change. Cambridge, MA: MIT Press. Liu, Shiru 1965 We`i Jı`n Na´nbeˇi Cha´o Lia`ngcı´ Ya´njiu# [Studies of Classifiers in Wei Jin Nanbei Dynasties.] Beijing: Zhonghua Shuju. Meillet, Antoine 1912 L’e´volution des formes grammaticales. Scientia 12, no. 26, 6.

212

Janet Zhiqun Xing

Ohta, Tatsuo 1987

Zho#ngguo´ Yuˇ Lı`shıˇ We´nfaˇ [Historical Grammar of the Chinese Language] translated by Jiang Shaoyu and Xu Changhua. Beijing: Beijing University Press.

Peyraube, Alain 1988 Syntactic change in Chinese: on grammaticalization. The Bulletin of the Institute of History and Philology, vol. LIX, 617–652. Taipei: Academia Sinica. Peyraube, Alain 1991 Some remarks on the history of Chinese classifiers. Santa Barbara Papers in Linguistics 3, 106–126. Peyraube, Alain 1998 Sha`ngguˇ, zho#ngguˇ ha`nyuˇ lia`ngcı´ de lı`shıˇ fa #zhaˇn [Historical development of classifiers in classical Chinese] Yuyanxue Luncong [Linguistic Inquiries], vol. 21. Beijing: Shangwu Yinshu Guan. Peyraube, Alain 1999 Historical change in Chinese grammar. Cahiers de Linguistique Asie Orientale, vol. 28 (2), 177–226. Tao, Liang 2006 Classifier loss and frozen tone in spoken Beijing Mandarin: The yi-ge` phono-syntactic conspiracy. Journal of Linguistics 44, 91– 133. Traugott, Elizabeth C. and Richard B. Dasher 2002 Regularity in Semantic Change. Cambridge: Cambridge University Press. Wang, Li 1980 Ha`nyuˇ Shıˇgaˇo [A Historical Study of Chinese Language]. Revised ed. Beijing: Zhonghua Shuju. Wang, Li 1989 Ha`nyuˇ Yu´faˇ Shıˇ [A History of Chinese Grammar]. Beijing: Shangwu Yinshu Guan. Wang, Shaoxin 1992 Ta´ngda`i shı #we´n xiaˇoshuo# zho#ng mı´nglia`ngcı´ de yu`nyo`ng [Classifiers in Tang dynasty novels]. In Suı´ Ta´ng Wuˇ Da`i Ha`nyuˇ Ya´njiu# [Chinese Studies – Sui Tang Five Dynasties], Xiangqing Chen (ed.), 327–386. Jinan: Shangdong Jiaoyu Chubanshe. Wu, Fuxiang 1999 Shı`lu`n xia`nda`i ha`nyuˇ do`ngbuˇ jie´go`u de la´iyua´n [On the development of verb-complement constructions in modern Chinese.] In Ha`nyuˇ Xia`nzhua`ng yu´ Lı`shıˇ de Yaˇnjiu# [Studies on the Current Status and Historical Development of Chinese Grammar], Jiang Lansheng & Hou Jinyi (eds.), 317–345. Beijing: Zhongguo Shehui Kexiu Chubanshe.

Semantic change in the grammaticalization of classifiers in Mandarin Chinese

213

Wu, Fuxiang, Shengli Feng, and James Huang 2006 Ha`nyuˇ ‘shu` þ lia`ng þ mı´ng’ ge´shı` de la´iyua´n [The development of ‘numeral þ classifier þ noun’ in Chinese]. Zhongguo Yuwen [Chinese Language] 5. Xing, Janet Zhiqun 2003 Grammaticalization of verbs in Mandarin Chinese. Journal of Chinese, vol. 31: 1, 101–144. Xing, Janet Zhiqun 2006 Mechanism of semantic change in Chinese. Studies in Language, Vol. 30: 3, 461–483. Xu, Shen 1981 2nd C. Shuo# We´n Jieˇ Zı` [Definition and Explanation of Characters and Words.] Shanghai: Guji Chubanshe. Zhang, Cheng 2008 Zho#ngguˇ ha`nyuˇ xı´ngzhua`ng lia`ngcı´ he´ to#ngyo`ng lia`ngcı´ de fa#zhaˇn [The development of size/shape and general classifiers in Middle Chinese.] Paper presented at the Symposium on Morpho-syntactic Change in Middle Chinese, March, Stanford University.

The repeater in Chinese and other languages Cheng Zhang

1. Introduction The ‘repeater’ is a type of nominal modifier which is formally identical to the noun that it modifies. The repeater can be found in many numeralclassifier languages in East Asia and Southeast Asia, including Dulong, Zaiwa, Achang, Jinuo, Susu, Thai, Burmese and so on. Historically, the repeater also existed in Early Archaic Chinese (1700 B.C.–1100 B.C.), although it disappeared in Middle Archaic Chinese (1100 B.C.–200 B.C.). Many scholars believe that the repeater is the earliest numeral classifier in Chinese (Wang 1958, Huang 1964, Peyraube 1998). Li (2000) proposes that all numeral-classifier languages in the Sino-Tibetan and Tai language groups involved evolution from a ‘‘repeater’’ stage to the ‘‘numeral classifier’’ stage. This paper provides evidence from the study of the historical development of Chinese numeral classifiers and other languages that shows that the numeral classifier in Chinese did not, in fact, originate from the repeater as suggested by earlier studies, but rather represents a class of its own. Many scholars have mentioned that a few sentences with ‘‘N1 Num N2’’ (N1 ¼ N2) can be found in Early Archaic Chinese documents (Wang 1958, Huang 1964, Peyraube 1998, Li 2000), but they fail to analyze the syntactic structure of the construction. The reason for their identifying the repeater as the earliest numeral classifier in Chinese is simply this: the repeater emerged earlier than other numeral classifiers and retained its full lexical meaning as a noun, whereas the classifier lost some of its lexical meaning as a noun. This paper discusses the view that the repeater is the earliest numeral classifier in Chinese by describing both the lexical meaning and the usage of the numeral classifiers that emerged in the initial formation stage of Chinese classifiers. Additionally, there are also many scholars who have studied the repeater in Tibeto-Burman languages and Tai languages (Liang 1983, Yang 1983, Xu 1990, 1993, 1994, Bisang 1999, Li 2000, Aikhenvald 2000, Ma 2003, Dai and Jiang 2005). This paper attempts to sort out the relevant data

216

Cheng Zhang

and describe the characteristics of the repeater in the languages in question. By comparing these languages with Chinese, this study suggests that the category of the numeral classifier emerges abruptly in all numeral classifier languages, and the repeater is one of the universal convenient ways to express this new grammatical category.

2. Numeral classifier in Archaic Chinese documents The ‘‘N1 þ Num þ N2’’-construction (N1 ¼ N2) appeared as early as the Oracle-Bone inscriptions and the Bronze inscriptions in Early Archaic Chinese. Following are a few examples: (1) 牛十牛。 (Early Archaic Chinese, Yınqı # `cuı`bia#n 190) Niu´ shı´ niu´ Cow ten cow (2) 羌百羌。 (Early Archaic Chinese, Yınqı # `cuı`bia#n 190) Qia#ng baˇi qia#ng Qiang (name of a tribe) hundred Qiang (3) 玉十玉 (Middle Archaic Chinese, Yıˇha`iguıˇmı´ng) Yu` shı´ yu` Jade ten jade Many scholars (Wang 1958, Huang 1964, Peyraube 1998, Li 2000 etc.) regard the second noun in ‘‘N1 þ Num þ N2’’ (N1 ¼ N2) as a repeater, and suggest that the repeater becomes a classifier when it is used to modify other nouns, such as 人 re´n ‘‘person/people’’. For example, 人 re´n ‘‘person/ people’’, found as a repeater in the Oracle-Bone script in (4), became a classifier when it was used to modify Qiang (the name of a tribe) in example (5). (4) 人十有六人。 (Early Archaic Chinese, He´jı´ 137) Re´n shı´ yo`u lı`u re´n. People sixteen people (5) 羌十人。 (Early Archaic Chinese, He´jı´ 26910) Qia#ng shı´ re´n. Qiang ten people However, except for 人 re´n, not many repeaters that appear in Early Archaic Chinese became classifiers. Even 人 re´n does not strictly qualify

The repeater in Chinese and other languages

217

as a classifier because it has not been used in the most typical and most frequent construction of numeral classifiers in the history of Chinese, namely ‘‘Num þ CL þ N’’. Scholars have di¤erent views on whether numeral classifiers existed in Middle Archaic Chinese (1100 B.C.–200 B.C.). Examples that might be considered classifiers in historical documents are rare and controversial. Sentences (6)–(8) are often cited by scholars as examples of numeral classifiers. Underscore indicates the elements considered numeral classifiers. (6) 莱人使正舆子赂夙沙卫以索马牛皆百匹。 (Middle Archaic Chinese, Zuoˇzhua`n Xia#nggo#ng e`r nia´n) La´i re´n shıˇ Zhe`ngyu´zıˇ lu` Lai people make Zhengyuzi bribe Su`sha#we`i yıˇ suoˇ maˇ nı´u jie# baˇi pıˇ. Sushawei to select horse cow all hundred CL:animal ‘The Lai people made Zhengyuzi bribe Sushawei to select one hundred horses and one hundred cows.’ (7) 子产以幄幕九张行。 (Middle Archaic Chinese, Zuoˇzhua`n Zha#ogo#ng shı´sa#n nia´n) Zıˇchaˇn yıˇ we´imu` jıˇu zha#ng xı´ng. Zichang with tents nine CL:stretchable-thing goes. ‘Zhichang went with nine tents.’ (8) 负服矢五十个。 (Middle Archaic Chinese, Xu´nzıˇ yı`bı #ng) Fu` fu`shıˇ wuˇshı´ ge`. Carry arrow fifty CL:general ‘Carry fifty arrows’ From the data used for this study, we know that the of ‘‘N1 þ Num þ N2’’-construction (N1 ¼ N2) began disappearing in Middle Archaic Chinese (1100 B.C.–200 B.C.). All examples in (6)–(8) have ‘‘N þ Num þ CL’’ instead. Peyraube (1998) points out that all examples cited by other scholars as using the numeral classifier can be analyzed as ‘‘N1 þ Num þ N2’’ (N1 A N2), and therefore it can be inferred that no real classifiers existed in Early Archaic Chinese or Middle Archaic Chinese. All scholars do agree, however, that classifiers existed in Late Archaic Chinese (200 B.C.–200 A.D.). According to investigations carried out by

218

Cheng Zhang

the present author,1 a total of fifty-seven classifiers were in common usage during this period. These are given in Table 1–4,2 which may however not be exhaustive lists of each type. In Late Archaic Chinese, not only did a large number of classifiers emerge, but also a system of classifiers formed according to the universal semantic parameters found in all numeral-classifier languages.3 Compared to Early and Middle Archaic Chinese, we find that the formation and development of classifiers was quite rapid in Late Archaic Chinese. 1. All the historical documents I have investigated are listed in the following data sources: Late Archaic Chinese (200 B.C.–200 A.D.) a. Historical works: 史记 Shıˇjı` ‘‘Record of History’’ 战国策 Zha`nguo´ce` ‘‘Policy of Warring States’’ b. Philosophical works: 淮南子 Hua´ina´nzıˇ ‘‘The Masters of Huainan’’, 论衡 Lu`nhe´n ‘‘The Balanced Inquiries’’ c. Buddhist documents: 修行本起经 Xı #uxı´ngbeˇn qıˇ jıng, # 长阿含十报法经 Zhaˇnga#ha´n shı´ ba`ofaˇ jıng, # 人本欲生经 re´n beˇn yu` she#ng jıng, # 一切流摄守因经 yı´qie` lı´u she` shoˇuyın# jıng, # 四谛经 Sı` dı` jı #ng, 中本起经 Zho#ngbeˇn qıˇ jıng, # 法镜经 Faˇjı`ng jıng, # 阿含口解十二因缘经 a#ha´n koˇu jieˇ shı´e`r yınyua # ´ n jıng # d. Bamboo and Silk Manuscripts: 尹湾汉墓简牍 Yıˇnwa#n ha`n mu` jiaˇnduˇ, 张家山汉墓竹简 Zha#ngjia# sha#n ha`n mu` zhu´jiaˇn Medieval Chinese (220 A.D.P618 A.D.) e. Agricultural work: 齐民要术 Qı´mı´nya`oshu` ‘‘Main techniques for the welfare of the people’’ f. Collection of poems: 玉台新咏 Yu`ta´ixıngyo # ˇ ng: ‘‘New Chants from a Jade Terrace’’ g. Collection of short novels: 世说新语 Shı`shuo#xı #nyuˇ ‘‘A New Account of the Tales of the World’’: h. Buddhist document: 贤愚经 Xia´n yu´jıng # 2. In Tables 1–4, I have also included some classifiers that were not detected in my data but have been mentioned by other scholars in their studies. 3. Typological investigation shows that animacy, function, and shape and size are universal semantic parameters used in numeral classifier systems, Aikhenvald (2000), Bisang (1999), Croft (1994) etc.

The repeater in Chinese and other languages

219

Table 1. General classifiers in Late Archaic Chinese Classifier

Usage

枚 me´i 个 ge` 物 wu` 只 zhı #

general classifier, used frequently general classifier, used frequently general classifier, seldom used general classifier, seldom used

Table 2. Classifiers for biological nature in Late Archaic Chinese Classifier

Usage

匹 pıˇ 驷 sı` 头 to´u 树 shu` 梃 tıˇng 本 beˇn 茎 jı`ng

for horses for horses for animals for trees for plants for plants for plants

Table 3. Classifiers for shape and size in Late Archaic Chinese Classifier

Usage

果 guoˇ 丸 wa´n 条 tia´o 节 jie´ 片 pia`n 合 he´ 具 ju` 口 koˇu

for small and round things for round things for long things for long things for slices for things that can be opened/closed for things that should be used in sets/series for things with mouths or things that can be opened

220

Cheng Zhang

Table 4. Numeral Classifiers for functions in Late Archaic Chinese Classifier

Usage

通 to#ng 牒 die´ 牍 du´ 卷 jua`n 章 zha#ng 言 ya´n 札 zha# 策 ce` 编 bia#n 家 jia# 篇 pia#n 两 liaˇng 领 lıˇng 领衣 lıˇngyı # 袭 xı´ 系 xı` 衣 yı # 衣衣 yıyı # # 称 che`n 裁 ca´i 件 jia`n 皮 pı´ 所 suoˇ 间 jia#n 区 qu# 炬 ju` 把 baˇ 两 liaˇng 乘 she`ng 发 fa# 封 fe #ng 骑 jı # 级 jı´ 曲 quˇ 张 zha#ng 艘 so#u 脔 lua´n 块 kua`i

for articles for documents for books for chapters for chapters for words for articles for books for books for school of philosophical or academic thought for articles for clothes for clothes for clothes for suits of clothes for clothes for clothes for clothes for clothes for fur for sheepskin for fur for buildings for rooms for buildings for fire for torches for carriages for carriages for ammunition for letters for cavalry for heads of captives for music and songs for bows for boats for pieces of meat for blocks of soil

The repeater in Chinese and other languages

221

Six types of constructions were used to express the quantity of things in Late Archaic Chinese: i. ‘‘Num þ N’’ (Number Noun); ii. ‘‘N þ Num’’ (Noun Number); iii. ‘‘Num þ CL þ N’’ (Number Classifier Noun); iv. ‘‘N þ Num þ CL’’ (Noun Number Classifier); v. ‘‘Num’’ (Number); and vi. ‘‘Num þ CL’’ (Number Classifier). Some examples are given below: (9) Type I: Num B N

一簪不得著身。 (1st Century BC, Shıˇjı` Nı`n xı`nglie`zhua`n) Yı # zha#n bu`de´ zhuo´ she#n. One hairpin cannot be in body ‘Not even one hairpin can be in/on his body.’

(10) Type I: Num B N

桓公三年，晋败我一将。 (1st Century BC, Shıˇjı` Qı´ngbeˇnjı` ) Hua´ngo#ng sa#n nia´n, Jı`n ba`i woˇ yı # jia`ng. Huangong three year, Jin defeat our one general ‘Jin defeated one of our generals in the third year of Huangong.’

(11) Type II: N B Num

智伯欲伐卫，遗卫君野马四百，白璧一。

(2nd Century BC, Zha`nguo´ce` So`ngwe`i) Zhı`bo´ yu` fa´ We`i, Zhibo want to attack Wei, we`i We`iju#n yeˇ maˇ sı` baˇi, ba´i bı` yı #. bestow Weijun wild horse four hundred, white jade one. ‘Zhibo was going to attack Wei, so he bestowed upon Weijun four hundred wild horses and one piece of white jade.’ (12) Type III: Num B CL B N

代三十六县，上党十七县，不用一领甲，不苦一民，皆秦之有也。

(2nd Century BC, Zha`nguo´ce` Qı´ngyı` ) Da`i sa#nshı´lı`u xia`n, Sha`ngdaˇng shı´qı # xia`n, Dai thirty-six county, Shangdang seventeen county, bu` yo`ng yı # Ling jiaˇ, bu` kuˇ yı # mı´n, not use one CL:clothes armour, not su¤er one people, jie# Qı´n zhı # yoˇu yeˇ. all Qin of belonging PART ‘Thirty-six counties of Dai and seventeen counties of Shangdang all became the land of the Qin, and not one soldier of Qin nor one person of Qin su¤ered from this.’

222

Cheng Zhang

(13) Type IV: N B Num B CL

故世子作《养书》一篇。(1st Century AD, Lu`nhe´ngBeˇnxı`ngpia#n) Gu` shı`zıˇ zuo` Yaˇngshu# yı # pia#n. So Shizi write Yangshu one Cl:article ‘So Shizi wrote a book called Yangshu.’

(14) Type V: Num

从比丘众六十二万。 (3rd Century AD, Xı´uxı´ngbeˇnqıˇjıng) #

Co´ng bıˇqiu# zho`ng liu`shı´ e`r wa`n. Follow Bhiksu people sixty two ten-thousand. ‘Bhiksus who followed Buddha were six hundred twenty thousand.’ (15) Type VI: Num B CL

连弩车五百六十四乘。

(1st Century BC, Yıˇnwa#nha`nmu`jiaˇndu´ Lı`u zhe`ng) Lia´nnuˇche# wuˇ baˇi liu`shı´-sı` she`ng. bow-carriage five hundred sixty-four CL:carriage. ‘There were five hundred sixty-four carriages with bows. Type II ‘‘N þ Num’’ and Type IV ‘‘N þ Num þ CL’’ should be analyzed as ‘‘subject þ predicate’’ as shown in (11) and (13). The syntactic functions of ‘‘Num’’ and ‘‘Num þ CL’’ in these two types of constructions are the same as ‘‘Num’’ and ‘‘Num þ CL’’ in Types V and VI as illustrated in (14) and (15). ‘‘Num’’ and ‘‘Num þ CL’’ in Type II and Type IV can be separated from the nouns they modify by inserting elements which are not part of the noun phrase. For example, in (16) below, the adverb 各 ge` ‘‘each’’, which is used as part of the predicate in Chinese, was inserted between the noun and the numeral and between the noun and the construction of ‘‘Number þ Classifier’’ in (17). (16) Type II: N B Num

其河、湫、汉水加玉各一。

(1st Century BC, Shıˇjı` Nı`n xı`nglie`zhua`n) Qı´ He´ Jıu# Ha`nshui jia# yu` ge` yı.# Those He Jiu Hanshui add jade each one ‘The Huanghe, Jiushui, Hanshui were added one piece of jade each as sacrificial o¤erings.’

The repeater in Chinese and other languages

223

(17) Type IV: N B Num B CL

河内女子发老屋，得逸《易》、 《礼》、 《尚书》各一篇，奏之。

(1st Century AD, Lu`nhe´ngZhe`ngshuo#pia#n) He´ne`i nuˇ¨ zıˇ fa# laˇo wu#, de´ yı` Yı`, Henei women dig old house, get lost Yi, Lıˇ, Sha`ngshu# ge` yı # pia#n, zo`u zhı.# Li, Shangshu each one CL:article, report it. ‘A woman in Henei dug in her old house, and found one chapter of each of the lost books Yi, Li, Shangshu and reported this.’ The six constructions can be divided into two groups, with one using classifiers and the other not. Table 5 shows the frequencies of classifier use in di¤erent written works in Late Archaic Chinese (200 B.C.–200 A.D.). The frequency of constructions using classifiers in late Archaic Chinese was very low while constructions without classifiers were dominant. This situation did not change until Medieval Chinese. Table 5 also shows that the classifiers were used more often in Bamboo and Silk Manuscripts (28%) than in other documents (less than 13%). Table 6 gives the statistics of the frequency of classifier use in Medieval Chinese (220 A.D.–618 A.D.). Table 6 shows that, except for Qı´mı´nya`oshu` (an agricultural work from the 6th Century), in all documents the frequency of constructions with a classifier is lower than that of constructions without a classifier. For Qı´mı´nya`oshu`, in addition to the fact that the frequency of constructions with classifiers exceeds that of the constructions without classifiers, the constructions with ‘‘N þ Num þ CL’’ and ‘‘Num þ CL’’ are more common than those with ‘‘N þ Num’’ and ‘‘Num’’. This indicates that the classifier prevailed in the ‘‘N þ Num’’ and ‘‘Num’’ construction before it spread to other constructions. Table 5. Frequency of Classifiers in Late Archaic Chinese w/ classifier

w/o classifier

Quantity

Percentage

Historical works

164

13%

996

86%

Philosophical works

144

12%

1098

88%

7 76

2% 28%

290 196

98% 72%

391

13%

2580

87%

Buddhist documents Bamboo/Silk Manuscripts Total

Quantity

Percentage

224

Cheng Zhang

Table 6. Frequency of Classifiers in Medieval Chinese Qı´mı´nya`oshu` (A.D. 6th)

Yu`ta´ixınyo # ˇ ng (A.D. 6th)

Shı`shuo#xınyu # ˇ (A.D. 5th)

Xia´n yu´jı #ng (A.D. 5th)

w/CL

w/o CL

w/CL

w/o CL

w/CL

w/o CL

w/CL

w/o CL

Num N

30

184

15

45

11

130

23

773

N Num

101

15

2

0

6

0

6

4

Num

198

15

13

4

11

8

25

23

Total

329

214

30

49

28

138

54

800

Type

Two facts are needed to support the view that ‘‘N þ Num þ CL’’ is actually derived from ‘‘N1 þ Num þ N2’’ (N1 ¼ N2), and the view that the numeral-classifier evolves from the repeater. Semantically then, if the earliest numeral classifiers were derived from the repeater, many classifiers should have been derived from the nouns that refer to matters as a whole in the initial formative stage of classifiers, e.g. nouns like ‘‘people\cow\jade’’ (see examples (1)–(2) and (4)). This is the case because only nouns of this sort can be used in the ‘‘N1 þ Num þ N2’’ (N1 ¼ N2)4 construction. Though this might seem a logical conclusion, in fact, in Late Archaic Chinese and Medieval Chinese, few classifiers were derived from these nouns. The three classifiers 匹 pıˇ, 张 zha#ng and 个 ge`, as shown in examples (6)–(8), are commonly considered the earliest instances in the numeral classifier construction. However, none of these three classifiers can be used as a repeater because of their synchronic meanings: at that time, 匹 pıˇ , used as classifier for horses, meant ‘‘match’’; 张 zha#ng, used as a classifier for things that can be stretched, meant ‘‘stretch’’, and 个 ge`, used to classify arrows, had the meaning ‘‘bamboo’’. In Late Archaic Chinese, the initial stage in the development of classifiers, only nine out of the fifty-seven classifiers were derived from nouns that refer to objects as a whole. 牒 die´ meant ‘‘a piece of wood or bamboo used for writing’’ and was used as a classifier for documents; 家 jia# means ‘‘family’’ but was then used as a classifier for schools of thought; 树 shu` means ‘‘tree’’ and hence was a classifier for trees; 衣 yı # means ‘‘clothes’’ and was a classifier for clothes; 艘 so#u meant ‘‘boat’’ and was a classifier for boats; 炬 ju` means ‘‘torch’’ and was a classifier for fire; 所 suoˇ means ‘‘place’’ 4. Most of this sort of noun corresponds to a basic level of categorization.

The repeater in Chinese and other languages

225

and is a classifier for places and buildings; 块 kua`i meant ‘‘a block of soil’’ and was a classifier for blocks of soil; 脔 lua´n meant ‘‘a piece of meat’’ and was a classifier for pieces of meat. Most classifiers were not derived from nouns that refer to matters as a whole. Examples include: 茎 jı`ng meaning ‘‘stem’’ and used as a classifier for plants; 领 lıˇng referring to ‘‘collar’’ and used as a classifier for clothes; 口 koˇu meaning ‘‘mouth’’ and used as a classifier for things with mouths; 条 tia´o meaning ‘‘branches’’ and used as a classifier for long things. Some classifiers were derived from verbs. Relevant examples are: 乘 che´ng derived from ‘‘take a carriage’’5 was a classifier for carriages pronounced as she`ng; 篇 pia#n meant ‘‘to join pieces of wood slips or bamboo slips on which characters have been written’’ and is a classifier for articles; 合 he´ means ‘‘close’’ and was once a classifier for things that could be opened or closed. The words 乘 che´ng and 篇 pia#n are found more than one hundred times in our Late Archaic Chinese data. The frequency of use of these two classifiers is the highest among all classifiers used during the Han Dynasty. It seems that the nine classifiers that refer to objects as a whole were neither the earliest nor the most frequently used in the initial formative stage of the classifying system . Thus, we argue that these nine classifiers cannot be regarded as a proof that repeaters emerged earlier than numeral classifiers in Chinese. It stands to reason that if, indeed, ‘‘N þ Num þ CL’’ was derived from ‘‘N1 þ Num þ N2’’ (N1 ¼ N2), classifiers should have emerged and been used in nominal postposition first. We have stated earlier that ‘‘N þ Num þ CL’’ should be analyzed as ‘‘Subject (N) þ Predicate (Num CL)’’. When referring to the quality of things, a construction should consist of a number and a classifier, rather than the combination of a noun, a number and a classifier. Similarly, since the original construction of ‘‘N þ Num þ CL’’ is ‘‘N þ Num’’, we would consider ‘‘N þ Num’’ as ‘‘Subject (N) þ Predicate (Num)’’, in which ‘‘Num’’ is used to express the quality of things. Additionally, as we have seen in Table 5 and Table 6, classifiers prevail in the ‘‘N þ Num’’ and ‘‘Num’’ construction before they spread to other constructions, so it is likely that the numeral classifier first emerged in the position of the predicate in Chinese. The ‘‘N þ Num’’-construction in Archaic Chinese or ‘‘N þ Num þ CL’’ in Medieval Chinese was often used when people counted things. This could be the reason why we can also find evidence that constructions with classifiers are used more frequently in the Bamboo and Silk Manuscripts and Qı´mı´nya`oshu` than in 5. 乘 che´ng can be used as a verb or a noun. Its verbal usage was the earliest from which developed the nominal usage and then the classifier usage.

226

Cheng Zhang

any other documents (see Table 5 and Table 6). Much of the content of the two documents (the Bamboo and Silk Manuscripts and Qı´mı´nya`oshu`), is a record of the quantity of things. In this section, we have discussed two issues: one is that most numeral classifiers have no relation to repeaters with regard to the origin of their lexical meaning, and the other is that ‘‘N þ Num þ CL’’ is derived from ‘‘N þ Num’’, not from ‘‘N1 þ Num þ N2’’ (N1 ¼ N2). Both ‘‘N þ Num þ CL’’ and ‘‘N þ Num’’ are constructions specifically used for counting. There is no evidence suggesting a direct relationship between repeaters and classifiers.

3. Repeaters in other languages6 The repeater is found in many modern Sino-Tibetan languages and Tai languages including Dulong, Zaiwa, Achang, Jinuo, Susu, Lahu, Hani, Naxi, Nu, Dai, Bai, Lingao, Burmese, Thai, and Laotian. Except for the last three, these languages can be found in the south of China, and they all have two types of nominal modifiers: numeral classifiers and repeaters as shown in (18)–(21). Tables 7 and 8 list the languages mentioned above which use both repeaters and classifiers. For the purposes of comparison, related languages which do not use repeaters are also listed. Dulong language: (18) lam31ɹɔʔ55a31 sɯm53jɔ ʔ55 friend three CL ‘three friends’ (19) c¸in55 tiʔ55 c¸in55 grass one grass ‘one blade of grass’ Dai language: (20) paı11 sa:m1 to1 fish three CL ‘three fish’

pɯ31tc¸iʔ55a31 sɯm53 guÐ55 bird three CL ‘three birds’

c¸ɯÐ55c¸i55 tiʔ55 c¸i55 fruit one fruit ‘one fruit’ sə2 sɔÐ1 phɯn1 clothes two CL ‘two clothes’

6. Many scholars have described repeaters in Sino-Tibetan languages and Tai languages; this section’s study is based on their data.

The repeater in Chinese and other languages

(21) som3 som3 nɯÐ room room one ‘one room’

227

kɔÐ3 kɔÐ3 nɯÐ6 bottle bottle one ‘one bottle’

Tables 7 and 8 reveal two facts: First, if a repeater exists in a language, so does a numeral classifier. There is no single language that has only repeaters and no numeral classifiers. Second, all languages which use repeaters (except for Lingao hua and Longzhou hua) use the ‘‘N þ Num þ CL’’ construction. The first fact has been confirmed by cross-linguistic investigations in di¤erent countries (see Aikhenvald 2000). It should be noted here that while some non-classifier languages use classifiers occasionally,

Table 7 7. Repeaters in Sino-Tibetan languages Group

Language

N1 Num N2

CL

Num CL N

N Num CL

N Num

Qiang

Jiarong









þ

Qiang

Muya

þ

þ



þ



Qiang

Qiang

þ

þ



þ



Jingpo Jingpo

Jingpo Donglong

 þ

 þ

 

 þ

þ 

Burmese

Burmese

þ

þ



þ



Burmese

Zaiwa

þ

þ



þ



Burmese

Achang

þ

þ



þ



Yi

Naxi

þ

þ



þ



Yi

Susu

þ

þ



þ



Yi

Yi



þ



þ



Yi Yi

Hani Jinuo

þ þ

þ þ

 

þ þ

 

Yi

Lahu

þ

þ



þ



Bai

Bai

þ

þ



þ



Nu

Nu

þ

þ



þ



7. ‘‘’’ means there is no targeted phenomenon in the language, and ‘‘þ’’ suggests the opposite condition.

228

Cheng Zhang

Table 8. Repeaters in the Kradai languages and Tai languages N1 Num N2

CL

Num CL N

N Num CL

N Num

Longzhouhua

þ

þ

þ

þ



Dai

þ

þ



þ



Thai

þ

þ



þ



Laotian Linggao hua

þ þ

þ þ

 þ

þ þ

 

such languages always use numeral classifiers, but not repeaters. For example, though neither Jingpo nor Jiarong is a numeral-classifier language, there are a few cases of the use of a numeral classifier as illustrated (underlined) in (22)–(25). Jingpo language: (22) ma˘31naÐ33 ma˘31 sum33 friend three ‘three friends’ (23) u31 khum31 mji33 hen CL one ‘one hen’ (khum31 means ‘body’ as a noun)

u31 ma˘31sum33 bird three ‘three birds’

mai31sau31 lap31 mji33 paper CL one ‘a piece of paper’ (lap31 means ‘leaf ’ as a noun)

Jiarong language: (24) tərmi kəsam people three ‘three people’ (25) tərmi tərgi people one CL ‘one person’ (tərgi is a phrase borrowed from the Chinese phrase 一个 yı´ge` ‘one CL:general classifier’) The reason for the occurrence of the numeral classifier in Jingpo and Jiarong is that they are a¤ected by their proximity to other related neighboring languages, such as Dulong, Muya, and Qiang. If such influence

The repeater in Chinese and other languages

229

and language contact continues and/or grows, it is possible that Jingpo and Jiarong will themselves eventually evolve into authentic numeralclassifier languages. As mentioned above, most of the classifiers in Chinese are not derived from nouns that refer to matters as a whole and they cannot be used in the ‘‘N1 þ Num þ N2’’ (N1 ¼ N2) construction. This is true in other SinoTibetan languages as well. Yang (2005) points out that the original nouns of almost all classifiers in the Dulong language can be traced to their early usages as making reference to parts or characteristics of objects/things. Here are some examples from his article. Noun vng31geu55 (N: body) vng31keui55 (N: root) vk31klong53 (N: root) vk31pong53 (N: sphere) vng31puq55 (N: wide) vng31gyoq55 (N: human being) vk31klvng53 (N: drop) vng31lung55 (N: cube)

Numeral classifiers geu55 (CL: long things, horses) keui55 (Cl: long things) klong53 (CL: long things) pong53 (CL: round things) puq55 (CL: flat things) gyoq55 (CL: people) klvng53 (CL: drops of liquid) lung55 (CL: round things, pieces, or large solid things)

None of these words, however, can be used in the ‘‘N1 þ Num þ N2’’ (N1 ¼ N2) construction. Therefore, we must conclude that neither in Chinese nor in Dulong is the numeral classifier derived from the repeater. Tables 7 and 8 show that all languages which have repeaters use the construction ‘‘N þ Num þ CL’’. They also show a number of other traits including the following: The dominant word order in Tibeto-Burman languages is ‘‘N þ Num’’ or ‘‘N þ Num þ CL’’. In the Jingpo language, the order is ‘‘N þ Num’’ (without a classifier), and in the Dulong language (which belongs to the same language family as Jingpo but has abundant numeral classifiers), the dominant word order is ‘‘N þ Num þ CL’’. In the Qiang language family, only the Jiarong language does not use the numeral classifier but uses the construction ‘‘N þ Num’’ to express a quantity of things instead. Finally, it can be seen that numeral classifier languages in the two language families use repeaters and ‘‘N1 þ Num þ N2’’ (N1 ¼ N2) just like other numeral classifier languages. Based on these facts, we may conclude that ‘‘N þ Num’’ is the original structure, from which the ‘‘N þ Num þ CL’’ and ‘‘N1 þ Num þ N2’’ (N1 ¼ N2) constructions were derived. Dai Qingxia and Jiang Ying (2005) investigated the circumstances

230

Cheng Zhang

under which repeaters emerged in some Tibeto-Burman languages and claimed that ‘‘N1 þ Num þ N2’’ (N1 ¼ N2) and ‘‘N þ Num þ CL’’ related to the emphasis of quantity. Similarly, as this paper stated earlier, the same conclusion can be reached with respect to the development of classifiers in Archaic and Medieval Chinese. Delancey (1986) reconstructs the Proto-Tai classifier system. According to Delancey (1986), the earliest classifiers in the Tai languages were not derived from repeaters. Delancey uses the term ‘‘class term’’ to refer to nouns that can be used as repeaters or semi-repeaters, and goes on to discuss the relationship between class term and the numeral classifier. He points out that some classifiers never function as either nouns or class terms, although class terms are a source of new classifiers in Tai languages. Moreover, he suggests that the history of those classifiers that are most frequently used and most widespread in the Tai languages is longer than that of the classifiers derived from class terms. Delancey then infers what the earliest five classifiers in the Proto-Tai language might have been based on these. The result of this study supports Delancey’s view.

4. Conclusion and discussion This paper has discussed the relationship between the repeater and the numeral classifier through the analysis of the diachronic evolution of numeral classifiers in Chinese and of the semantic and syntactic features of numeral classifiers in other numeral classifier languages. From a semantic perspective, both diachronic and synchronic evidence shows that most numeral classifiers were not derived from repeaters. The system of numeral classifiers in Chinese was established in Late Archaic Chinese. At the earliest stage of development in Archaic Chinese, there were few classifiers that may have been derived from repeaters. However, the meaning of most numeral classifiers as nouns makes it unlikely that they were used in the construction ‘‘N1 Num N2’’ (N1 ¼ N2) at the time. This observation is supported by evidence taken from the study of a number of contemporary Tibeto-Burman and Tai languages, which have both numeral classifiers and repeaters. From a syntactic point of view, all evidence shows that both the ‘‘N þ Num þ CL’’ construction and the ‘‘N1 þ Num þ N2’’ (N1 ¼ N2) construction are derived from constructions of the form ‘‘N þ Num’’. As indicated in Section 2, ‘‘Num’’ in ‘‘N þ Num’’ and ‘‘Num þ CL’’ in ‘‘N þ Num þ

The repeater in Chinese and other languages

231

CL’’ are both predicates and not nominal modifiers and are used in lists to emphasize a quantity of things. Consequently, the numeral classifier emerged in the predicate position in Chinese. Cross-linguistic investigations of contemporary Tibeto-Burman and Tai languages reveal that there are no languages which have only the repeater without also having the classifier. Furthermore, there are a few languages (such as Jingpo and Jiarong) in which numeral classifiers have just begun to emerge. In those languages there exist only a few numeral classifiers, but as yet no repeaters. Although ‘‘N1 þ Num þ N2’’ (N1 ¼ N2) and ‘‘N þ Num þ CL’’ coexist in many Tibeto-Burman and Tai languages, this observation is not su‰cient to confirm that the repeater emerged earlier than the numeral classifier in these contemporary languages. Early Archaic Chinese (1700 B.C.–1100 B.C) may be the only indication that there was a stage in which only ‘‘N1 þ Num þ N2’’ (N1 ¼ N2) existed. However, this construction disappeared in Middle Archaic Chinese (1100 B.C.–200 B.C.) and in Late Archaic Chinese (200 B.C.–200 A.D.) numeral classifiers emerged. Nevertheless, except for the chronologic order of their occurrence, there is no direct evidence to support the hypothesis that the repeater emerged earlier than the numeral classifier. Even with regard to the time of occurrence, the use of repeaters should not be considered a continual process, hence there is not su‰cient evidence showing that the numeral classifier was derived from the repeater, because the latter had disappeared before 1100 B.C., while numeral classifiers emerged and became established between 200 B.C–200 A.D. The fact that numeral classifiers emerged and developed in Late Archaic Chinese, and that repeaters always coexist with the classifier in contemporary languages, leads us to believe that the emergence and development of numeral classifiers is abrupt, with repeaters and numeral classifiers appearing at approximately the same time. Based on their function, we conclude that the repeater is just one convenient way to express this grammatical category – the one of numeral classifier – but the repeater is neither the origin nor the pathway followed by most of the numeral classifiers investigated in this study.

References Aikhenvald, Alexandra Y. 2000 Classifiers – A Typology of Noun Categorization Devices. Oxford: Oxford University Press.

232

Cheng Zhang

Bisang, Walter 1999

Classifiers in East and Southeast Asian Languages Counting and Beyond, in Gvozdanovic and Jadranka (ed.), Numeral types and changes worldwide, pp. 113–185. Berlin: Mouton de Gruyter.

Croft, Willam 1994 Semantic Universals in Classifier Systems. Word 45: 145–171. Dai, Qingxia and Jiang Ying 2005 Lu`n za`ng miaˇn yuˇ de fa´nxiaˇng xı´ng mı´nglia`ngcı´ [On the Echo Classifier of Nouns in Tibeto-Burman Languages.] Zho#ngya#ng Mı´nzu´ Da`xue´ Xue´ba`o [Journal of Minzu University of China], 2: 124–129. Delancey, Scott 1986 Toward a History of Tai Classifiers Systems. In Noun Classes and Categorization, Colette Craig (ed.), pp. 437–452. Amsterdam/ Philadelphia: John Benjamins. Huang, Zaijun 1964 Co´ng jiaˇwe´n, jı #nwe´n lia`ngcı´ de yı`ngyo`ng kaˇocha´ ha`nyuˇ lia`ngcı´ de qıˇyua´n he´ fa#zhaˇn [A study on the source and development of the Chinese classifiers from their usages in oracle and bronze inscription.] Zho#ngguo´ Yuˇwe´n [Studies of the Chinese Language], 6: 443–43. Li, Yuming 2000 Kaˇobe`i xı´ng lia`ngcı´ jı´qı´ za`i ha`nza`ng yuˇxı` fa#zhaˇn zho#ng de dı`we`i [Copied quantifiers and their status in the development of SinoTibetan quantifiers.] Zho#ngguo´ Yuˇwe´n [Studies of the Chinese Language], 1: 27–34. Liang, Min 1983 Zhua`ngdo`ng yuˇzu´ lia`ngcı´ de chaˇnshe#ng he´ fa#zhaˇn [The Origin and Development of the Classifiers in Zhuang-Dong (Kam– Tai) Languages]. Mı´nzu´ Yuˇwe´n [Minority Languages of China], 3: 8–16. Ma, Xueliang (ed.) 2003 Ha`n za`ng yuˇ ga`ilu`n [An introduction to Sino-Tibetan Language]. Beijing: Mı´nzu´ chu#baˇnshe`. Peyaraube, Alain 1998 Zho#ngguˇ, sha`ngguˇ ha`nyuˇ lia`ngcı´de fa#zhaˇn [Historical development of classifiers in Archaic and Medieval Chinese]. Yuˇya´nxue´ Lu`nco´ng [Linguistic Inquiries], 21: 99–21. Beijing: Sha#ngwu` Yı`nshu#guaˇn. Wang, Li 1980 Ha`nyuˇ shıˇgaˇo [A draft history of the Chinese language]. Revised ed. Beijing: Zho#nghua´ Shu#ju´. Xu, Xijian 1990 Jıˇngpo# yuˇ lia`ngcı´ de chaˇnshe#ng [The Origin and Development of the Terms for Measurement of the Jingpo Language], Zho#ngya#ng Mı´nzu´ Xue´yua`n Xue´ba`o [Journal of Minzu College of China], 2: 85–90.

The repeater in Chinese and other languages Xu, Xijian 1993 Xu, Xijian 1994

233

Zaˇiwaˇ yuˇ de lia`ngcı´ [Classifier of Zaiwa Language.] Mı´nzu´ yuˇwe´n [Minority Languages of China], 4: 58–62. Yı´ miaˇn yuˇ lia`ngcı´ de chaˇnshe#ng he´ fa#zhaˇng [The Emergence and Development of the Classifiers in the Yi-Burmese Languages.] Yuˇya´n ya´njıu# [Studies in Language and Linguistics], 1: 185–190.

Yang, Huandian 1983 Na`xı # yuˇ de shu`lia`ng cı´ [The Numeral–Classifier Structure in Naxi Language.] Mı´nzu´ Yuˇwe´n [Minority Languages of China], 4: 61–67. Yang, Jiangling 2005 Du´lo´ng yuˇ ge`tıˇ lia`ngcı´ he´ faˇnxiaˇngxı´ng jie´go`u de gua#nxı` – jia#nlu`n Za`ngMiaˇn yuˇ ge`tıˇ lia`ngcı´ de fa#zhaˇn te`diaˇn [The relationship between numeral classifier and repeater in Dulong yu.] The 38th Symposium on Sino-Tibetan Languages, Xianmen China, Nov. 2005.

Lexicalization in the history of the Chinese language Xiufang Dong 1. Introduction Like many other linguistic terms, lexicalization has more than one meaning and can mean di¤erent things when used by di¤erent scholars. In this section, the di¤erent definitions of lexicalization are reviewed. Then the definition that we use in this article is put forward. 1.1. Definitions of lexicalization from the synchronic perspective Lexicalization can be used to refer to the process of transferring concepts into words, and di¤erent languages are di¤erent in this regard (Talmy 1985, 2000). In formal linguistics, lexicalization can mean the phonological representation of a functional category. From the perspective of language production, lexicalization can mean the process of selecting a suitable lexical form to express an idea. Some scholars, like van der Auwera (2002), to name just one, equate lexicalization with ordinary word formation. The above is, by no means, a complete list of definitions of lexicalization from a synchronic perspective. There may be other definitions with subtle di¤erences from the ones mentioned here. In this paper, however, we are not concerned with the kind of lexicalization that is defined synchronically. 1.2. Definitions of lexicalization from the diachronic perspective From a diachronic perspective, the denotations of lexicalization are not unified. Some scholars (Ramat 1992, 2001; Hopper & Traugott 1993; Moreno Cabrera 1998; van der Auwera 2002, among others) contrast lexicalization with grammaticalization, and regard lexicalization as the reversal of grammaticalization. That is, they see lexicalization as a process in which a grammatical form becomes a lexical one, or a more grammatical form becomes a less grammatical one. Empirically, although this kind of lexicalization exists, it is rare and unsystematic across languages. Some other scholars do not contrast lexicalization with grammaticalization (Lehmann 1995 [1982], 2002; Dong 2002; Himmelmann 2004; Brinton

236

Xiufang Dong

& Traugott 2005). For example, Lehmann (1995 [1982]) interprets lexicalization as a change from a regular, analytic structure to an idiosyncratic holistic structure, and he regards lexicalization as orthogonal to grammaticalization to some degree (Lehmann 2002: 15). Dong (2002) defines lexicalization as a diachronic process in which a non-word form becomes a word. This definition is a broad one, as any process that changes a form into a member of the lexicon can be regarded as lexicalization. The lexicon in this definition contains not only lexical items but also functional items, thus the change from non-word form to grammatical word can be considered lexicalization. Dong’s (2002) definition of lexicalization is even broader than that of Brinton & Traugott (2005). Brinton & Traugott (2005) argue against the view that lexicalization is the reversal of grammaticalization, and they furthermore regard lexicalization as a process of conventionalization, like grammaticalization, sharing many parallel characteristics and mechanisms with the latter. In this respect, Brinton & Traugott (2005) have the same opinion as Dong (2002). However, they exclude grammatical words from the lexicon and attribute all the changes resulting in functional items to grammaticalization.1 Dong (2002, 2006a), on the other hand, does not simply distinguish grammaticalization and lexicalization in terms of the result of the change. Himmelmann (2004) and Dong (2006a) argue that grammaticalization and lexicalization should be distinguished through the characteristics of the process. According to Dong (2002), if a grammatical word is derived from two or more separate words, the change should be regarded as both lexicalization and grammaticalization. Thus, in Dong’s opinion, lexicalization can coincide with grammaticalization. The process that changes combinations of two (or more) words into a single lexical word is pure lexicalization. The process that changes a lexical word into a functional item (grammatical word or a‰x) is pure grammaticalization, since it does not involve the emergence of a new word, but only adds new meaning to an already existing word form. The definition of lexicalization used in this contribution is that of Dong (2002). We believe that what di¤erentiates grammaticalization from lexicalization is not merely the result (one is a functional form and the other a 1. For example, the change that turned ‘be going to’ into a marker for the future tense is regarded as an instance of grammaticalization (Hopper & Traugott 2003). However, if we deal with this example in the way of Dong (2002), the process can be said to include both lexicalization and grammaticalization: it is a type of lexicalization since it involves a change from three separate words to an idiom-like unit.

Lexicalization in the history of the Chinese language

237

lexical form), but also the characteristics of the process. Those changes that turn two or more words into a single one, whether the resultant form is a grammatical word or a lexical word, are instances of lexicalization, though they may also be considered grammaticalization when the result is a functional form. Lexicalization as defined here is di¤erent from ordinary word formation in that the former is a natural change whereas the latter is an intentional creation. We find that lexicalization as defined above is very common in the history of the Chinese language. In other languages, this kind of lexicalization is also widely attested.

2. Types of lexicalization in the history of the Chinese language In the lexicon of Modern Mandarin, disyllabic words constitute the majority, most of them being compounds. However, in the lexicon of Old Chinese, monosyllabic words were dominant. Many originally monosyllabic words were replaced by disyllabic ones at a later stage. The change from monosyllabic to disyllabic has been called ‘disyllabification’ in the literature of Chinese studies, and it is one of the most evident changes in the history of the Chinese language. The process of disyllabification was remarkable starting in the Eastern Han dynasty (Cheng 1992 among others). Some disyllabic words in the lexicon of Modern Mandarin were formed through rules of compounding. However, some other disyllabic words were historically derived from syntactic constructions or some syntactically unrelated word strings. Such a change constitutes lexicalization, which turns the form comprising two independent monosyllabic words into a single word. In fact, the rules of compounding in contemporary Chinese emerged late in history (probably after the Tang Dynasty, i.e. after the 10th Century) after many disyllabic words had already been lexicalized. In the history of the Chinese language, three main processes in which disyllabic words emerged and developed through lexicalization have been identified. In this section, we will discuss these three types of lexicalization. 2.1. From lexical phrases to words Most disyllabic words emerged through the reanalysis of phrases containing two monosyllabic lexical words. Historical evidence shows that all of the five main types of Chinese phrases (i.e. coordinate phrases, subordinate phrases, verb-object phrases, verb-result phrases, and subject-predicate

238

Xiufang Dong

phrases) were capable of developing into words, although some phrasal types lexicalized more easily than others. The evolution from phrase to word requires a necessary middle stage, the one of idioms. Thus, idiomization can be seen as an early stage of lexicalization. Over time, the words of the original phrases gradually lose their independent status and become more and more bound. In the following discussion, two stages of the lexicalization process will be discussed: 1) prelexicalization, i.e. before a phrase becomes lexicalized, and 2) post-lexicalization, i.e., after a phrase has become lexicalized. In the first stage, the forms discussed will be written in two words separated by a space. In the second stage, the word boundary has been lost, so the same form will be written without a space. 2.1.1. Lexicalization of coordinate phrases The semantic relationship between words comprising coordinate phrases can be synonymous or antonymous. One of the characteristics of coordinate phrases is that their internal word order can be changed, yet their meaning remains the same. Of course, some idiom-like coordinate phrases have rigid word order due to conventional usage. After a coordinate phrase undergoes lexicalization, the internal word order will be fixed, because syntactic operations cannot be applied to a word, as stated by the Lexical Integrity Hypothesis (Lapointe 1979). For example, xue´xı´ 学习 ‘study’ is a word lexicalized from a coordinate phrase. When it first appeared in the language, xue´ xı´ was a coordinate phrase, meaning ‘to study and exercise’, and its internal word order could be changed, so the form xı´ xue´ 习学 ‘to exercise and study’ was also acceptable. The forms xue´ xı´ and xı´ xue´ were both legitimate from the 2nd Century BC until the 13th Century AD. After that, the form xı´ xue´ was lost and only xue´ xı´ remained in the language. On this basis, we can draw the following conclusions: At the beginning, xue´ xı´ was a coordinate phrase, so its internal word order was free. After the 13th Century, xue´ xı´ became lexicalized and its internal word order became fixed. As a result, xı´ xue´ disappeared. After lexicalization, xue´xı´ came to mean ‘to study’, and the original meaning of xı´ ‘exercise’ was lost. Some nominal coordinate phrases became lexicalized through metaphorical change. For example, shoˇulıˇng 首领 ‘leader’ was initially a coordinate phrase, referring to ‘head and neck’, first seen during the time of the Western Han Dynasty (206 BC to 9 AD):

Lexicalization in the history of the Chinese language

239

(1) 当此之时，百姓不保其首领。(桓宽《盐铁论》) Da#ng cıˇ zhı # shı´, at this POSS time baˇixı`ng bu` baˇo qı´ shoˇu lıˇng. commoner not keep their head neck ‘At that time, common people could not protect their own heads and necks (i.e. common people could not protect their own lives).’ (1st C. BC, Yantielun) During the Tang Dynasty (618 AD–907 AD), shoˇu lıˇng underwent a metaphorical change. It began to refer to the leader of a group, and thus became lexicalized. For example: (2) 黔安首领田罗驹阻清江作乱。（《隋书  郭荣传》） Qia´n’a#n shoˇulıˇng Tia´n Luo´ju# zuˇ qıngjia# # ng zuo`lua`n. Place Name leader Person’s Name rely on Place Name rebel ‘The leader Tian Luoju of Qian’an relied on the Qı #ng River to rebel.’ (7th Century, Suishu) 2.1.2. Lexicalization of modifier-head phrases Modifier-head phrases are very common in Chinese, just as in other languages. There are many disyllabic words which became lexicalized from this kind of phrase. Modifier-head phrases can be divided into two types: nominal modifier-head phrases, in which the head is a noun, and verbal modifier-head phrases, in which the head is a verb. Both kinds can be lexicalized, but nominal modifier-head phrases are more likely to become lexicalized. Below, we will give two examples to illustrate the lexicalization of the two kinds of modifier-head phrases respectively. Shoˇushı` 首饰 ‘jewelry’ originally was a nominal modifier-head phrase, in which shoˇu 首 meant ‘head’ and shı` 饰 meant ‘ornaments’, thus the whole meaning was ‘ornaments on the head’. Shoˇu shı` as a phrase first appeared during the time of the Western Han: (3) 副者，后、夫人之首饰。(《诗》毛传) Fu` zheˇ, ho`u, fu#re´n zhı # shoˇu shı`. Fu Topic Marker queen noblewoman POSS head ornament ‘Fu is a kind of ornament worn on the heads of queens or noblewomen.’ (2nd Century BC, Mao’s notation of Shijing)

240

Xiufang Dong

In contemporary Chinese, shoˇushı` is a word that refers to ornaments – especially jewelry – worn on the body and not necessarily restricted to the head. After lexicalization, the original meaning of shoˇu ‘head’ was lost, and shoˇu became an internal phonological element of the word. Ho`uhuıˇ 后悔 ‘regret’ originally was a verbal modifier-head phrase, in which ho`u 后 meant ‘later’ and huıˇ 悔 meant ‘regret’. For example: (4) 作而后悔，亦无及也。（《左传  哀公六年》） Zu` e´r ho`u huıˇ, yı` wu´ jı´ yeˇ. do and later regret too not catch PART ‘It is too late to regret an act already done.’ (5th century BC, Zuozhuan) At first glance, it seems that this ho`u huıˇ in (4) is the same as that in contemporary Chinese. However, the following examples show that the ho`u huıˇ in (4) is, in fact, a phrase, since in the literature of the same period and a later period, some adverbs, auxiliaries or topic markers could be inserted between ho`u and huıˇ: (5) 不我以，其后也悔。（《诗  召南  江有汜》） Bu` woˇ yıˇ, qı´ ho`u yeˇ huıˇ. not I take DET later PART regret ‘If (you) do not take me with you, you will regret it later.’ (between 11th century BC and 6th century BC, Shijing) (6) 赢缩转化，后将悔之。（《国语  越语下》） Yı´ng suo# zhuaˇn hua`, ho`u jia#ng huıˇ zhı.# advance retreat turn change later will regret it ‘Times change, and later you will regret it.’ (5th century BC, Guoyu). (7) 今不灭越，后必悔之。（《史记  越王勾践世家》） Jı #n bu` mie` yue`, ho`u bı` huıˇ zhı # now not destroy Country’s Name later must regret it ‘If (you) do not destroy Yue now, you will certainly regret it later.’ (1st Century BC, Shiji) By the Tang Dynasty (618 AD–907 AD), we find the usage of ho`uhuıˇ as a single word, indicating that it had become lexicalized. For example: (8) 夫机事不先，鲜不后悔。（《晋书  孔愉传》） Fu# jıshı # ` bu` xia#n, xiaˇn bu` ho`uhuıˇ. PART important thing not first few not regret ‘If the important things are not prioritized, few will not regret it later.’ (7th Century, Jinshu)

Lexicalization in the history of the Chinese language

241

In (8), the negative bu` (not) precedes ho`uhuıˇ, which suggests that ho`uhuıˇ is a single verb with ho`u as a word-internal element and not an independent adverb. If ho`uhuıˇ were still a phrase and ho`u a temporal adverb, correct word order would demand that the negative bu` follow ho`u rather than precede it. The examples in (9)–(11) show the correct word order when using negatives with temporal adverbs (the related temporal adverbs in the examples are in boldface): (9)

梗阳人有狱，将不胜。（《国语  晋语九》） Geˇngya´ng re´n yoˇu yu`, jia#ng bu` she`ng. Place name person have lawsuit will not win ‘A person in Gengyang has brought a lawsuit, and will not win.’ (3rd Century BC, Guoyu)

(10) 向不怒而今怒。（《淮南子  诠言》） Xia`ng bu` nu` e´r jı #n nu`. Past not angry but now angry. ‘(You) were not angry last time but now you are angry.’ (1st Century BC, Huainanzi) (11) 一举而欺之，后不可复用也。（《淮南子  人间》） Yı # juˇ e´r qı # zhı,# One act PART cheat PRON ho`u bu` keˇ fu` yo`ng yeˇ. afterwards not can again use PART ‘If (you) cheat them (people) on one thing, afterwards you will not make use of them again.’ (1st Century BC, Huainanzi) Note that in (11), the negative bu` follows the temporal adverb ho`u, showing the same word order as when bu` co-occurs with other temporal adverbs as illustrated by (9) and (10). By contrast, in (8), bu` precedes ho`uhuıˇ. Thus we can conclude that the ho`uhuıˇ in example (8) has already become lexicalized and herein functions as a single verb. It is argued in Dong (2002) that the closer two elements are to each other at the syntactic level, the more likely it is for them to become lexicalized into a word. The syntactical closeness of two elements is determined in the following way: the fewer syntactic elements that separate them, the closer they are to each other. In the case of the modifier-head phrases, multiple modifiers have fixed word order at the syntactic level. For instance:

242

Xiufang Dong

(12) 那一件漂亮的红色短袖衬衣 na` yı # jia`n pia`olia`ng de ho´ngse` duaˇnxiu` che`nyı # that one CL pretty POSS red short-sleeved shirt ‘That pretty red short-sleeved shirt.’ In (12), there are several modifiers that precede the head noun che`nyı #

衬衣 ‘shirt’. The modifier duaˇnxiu` 短袖 ‘short-sleeved,’ which is closest

to the head, is an adjective that can be regarded as classifying the referent of the head noun. The modifier ho´ngse` 红色 ‘red,’ which is a little bit farther from the head noun, provides an objective description of the property of the head noun. Even farther from the head is the modifier pia`olia`ng 漂亮 ‘pretty’ which can be seen as a subjective description of the property of the head noun. The modifiers that are located at the outermost edge are the demonstrative determiner na` 那 ‘that’ and the numeral classifier phrase yı # jia`n 一件. The modifiers in (12) must appear in that order.2 The constraints on word order in sentences containing multiple modifiers have been discussed by many grammarians (Halliday 1985; Lu 1988; Fang 1999, among others), so we will not go into the details here. We found that the types of modifiers that are prone to appear closer to the head are more likely to undergo lexicalization with the head. For example, the modifiers that classify the referent of the head noun are more likely to undergo lexicalization with the head noun than modifiers describing the head noun from subjective perspectives. The situation is the same with the modifiers of verbal heads (a detailed discussion can be found in Dong 2002). 2.1.3. Lexicalization of verb-object phrases There are many disyllabic verb-object words in Modern Mandarin. Verbs and their objects are usually the focus of a sentence, and one may therefore rightly think that this would make it di‰cult for them to be lexicalized. This di‰culty is indirectly supported by the following fact: disyllabic verb-object words are likely to be susceptible to specific syntactic processes. Some frequently used disyllabic verb-object words can be separated by the insertion of other elements, such as aspectual markers. This calls into question their status as a single word. In the literature (Lu 1957 and others), those verb2. The position of ‘demonstrative þ numeral classifier’ has some (degree of ) flexibility. Sometimes, it can appear close to the head noun. In this kind of situation, it can be related to richer contextual information and has the function of identifying the referent of the head noun (Fang 1999).

Lexicalization in the history of the Chinese language

243

object words whose internal structure is to some degree visible to specific syntactic processes are called lı´he´ cı´ 离合词 (literally ‘separate-unified words’). There are many semantic constraints on those verb-object phrases that are candidates for lexicalization: the verb in the phrase should not refer to a physical activity; the object must be more or less abstract; the e¤ect of the verb on the object must be invisible, that is to say, the object will not experience visible changes due to the action denoted by the verb. All disyllabic verb-object words in the lexicon of Modern Mandarin show the semantic constraints above, and the same constraints can be seen in documented historical cases of lexicalized verb-object phrases. The emergence of the disyllabic word hua´iyı´ 怀疑 ‘suspect’ provides a relevant example. Hua´i yı´ was a verb-object phrase when it appeared in Middle Chinese, in which hua´i 怀 meant ‘hold, possess’ and yı´ 疑 meant ‘suspicion’, as shown in the following example from the period of Three Kingdoms (220 AD–280 AD): (13) 子犹怀疑，求之明据。（三国魏曹植《王仲宣诔》） Zıˇ yo´u hua´i yı´, qiu´ zhı # mı´ng ju`. you still have suspicion request PART clear evidence ‘You still have suspicions, so you requested clear evidence.’ (3rd Century, Cao Zhi’s essay) The phrasal status of hua´i yı´ in (13) can be shown with the existence of hua´i yı´ huo` 怀疑惑 ‘have suspicions and confusions’, hua´i yı´ ju` 怀疑惧 ‘have suspicions and fears’, hua´i yı´ fe`n 怀疑忿 ‘have suspicions and angers’, hua´i yı´ e`r 怀疑贰 ‘have suspicions and disloyalties’, hua´i yı´ we`i 怀疑畏 ‘have suspicions and dread’, hua´i yı´ lu¨` 怀疑虑 ‘have suspicions and misgivings’, hua´i yı´ jı` 怀疑忌 ‘have suspicions and scruples’ and other similar phrases taken from historical documents. In these phrases, yı´ forms a coordinate phrase with a word of similar meaning and the combination serves as the object of the whole phrase. If hua´i yı´ were a word, with yı´ as a part of the word, it would not be able to participate in this coordinating syntactic operation. As a phrase, hua´i yı´ can not take an object because the position of the object is already taken by yı´. After a long period of frequent use, hua´iyı´ became lexicalized and began to be used as a single verb. This happened in the late Qing Dynasty (second half of the 19th Century). Once hua´iyı´ became lexicalized, it was used as a transitive verb and was able to take a nominal object or a clausal object.

244

Xiufang Dong

2.1.4. Lexicalization of subject-predicate phrases Disyllabic subject-predicate words are rare in the contemporary Chinese lexicon. Since the subject is an external argument of the verb in a sentence and the object is an internal one, the relationship between subject and verb is more distant than the one between object and verb. This explains the fact that the verb is more likely to undergo lexicalization with the object rather than with the subject. However, there do exist some disyllabic words lexicalized from subject-predicate phrases. Dı`zhe`n 地震 ‘earthquake’ is a case in point. (14) 九月癸酉，地震。（《左传  文公九年》） Jiuˇyue` guıˇyoˇu, dı` zhe`n. September Guiyou earth shake ‘On the Guiyou day of September, the earth shook.’ (5th century BC, Zuozhuan) In (14), dı` zhe`n is a verb phrase, which can function as a clause. This is supported by the fact that there were some forms in which zhe`n was coordinated with synonymous words, such as dı` zhe`n do`ng 地震动 ‘earth shake and move’, dı` zhe`n lie` 地震裂 ‘earth shake and split’, as illustrated in (15) and (16) below: (15) 西周镐京地震动。（《史记  周本纪》集解） Xı #zho#u ha`ojıng # dı` zhe`n do`ng. Dynasty Name Place Name earth shake move ‘In Haojing, Western Zhou, the earth shook and moved.’ (3rd Century, annotation of Shiji) (16) 九月戊辰，地震裂。（《后汉书  光武帝纪下》） Jiuˇyue` wu`che´n, dı` zhe`n lie`. September Day’s Name earth shake split ‘On the Wuchen day of September, the earth shook and split.’ (5th Century, Houhanshu) Although dı` zhe`n had some nominal usages in historical texts, in most cases dı` zhe`n was verbal, whereas in Modern Mandarin, dı`zhe`n is used more frequently as a noun than a verb. This di¤erence is simply an e¤ect of its lexicalization. Dong (2002) points out that there are some semantic constraints on the lexicalization of subject-predicate phrases. The subject part must be an inanimate noun and the predicate part must be an intransitive verb or an

Lexicalization in the history of the Chinese language

245

adjective whose meaning must have the characteristic of ‘involitional’ (for the concepts of ‘volitional verb’ and ‘involitional verb’, see Ma 1998 and references therein), i.e. denoting some activity or state which cannot be controlled volitionally. Due to limitations of space, we will not discuss these constraints in detail herein. 2.1.5. Lexicalization of verb-result phrases Resultative compounds are a special kind of compound in Chinese. In this kind of compound, the first part is a verb (usually a transitive verb, seldom an intransitive verb or an adjective), and the second part is an adjective or an intransitive verb. Semantically, the second part expresses the result of the action represented by the preceding verb. For example, kuo`da` 扩大 means ‘enlarge, extend’ (literally ‘expand-big’). Historically, resultative compounds were lexicalized from verb-result phrases which were reanalyzed from coordinate verb phrases. There is a lot of controversy regarding the exact time at which these resultative constructions emerged (cf. Jiang 1996). We believe that it happened around the 5th or 6th Century BC. The semantic condition for the lexicalization of a verb-result phrase requires that the result part, to some degree, be predictable from the meaning of the first verb. In some cases, the meaning of the result part is already implied in the meaning of the verb. For example, in kuo`da` 扩大 ‘enlarge, expand’, the result part, da` 大, which means ‘big’, is entailed by the meaning of the verb kuo` 扩, which means ‘expand’. In such cases, the predictability of the result is highest. In other cases, the meaning of the result is not entailed by the meaning of the verb, but it is to some degree an expected result, and therefore, also to some degree predictable. For example, in shua#isuı` 摔碎 ‘throw down and cause to become broken’, the result part, suı` 碎 ‘broken,’ is an expected result of the verb shua#i 摔 ‘throw down’, though the former is not implied by the latter. When the second part of a verb-result phrase expresses an unexpected result, the phrase is unlikely to undergo lexicalization. For example, Lu (1990) points out that in the construction wa# qiaˇn 挖浅, literally ‘dig-shallow’, qiaˇn 浅 ‘shallow’ is an unexpected result of the action of wa# 挖 ‘dig’, in other words, it refers to an unsatisfactory result. This semantic characteristic can be used to explain the fact that wa# qiaˇn cannot take an object though it can still be regarded as a verb phrase, as shown in example (17), in which the patient of wa# can only appear as the subject:

246

Xiufang Dong

(17) 沟挖浅了。 go#u wa# qiaˇn le. trench dig shallow PART ‘The trench was dug, but it turned out to be too shallow.’ (The implication of the sentence is: ‘the trench does not meet the depth requirement’) We argue that all resultative constructions that can take objects must be regarded as syntactic words (Dong 1998, 2007a). Some of them became further lexicalized into lexical words because of semantic idiosyncrasy or high frequency of use. For example, tı´ga#o 提高 (literally ‘raise high’), which means ‘improve’, is a lexical word since its meaning cannot be predicted from its components and it is included in Chinese dictionaries. 2.1.6. Some generalizations about lexicalization of lexical phrases Historical facts show that some phrasal types become lexicalized more easily than others. Syntactic di¤erences can explain the asymmetry with respect to the probability of lexicalization among the five phrasal types (i.e. coordinate phrases, subordinate phrases, verb-object phrases, verbresult phrases, and subject-predicate phrases). One conclusion that can be drawn is that phrases which have a tendency to appear as a core constituent of a sentence are less likely to become lexicalized than those that are prone to appear as a modifying constituent. This is because the core constituent (e.g. the predicate) is the most important part of the sentence and attracts the attention of language users. Verb phrases usually undergo lexicalization when another verbal element which functions as the focus of the sentence is presented.3 Lexicalization is natural in language and, like grammaticalization, it develops outside the consciousness of language users. That is to say, only after the change has happened, and by contrasting the new usage with the old one, do language users notice the di¤erence. Therefore, if a phrase is central, it is less likely to undergo lexicalization. The generalization concerning the principle of lexicalization of all types of phrases is: only the syntactically and semantically less typical members of phrases are candidates for lexicalization. For example, the verb-object phrases formed by verbs referring to physical action are more typical 3. Some verbal coordinate phrases may seem exceptional, since some of them can become verbs in the predicate position. Probably some verbal coordinate compounds are not the result of lexicalization but word formation.

Lexicalization in the history of the Chinese language

247

than those formed by verbs referring to abstract action, thus they are less likely to become lexicalized. 2.2. From functional phrases to words A functional phrase contains a grammatical word. Some disyllabic words were lexicalized from their original syntactic structures which were comprised of a grammatical word and a lexical word (i.e. a functional phrase). Some originally very active grammatical words gradually lost their syntactic function over time and, as a result, some of the constructions involving that grammatical word became ‘fossilized’ and were reanalyzed as words. The result of this kind of lexicalization is often a grammatical word. In this section, we illustrate this type of lexicalization with some examples and explain the mechanisms behind this kind of change. 2.2.1. Lexicalization of suoˇyoˇu 所有 Suoˇ 所 was a nominalization marker in Old Chinese (Zhu 1983). It could attach to a VP and transform it into an NP. In most cases, the meaning of ‘suoˇ þ VP’ is the argument of the VP, especially the patient of the verb (as for the detailed study of the semantic function of suoˇ, please see Yin 2006). For example, suoˇ sha# 所杀 ‘suoˇ þ kill’ means ‘those that are killed’. This function of suoˇ was very frequent in Old Chinese, but declined after Middle Chinese (Dong 2002; Wei 2003). In Modern Mandarin, the function of suoˇ has been taken over by the particle de 的. Suoˇ remains only in written expressions and its use has become optional. In Modern Mandarin, the meanings of (18a) and (18b) are the same, indicating that suoˇ is no longer an obligatory element, and its occurrence only gives the sentence a written flavor. (18) a. 他所写的书 ta# suoˇ xieˇ de shu# he PART write POSS book ‘the book that he wrote’ b. 他写的书 ta# xieˇ de shu#. he write POSS book ‘the book that he wrote’ After the decline of the function of suoˇ, some ‘suoˇ þ verb’ constructions became lexicalized. Suoˇyoˇu is a case in point. Suoˇyoˇu means ‘all’ in Modern

248

Xiufang Dong

Mandarin, but originally, it was a regular nominalization construction involving suoˇ and the verb yoˇu 有 ‘have’, and it meant ‘those that (one/some people) has/have’, as indicated in (19): (19) 古之为市也，以其所有，易其所无者。（《孟子  公孙丑下》） Guˇ zhı # we´i shı` yeˇ, yıˇ qı´ suoˇ yo`u, ancient POSS do market PART with DET PART have yı` qı´ suoˇ wu´ zheˇ. exchange DET PART not have PART ‘The building of markets in ancient times was to exchange what one has for what one does not have’ (3rd Century BC, Mencius) In some Middle Chinese contexts, suoˇ yoˇu can imply ‘all the things that someone has’: (20) 死至命尽，所有财物，官爵俸禄，故在世间。

（三国吴支谦译《四愿经》）

Sıˇ zhı` mı`ng jı`n, suoˇ yoˇu ca´iwu`, death arrive life end PART have properties gua#njue´ fe`nglu`, gu` za`i shı`jia#n. o‰cial-rank salary still exist world ‘When death comes and life ends, the properties that one has, the o‰cial ranks and salaries, will still exist in the world.’ (3rd Century, Buddhist text) In (20), suoˇ yoˇu can be interpreted as ‘what (one) has’, but it can also be interpreted as ‘all that (one) has’ in the context by inference. As this kind of usage increased over time, the context-obtained meaning ‘all’ became the inherent meaning of suoˇ yoˇu. By the Tang Dynasty (618 AD–907 AD), suoˇ yoˇu lost its original meaning altogether and became lexicalized into an adjective meaning ‘all’, used as a quantifier to modify a nominal head. Other forms containing suoˇ and a verb that also experienced lexicalization include suoˇwe`i 所谓 ‘so called’ (literally ‘suoˇ þ call’), suoˇza`i 所在 ‘location’ (literally ‘suoˇ þ exist’), etc. 2.2.2. Lexicalization of qı´shı´ 其实 ‘in fact’ In traditional studies, qı´ 其 is regarded as a demonstrative pronoun which functions as a modifier of NPs. Based on evidence from Old Chinese texts, we propose that qı´ also be considered as a definite marker in some cases

Lexicalization in the history of the Chinese language

249

(especially after the Han period), with a function similar to that of the definite article ‘the’ in English though its use is not obligatory (so qı´ is not a full-fledged definite article). Therefore, we can regard qı´ as a determiner in Classical Chinese. For example:4 (21) 公闻其期，曰：‘‘可矣。 ’’（《左传  隐公元年》） Go#ng we´n qı´ qı,# yue#: ‘‘keˇ yıˇ.’’ duke hear DET date say can PART ‘The duke heard the date, and said ‘‘OK now.’’’ (5th Century BC, Zuozhuan) Over time, the function of qı´ as a definite marker gradually declined, which led some constructions comprised of qı´ and a noun to undergo lexicalization. In this way qı´shı´ became a grammatical word. In Old Chinese, qı´ shı´ meant ‘the essence, the fact or the reality’. It was a construction containing the definite marker qı´ and a noun shı´ 实, which meant ‘essence, reality’. Example (22) illustrates this: (22) 吾有卿之名，而无其实。（《国语  晋语八》） Wu´ yoˇu qıng # zhı # mı´ng, e´r wu´ qı´ shı´. I have Qing POSS name but not have DET substance ‘I am a Qing in name only; I do not have the materials of a Qing.’ (3rd Century BC, Guoyu) In (22), qı´ shı´ is the nominal object of the negative verb wu´ 无 ‘not have’. But when qı´ shı´ functions as the subject preceding a VP, it is possible for it to be reanalyzed as an adverb, and such a reanalysis took place in the Song Dynasty , as shown in (23): (23) 其实后即叟之长女也。（《旧五代史  后妃列传》） Qı´shı´ ho`u jı´ soˇu zhı # zhaˇngnuˇ¨ yeˇ. in fact queen be old-man POSS oldest-daughter PART ‘In fact, the queen was the oldest daughter of the old man.’ (10th Century, Jiu Wudai Shi)

4. The reviewer pointed out that in example (21) and (22), qı´ refers back to a situation in the preceding context and still has the possessive function ‘the X of it > its X’. However, if this function can be regarded as possessive, it is rather atypical. It is under just this kind of circumstance that qı´ obtained its new function as a definite marker.

250

Xiufang Dong

After reanalysis, qı´shı´ became an adverb meaning ‘in fact’. As an adverb, qı´shı´ can either precede or follow the subject. Qı´zho#ng 其中 ‘among (which, them, etc.), inside’, qı´yu´ 其余 ‘the others’, qı´cı` 其次 ‘next, then’ and others are words in Modern Mandarin that emerged in a similar way. 2.2.3. Brief descriptions of some other examples The postposition jia#n 间 indicating location or time was used frequently in Middle Chinese, but its function as a postposition disappeared in Modern Mandarin and it became a bound morpheme. Some frequently used postpositional constructions in the form of ‘noun þ jia#n’ in Middle Chinese became lexicalized and were retained in the language as a relic of a vanished syntactic phenomenon. For example, shı`jia#n 世间 ‘world’ and ye`jia#n 夜间 ‘evening’ are words that emerged in this way. Some negative words, such as bu` 不 ‘not’, wu´ 无 ‘not have; not’, fei# 非 ‘not be’, we`i 未 ‘did not, have not’, became lexicalized with the word that they negated. Among the many words that emerged in this way are bu`miaˇn 不免 ‘unavoidable’ (literally ‘not avoid’), bu`lu`n 不论 ‘no matter’ (literally ‘not consider’), bu`guaˇn 不管 ‘no matter; despite’ (literally ‘not be in charge of ’), bu`maˇn 不满 ‘be critical of; discontent; disgruntled’ (literally ‘not content’), wu´lu`n 无论 ‘no matter’ (literally ‘no considering’), fe#icha´ng 非常 ‘extremely’ (literally ‘not common’) , and we`imiaˇn 未免 ‘rather; a bit too’ (literally ‘have not avoided’). These words belong to di¤erent syntactic categories: adverbs, conjunctions, verbs, adjectives and nouns. This kind of lexicalization is systematic to some degree, since it creates a series of words having the same internal structure. We regard this kind of change as lexicalization but not grammaticalization, since these words involving a negative element all emerged independently over time, and the negative element in these words cannot be regarded as a prefix since its use is not regular and new words cannot be created by attachment of the negative morpheme based on morphological rules. 2.2.4. Motivations and characteristics of lexicalization of functional phrases The most important motivation for the lexicalization of functional constructions is the decline in use of the grammatical item involved. When the function of the grammatical item fades, it can no longer collocate with specific types of lexical items. Constructions involving the faded grammatical item may not disappear entirely. In fact, some commonly used collocations are sustained in the language through lexicalization.

Lexicalization in the history of the Chinese language

251

This kind of lexicalization often results in a series of grammatical words with the same internal structure, such as the words involving the same negative word bu` mentioned in the preceding section. Therefore, this kind of lexicalization, to some degree, exhibits systematic characteristics. At first glance, the common part in a series of words which become lexicalized in the same way seems to be an a‰x, but actually it is not. This kind of change cannot be regarded as grammaticalization based solely upon the syntactic category of the resultant form. At best, we can say, this kind of change constitutes lexicalization as well as grammaticalization when the result is a grammatical word. Lexicalized functional phrases are usually quite opaque since one of the members, i.e. the original grammatical word, like suoˇ in the word suoˇyoˇu, often loses its original function, thus making no (or a very minimal) semantic contribution to the whole word. 2.3. From syntactically unrelated word string to word Two independent words that originally were not at the same syntactic level (and therefore did not form one linguistic unit), but were only adjacent in linear order, can also become lexicalized over time. This kind of lexicalization is special in so far as the source form is not a linguistic unit. Disyllabic words that emerged in this way are usually grammatical words and their internal structure usually became opaque over time. 2.3.1. Adverbs lexicalized from syntactically unrelated word strings Some adverbs emerged through the union of two adjacent though syntactically unrelated words. For instance, jıhu# # 几乎 means ‘almost’ in Modern Mandarin, whereas originally it consisted of two separate words that did not have an immediate syntactic relationship: jı # 几 was a verb, meaning ‘near’; hu# 乎 was a preposition that could introduce an NP after the verb. This is illustrated by (24): (24) 如知为君之难也，不几乎一言而兴邦乎？（《论语  子路》） Ru´ zhı # we´i ju#n zhı # na´n yeˇ, if know do emperor POSS di‰culty PART bu` jı # hu# yı # ya´n e´r xıng # ba#ng hu#? not near to one utterance and strengthen country PART ‘If one knows the di‰culties of being an emperor, isn’t his utterance near to ‘‘the one utterance that can strengthen the country’’?’ (5th Century BC, Lunyu)

252

Xiufang Dong

(25) 《易》不可见，则乾坤或几乎息矣。（《易  系辞上》） Yı` bu` keˇ jia`n, ze´ qia´nku#n huo` jı # hu# xı # yıˇ Yı` not can see then heaven earth probably near to stop PART ‘(If ) Yi cannot be seen, then heaven and earth are probably near the end.’ (3rd Century BC, Yizhuan) In Chinese, a verb can become a noun through conversion without any overt phonological change. Thus, VPs can appear in NP positions without changing their forms. When a VP appears after the verb jı # and the preposition hu#, it produces the linear order jı # hu# þ VP, in which jı # hu# takes the syntactic position of an adverb. Under this condition, jı # hu# was reanalyzed and became lexicalized as an adverb meaning ‘almost’. This reanalysis, based on the data used in this study, happened during the Ming Dynasty (1368AD–1644AD). Apparently, the meaning of the lexicalized jı #hu# is closely related to the meaning of the source construction ‘jı # (‘near’) þ hu# (‘to’) þ VP’, i.e. ‘near to some situation’. An example similar to jıhu# # is zho#ngyu´ 终于, which is also derived from a combination of a verb and a preposition. Zho#ng originally was a verb and meant ‘end’. Yu´ was a preposition, introducing an NP after the verb. Originally, zho#ng yu´ was not a syntactic unit but formed part of the construction [zho#ng [ yu´ NP]], as shown in (26): (26) 愚者始於乐而终於哀。（《淮南子  主术》） Yu´ zheˇ shıˇ yu´ le` e´r zho#ng yu´ a#i. stupid person begin PREP happiness and end PREP sadness ‘Stupid people begin with happiness and end with sadness.’ (2nd Century BC, Huainaizi) In Modern Mandarin, zho#ngyu´ has become lexicalized into a single word meaning ‘at last, finally’. It is evident that the meaning of the adverb zho#ngyu´ is closely related to the meaning in the source construction [zho#ng [ yu´ NP]], which can be interpreted as ‘[end [in some situation]]’. Here it involves a change of ‘subjectification’ (for the concept of ‘subjectification’, see Traugott 1989). The source construction describes an objective end point of an event, whereas the lexicalized form of zho#ngyu´ expresses the speaker’s attitude in addition to expressing the temporal meaning of ‘in the end’. Let’s look at another instance of this type of evolution: he´bı` 何必. He´bı` is an interrogative adverb in Modern Mandarin, but originally it was just a combination of two adjacent words, one of which was the interrogative adverb he´ 何 ‘why’, and the other the modal verb (originally a verb) bı` 必

Lexicalization in the history of the Chinese language

253

‘must’. In Old Chinese, all the documented examples of he´ bı` were not syntactic units but merely word strings, as shown in (27): (27) 天下多美妇人，何必是？（《左传  成公二年》） Tia#nxia` duo# meˇi fu`re´n, he´ bı` shı`? world many pretty woman why must this ‘There are many pretty women in the world, why must you have this one?’ (5th Century BC, Zuozhuan) In (27), bı` is at the same syntactic level as the succeeding VP (a verbalized NP), while he´ is at a higher level of the clause. The internal structure of he´ bı` shı` 何必是 in (27) can be analyzed as [ IP he´ [vp [bı` shı` ]]. Clearly, he´ and bı` are not at the same syntactic level, thus they do not constitute a syntactic unit. By early Middle Chinese, however, he´ bı` underwent lexicalization and became an interrogative adverb, which could be used to dissuade others from doing something, as shown in example (28): (28) 何必以言取怨？（《后汉书  乐恢传》） He´bı` yıˇ ya´n quˇ yua`n? why must with speech obtain complain ‘Why must you incur odium with your words?’ (5th Century, Houhanshu) Before lexicalization, the construction ‘[IP he´ [vp [bı` shı` ]]’ was used to question why there must be a selection among some possibilities as shown in (27). After lexicalization, [he´bı` VP] came to be used to politely deny the necessity of doing something. Evidently, the meaning of he´bı` as a single word is closely related to the meaning of the source construction ‘[he´ [bı` VP]]’, since asking the reason why one must do something often implies disagreement. Therefore, we can conclude that the interrogative adverb he´bı` was derived from the syntactically unrelated he´ bı` in the construction ‘[he´ [bı` VP]]’, and the two kinds of he´ bı` are not accidentally homophonous forms. 2.3.2. Conjunctions lexicalized from syntactically unrelated word strings Some conjunctions were lexicalized from syntactically unrelated word strings. In this section, we will analyze some examples that illustrate this kind of lexicalization. Foˇuze´ 否则 is a conjunction in Modern Mandarin that means ‘otherwise’. Historically, foˇuze´ is documented as two separate words in Old

254

Xiufang Dong

Chinese, i.e. a predicative negative foˇu 否 and a conjunction ze´ 则. Apparently, in Old Chinese, foˇu and ze´ did not constitute a single linguistic unit, as shown in (29): (29) 格则承之庸之，否则威之。（《尚书  益稷》） Ge´ ze´ che´ng zhı # yo#ng zhı #, foˇu ze´ we#i zhı #. obey CONJ recommend PRO use PRO not CONJ overawe PRO ‘If (they) obey (the rules), we will recommend and appoint them; if not, we will intimidate them (with punishment).’ (3rd Century BC, Shangshu) In (29), foˇu is a predicative negative, forming a clause, ze´ is a conjunction introducing another clause. After lexicalization in early Modern Chinese, foˇuze´ became a single conjunction, which can be clearly shown with the following fact: before lexicalization, ze´ was closely related to the clause that it introduced, but in Modern Mandarin, foˇuze´ can be followed by the particle dehua` 的话 and a pause, which separates ze´ and the following clause: (30) 你必须来，否则的话，这个会就开不成了。 Nıˇ bı`xu# la´i, foˇuze´ dehua`, you must come otherwise PART zhe` ge` huı` jiu` ka#i bu` che´ng le. this CL meeting then open not succeed PART ‘You must come; otherwise it is not possible to hold the meeting.’ 2.3.3. Verbs lexicalized from syntactically unrelated word strings Yıˇwe´i 以为 is a verb in Modern Mandarin meaning ‘to think or regard’. Historically, yıˇwe´i was derived from the syntactically unrelated combination of yıˇ 以 and we´i 为. Yıˇ was a preposition (derived from a verb) and had a general meaning, roughly equal to ‘take; with’. We´i was a verb and meant ‘be’. In Old Chinese texts, the object of yıˇ is often omitted due to its appearance in the antecedent clause, thus yıˇ and we´i could be adjacent to each other. For example: (31) 人 i 之无良，我以j i为兄!（《诗  鄘风  鹑之奔奔》） Re´n zhı # wu´ lia´ng, woˇ yıˇ we´i xio#ng. person PART not good I PREP be brother ‘That person is not good, and I regard him as my brother!’ (between 11th Century BC to 6th Century BC, Shijing)

Lexicalization in the history of the Chinese language

255

In (31), the object of yıˇ 以 ‘take’ is an empty category and it is co-referent with the noun re´n 人 ‘the person’ in the preceding clause. The same subscript indicates co-reference. In the middle of the Warring States period (475BC–221BC), yıˇwe´i underwent lexicalization and became a cognitive verb (Yao 1997). 2.3.4. Prepositions lexicalized from syntactically unrelated word strings Zhı`yu´ 至于 is a preposition, meaning ‘concerning, with regard to’ in Modern Chinese. Historically, it was derived from a syntactically unrelated combination of the verb zhı` 至 and the preposition yu´ 于. In the following example from Shangshu, zhı` yu´ is such a combination, in which zhı` means ‘arrive’ and yu´ introduces a locative NP and thus forms a PP serving as the complement of the verb zhı`: (32) 王命众，悉至于庭。（《尚书  盘庚上》） Wa´ng mı`ng zho`ng, xı # zhı` yu´ tı´ng. king order mass all come PREP court ‘The king ordered the masses all to come to the court.’ (3rd Century BC, Shangshu) Zhı`yu´ was lexicalized into a preposition by the end of the Warring States period and obtained the meaning of ‘concerning, as for’ used to introduce a topic. The semantic change of zhı`yu´ resulting from lexicalization can be explained by metaphoricalization: It is a change from the original meaning ‘arrive at’, which is related to movement in physical space, i.e. change of location, to the extended meaning ‘as for’, which is related to a transfer in the field of discourse, i.e. change of topic. 2.3.5. Characteristics of lexicalization of syntactically unrelated word strings and constraints on them The words lexicalized from syntactically unrelated word strings are opaque in their internal structure from the perspective of synchronic analysis. Those word strings that can be lexicalized often include a functional item which was semantically light and often phonologically weak. Those word strings that appear in the sentence-initial position are more likely to be lexicalized than word strings that appear in other positions.

256

Xiufang Dong

Nevertheless, not just any two syntactically unrelated words can undergo lexicalization. There are strict constraints on this kind of lexicalization, and the constraints are even stricter than those governing other kinds of lexicalizations: since the lexicalization of syntactically unrelated word strings is a change from a non-linguistic form to a linguistic unit, it is less likely than a change from one kind of linguistic unit to another. There are three main constraints on the lexicalization of syntactically unrelated word strings: A. The sentence type in which the word string undergoing lexicalization appears should be frequently used and fixed to some degree, that is to say, it can be regarded as a ‘construction’ in the sense of Goldberg (1995). B. A word string that can be lexicalized should be representative of the frame of the construction in which it appears. In other words, the two separate words that undergo lexicalization must be key words in the construction where they are located. For example, foˇu and ze´ are two important words in the construction ‘X ze´ VP1, foˇu ze´ VP2’ (‘X, then VP1, not then VP2’). Semantically, the meaning of the word string should be able to activate the whole meaning of the construction, that is to say, the meaning of the word string can, to some degree, be associated with the meaning of the whole construction. C. The functional item in the word string undergoing lexicalization usually experiences some functional weakening.

3. The degree of lexicalization Like grammaticalization, lexicalization is a gradual, step-by-step process. A disyllabic form in Chinese can show di¤erent degrees of lexicalization during its development. The degree of lexicalization of a word is related to the developmental stages that it has gone through. Dong (2002) investigated the degree of lexicalization of coordinate verbal compounds and found that there are four stages that coordinate verbal compounds typically go through during the process of lexicalization. Stage 1: Internal word order becomes fixed. At this stage, in the language system there exists a monosyllabic variant expressing the same meaning. For example, as previously discussed, xue´ xı´ originally was a coordinate phrase, thus its internal word order could be changed. After lexicalization, the internal word order of xue´ xı´ became fixed, and xı´ xue´ was no longer

Lexicalization in the history of the Chinese language

257

an acceptable form. The unacceptability of the form xı´ xue´ indicates that xue´xı´ cannot be regarded as a coordinate phrase anymore and must be seen as a single word. In Modern Mandarin, the monosyllabic form xue´ is a word and can express almost the same meaning as xue´xı´, thus xue´ can be regarded as a monosyllabic variant of xue´xı´. The existence of a monosyllabic variant indicates the low degree of lexicalization achieved by xue´xı´. It indicates that the earlier internal structure of xue´xı´ is still clear to some degree and one of the elements remains syntactically independent. Stage 2: Monosyllabic variant disappears. For example, miaˇnlı` 勉励 ‘encourage’ was derived from a coordinate verb phrase. In Modern Mandarin, not only is its internal word order fixed, but also have its members miaˇn 勉 and lı` 励 lost word status (in Old Chinese, both miaˇn 勉 and lı` 励 could be used independently). Today, there is no monosyllabic variant for miaˇnlı`. Stage 3: The expression undergoes semantic change. For example, duo`luo` 堕落 ‘corrupt; corruption’ was derived from a coordinate verb phrase. After lexicalization, its internal word order became fixed. Moreover, it underwent semantic change. At the beginning of lexicalization, duo` luo` meant ‘fall’, referring to the (physical) event of dropping from a high location to a low one. At a later stage, duo`luo` came to mean ‘corrupt; corruption’, metaphorically referring to degeneration of morality, which, in abstract terms, represents moving from a good (high) status to a bad (low) one. Under the newly obtained meaning, duo`luo` has no monosyllabic variant, that is to say, neither duo` 堕 nor luo` 落 can express the meaning of ‘corrupt, corruption’. Stage 4: The expression undergoes syntactic change, i.e. categorical change such as transforming from a VP to a noun. For example, xue´we`n 学问 originally was a coordinate verb phrase meaning ‘study and inquiry’. After lexicalization, xue´we`n became a noun referring to ‘knowledge; learning’. It can be seen that, with the change in syntactic category, xue´we`n underwent a corresponding semantic change. Also, there is no monosyllabic variant for the noun xue´we`n. The four stages distinguished above correspond to four degrees of lexicalization, from low to high. The higher the degree of lexicalization of a word, the more opaque it is in terms of semantics and internal structure. Similarly, other kinds of disyllabic words besides coordinate verbal ones can also be identified as having di¤erent degrees of lexicalization. A high degree of lexicalization can lead to phonological reduction, usually reflected in the loss of the tone of the second syllable.

258

Xiufang Dong

4. Constraints on lexicalization 4.1. Prosodic constraints The form that undergoes lexicalization must be disyllabic. According to the theory of Prosodic Morphology (McCarthy & Prince 1990, 1993), a prosodic word must be at least a foot; and as argued by Feng (1996), in Chinese a foot is disyllabic. Therefore, disyllabic words are prosodic words in Chinese. A disyllabic form meets the phonological requirement of a word, thus it can become lexicalized. A possible exception to this prosodic constraint is the lexicalization of discourse markers: Multi-syllabic forms can be lexicalized into discourse markers (Dong 2007). After disyllabic words became predominant in the lexicon of Chinese, lexicalization of trisyllabic forms based on disyllabic words became possible. Dong (2004a) argues that some disyllabic adverbs or conjunctions in Modern Mandarin can incorporate the copular shı` 是 to form synonymous adverbs or conjunctions. For example, suıra # ´ nshı` 虽然是 ‘although þ copular’, which can be synonymous to suıra # ´ n 虽然 ‘although’; and jıng# cha´ngshı` 经常是 ‘often þ copular’, which can be synonymous to jı #ngcha´ng 经常 ‘often’, etc. 4.2. Syntactic constraints Adjacency in linear order is the basic requirement for two separate words to become a single word. However, empty categories will not interfere with the occurrence of lexicalization. That is to say, if there is an empty category between two words, the two words can still experience lexicalization. For example, the word tıngshuo# # 听说 ‘hear about, it is said that’ was historically derived from the construction ‘‘tıng # 听 ‘listen to, hear’ þ empty category þ shuo# 说 ‘say’’’, which meant ‘hear (somebody) say (something)’, as illustrated in (33): (33) 闲听说五岳。（唐李洞《题玉芝赵尊师院》） Xia´n tıng # shuo# wuˇ yue`. leisure listen say five mountain ‘(He) listened leisurely to somebody talk about the Five Mountains.’ (9th Century, Lidong’s Poem) In (33), the reference of the empty category following tı #ng is not overtly expressed because it is irrelevant to the discourse.

Lexicalization in the history of the Chinese language

259

The empty category could be expressed in the preceding context, as shown in (34): (34) 行者在旁道：‘‘师父，我前日在包袱里，曾见那领袈裟， 不是件宝贝？拿与他看看如何？’’ 众僧听说袈裟，

一个个冷笑。（《西游记》第十六回）

Xı´ngzheˇ za`i pa´ng da`o: ‘‘Shı #fu`, woˇ qia´nrı` za`i ba#ofu# lıˇ, Xı´ngzheˇ at aside say master I day-before-yesterday at luggage in ce´ng jia`n na` lıˇng jia#sha#, bu`shı` jia`n baˇobe`i? ever see that CL cassock, not CL treasure? Na´ yuˇ ta# ka`nka`n ru´he´ ?’’ Zho`ng se#ng tı #ng shuo# jia#sha#, Take give he look-look how many monk listen say cassock yı # ge` ge` leˇngxia`o. one CL CL sneer ‘Xingzhe said: ‘‘Master, I saw a cassock in the luggage the day before yesterday. Isn’t it a treasure? How about taking it out and letting them see it? ’’ The monks heard (Xı´ngzheˇ ) say this and they all sneered.’ (17th Century, Xiyouji) In (34), the empty category following tı #ng refers to Xı´ngzheˇ (Su#n Wu`ko#ng), that appears in the previous context. The existence of the empty category between tı #ng and shuo# did not block lexicalization. In the Qing Dynasty (1644–1840), tı #ngshuo# became lexicalized. Loss of syntactic function in one element in the combination of words is often a factor that increases the probability of lexicalization of the combination, as illustrated in lexicalization involving grammatical words. However, this is not a necessary condition. We can find some combinations involving frequently used grammatical words that underwent lexicalization. For example, haˇole 好了 ‘good þ aspectual marker’ in Modern Mandarin is lexicalized in some contexts and is roughly equivalent to ‘OK’, functioning as a discourse marker to finish the current topic and introduce a new topic. One of its components, le 了, is an aspectual maker and has a high frequency of use in Modern Chinese. 4.3. Semantic constraints Di¤erent types of lexicalizations have di¤erent semantic constraints. Dong (2002) analyzes the semantic conditions for di¤erent types of lexicalization in detail, some of which were mentioned in Sections 2 and 3. In what follows, we summarize some general semantic constraints on lexicalization.

260

Xiufang Dong

There are mainly two semantic constraints on the lexicalization of phrases comprising lexical items. A. The components of a phrase that can undergo lexicalization are usually atypical members of the syntactic category they belong to. Many studies show that a concrete noun referring to a three-dimensional object is more typical than an abstract noun as far as the syntactic function is concerned, and a verb referring to physical action is more typical than a verb referring to abstract action. Those phrases that can undergo lexicalization often contain abstract nouns and/or abstract verbs. For example, most of the lexicalized verb-object phrases involve abstract nouns, such as the verb de´zuı` 得罪 ‘o¤end’, which was originally a verb-object phrase containing an abstract noun zuı` 罪 ‘crime’ and an abstract verb de´ 得 ‘obtain’. We can imagine that phrases like daˇ ta# 打他 ‘hit him’ would be nearly impossible, or at least very unlikely to undergo lexicalization. If those phrases that underwent lexicalization contain a concrete noun, the noun must be non-referential or generic, because if a noun has specific reference, its independent status as a noun is less likely to be lost, while losing independent word status is a requirement of lexicalization. If those phrases that underwent lexicalization contain a verb that denotes a concrete activity, the verb is either used metaphorically or metonymically, or refers to a general type of action rather than a specific action related to a specific time and location. For example, lexicalized da#qia´o 搭桥 means ‘establish a connection’, which is a metaphorical extension of the literal sense of ‘build a bridge’; lexicalized gua`pa´i 挂牌 means ‘to start business, to open a private practice’, which is a metonymical extension of the literal sense ‘to hang a sign’; lexicalized jiaˇnzhıˇ 剪纸 ‘paper cutting’ refers to a type of action rather than a specific event of cutting paper occurring at a specific time and in a specific location. B. The whole meaning of the phrase that can undergo lexicalization must be general to some degree. For example, jı´bı`ng 疾病 was derived from a coordinate noun phrase containing the noun jı´ 疾 ‘a slight illness’ and the noun bı`ng 病 ‘a severe disease’. The lexicalized item jı´bı`ng means ‘(all kinds of ) illness, disease’ which reflects a general concept. In the lexicalization of lexical phrases (especially coordinate phrases) we can often observe a process of ‘semantic generalization’. Lexicalization of functional phrases depends heavily on syntactic change, i.e. the decline of the function of the functional item. The semantic constraint here is that the whole meaning of the combination must be general,

Lexicalization in the history of the Chinese language

261

which also applies to other cases of lexicalization. For example, the lexicalization of suoˇyoˇu 所有 ‘all’ was related to the functional weakening of the functional word suoˇ 所. On the other hand, the verb yoˇu 有 has a general meaning which leads to the generality of the meaning of the whole combination. Lexicalization of syntactically unrelated word strings often involves a highly grammaticalized functional word which has the lightest semantic content, for example, yu´ 于 in zhı`yu´ 至于 ‘as far as’, hu# 乎 in jıhu# # 几乎 ‘almost’ , etc. The word string that can undergo lexicalization should be able to activate the meaning of the whole construction, as discussed in Section 2.3. 4.4. Pragmatic constraints High frequency is the fundamental constraint applied to all kinds of lexicalization. Like grammaticalization, lexicalization is a process of conventionalization, through which a repeatedly used form is ‘ritualized’ and obtains a fixed meaning. There are, however, exceptions to this generalization. We find some cases of lexicalization triggered by literary quotation. A phrase from the classics in Old Chinese might not have had a high frequency at the time, but afterwards it may have been quoted repeatedly because scholars of the past were usually very familiar with those classics, and as a result, the phrase could become fixed and lexicalized. For example, the word we`njın# 问津 ‘make inquiries’ originally was a verb phrase comprising a verb we`n 问 ‘ask’ and its nominal object jın# 津 ‘ferry’, which was first seen in Lunyu, a Confucian classic: (35) 长沮、桀溺耦而耕，孔子过之，使子路问津焉。 （《论语  微子》） Cha´ngju# Jie´nı` oˇu e´r ge#ng, Koˇngzıˇ guo` zhı,# Name Name pair and plough Confucius pass them shıˇ zıˇlu` we`n jın# ya#n. let Name ask ferry PART ‘Changju and Jieni ploughed together. Confucius passed by, and he let Zilu ask them about the ferry.’ (5th Century BC, Lunyu) In (35), we`n jı #n was just a freely combined verb-object phrase. Because of the core status of Lunyu as a Confucian classic, many scholars in later historical periods could recite it and they liked to quote the utterances in Lunyu (and other classics) when they wrote. We`njın# was quoted frequently and thus became lexicalized into a verb meaning ‘make inquiries’.

262

Xiufang Dong

Lexicalization of syntactically unrelated word strings can also be caused by creative literary quotation. For example, yoˇuyu´ 友于, as a word referring to ‘brothers’, was lexicalized from a syntactically unrelated word string, in which yoˇu 友 was a denominal verb meaning ‘be friendly (to somebody)’ and yu´ 于 a preposition introducing an NP. The combination was first observed in Shangshu, one of the thirteen classics, as shown in (36): 5 (36) 惟孝友于兄弟。（《尚书  君陈》） We´i xia`o yoˇu yu´ xio#ngdı`. PART filial piety friend PREP brother ‘Those who have the virtue of filial piety can be friendly to their brothers.’ (before 5th Century BC, Shangshu)

Scholars at later stages creatively used yoˇuyu´ to refer to xio#ngdı` 兄弟 ‘brothers’, which appeared close to yoˇuyu´ in the original text. This kind of usage was a witty demonstration of the education between scholars, and gradually the new meaning of yoˇuyu´ was conventionalized. This kind of lexicalization induced by literary quotation is very di¤erent from the ordinary type of lexicalization discussed earlier in this paper, in that the former was consciously created by language users, whereas the latter is a natural change that occurs over time and is unconscious to language users. Those ordinary processes of lexicalization must be based on frequently used forms, but lexicalization induced by literary quotation can be excepted from the high frequency constraint, since in this kind of lexicalization the selection of the input form is based on the fame of the original user or the work in which the form first appeared, rather than the frequency of use of the form.

5. Relation between syntactic change and lexicalization Lexicalization is a kind of change in which a regular, analyzable form becomes an irregular, unanalyzable form. The loss of the regularity and analyzability can be caused by relevant syntactic change. Based on abundant data in the history of the Chinese language, Dong (2009) points out the 5. The book Shangshu includes articles which were written at di¤erent times. Some articles were written in Old Chinese, and some were forged old texts. The article Junchen belongs to the latter category. However, the sentence in example (36) was also cited in Lunyu, so it should be an utterance in Old Chinese.

Lexicalization in the history of the Chinese language

263

close relation between some syntactic changes and lexicalization. In this section, we outline the impact that syntactic changes have on lexicalization. 5.1. Lexicalization related to changes of syntactic function of lexical words Lexical words may exhibit di¤erent types of syntactic behavior at di¤erent historical stages. Syntactic changes in lexical words can be reflected in their collocations with other words, and a possible result is that a specific kind of collocation disappears from the syntax. However, some instances of that kind of collocation can survive by way of entering the lexicon as a lexical item. Let us examine one syntactic change in nouns and its consequences on lexicalization. In Old Chinese, nouns could have an adverbial function and modify verbs. One kind of nominal adverbial which is obsolete in Modern Mandarin expressed a metaphorical manner. It is shown below: (37) 豕人立而啼。（《左传  庄公八年》） Shıˇ re´n lı` e´r tı´. pig human stand and cry ‘The pig stood up like a human and cried.’ (5th century BC, Zuozhuan) When this kind of ‘nominal adverbial þ verb’ combination disappeared at the syntactic level, some combinations survived and entered the lexicon. One such example is waˇjieˇ 瓦解 ‘disintegration, collapse’, whose internal structure is ‘N þ V’ (literally ‘tile þ unbind’), and it originally meant ‘disintegrate like tiles’. Another example is xiaˇngyı`ng 响应 ‘respond’, whose internal structure is ‘N þ V’ (literally ‘echo þ answer’), and it originally meant ‘respond (quickly) as an echo’. These forms survived in the lexicon because they were frequently used and often quoted by later generations of scholars. 5.2. Lexicalization related to syntactic changes of grammatical words When the function of a grammatical word weakens, the combination containing this grammatical word and a related lexical word can become lexicalized because it is no longer legitimate at the syntactic level. This kind of lexicalization has already been illustrated in previous discussions when we analyzed lexicalizations involving grammatical words.

264

Xiufang Dong

There is another kind of change in grammatical words that can also induce lexicalization. That is, a grammatical word originally obligatory in a specific construction is no longer needed. In other words, a grammatical word is lost from a specific construction. Lexicalization induced by this kind of change can be divided into two types. One type is the lexicalization of old combinations involving the grammatical word whose function is lost. The other type is the lexicalization of new constructions introduced after the loss of the grammatical word. Below, we will examine each type of lexicalization using examples documented in the history of the Chinese language. 5.2.1. Loss of e´r 而 ‘and’ in successive verbal constructions and lexicalization of some combinations involving this kind of e´r In Old Chinese, e´r was a widely used conjunction and it could appear between two verb phrases to express various semantic relations including coordination, succession, manner and action, among other things. In Modern Mandarin, the function of e´r has declined: it can no longer appear between two verb phrases to express successive relation. In other words, e´r in successive verbal constructions was lost over time, specifically around the time of Middle Chinese. For example, e´r in the following examples of Old Chinese cannot be used in equivalent sentences in Modern Mandarin: (38) 匠人斫而小之。（《孟子  梁惠王下》） Jia`ngre´n zhuo´ e´r xiaˇo zhı.# craftsman cut and small it ‘The craftsman cut it and made it small.’ (3rd Century BC, Mencius ) (39) 亡羊而补牢，未为迟也。（《战国策  楚策四》） Wa´ng ya´ng e´r buˇ la´o, we`i we´i chı´ yeˇ. lose sheep and repair sheepfold not be late PART ‘It is not too late to repair the sheepfold after losing sheep’ (3rd Century, Zhanguoce) After Middle Chinese, e´r no longer appeared in successive verbal constructions, and consequently some syntactically unrelated word strings involving e´r in successive verbal constructions became lexicalized. For example, e´rho`u 而后 ‘then’ is a word formed in this way. Originally, e´r ho`u appeared in successive verbal constructions, in which e´r was the conjunction between two VPs and ho`u ‘later’ was a temporal adverb that appeared in the second VP, as shown in (40):

Lexicalization in the history of the Chinese language

265

(40) 先号咷而后笑。（《易  同人卦》） Xia#n ha´ota´o e´r ho`u xia`o. first cry loudly and later laugh ‘First, cry loudly and later, laugh.’ (9th Century BC, Yijing) After the Western Han Dynasty (206BC–8AD), e´rho`u became lexicalized into a temporal adverb and obtained the meaning ‘then, later’. 5.2.2. Loss of yu´ 于/於 between verb and non-patient NPs and lexicalization of verb-object phrases In Modern Mandarin, many semantic roles can appear in the object position. In contrast, in Old Chinese, generally only the semantic role of the typical patient could serve as an object of verbs (Dong 2006b). Other semantic roles which can be realized as objects in Modern Mandarin had to be introduced by yu´ (written as 于 or 於) in Old Chinese. For example: (41) 国将兴，听于民。（《左传  庄公三十二年》） Guo´ jia#ng xıng, # tı #ng yu´ mı´n. country will thrive listen PREP people ‘When a country thrives, it listens to (the voice of ) its people.’ (5th century BC, Zuozhuan) (42) 三年无改於父之道，可谓孝矣。（《论语  学而》） Sa#n nia´n wu´ gaˇi yu´ fu` zhı # da`o, three year not change PREP father POSS way keˇ we`i xia`o yıˇ can call filial-piety PART ‘Not changing the father’s way for three years can be regarded as filial piety.’ (5th Century BC, Lunyu) The construction of ‘verb þ yu´ þ NP’ exemplified in (41) and (42) was not a verb-object phrase due to the existence of a preposition between the verb and the NP, but it can be expressed by ‘verb þ NP’ in Modern Mandarin. That is to say, in Modern Mandarin, NPs expressing semantic roles other than the typical patient can directly follow the verb to form verb-object phrases. This kind of verb-object phrase in Modern Mandarin has the characteristic of low transitivity, and is not typical compared with verb-object phrases containing a typical patient object. The loss of yu´ between the verb and non-typical patient NPs happened after the Han Dynasty (i.e. at the beginning of Middle Chinese). From Old

266

Xiufang Dong

Chinese to Middle Chinese, an obvious tendency was that more and more semantic roles could be tolerated in the object position. At the beginning, this may have been due to pragmatic reasons: the way to emphasize a nontypical patient NP is to place it in the object position. However, after repeated use, the pragmatic object can become a syntactic object. Those verb-object phrases that involve non-typical patient objects are potential candidates for lexicalization (Dong 2002). Therefore, it can be said that the loss of yu´ between a verb and a non-typical patient NP introduced more semantic relations into verb-object phrases, and consequently induced some lexicalizations of verb-object phrases that are low in transitivity. 5.3. Lexicalization related to word order change When word order changes, combinations conforming to the old word order rules cannot exist at the syntactic level, but some of them can survive at the lexical level as fossilized forms. Below, we examine two kinds of lexicalization that arose out of the change of word order in the history of the Chinese language. 5.3.1. Lexicalization related to disappearance of preverbal pronominal objects In Old Chinese, pronominal objects could precede verbs, while nominal objects could only follow the verb. After Middle Chinese, pronominal objects could no longer precede verbs but had to follow verbs instead, just like nominal objects. After this change of word order, some previous VP constructions containing preverbal pronominal objects did not disappear, but became lexicalized and were integrated into the lexicon as fossilized forms due to high frequency of use. For instance, zı` 自 ‘self ’ was a reflexive pronoun in Old Chinese, and it preceded the verb when functioning as an object. Some verb phrases involving the preverbal object zı` became lexicalized after the OV word order completely disappeared from the language. The word zı`sha# 自杀 ‘commit suicide’ is such an example. Its internal structure is ‘self þ kill’, and originally meant ‘kill oneself ’. 5.3.2. Lexicalization tendency of ‘verb þ preposition’ related to the word order change in PPs The prepositional phrase underwent a change of word order in the history of the Chinese language from predominantly postverbal to predominantly preverbal. There are still a few prepositional phrases that can appear in

Lexicalization in the history of the Chinese language

267

the postverbal position in Modern Mandarin, but semantically this category is very restricted. As a result, the verb and the following preposition have a strong tendency to undergo lexicalization. Although most of the combinations of ‘verb þ preposition’ cannot be regarded as lexical items at the present time, they can already be seen as incorporated words created on site in the sentence, as evidenced by the fact that the aspect marker le 了 can only appear after a ‘verb þ preposition’ combination instead of immediately following the verb: (43) 他睡在了沙发上。 Ta# shuı` za`i le sha#f a# sha`ng. he sleep at ASP sofa on ‘He slept on the sofa.’ In (43), the aspect marker le follows the preposition za`i 在 ‘at’ instead of immediately following the verb shuı` 睡 ‘sleep’. This can only be explained by assuming that shuı`zaı` acts like a verb in the sentence. Phonologically, the monosyllabic verb and the preposition following it form a prosodic word, and no pause is allowed between them. But in terms of semantics, the verb and the preposition are independent of each other in most cases, so we cannot say that the combination is already lexicalized. It is obvious that the tendency of the verb and the following preposition to become lexicalized is related to the change of word order in PPs: When most PPs appear in the preverbal position, those PPs that remain postverbal seem to be irregular at the syntactic level, resulting in the tendency to transfer into the lexicon. This kind of lexicalization is ongoing and we need to observe its future development. However, there are a few frequently used disyllabic forms of ‘verb þ preposition’ that can be said to have lexicalized to some degree, such as la´izı` 来自 ‘come from’, whose internal structure is ‘come þ PREP’, yua´nyu´ 源于 ‘to be derived from’, whose internal structure is ‘derive þ PREP’, etc. 5.4. Relation between syntax and lexicon Syntactic changes can induce lexicalization, but syntactic change is not a necessary condition for lexicalization. Some types of lexicalization take place without relevant syntactic change. From the types of lexicalization that are related to syntactic change, we can see the relation between syntax and lexicon. We have shown that a syntactic form can enter the lexicon after some historical change which provides evidence for the interrelationship between the subsystems of a language.

268

Xiufang Dong

Can lexicalization a¤ect syntactic change? The answer is likely to be yes. However, not much work has been done to explore this, and the issue is worthy of future inquiry.

6. Chinese typology and lexicalization Chinese is a typical isolating language. The typological characteristics of Chinese have some e¤ects on the types of lexicalization that occur in this language. The most evident phenomenon is that there is a kind of lexicalization involving grammatical words. In this kind of lexicalization, an originally grammatical word becomes an unanalyzable word-internal element. We saw this type of process when we discussed the lexicalization of functional phrases and the lexicalization of syntactically unrelated word strings. The occurrence of this kind of lexicalization is relevant to the characteristics of Chinese as an isolating language. In languages that have inflectional categories, a possible path for the change of a grammatical word is to become a clitic and further become an inflectional a‰x, as described in the grammaticalization cline proposed by Hopper and Traugott (1993: 7): content item > grammatical word > clitic > inflectional a‰x According to this cline, grammatical words can experience a process of morphologicalization. This kind of change is observed in languages with a rich morphology. However, there is no inflectional a‰x in Chinese, so the destination of grammatical words in Chinese is not to become morphological elements but rather to become fused, through lexicalization, into unanalyzable word-internal elements with their adjacent word. In other words, in Chinese the possible subsequent change for a grammatical word or clitic is lexicalization instead of further grammaticalization (Dong 2003, 2004a, 2004b; Wu 2005). To illustrate this point, we give another example here. There are some conjunctions and adverbs in Modern Mandarin involving shı` 是. Synchronically, shı` in these words cannot be analyzed, and it makes no semantic contribution to the whole word. But historically, shı` was a copular, and this kind of shı` is still used in contemporary Chinese. It was through lexicalization with the adjacent word that the independent copular shı` became a word-internal element. In some cases, the change went through a middle stage where shı` became a focus marker and consequently the focus marker shı` became word-internal. Let us take the lexicalization of ruo`shı` 若是 ‘if ’

Lexicalization in the history of the Chinese language

269

as an example. Ruo` shı` was not a single word but two independent words when it first appeared in historical texts where ruo` was a conjunction meaning ‘if ’ and shı` was a copula. For example: (44) 汝若是神，速听明教。（唐载孚《广异记  狄仁杰》） Ruˇ ruo` shı` she´n, su` tı #ng mı´ng jia`o. you if be god quickly listen clear teach ‘If you are a God, (let me) hear your instructions immediately.’ (8th Century, Guangyiji) By the Song Dynasty, ruo`shı` had become lexicalized, and shı` had lost its independence and become a meaningless word-internal element. After lexicalization, the meaning of ruo`shı` became roughly the same as that of ruo`. For example: (45) 若是三日而不活，则诚死矣。（宋曾敏行《独醒杂志》卷七） Ruo`shı` sa#n rı` e´r bu` huo´, ze´ che´ng sıˇ yıˇ. if three day and not alive then indeed dead PART ‘If after three days (he) does not revive, then (he) is certainly dead.’ (12th Century, Du Xing Za Zhi) Based on the data found in the history of the Chinese language, we conclude that the changing path of grammatical items is influenced by the typological features of the language. Another characteristic of lexicalization in Chinese is that the result of lexicalization tends to be a compound or word with no internal structure, instead of derivatives. In Chinese, when an independent word loses its independence and becomes a bound morpheme, it will usually maintain its phonological form and most of its lexical meaning, thus it retains the ability to combine with other morphemes in a rather free way, which is remarkably di¤erent from the situation in English (Packard 2000; Dong 2002). As a result, there are many root morphemes in Chinese, but derivational a‰xes are extremely rare and some so-called ‘derivational a‰xes’ are actually atypical compared with derivational a‰xes in languages with a rich morphology (Dong 2005). Therefore, becoming root compounds (i.e. compounds containing two root morphemes) is a natural destination of lexicalization in Chinese, resulting in the predominant status of compounds in the lexicon of Modern Mandarin and the rare occurrence of derivatives. A compound in Chinese can further turn into a simple word with no internal structure when one component totally loses its original meaning.

270

Xiufang Dong

7. Conclusion Lexicalization from a diachronic point of view can be regarded as a process that changes a non-word form (containing more than one word) into a word. The source form can vary: lexical phrases, constructions involving a grammatical word, syntactically unrelated word strings, etc. Lexicalization is a naturally occurring change in language, thus it is pervasive like grammaticalization, which can be seen very clearly from diachronic Chinese data. Like grammaticalization, lexicalization is a continuous process. Lexicalized forms can have di¤erent degrees of lexicalization. There are many constraints on the forms that can experience lexicalization, which strongly suggests that lexicalization is not a random phenomenon, and that there are regularities in its occurrence. Syntactic changes can be closely related to lexicalization. In some situations, lexicalization can be regarded as a consequent change of some syntactic change, and it accepts the illegitimate syntactic form as its input. Compared with grammaticalization, lexicalization might have more idiosyncratic and language specific features that are not, as yet, well understood by scholars. The relationship between the characteristics of lexicalization and language typology is an interesting topic that requires further examination. References Brinton, Laurel J. and Elizabeth Closs Traugott 2005 Lexicalization and Language Change. Cambridge: Cambridge University Press. Cheng, Xiangqing 1992 Lu`nhe´ng fu`yı #ncı´ ya´njiu#. In LiaˇngHa`n Ha`nyuˇYa´njiu# [Studies on Chinese in the Han Dynasty], Cheng, Xiangqing (ed.), Jıˇna´n: Sha#ndo#ng Jia`oyu` Chu#baˇnshe` [Shandong Educational Publisher]. Dong, Xiufang 1998 Shu` buˇ da`i bı #n ju`shı` zho#ng de yu`nlu`¨ zhı`yue # [The prosodic constraints on the Predicate-Complement-Object construction.] Yuˇya´n Ya´njiu# [Linguistics Study] 1, 55–62. Dong, Xiufang 2002 Cı´huı`hua`: Ha`nyuˇ Shua#ngyı #ncı´ de Yaˇnshe #ng he´ Fa#zhaˇn [Lexicalization: The Origin and Development of Chinese Disyllabic Words]. Chengdu: Sı`chua#n Mı´nzu´ Chu#baˇnshe` [Sichuan Nationality Publisher].

Lexicalization in the history of the Chinese language Dong, Xiufang 2003

Dong, Xiufang 2004a

Dong, Xiufang 2004b Dong, Xiufang 2005

Dong, Xiufang 2006a

Dong, Xiufang 2006b

Dong, Xiufang 2007a

Dong, Xiufang 2007b

Dong, Xiufang 2009

Fang, Xi 1999

271

Lu`n ‘X zhe’ de cı´huı`hua` [On lexicalization of the structure ‘X zhe 着’]. Yuˇya´nxue´ Lu`nco´ng [Essays on Linguistics] 28, 138–151. Beijing: Sha#ngwu` Yı`nshu#guˇan [The Commercial Press]. Shı` de jı`nyı #bu` yuˇfaˇhua` [Further grammaticalization of shı` 是: From functional word to word-internal element]. Da#ngda`i Yuˇya´nxue´ [Contemporary Linguistics] 1, 35–44. Ha`nyuˇ de Cı´ku` yuˇ Cı´faˇ [The Chinese Lexicon and Morphology]. Beijing: Beˇijı #ng Da`xue´ Chu#baˇnshe` [Peking University Press]. Ha`nyuˇ cı´zhuı` de xı`ngzhı` yuˇ ha`nyuˇ cı´faˇ te`diaˇn [The properties of Chinese a‰xes and the characteristics of Chinese morphology]. Ha`nyuˇ Xue´xı´ [Chinese Language Learning] 6, 13–19. Cı´huı`hua` yuˇ yuˇfaˇhua` de lia´nxı` yuˇ qu#bie´: yıˇ ha`nyuˇshıˇ zho#ng de yı #xie # cı´huı`hua` we´i lı` [Comparing lexicalization with grammaticalization: Taking some lexicalizations that occurrred in the history of the Chinese language as examples.] Ershı´yı # Shı`jı` de Zho#ngguo´ Yuˇya´nxue´ (e`r) [Chinese Linguistics in the 21st century (second volume.)] Beijing: Sha#ngwu` Yı`nshu#guˇan [The Commercial Press]. Guˇ ha`nyuˇ zho#ng do`ng mı´ng zhı #jia#n ‘yu´’ de go#ngne´ng za`i re`nshı´ [Reconsidering the function of yu´ between a verb and a noun in Old Chinese]. Guˇha`nyuˇ Ya´njiu# [Research in Ancient Chinese Language] 2, 2–8. Co´ng cı´huı`hua` de jiaˇodu` ka`n nia´nhe´shı` do`ngbuˇ jie´go`u de xı`ngzhı` [On the property of resultative constructions: From the perspective of lexicalization]. Yuˇya´n Ke#xue´ [Linguistic Sciences] 1, 40–47. Cı´huı`hua` yuˇ hua`yuˇ bia#ojı` de xı´ngche´ng [Lexicalization and the origin of discourse markers]. Shı`jie` Ha`nyuˇ Jia`oxue´ [Chinese Teaching in the World] 1, 50–61. Ha`nyuˇ de ju`faˇ yaˇnbia`n yuˇ cı´huı`hua` [Syntactic change and lexicalization in Chinese]. Zho#ngguo´ Yuˇwe´n [Chinese Language] 5, 399– 409. Yoˇudı`ng yuˇ xia`ngxı #n jie´go`u de yuˇxu` [Definiteness and the word order of endocentric structures]. Yuˇwe´n Ya´njiu# [Linguistic Researches] 1, 1–8.

272

Xiufang Dong

Feng, Shengli 1997

Ha`nyuˇ de Yu`nlu`¨ , cı´faˇ yuˇ ju`faˇ [Interactions between Morphology, Syntax and Prosody in Chinese]. Beijing: Beˇijı #ng Da`xue´ Chu#baˇnshe` [Peking University Press]. Goldberg, Adele Elizabeth 1995 A Construction Grammar Approach to Argument Structure. Chicago: The Chicago University Press. Halliday Michael. A. K. 1985 An Introduction to Functional Grammar. London: Edward Arnold. Himmelmann, Nikolaus P. 2004 Lexicalization and grammaticization: Opposite or orthogonal? In What Makes Grammaticalization?: A Look from its Fringes and its Components, Bisang, Walter, Nikolaus P. Himmelmann, Bjo¨rn Wiemer (eds.), Berlin & New York: Mouton de Gruyter. Hopper, Paul, and Elizabeth Closs Traugott 1993 Grammaticalization. Cambridge: Cambridge University Press. Jiang, Shaoyu 1996 Jı`nda`i Ha`nyuˇ ya´njiu# ga`ikua`ng [Outlines of Studies on Early Modern Chinese]. Beijing: Beˇijı #ng Da`xue´ Chu#baˇnshe` [Peking University Press]. Lapointe, Steven G. 1979 A lexical analysis of the English auxiliary verb systems. In Lexical Grammar, Teun Hoekstra, H. van der Hulst, and M. Moorgat (eds.), Dordrecht: Foris. Lehmann, Christian 1995 [1982] Thoughts on Grammaticalization (LINCOM Studies in Theoretical Linguistics 1). Munich: LINCOM EUROPA. Lehmann, Christian 2002 New reflections on grammaticalization and lexicalization. In New Reflections on Grammaticalization – Proceedings from the International Symposium on Grammaticalization, Wischer, Ilse and Gabriele Diewald (eds.), 1–18. Amsterdam/Philadelphia: Benjamins. Lipka, Leonhard 2002 [1990] English Lexicology: Lexical Structure, Word Semantics and Word-Formation. Tu¨bingen: Max Niemeyer Verlag, 3rd revised edition of An Outline of English Lexicology. Lu, Bingfu 1988 Dı`ngyuˇ de wa`iya´nxı`ng, ne`iha´nxı`ng he´ che #ngwe`ixı`ng jı´qı´ shu`nxu` [The nominal modifier’s properties in denotation, connotation, denomination and their word order]. Yuˇfaˇ Ya´njiu# yuˇ Ta`nsuˇo (sı` ) [Studies and Explorations in Grammar, Vol. 4]. Beijing: Beˇijı #ng Da`xue´ Chu#baˇnshe` [Peking University Press]. Lu, Jianming 1990 ‘VA le’ shu`buˇ jie´go`u yuˇyı` fe #nxı # [Semantic analysis of the construction of VA le]. Ha`nyuˇ Xue´xı´ [Chinese Language Learning] 1, 1–6.

Lexicalization in the history of the Chinese language Lu, Zhiwei 1957 Ma, Qingzhu 1998

273

Ha`nyuˇ de go`ucı´faˇ [Chinese Word-Formation Rules]. Beijing: Ke #xue´ Chu#baˇnshe` (Science Press).

Zı`zhuˇ do`ngcı´ he´ fe#izı`zhuˇ do`ngcı´ [Volitional verbs and involitional verbs]. Zho#ngguo´ Yuˇya´n Xue´ba`o [Journal of Chinese Linguistics] 3, 157–180. McCarthy, John, and Alan Prince 1990 Foot and word in prosodic morphology: The Arabic broken plural. Natural Language and Linguistic Theory 8, 209–283. McCarthy, John, and Alan Prince 1993 Prosodic Morphology I: Constraint interaction and satisfaction. MS., University of Massachusetts at Amherst and Rutgers University, New Brunswick, HJ. Moreno Cabrera, Juan C. 1998 On the relationships between grammaticalization and lexicalization. In The Limits of Grammaticalization, Giacalone Ramat, Anna, and Paul J. Hopper (eds.), Amsterdam: Benjamins. Packard, Jerome L. 2000 The Morphology of Chinese: A Linguistic and Cognitive Approach. Cambridge: Cambridge University Press. Ramat, Paolo 1992 Thoughts on degrammaticalization. Linguistics 30: 549–560. Ramat, Paolo 2001 Degrammaticalization or transcategorization? In Naturally! Linguistic Studies in Honor of Wolfgang Ulrich Dressler Presented on the Occasion of his 60th Birthday, Chris Schaner-Wolles, John Rennison, and Friedrich Neubarth (eds.), 393–401. Torino: Rosenbach and Sellier. Talmy, Leonard 1985 Lexicalization patterns: Semantic structure in lexical forms. In Language Typology and Syntactic Description: Vol. 3, Grammatical Categories and the Lexicon, Shopen, T. (ed.), Cambridge: Cambridge University Press. Talmy, Leonard 2000 Toward a Cognitive Semantics, 2 Vols. Cambridge: MIT Press. Traugott, Elizabeth Closs 1989 On the rise of epistemic meanings in English: An example of subjectification in semantic change. Language 65, 31–55. van der Auwera, Johan 2002 More thoughts on degrammaticalization. In New Reflections on Grammaticalization – Proceedings from the International Symposium on Grammaticalization, Wischer, Ilse and Gabriele Diewald (eds.), 19–29. Amsterdam/Philadelphia: Benjamins.

274

Xiufang Dong

Wei, Peiquan 2003

Wu, Fuxiang 2005

Yao, Zhenwu 1997

Sha`ngguˇ ha`nyuˇ da`o zho#ngguˇ ha`nyuˇ yuˇfaˇ de zho`ngya`o fa#zhaˇn [Important syntactic changes from Old Chinese to Middle Chinese]. In Guˇ Jın# To#ng Sa`i: Ha`nyuˇ de Lı`shıˇ yuˇ Fa#zhaˇn. [Historical Development of the Chinese Language], He Da’an (ed.), Taipei: Institute of Linguistics, Academia Sinica. Ha`nyuˇ yuˇfaˇhua` yaˇnbia`n de jıˇ ge` le`ixı´ngxue´ te`zhe #ng [Some typological features of grammaticalization in Chinese]. Zho#ngguo´ Yuˇwe´n [Chinese Language] 6, 483–494. Yıˇwe´i de xı´ngche´ng jı´ xia#nggua#n we`ntı´ [The origin of yıˇwe´i and some related issues]. Guˇha`nyuˇ Ya´njiu# [Research in Ancient Chinese Language] 3, 25–32.

Yin, Guoguang 2006 ‘Suoˇ’ zı` jie´go`u de zhuaˇnzhıˇ duı`xia`ng yuˇ do`ngcı´ peı`jia` – Zhua#ngzıˇ suoˇ zı` jie´go`u de kaˇocha´ [The semantic reference of the Suoˇ construction and the valence of the verb: Study on suoˇ constructions in Zhuangzi ]. Yuˇya´n Ya´njiu# [Linguistics Study] 3, 30–36. Zhu, Dexi 1983 Zı`zhıˇ he´ zhuaˇnzhıˇ – ha`nyuˇ mı´ngcı´hua` bia#ojı` de, zheˇ, suoˇ, zhı # de yuˇfaˇ go#ngne´ng he´ yuˇyı` go#ngne´ng [Self-designation and otherdesignation: Grammatical function and semantic function of the nominalization marker de, zhe, suoˇ, zhı # in Chinese]. Fa#ngya´n [Dialact] 3, 16–31.

Argument structure change, reanalysis and lexicalization: Grammaticalization of transitive verbs into ditransitive verbs in Chinese, Japanese and English Feng-fu Tsao 1. Introduction Verbs are traditionally categorized in accordance with the number of arguments they necessarily entail. Intransitive verbs such as walk, sleep, and fly take only one argument. Actions depicted by intransitive verbs can be accomplished by a single agent nominal; no other participants are obligatory. Transitive verbs such as tell (the story) and bake (a cake), however, should include two participants in the activity depicted by the verb. If there is only one nominal available in the sentence, the action of telling and baking remains incomplete. We need another argument, either AGENT or THEME, to complete the argument structure. There are also verbs that require three nominals in a single sentence to satisfy their semantic needs. These predicates, such as send and give, are realized as ditransitive verbs, and, in addition to the subject, two nouns should follow in the object positions. The nominal which is semantically closer to the verb functions as the direct object and the one semantically further from the verb functions as the indirect object, specifying the GOAL to which the THEME is transferred. Due to the complexity of the argument structure of ditransitive verbs, there are two possibilities in English for arranging the required object nominals. With the verb send, for example, the direct object can either precede the indirect object, which is itself preceded by the preposition to, or the indirect object may directly follow the predicate and be followed by the direct object. In this case, as long as the indirect object is strictly adjacent to the predicate, the preposition introducing the indirect object will be systematically absent. (1a) and (1b) show the two possibilities. (1) a. b.

Michael sent a birthday gift to his daughter last year. Michael sent his daughter a birthday gift last year.

In addition to three-place ditransitive predicates, however, there is a large number of two-place predicates that also include an additional nominal in

276

Feng-fu Tsao

their structures and consistently exhibit the same structural variation as the examples in (1). For instance, bake, which is traditionally categorized as a two-place predicate, can also behave like a ditransitive verb. The examples in (2) illustrate this kind of behavior. (2) a. b.

Mom baked a chocolate cake for me. Mom baked me a chocolate cake.

In (2a), bake is followed by a prepositional phrase indicating the recipient of the action. Even though the predicate bake, unlike the verb send, does not imply transfer, it can still occur in the double object construction as illustrated in (2b). The same phenomenon also exists in Japanese and in Mandarin Chinese. In Japanese, some transitive verbs such as katte ‘buy’ can appear in the following two types of sentences, which express roughly the same meaning.1 (3) a.

b.

taro-wa hanako-ni doresu-o katte agera Taro-NOM Hanako-DAT dress-ACC buy give ‘Taro bought a dress and gave it to Hanako. / Taro bought a dress to give to Hanako.’ taro-wa hanako-ni doresu-o katta Taro-NOM Hanako-DAT dress-ACC buy ‘Taro bought Hanako a dress.’

In Mandarin Chinese a transitive verb may include an additional noun phrase as the recipient of the THEME. Li and Thompson (1981) list a number of verbs which systematically accept the special orderings of the objects in a sentence where the direct object may either follow or precede the indirect object. Yet, the arrangement of noun phrases in Chinese is somewhat more puzzling than the corresponding structural variations in English. In Chinese, when the indirect object directly follows the predicate, the presence of the coverb introducing the indirect object must be taken into consideration. In some cases, the coverb must be present. In other cases, the presence or absence of the coverb is semantically immaterial, and in yet another case, in the presence of certain predicates there must be no coverb at all. The examples in (4), (5), and (6) show these irregularities respectively. 1. This is just a rough characterization of the process involved in the change. A more detailed characterization of the process and the constraints involved will be given in a later section.

Argument structure change, reanalysis and lexicalization

277

(4) a. 他帶一包糖給張三。 ta# da`i yı` ba#o ta´ng geˇi Zha#ngsa#n 3sg bring one bag candy to Zhangsan ‘S/he brought a bag of candy to Zhangsan.’ b. 他帶給張三一包糖。 ta# da`i geˇi Zha#ngsa#n yı` ba#o ta´ng 3sg bring to Zhangsan one bag candy ‘S/he brought a bag of candy to Zhangsan.’ (5) a. 我送一瓶酒給他。 woˇ so`ng yı` pı´ng jiuˇ geˇi ta# I give one bottle wine to 3sg ‘I gave a bottle of wine to him/her.’ b. 我送(給)他一瓶酒。 woˇ so`ng (geˇi) ta# yı` pı´ng jiuˇ I give to 3sg one bottle wine ‘I gave a bottle of wine to him/her.’ (6) a.

*他搶了兩萬塊錢給銀行。 *ta# qiaˇng le liaˇng wa`n kua`i qia´n geˇi yı´nha´ng 3sg rob PFV two ten thousand dollar money to bank ‘S/he robbed $20,000 from the bank.’

b. 他搶了銀行兩萬塊錢。 ta# qiaˇng le yı´nha´ng liaˇng wa`n kua`i qia´n 3sg rob PFV bank two ten thousand dollar money ‘S/he robbed $20,000 from the bank.’2 From the examples in (6) above, it is clear that the coverb need not be present when the indirect object immediately follows the predicate. Moreover, we detect that the presence of the coverb is not allowed in (6a), where the indirect object follows the direct object. This peculiarity reveals that Chinese is not only di¤erent from English in terms of the necessity of a coverb in the double object construction but also exhibits di¤erent conditions which govern the presence or absence of the coverbs. What is the semantic requirement for the development of transitive verbs into their ditransitive counterparts in the three languages? How did the double object construction develop, and why does the coverb behave 2. (6a) is grammatical if it is taken in the sense of her robbing someone of $20,000 and giving it to the bank.

278

Feng-fu Tsao

asymmetrically in Mandarin, but systematically in English and in Japanese? These are issues that will be addressed in this discussion. But before moving on to our main analysis, a review of the chapter on double object constructions in Li and Thompson’s Chinese Grammar will be given to provide som background. 2. Ditransitive Constructions in Chinese In their grammar book, Li and Thompson (1981) have attempted to explain the asymmetry of the coverb inside the double object constructions in Mandarin. In this paper, their account of the distribution of geˇi will be presented first. Next, their explanations for the distribution of the indirect object marker geˇi will be examined and potential problems in their analysis will be addressed. And, lastly, their analysis of how transitive verbs evolve to perform ditransitively will be explained. 2.1. Distribution of the coverb geˇi 2.1.1. Li and Thompson’s (1981) account In Chinese, when an indirect object is included in a sentence, consideration must be given as to whether geˇi is required. The verbs in Table 1 all require that geˇi be placed before the indirect object when the indirect object is preceded by the verb. Table 1. Verbs with the obligatory geˇi 3 Mandarin

Gloss

Mandarin

Gloss

dı` 遞

‘bring to’

diu#, re#ng 丟, 扔

‘toss, throw’

fe #n 分 na´, da`i 拿, 帶 jı` 寄

‘allocate’

shu# 輸 xieˇ 寫

‘lose’

‘bring to’ ‘mail’ ‘deliver, hand in’

zu# 租 liu´ 留

‘rent’ ‘keep, save’

‘kick’

ba#n 搬

‘move’

‘sell’

tue# 推

‘push’

jia#o 交 tı # 踢 ma`i 賣

‘write’

3. The boldfaced verbs in Tables 1, 2, 3 are those that are in one sense or another misplaced. We will explain why they are regarded as misplaced in a later discussion.

Argument structure change, reanalysis and lexicalization

279

The following sentences exemplify the ordering variations of the object nominals. The examples in (7) demonstrate that if the indirect object precedes the direct object, the coverb geˇi must be present; otherwise, the sentence would be ungrammatical, as shown in (7b). (7) a. 他丟了一包糖給張三。 ta# diu# le yı` ba#o ta´ng geˇi Zha#ngsa#n 3sg throw PFV one bag candy to Zhangsan ‘S/he threw a bag of candy to Zhangsan.’ b. 他丟*(給)了張三一包糖。 ta# diu# *(geˇi) le Zha#ngsa#n yı` ba#o ta´ng 3sg throw to PFV Zhangsan one bag candy ‘S/he threw a bag of candy to Zhangsan.’ Table 2 displays verbs allowing an optional geˇi in the double object construction and the following examples illustrate the characteristics of these non-compulsory coverbs. Table 2. Verbs with optional geˇi Mandarin

Gloss

Mandarin

Gloss

hua´n 還 pe´i 賠

‘return’

fu` 付 xuˇ 許 jie` 借

‘pay’

so`ng, ze`ng 送, 贈

‘send’

jia#o 教 shaˇng, cı` 賞, 賜

‘teach’

jia# 加 chua´n 傳

‘add on’ ‘pass’

‘bestow’

‘compensate, pay back’ ‘promise to give’ ‘lend’

(8) a. 我送了一瓶酒給他。 woˇ so`ng le yı` pı´ng jiuˇ geˇi ta# I give PFV one bottle wine to 3sg ‘I gave a bottle of wine to him/her.’ b. 我送(給)了他一瓶酒。 woˇ so`ng (geˇi) le ta# yı` pı´ng jiuˇ I give to PFV 3sg one bottle wine ‘I gave a bottle of wine to him/her.’ In (8b) we notice that the presence or absence of the coverb does not a¤ect the grammaticality of the sentence. To put it di¤erently, the coverb is optional when the indirect object is directly preceded by a verb of this category.

280

Feng-fu Tsao

Table 3. Verbs barring geˇi 4 Mandarin

Gloss

Mandarin

Gloss

geˇi 給 duo´ 奪

‘give’

to#u 偷 qiaˇng 搶

‘steal’

‘snatch’

‘rob’

Table 3 presents verbs that do not need the coverb geˇi in front of the indirect object when the indirect object is followed by the direct object nominal. The sentences in (9) are examples where geˇi is not allowed. (9) a.

*他搶了兩萬塊錢(*給)銀行。 ta# qiaˇng le liaˇng wa`n kua`i qia´n (*geˇi) yı´nha´ng 3sg rob PFV two ten thousand dollar money to bank ‘S/he robbed $20,000 from the bank.’

b. 他搶了(*給)銀行兩萬塊錢。 ta# qiaˇng le (*geˇi) yı´nha´ng liaˇng wa`n kua`i qia´n 3sg rob PFV to bank two ten thousand dollar money ‘S/he robbed $20,000 from the bank.’ Regardless of whether the direct object precedes or follows the indirect object(s), the presence of the coverb geˇi is prohibited with the verbs listed in Table 3. These verbs di¤er from the verbs of the foregoing two categories, where the presence of the coverb with the indirect object at the end of the sentence is not problematic. Briefly, according to Li & Thompson (1981) the presence or absence of the Chinese coverb geˇi is determined by the verb and the ordering of the two objects in the double object construction, but the use of the coverb is arbitrary. Thus, the verb (with or without the coverb under the structure in question) must be acquired by memorization. This conclusion is counterintuitive, to say the least and, after close examination, we have come to a completely di¤erent conclusion. 2.1.2. Inadequacies in Li and Thompson’s explanation for the indirect object facts The coverb geˇi in Li and Thompson’s analysis is considered to be closely related to the verb geˇi (give) in terms of its semantic interpretation. The 4. Table 3 as originally given by Li & Thompson (1981) includes some verbs of communication. As these verbs involve transfer of abstract entities instead of concrete objects, they are omitted in the present discussion.

Argument structure change, reanalysis and lexicalization

281

verb geˇi singles out the sense of transaction, as does its functional counterpart. Due to the reading of the verb geˇi, the coverb geˇi, which inherits the interpretation of its verbal equivalent, must appear with predicates that depict actions of transfer. The goal of the transaction named by the verb is marked by the coverb. Yet, there is no way to distinguish verbs of different categories: those verbs with an obligatory geˇi, an optional geˇi or without geˇi. Evidently the meaning of geˇi is not the only factor that determines its distribution. To illustrate this, let us begin with Table 3, which comprises only two classes of verbs, i.e. verbs of taking and verbs of giving. Sentences involving a verb of taking such as qiaˇng ‘rob’, to#u ‘steal’ and duo´ ‘snatch’ will have a goal NP as their subject and a source NP as their indirect object. As it is generally assumed that geˇi ‘give’ is a goal marker, and since a subject in Chinese, as I have repeatedly pointed out elsewhere (Tsao 1979, 1988, 1990), is always unmarked, there is no place for the geˇi coverb to appear in this type of sentence. Li and Thompson have correctly described and demonstrated this in (9). But they fail to give a satisfactory account for the absence of the geˇi coverb in the case of a sentence that has a verb of transfer, like geˇi ‘give’. In this construction a sentence is interpreted as having a source subject and a goal indirect object, as shown in (10). (10) a. 他給妹妹一塊田。 ta# geˇi me`imei yı´ kua`i tia´n he give younger sister one piece rice paddy ‘He gave his younger sister a portion of the rice paddy.’ *b. 他給一塊田給妹妹。 ta# geˇi yı´ kua`i tia´n geˇi me`imei he give one piece rice paddy to younger sister ‘He gave a portion of the rice paddy to his younger sister.’ Upon comparison with other Chinese dialects, such as Hakka, it seems clear that there is an idiosyncratic constraint in Mandarin Chinese that bars a coverb derived from a verb from having the same etymon to mark the indirect object in the same ditransitive sentence. The constraint has been termed haplology by Chao (1968). Looking back at Table 1, the boldfaced verbs are regarded as misplaced for di¤erent reasons. Fe`n ‘allocate’ and zu# ‘rent’ are misplaced because they belong to a class called bi-directional verbs. Their directionality of transaction is unmarked, i.e. zu# ‘rent’ can be used to mean ‘rent from someone’ or ‘rent to someone’, as shown in (11a) and (11b) respectively.

282

Feng-fu Tsao

(11) a. 他向我租一部車。 ta# xia`ng woˇ zu# yı´ bu` che# he from me rent one CL car ‘He rented a car from me.’ b. 他租一部車給我。 ta# zu# yı´ bu` che# geˇi woˇ he rent one CL car to me ‘He rented a car to me.’ In this case, the presence or absence of geˇi ‘give’ as a coverb determines the meaning of the main verb. Hence, its presence is not always required. Fe`n ‘allocate’ exhibits the same distributional pattern. In the case of ma`i ‘sell’ and shu# ‘lose’, since they both belong to the class of transactional verbs, their meanings strongly imply a goal (recipient) and hence they should be more properly classified as belonging to the class B verbs listed in Table 2. With all these misplaced verbs removed, let us take a closer look at the remaining verbs. All of them exhibit two usages, i.e. they can be used both transitively and ditransitively, as the following examples show: (12) a. 他寫了一封信。 ta# xieˇ le yı` fe#ng xı`n he write PFV one CL letter ‘He wrote a letter.’ b. 他寫給了我一封信。 ta# xieˇ geˇ le woˇ yı` fe #ng xı`n he write GEI PFV me one CL letter ‘He wrote me a letter.’ In addition, these verbs all fall naturally into three classes namely, verbs of movement, verbs of creation and verbs of acquisition. These verbs are listed below with some more relevant verbs added.5

5. Some of the verbs listed here admit cross-classification. Tou ‘steal’ and qiang ‘rob’ are both verbs of deprivation and verbs of acquisition, while jie ‘borrow/ lend’ and zu ‘rent’ are both bi-directional verbs of transaction and verbs of acquisition.

Argument structure change, reanalysis and lexicalization

283

Class I: Verbs of movement dı` (遞) ‘bring to’; jı` (寄) ‘mail’; jia#o (交) ‘deliver, hand in’; diu#, re#ng (丟, 扔) ‘toss, throw’; ba#n (搬) ‘move’; tue# (推) ‘push’; na´, ta`i (拿,帶) ‘bring to’ etc. Class II: Verbs of acquisition liu´ (留) ‘keep, save’; sı` (賜) ‘bestow’; maˇi (買) ‘buy’; to#u (偷) ‘steal’; qiaˇng (搶) ‘rob’; jie` (借) ‘borrow/lend’; zu# (租) ‘rent’ Class III: Verbs of creation xieˇ (寫) ‘write’; hua` (畫) ‘draw, paint’; za`o (造) ‘make, build’; ho#ng (烘) ‘bake’; ga`i (蓋) ‘build’ Moving on to Table 2, we find two boldfaced verbs: jia#o ‘teach’ and jie` ‘lend/borrow’. Jie` ‘lend/borrow’ should be excluded because, like fe#n ‘allocate’ and zu# ‘rent’, it is a bi-directional verb of transaction which in Chinese admits of two interpretations, ‘borrow’ and ‘lend’. Only when it is used in the sense of ‘lend’ does it take a goal/recipient role, so for the present analysis we will exclude it from consideration. Jia#o ‘teach’ will also be disregarded, since what is transferred in this case is something abstract, and jia#o ‘teach’ as a ditransitive verb exhibits some idiosyncratic properties which may complicate our analysis here.6 Now with the two improperly placed verbs removed, we can easily see that the remaining verbs are all verbs of transfer which share a distinctive feature, namely that their meanings strongly imply the existence of a goal/ recipient role, i.e. when you hua´n ‘return’, it implies the existence of a recipient to whom you are returning something. Similarly, when an emperor says cı`zuo` ‘bestow seat’, he is addressing a certain individual who is the designated recipient of the seat. 2.1.3. Our account With all the problematic cases removed, we can now make the following three generalizations: (1) Table 1 contains only transitive verbs of acquisition, verbs of movement, and verbs of creation; (2) Table 2 contains only verbs of transfer which imply the existence of a goal/recipient role; and finally, (3) Table 3 is comprised of verbs of deprivation which do not syntactically subcategorize a goal/recipient NP. With these classes of verbs properly set up, we can now give a reasonable account of why in Table 2, geˇi is optional while in Table 3 it is obligatorily absent. The goal/recipient marker geˇi is optionally present because there is no need to mark the con6. See Tsao (1988) for an attempt to characterize some of its idiosyncratic properties.

284

Feng-fu Tsao

cerned NP as goal/recipient since the meaning of this class of verbs implies its existence. In Table 3 the goal marker geˇi is absent because no goal/ recipient NP exists in the subcategorization frame of the verb. The reason for the obligatory presence of geˇi as a goal/recipient marker in Table 1 is more complicated, as it involves the process of category shift, reanalysis and lexicalization. A full account of this will be given in the next section.7 2.2. Grammaticalization of transitive verbs into ditransitive verbs As far as we can determine, there are three grammatical processes involved in turning a transitive verb into a ditransitive one: (1) goal phrase fronting, (2) reanalysis and (3) lexicalization. We will discuss each in detail in this section. In addition, there are two issues that we need to look into in this grammaticalization process. The first involves the status of the goal phrase and the second has to do with the motivation for the fronting process. In the following section we will discuss each of these as well. 2.2.1. The grammatical status of the goal phrase In Chinese, the grammatical status of the goal phrase in a construction with a Class I ditransitive verb (i.e. a verb of movement), as in (13a), is generally analyzed as an adjunct phrase, much like its corresponding phrase in an English ditransitive sentence such as (14a). (13) a. 李四扔一根骨頭給汪汪。 Lıˇsı` re#ng yı` ge#n guˇto´u geˇi Wa#ngwa#ng Lisi throw one CL bone to Wangwang ‘Lisi threw a bone to Wangwang.’ (14) a.

John brought a cake for Mary.

As simple as this looks, on closer examination it becomes clear that this is not the only possible analysis. In fact, in both languages, the goal phrase can be followed by another verb as shown in (13b) and (14b). (13) b. 李四扔一根骨頭給汪汪當點心。 Lıˇsı` re#ng yı` ge#n guˇto´u geˇi Wa#ngwa#ng da#ng diaˇnxın# Lisi throw one CL bone for Wangwang serve-as snack ‘Lisi threw a bone for Wangwang to serve as a snack.’ (14) b.

John brought a cake for Mary to take along on her hiking trip.

7. Liu (2006) has more or less come to the same conclusion on the basis of semantic considerations.

Argument structure change, reanalysis and lexicalization

285

When this situation arises, we are tempted to claim that in Chinese, geˇi serves as the second verb in a serial verb construction. But a deeper observation reveals that it serves as a preposition governing a PP which works as an optional adjunct in the sentence when it is not followed by any other verb as in (13a). Huang and Mo (1992) dubbed this a ‘‘mixed’’ analysis and then went on to challenge it saying that a major di‰culty lies in the fact that the verb occurring after the phrase geˇi Wa#ngwa#ng is optional in some contexts. (15) 李四扔一根骨頭給汪汪(吃)。 Lıˇsı` re#ng yı` ge#n guˇto´u geˇi Wa#ngwa#ng (chı )# Lisi throw one CL bone GEI Wangwang (eat) ‘Lisi threw a bone for Wangwang (to eat).’ Huang and Mo (1992) have presented an elaborate argument against the ‘mixed’ analysis and, on the basis of other arguments in its favor, they have posited what they called a unified account of serial verb construction. We feel, however, that such a unified account is not necessarily the best account. If we look at (13a), (13b) and (15) again from the perspective of the grammaticalization process, we find a chain of grammaticalization from discourse to syntax with (15) serving as an intermediary link between them. Before explaining in detail what is meant by an intermediary link, there is one generalization that must be addressed with regard to the optional verb occurring after the geˇi phrase. As pointed out in Tsao (1989), the verb in question can be omitted only when its meaning can be predicted on the basis of the meaning obtained from the previous clause. A few more examples will make this clear. (16) a. 張三寄一包粽子給李四吃。 Zha#ngsa#n jı` yı` ba#o zo`ngzi geˇi Lıˇsı` (chı )# Zhangsan send one CL rice-dumpling GEI Lisi (eat) ‘Zhangsan sent a package of rice-dumplings for Lisi (to eat).’ b. 張三寄一包粽子給李四轉交王五。 Zha#ngsa#n jı` yı` ba#o zo`ngzi geˇi Lıˇsı` zhuaˇng jia#o Zhangsan send one CL rice-dumpling GEI Lisi transfer to Wa´ngwu˘ Wangwu ‘Zhangsan sent a package of rice-dumplings for Lisi to give to Wangwu.’

286

Feng-fu Tsao

(17) a. 張三蓋一棟房子給李四住。 Zha#ngsa#n ga`i yı´ do`ng fa´ngzi geˇi Lıˇsı` (zhu`) Zhangsan build one CL house GEI Lisi (live) ‘Zhangsan built a house for Lisi (to live in).’ b. 張三蓋一棟房子給李四賣。 Zha#ngsa#n ga`i yı´ do`ng fa´ngzi geˇi Lıˇsı` ma`i Zhangsan build one CL house GEI Lisi sell ‘Zhangsan built a house for Lisi to sell.’ It is easily inferred that the most obvious reason for ‘sending someone a package of rice-dumplings’ is so that someone can eat them, and the most obvious reason for ‘building a house’ is so that someone can live in it. It is because of this prominence along with the generally expected collocation between the action and the purpose that makes the presence of a verb denoting the purpose optional. One could, of course, think of many other purposes like ‘‘to be transferred to someone else’’, or ‘‘to sell a house’’ rather than ‘‘to eat in it’’ or ‘‘to live in it’’ but they are all non-primary or unpredictable purposes and for that reason, have to be specified in Chinese.8 With the optionality of the verb explained, let us return to our proposed analysis. It says that, when the optional verb is present as in (16b) and (17b), we have a serial verb construction where geˇi serves as the second verb in the series, but when the optional verb is omitted as in (13a), the geˇi phrase will be analyzed as a prepositional phrase which occurs as a constituent in the clause. 2.2.2. Motivation for the goal phrase fronting Previously, we have seen that ditransitive constructions involving Class I or Class II verbs allow the goal phrase to be fronted, bringing it from its post-DO position to the position immediately after the main verb as shown in (18) and (19).

8. One can think of a parallel situation of object omission in English in the case of certain transitive verbs such as ‘‘smoke’’ and ‘‘drink’’ as in the following examples: (i) a. Do you smoke? b. Do you smoke marijuana? (ii) a. Do you drink? b. Do you drink vodka?

Argument structure change, reanalysis and lexicalization

287

(18) a. 李四寄一封信給張三。 Lıˇsı` jı` yı` fe#ng xı`n geˇi Zha#ngsa#n Lisi send one CL letter to Zhangsan ‘Lisi sent a letter to Zhangsan.’ b. 李四寄(給)張三一封信。 Lıˇsı` jı` (geˇi) Zha#ngsa#n yı` fe#ng xı`n Lisi send to Zhangsan one CL letter ‘Lisi sent (to) Zhangsan a letter.’ (19) a. 李四扔一根骨頭給汪汪。 Lıˇsı` re#ng yı` ge#n guˇto´u geˇi Wa#ngwa#ng Lisi throw one CL bone to Wangwang ‘Lisi threw a bone to Wangwang.’ b. 李四扔*(給)汪汪一根骨頭。 Lıˇsı` re#ng *(geˇi) Wa#ngwa#ng yı` ge#n guˇto´u Lisi throw (to) Wangwang one CL bone ‘Lisi threw (to) Wangwang a bone.’ Traditional grammar has assumed all along that the fronting movement does not a¤ect the meaning of the sentence, i.e. that (18a) and (18b) have the same meaning. But Fillmore (1972) and Radford (1988) have argued otherwise. They cite (20) and (21) to show that, rather than being semantically equal, the act of fronting has the e¤ect of profiling a certain argument. (20) a.

They sprayed paint onto the wall.

b.

They sprayed the wall with paint.

(21) a. b.

They loaded hay into the truck. They loaded the truck with hay.

According to the argument presented by Fillmore and Radford, ‘‘spraying paint’’ is the focus of (20a) while in (20b) ‘‘spraying the wall’’ is the focus. Similarly, in (21a) hay-loading is the main discussion while (21b) is all about truck-loading. They also point out that fronting has the e¤ect of turning the fronted phrase into a DO. This is confirmed by the tendency for the newly fronted DO to lose its status as prepositional phrase and to become a noun phrase, as demonstrated in the following examples.

288

Feng-fu Tsao

(22) a. b. c.

I argued with the tax collector about every point. I argued about every point with the tax collector. I argued every point with the tax collector.

This tendency is further illustrated by the following ditransitive sentences: (23) a. b.

Michael sent a birthday gift to his daughter last year. Michael sent his daughter a birthday gift last year.

(24) a. b.

Tina read the story for the children. Tina read the children the story.

(25) a. b.

Mom baked a chocolate cake for me. Mom baked me a chocolate cake.

Thus, we can clearly see from the previous discussion that when there is a need to profile the goal phrase in either Chinese or English, it can be fronted to the position immediately following the main verb. But when this happens, the preposition heading the goal phrase in Chinese, namely geˇi ‘give/to’, will be obligatorily or optionally present, depending upon whether the main verb belongs to the Class I or Class II ditransitive predicate. In English, the goal phrase, when fronted, will lose its preposition, turning it from a PP into an NP and making it a more DO-like constituent. 2.2.3. Further grammaticalization and lexicalization of the goal phrase As we have just discussed, a goal phrase headed by geˇi ‘give’ in Chinese or to/ for in English exhibits a tendency to being fronted, replacing the original DO phrase in many discourse contexts. In Chinese, this has the e¤ect of putting the geˇi-phrase in close proximity to the verb and thus creating a favorable condition for the verb to incorporate the coverb geˇi ‘give’ and form a lexical compound, as shown in the following diagram, using (4a) and (4b) repeated here as (26a) and (26b) as examples: (26) a.

ta# [da`i ]v [yı` ba#o ta´ng]NP ]]VP [geˇi]P [Zha#ngsa#n]NP ]]PP ]]]VP ]]]]VP ‘S/he brought a bag of candy to Zhangsan.’

b.

ta# [da´i]V [geˇi ]P [Zha#ngsa#n]NP ]PP ]]]]V’ [ yı` ba#o ta´ng]NP ]]]V’ ]]]]VP ‘S/he brought Zhangsan a bag of candy.’

And then the structure in (26b) undergoes reanalysis to become (26c) (26) c. ta# [da`i geˇi]V [Zha#ngsa#n]NP ]]V’ [ yı´ ba#o ta´ng] NP ]]] V’ ]]]]VP ‘S/he brought Zhangsan a bag of candy.’

Argument structure change, reanalysis and lexicalization

289

Parker (1976) has posited two general conditions that are necessary in order for the process of reanalysis to take place: 1) the elements involved must be strictly adjacent; 2) their original meanings must be preserved. It is quite clear from (26) that both conditions are met in this case. The three constituents involved, namely da`i ‘bring’, geˇi ‘give’ and Zha#ngsa#n ‘Zhangsan’ are strictly adjacent and the meanings of all three elements are preserved in the process although their constituent structure is changed. The likelihood that such a structural reanalysis has taken place is strengthened when we examine a case of similar structural change through reanalysis in English. Compare (27a) with (27b): (27) a. b.

Aunt Mary is looking after my daughter. My daughter is being looked after by Aunt Mary.

While (27a), an active sentence, is usually analyzed as containing a verb followed by a prepositional phrase, (27b), its passive counterpart, cannot be so analyzed because the NP ‘my daughter’ has been taken out of its governing domain PP and been moved to the subject position. In other words, the preposition ‘‘after’’ is structurally reanalyzed as part of the newly formed verb complex ‘look after’. Similarly in Chinese, the transitive verb and the following geˇi ‘give’ are reanalyzed as forming a new lexical item V þ geˇi. This automatically accounts for the fact that the originally transitive verb is always followed by geˇi ‘give’ in its ditransitive use. In addition to this argument, we give three arguments in support of the analysis that lexicalization has taken place: 1) no constituents can intervene between the verb and geˇi, 2) the V-geˇi combination shows lexical properties such as semantic shift and idiosyncratic gaps, 3) the attachment of geˇi is a lexical operation. To begin with, the verb complex V-geˇi exhibits the property of lexical integrity as no constituents can intervene between V-geˇi. (28) a. 張三扔給了王五一條毛巾。 Zha#ngsa#n re#n geˇi le Wa´ngwuˇ yı` tia´o ma´ojı #n Zhangsan throw GEI PFV Wangwu one CL towel ‘Zhangsan threw Wangwu a towel.’ (28) *b. 張三扔了給王五一條毛巾。 Zha#ngsa#n re#n le geˇi Wa´ngwuˇ yı` tia´o ma´ojın# Zhangsan throw PFV GEI Wangwu one CL towel ‘Zhangsan threw Wangwu a towel.’

290

Feng-fu Tsao

Secondly, the V-geˇi verbal complex shows such lexical properties as semantic shift and idiosyncratic gaps. This argument was originally put forth by Huang and Ahrens (1999) who cited (29b) from the Sinica Corpus to demonstrate the semantic shift. (29) a. 他冠夫姓。 ta# gua`n fu#x`ıng she cap husband family name ‘She adopts her husband’s family name (on top of her maiden name). b. 他冠給張三一個渾名。 ta# guan geˇi Zha#ngsa#n yı´ ge hu´nmı´ng s/he cap GEI Zhangsan one CL nickname ‘S/he imposed a nickname on Zhangsan.’ We see from (29a) that gua`n ‘cap’ which originally meant simply ‘to adopt (a family name)’ has acquired an additional meaning of ‘to use/apply certain expressions/names to someone’. Huang and Ahrens also cite as examples the synonymous pair pa`n and pa`nchuˇ, both meaning ‘to judge, to sentence’ and both sharing the same subcategorization frame as shown in (30a) and (30b). (30) a. 法院判處李鳳州死刑。 faˇyua`n pa`nchuˇ Lıˇ Fe`ngZho#u sıˇxı´ng court sentence Li FengZhou death penalty ‘The court sentences Li FengZhou to death.’ b. 中共判他死刑。 Zho#nggo`ng pa`n ta# sıˇxı´ng Chinese communist sentence him/her death penalty ‘The Chinese communists sentenced him/her to death.’ Only pa`n geˇi, however, is allowed as a verbal complex, as in (31a). Pa`nchuˇ geˇi is not allowed: (31b) is not grammatical. (31) a. 播音權已判給華視。 ´n buo#yıngqua # yıˇ pa`n geˇi hua´shı` broadcast right already judge GEI CTS ‘The broadcast rights have already been granted to CTS.’

Argument structure change, reanalysis and lexicalization

291

(31) *b. 播音權已判處給華視。 buo#yı #ngqua´n yıˇ pa`nchuˇ geˇi hua´shı` broadcast right already judge GEI CTS ‘The broadcast rights have already been granted to CTS.’ Finally, we can show that the incorporation of geˇi into a transitive verb to turn it into a ditransitive verb has become a part of Modern Chinese grammar by giving the following example in which the rule has been applied to an English loan word. (32) 美國telex給我們一筆訂單。 meˇiguo´ telex geˇi woˇman yı` bıˇ dı`ngda#n USA telex GEI us one batch order ‘The US (company) telexed us a batch of orders.’ 2.3. Summary of Section 1 Our lengthy discussion of Mandarin ditransitive constructions has led us to the following conclusions: 1. The distribution of geˇi is jointly determined by its meaning and the meaning of di¤erent verb clauses that go into this construction. The presence of geˇi is forbidden in a ditransitive construction that has a verb of deprivation as its main verb. Its presence is optional if the main verb is a verb of transfer other than geˇi where the absence of the co-verb geˇi is accounted for by the rule of haplology. Finally, its presence is obligatory in cases where the main verb’s origin as a transitive verb has been ditransitivised. 2. The transitive verbs that are ditransitivised fall into three basic semantic classes: verbs of movement such as re#ng ‘throw’ and ba#n ‘move’; verbs of acquisition such as maˇi ‘buy’ and qiaˇng ‘snatch by force’; and verbs of creation such as xieˇ (xı`n) ‘write (a letter)’ and ga`i ‘build’. 3. The long process of ditransitivization begins with the attachment of geˇi ‘give’ to the main verb phrase to form a serial verb construction. Then with the deletion of the verb due to semantic redundancy, the geˇi phrase is reanalyzed as a prepositional phrase denoting GOAL or RECIPIENT. When this happens, the goal/recipient phrase can, under certain conditions, be fronted to the position immediately adjacent to the main verb. Induced by this constant juxtaposition and encouraged by the general tendency of compounding in the evolution of the Chinese language (Wang 1958), the main verb and geˇi form a verbal com-

292

Feng-fu Tsao

plex which ensures the presence of geˇi when a transitive verb is used ditransitively. With these generalizations in mind, we are ready to move on to see what happens when a transitive verb is ditransitivised in Japanese and English.

3. Ditransitivization in Japanese and English 3.1. Japanese ditransitivization The ditransitivization of transitive verbs can also be observed in Japanese. Normally, the Japanese verbs yomu ‘read’, ensousuru ‘play’ and utau ‘sing’ can only be used transitively. If Japanese speakers would like to express that a story, a symphony or a song is read, played or sung for someone, the verb ageru, meaning ‘to give’, must be su‰xed to the relevant Japanese transitive verbs, thus turning the whole structure into a serial verb construction as shown in the following examples. (33) okaasan-wa kodomo-ni monogatari-o yonde ageta mother-NOM child-DAT story-ACC read give ‘The mother read the story to the child.’ (34) ookesutora-wa kare-ni kookyookyoku-o ensooshite ageta orchestra-NOM he-DAT symphony-ACC play give ‘The orchestra played a symphony for him.’ (35) Hanako-wa kare-ni uta-o utatte ageta Hanako-NOM he-DAT song-ACC sing give ‘Hanako sang a song for him.’ However, in sentences with the verb kau ‘buy’ and tateru ‘build’, we find alternations where the verb ageru is present in one sentence and absent in another. The following two examples indicate that verbs in Japanese are also developing in such a way that transitive verbs acquire an additional reading: a RECIPIENT/GOAL noun phrase is introduced. (36) a.

b.

Taroo-wa Hanako-ni doresu-o katte ageta Taro-NOM Hanako-DAT dress-ACC buy give ‘Taro bought a dress for Hanako.’ Taroo-wa Hanako-ni doresu-o katta Taro-NOM Hanako-DAT dress-ACC buy ‘Taro bought Hanako a dress.’

Argument structure change, reanalysis and lexicalization

(37) a.

b.

293

watashi-wa kazoku-ni ie-o tatete ageta I-NOM family-DAT house-ACC build give ‘I built a house for my family.’ watashi-wa kazoku-ni ie-o tateta I-NOM family-DAT house-o build ‘I built my family a house.’

The necessity of the RECIPIENT/GOAL NP in these sentences suggests that those previously transitive predicates have been turned into ditransitive ones. (38) watashi-wa Taroo-ni booru-o watashita I-NOM Taro-DAT ball-ACC pass ‘I passed Taro a ball.’ (39) kare-wa watashi-ni juubunna joohoo-o teikyooshita he-NOM I-DAT enough info-ACC provide ‘He provided me with enough information.’ (40) kare-wa Taroo-ni otooto-o syookaishita he-NOM taro-DAT brother-ACC introduce ‘He introduced Taro his brother.’ (intended meaning: He introduced his brother to Taro.) (41) kare-wa Hanako-ni kaado-o tewatashita he-NOM hanako-DAT card-ACC hand ‘He handed Hanako a card.’ (42) watashi-wa gakkoo-ni okane-o kifusshita I-NOM school-DAT money-ACC donate ‘I donated school the money.’ (intended meaning: I donated money to the school.) (43) ame-ga watashitati-ni kooun-o motarashita rain-NOM I-DAT good luck-ACC bring ‘The rain brought me good luck.’ (44) sensei-wa watashitati-ni syashin-o miseta teacher-NOM I-DAT photo-ACC see ‘The teacher showed me the photo.’

294

Feng-fu Tsao

(45) watashi-wa kere-ni kuruma-o yakusokushita I-NOM he-DAT car-ACC promise ‘I promised him a car.’ (46) kare-wa tomodati-ni okane-o kashita he-Nom friend-DAT money lend ‘He lent his friends some money.’ To summarize, we have come to the following generalizations with regard to ditransitivization in Japanese: 1. When the THEME NP is moved backward, the marker ni indicating GOAL/RECIPIENT must always be present to mark the GOAL/ RECIPIENT nominal. In this case, the verb is interpreted as ditransitive. 2. The transitive verbs that allow this process of argument increment fall into three semantic classes: verbs of movement, verbs of acquisition and verbs of creation as shown in (47). (47) a. b. c.

Verbs of creation: yomu ‘read’, ensoosuru ‘play’, utau ‘sing’, tateru ‘build’, tsukuru ‘make’ Verbs of acquisition: kau ‘buy’, teikyoosuru ‘provide’ Verbs of movement: tewatasuru ‘hand’, sosogiireru ‘pour’, watasu ‘pass’, mottekuru ‘bring’, nageru ‘throw’

3.2. Ditransitivization in English Ditransitivization in English can be clearly seen in the following examples. (48) a.

Michael sent a birthday gift to his daughter last year. Michael sent his daughter a birthday gift last year.

(49) a.

Tina read a story for the children. Tina read the children a story.

(50) a.

Mom baked a chocolate cake for me. Mom baked me a chocolate cake.

Like their counterparts in Japanese and Chinese, read and baked would normally be used as transitive verbs, which can be followed by an optional beneficiary phrase headed by the preposition ‘for.’ But in some discourse contexts the ‘for’ phrase can be fronted to the position immediately following the main verb. As a result, the preposition ‘for’ is absorbed by the verb,

Argument structure change, reanalysis and lexicalization

295

leaving the verb to be followed by two NPs and turning the sentence into a ditransitive one. Similar to what we have just shown in Chinese and Japanese, English only allows transitive verbs of creation, verbs of acquisition and verbs of movement to be ditransitivised as shown in (49) and (50) and in the following examples. (51) a. b.

We will buy a house for our son next year. We will buy our son a house next year.

(52) a. b.

The maid will bring an umbrella to you in a minute. The maid will bring you an umbrella in a minute.

(53) a. b.

I will make some co¤ee for you. I will make you some co¤ee.

(54) a. b.

I will pour a cup of hot tea for you. I will pour you a cup of hot tea.

These similarities notwithstanding, there are features that distinguish ditransitivization in English from that in Chinese and Japanese. For example, while English allows types of IO phrases, i.e. some headed by ‘to’ and others by ‘for’, to be fronted and to have the preposition absorbed by the main verb, both Japanese and Chinese allow only one type of adposition, namely ni in Japanese and geˇi in Chinese. We have posited explicitly that what is absorbed by the verb is a beneficiary ‘for’ rather than a recipient ‘to’. Goldberg (1995), however, posits that ditransitive sentences like the b-examples in (51) to (54) always have a RECIPIENT role implied. Clearly, disagreements like this indicate the need to further explore the absorption of prepositions.

4. Summary and theoretical implications 4.1. Summary Our previous discussion can be summarized as shown in the following table. From Table 4, it is clear that all three languages allow only three classes of transitive verbs to take part in the process of ditransitivization: verbs of movement, acquisition and creation. Table 5 presents verbs that undergo this process.

296

Feng-fu Tsao

Table 4. Changes that occur in the ditransitivization of transitive verbs in Chinese, Japanese and English Features

Language Chinese

Japanese

English

Only transitive verbs of movement, acquisition & creation are allowed.

Z

Z

Z

There is an overt argument increment.

Z

Z

Z

Word order changes.

Z

Z

Z

Morphological changes are involved.

Z

Y

Z

SVC serves as an intermediate stage.

Z

Z

?

(SVC ¼ serial verb construction) Table 5. Classes of transitive verbs that can undergo ditransitivization Language

English

Japanese

Chinese

Verbs of creation 1.

bake

yaku

2.

promise

yakusokusuru

kaˇo 烤 che´ng-nuo` 承諾

3.

make

tsukuru

zuo` 做

Verbs of acquisition 1. donate

kifusuru

2.

buy

kau

jua#n 捐 maˇi 買

1. 2.

hand pour

tewatasu sosogiireru

3.

pass

watasu

na´ 拿 da`o 倒 chua´n 傳

4.

bring

mottekuru

da`i 帶

5.

throw

nageru

6.

lend

kasu

diu# 丟 jie` 借(給)

Verbs of movement

A question that may arise here is why it is that only these classes of transitive verbs are allowed. A possible answer suggests itself, one that lies along the line of facilitative conditions that underlie successful completion of an event of transfer. For a transitive predicate to successfully function

Argument structure change, reanalysis and lexicalization

297

as a ditransitive one, the Agent of the action must literally or metaphorically possess or create the THEME/PATIENT of the verb and the THEME/ PATIENT of the predicate must be moved to a place where the action denoted by the verb is completed. As reasonable as this semantic account seems to be, it does not imply that all the transitive verbs that fall into these classes can undergo the process of ditransitivization. Other factors such as frequency of use and cultural preference must also be taken into account, but limitations of space prevent us from exploring this path any further. Two other features are common to all three languages: (1) overt increment of argument and (2) grammaticalization. The overt increment of argument is expressed in all three languages. In Japanese it is the NI phrase, in Chinese the GEI phrase, and in English the TO/FOR phrase. Even though there is a change in word order when the above-mentioned phrase is added, the change in Japanese is minimal since when the THEME NP moves to the back, its original position is taken up by the NI phrase. Morphological changes can be observed in Chinese and English. The geˇi-phrase in Chinese is concatenated with the verb to form a verbal-complex as it is habitually fronted to a position immediately adjacent to the latter. FOR/TO phrases in English incorporate their prepositions into the verb, turning a PP into an NP as the phrase is placed immediately adjacent to the verb. In Japanese no morphological changes take place. Only the NI-phrase is added to the argument structure. As for the grammaticalization process of ditransitivization: the question is raised as to whether SVC is involved in the process. We have presented evidence to show that SVC is involved in both Chinese and Japanese, but in English our evidence is insu‰cient. Much work needs to be done before a firm conclusion can be drawn. 4.2. Theoretical implications A very influential book about argument structure was published in the mid90’s. That was A.E. Goldberg’s seminal work on Construction Grammar in which she discusses argument structure change. According to her theory, argument structures are meaning-oriented. They o¤er ‘‘the basic means of clausal expressions’’ (Goldberg 1995: 3) in the language. When a predicate is put into a frame or a construction, the verb inherits the interpretation possessed by the construction itself. The roles inside a construction are determined by two principles: the Semantic Coherence Principle and the Correspondence Principle. The two principles are defined as follows (Goldberg 1995: 50):

298

Feng-fu Tsao

Semantic Coherence Principle Only roles which are semantically compatible can be fused. Two roles r1 and r2 are semantically compatible if either r1 can be construed as an instance of r2 or r2 can be construed as an instance of r1. For example, the kicker participant of the kick frame may be fused with the agent role of the ditransitive construction because the kicker role can be construed as an instance of the agent role. Whether a role can be construed as an instance of another role is determined by general categorization principle. Correspondence Principle Each participant role that is lexically profiled and expressed must be fused with a profiled argument role of the construction. If a verb has three profiled participant roles, then one of them may be fused with a nonprofiled argument role of a construction. According to Goldberg, the ditransitive construction would look like this:

In the presentation given above, the arguments of the verb are connected by a solid line, which means that the agent and patient are subcategorized by the verb. The dashed line indicates that the recipient role is integrated into the construction and is not syntactically required by the verb. That is to say, the recipient role is an adjunct inside this construction but the agent and the patient role are obligatory arguments. Take the verb bake for example. Bake without any additional modification is regarded as a transitive verb. Two noun phrases are compulsory to satisfy the predicate’s semantic and syntactic requirements. In the sentence ‘‘Teresa baked a cake’’, ‘‘Teresa’’ is the agent and is nominative casemarked, and the ‘‘cake’’ is the patient, which refers to something that undergoes the action of baking. If we put the verb ‘‘bake’’ into a ditransitive framework, we can place an additional noun phrase between these two arguments viz. ‘‘Teresa baked Tom a cake.’’ The additional argument is

Argument structure change, reanalysis and lexicalization

299

interpreted as the GOAL for which the cake is baked. The sentence becomes ‘‘Teresa baked Tom a cake.’’ Following Goldberg’s analysis, the interpretation of the verb ‘‘bake’’ is unmodified; rather the extra RECEPIENT nominal is contributed by the ditransitive construction itself. 4.2.1. Problems with Goldberg’s account Goldberg’s idea that construction grammar enables the verb to extend its reading sounds reasonable. Within this framework, it would be very e‰cient to create a verb with di¤erent interpretations by assigning to it a set of arguments and putting the verb into a fixed construct. Such a ready-made schema could enhance a child or student’s learning processes. Yet, we find it hard to explain why some of the transitive verbs cannot be fit into the framework above in spite of the fact that these verbs meet every requirement for the ditransitive structure. Some unacceptable examples are given below. (55) a. I swept the floor every weekend. b. ??I swept Maria the floor every weekend. (56) a. He did the dishes last night. b. ??He did his son the dishes last night. In the illustrations given above, the English verbs sweep and do are indeed applied transitively. Each verb takes one AGENT and PATIENT/ THEME nominal as their basic arguments. According to Goldberg, the insertion of an additional RECIPIENT role should result in a comprehensible sentence. The additional nominal should be the beneficiary of the action depicted by the verb. The results, however, do not meet our expectations. Sentence (55b), and (56b) are not natural or, indeed, even sensible whereas (57) is. (57) Transitive application of bake: Mom baked a chocolate cake for me. Ditransitive application of bake: Mom baked me a chocolate cake. Obviously, Goldberg’s theory is in need of revision before it can give an adequate explanation for those structurally acceptable but semantically unnatural instances. 4.2.2. Our proposed account Returning now to our own account, our investigation of the ditransitivization process which is based on both semantic and syntactic considerations

300

Feng-fu Tsao

has led us to the conclusion that the process involves only three naturally defined verb classes: verbs of movement, creation and acquisition. We have also argued that lexicalization is involved in the three languages that we have investigated. Furthermore, lexicalization, as it is usually defined, involves some poorly understood pragmatic factors such as frequency of use and cultural preference, and these factors can often account for the idiosyncratic properties of the lexicalization process. With these points accounted for, we can now go back and re-examine the counterexamples (55b) and (56b). The latter is not acceptable for the obvious reason that ‘‘do’’ in the sense of ‘‘do the dishes’’ is neither a verb of movement, nor is it a verb of creation or acquisition, and since the semantic-syntactic condition is not met, there is no way for it to become a ditransitive verb. The fact that (55b) fails to be ditransitivised, however, calls for a more involved explanation. The verb ‘‘sweep’’ may look like a verb of movement but, on closer examination is found not to be one. A common feature of the class of movement verbs is that a ‘theme’ argument has to be included in the argument structure. Even though ‘‘floor’’ in (55b) may look like a ‘theme’ argument, it is not. It is a ‘locative’ argument. In other words, it is not something that is moved, rather, it is a place where a certain action occurs. Hence, Goldberg’s theory fails in this example as well. To sum up, it seems that our theory of ditransitivization in terms of grammaticalization and lexicalization goes a long way toward explaining the syntactic as well as the semantic phenomena in association with ditransitive constructions and the ditransitivization process in Chinese, Japanese and English. We therefore suggest that it deserves a place in the universal consideration of human grammar.

Reference Cheng, Ya-hsia, and Chu-Ren Huang 1988 Guo´yuˆ zho#ng de Shua#ngbın# Do`ngcı´ [Double Objects Verbs in Mandarin]. The Word of Chinese Language 48: 34–48. Chao, Yuen-Ren 1968 A Grammar of Spoken Chinese. Berkeley: University of California Press. Fillmore, C. J. 1968 The case for case. In Universals in Linguistic Theory, E. Bach & R. T. Harms (eds.), 1–88. London: Holt, Rinehart & Winston.

Argument structure change, reanalysis and lexicalization Fillmore, C. J. 1971 Givon, T. 1975

301

Some problems for case grammar. Working Papers in Linguistics 10: 245–265. Ohio State University. Serial verbs and syntactic change: Niger-Congo. In Word Order and Word Order Change, D. Li (ed.), 47–111. Austin: University of Texas Press.

Goldberg, A. E. 1995 Constructions: A Construction Grammar Approach to Argument Structure. University of Chicago Press. Huang, Chu-Ren 1990 Mandarin Double Object Construction and Morpho-lexical Rules. Paper presented at the 23rd International Conference on Sino-Tibetan Languages and Linguistics. Arlington, Texas. Huang, Chu-Ren, and Kathleen Ahrens 1999 The Function and Category of gei in Mandarin Ditransitive Constructions. Journal of Chinese Linguistics 27.2: 1–26. Huang, Chu-Ren, and Ruo-ping Mo 1992 Mandarin ditransitive construction and the category of gei. Proceedings of the Berkeley Linguistics Society 18: 109–122. Li, Charles N., and Sandra A. Thompson 1981 Mandarin Chinese: A Functional Reference Grammar. Berkeley: University of California Press. Liu, Feng-hsi 2006 Dative Constructions in Chinese. Language and Linguistics 7.4: 863–904. Lyons, J. 1968 Introduction to Theoretical Linguistics. London & New York: Cambridge University Press. Parker, Frank 1976 Language change and the passive voice. Language 52: 449–460. Radford, Andrew 1988 Transformational Grammar: A First Course. New York: Cambridge University Press. Tang, Ting-Chi 1979 Double Object Constructions in Mandarin. Guoyu Yufa Yanjiu Lunji [Studies in Chinese Syntax], 449–537. Taipei: Student. Tsao, Feng-fu 1979 A Functional Study of Topic in Chinese: The First Step toward Discourse Analysis. Taipei: Student. Tsao, Feng-fu 1988 The Function of Mandarin Gei and Taiwanese Hou in the Double Object and the Passive Constructions. In The Structure of Taiwanese: A Modern Synthesis, Cheng, R. L. and Huang, S. F. (eds.), 165–208.

302

Feng-fu Tsao

Tsao, Feng-fu 1989

Tsao, Feng-fu 1990 Wang, Li 1958 Xu, Dan 1994

Serial Verb Constructions in Chinese. Proceedings of the Second International Conference on Sinology, 653–678. Taipei, Taiwan: Academia Sinica. Sentence and Clause Structure in Chinese: A Functional Perspective. Taipei: Student. Ha`nyuˆ Shıˆ Gaˆo [A History of the Study of the Chinese Language]. Beijing: Zhonghua Shuju. The Status of the Marker Gei in Mandarin Chinese. Journal of Chinese Linguistics 22.2: 363–394.

Subject index aboutness 74–75, 79 accretion (of meaning) 4 accessible (referent) 71–72 accessibility (of meaning) 66 adjective 5, 10, 28, 33, 67–68, 123, 129, 242, 245, 250 evaluative adjective 127 gradable adjective 127–128 non-gradable adjective 127–128 predicative adjective 115 reduplicative adjective 140 African languages 1–3 agent 275, 297, 298, 299 agreement marker 2, 10, 174 analytical (language) 2 anaphoric (function) 57, 60, 79 anterior 38, 46, 48 (also see perfect aspect marker) lexical source 24 young 47, 49–50 old 47, 49 Archaic Chinese 10, 14, 23–25, 29, 47, 90–93, 149, 157, 215–225, 230– 231 (also see Old Chinese) argument structure 275, 297 aspectual marker 23–24, 29–30, 64, 259 Austro-Asiatic languages 189 baˇ 把 3, 172, 175–179, 183–184, 187– 188, 198 be`i 被 8, 195 Beijing Mandarin 12, 55–56, 60, 80– 84, 191–192 classifier (numeral) 13–14, 80–81, 125, 169–173, 215–231, 242 classifier meaning 13, 176–181, 188– 189, 195–196 clause 27, 31–34, 36–37, 46–47, 74, 244, 254–255, 291

construction grammar 15, 117, 120, 193, 297, 299 constructional operator 13, 116–117, 141–142 Contemporary Chinese 156, 161, 240 content word 87, 91, 118 conversational implicature 192 co-referential 61–65, 69 current relevance 24, 33–35, 38, 45 declarative sentence 64 definite article 12, 55, 71, 73, 79, 83, 249 definiteness 5, 12, 83 degree adverb 15, 115, 117–118, 120, 126–127, 134–135, 141–142 degree evaluative construction (DEC) 13, 117, 122, 138, 141–142 deictic center 55, 60 demonstrative 12, 116, 121–124, 135– 138, 171, 242, 248 distal demonstrative 55, 73–75, 77, 83 pronominal demonstrative 63, 74 proximal demonstrative 55, 73–75, 77–78, 83 diphthong 60 direct object 275–277, 279–280 discourse/contextual (function) 58 discourse mode 55, 74 disyllabic word 10–11, 237, 239, 247, 258 disyllabification 237 ditransitive (construction) 14, 275– 278, 284, 288–291 ditransitivation 291–297, 300 duration/continuation (marker) 106– 109 dynamic yeˇ 12, 22–24, 29–50 even, 7

304

Subject index

frequency 13, 31, 40, 42, 120, 171– 172, 189–191, 197, 223–225, 261– 262, 299–300 function word 3, 87, 91, 186 functional phrase 247 future (meaning/marker) 13, 39, 46– 47, 158, 162–165 gaˇn 敢 149, 152 gei 給 14, 277–294, 297 generic (function) 60, 72, 172, 260 given information 78 goal phrase 284, 292–297, 300 gram 47, 49, 178 grammaticalization 1–5, 10–15, 55– 56, 71, 83, 87–88, 90–92, 117, 143, 170, 195–196, 235–237, 284– 291, 297 guo` 過 10–11 haplology 281, 291 heˇn 很 (see constructional operator) 12–13, 115–141 hortative marker la´i 来 95–97 huı` 會 5–6 identifiable (referent) 60–63, 66–67, 71, 74 immediate future (function) 46–47 imperative sentence 95–97, 110 indefinite article 55, 71, 79–80, 83 indirect object 275–281 Indo-European language 1, 3, 24 inflectional (language) 2, 9 intention (meaning) 13, 152, 157–158, 161–162, 164–165 isolating (language) 4, 9, 174, 268 jia#ng 将 39, 92, 98, 149, 157–158, 165 jia`o 叫 8, 132 jiu` 就 9–10, 195 ken 肯 149, 153 la´i 来 12, 45, 87–111

lexicalization 1–2, 10–11, 14, 235– 239, 268–270, 284, 288–291, 300 constraint 258–262 definition 235–237 degree 256–257 syntactic change 262–267 lexical meaning 215, 226 lexical semantic 3, 115 lia´n 连 6–7 measure word 171–172, 183 mechanism (of semantic change) 13, 182–189, 197 Medieval Chinese 23–24, 39–40, 90, 149, 218, 223–225, 230, 232 (also see Middle Chinese) mental domain 165 metaphor (or metaphoricalization) 3, 13, 98, 124–126, 130, 182–186, 197, 238–239 metonymy (or metonymization) 13, 180–182, 184–186, 197, 260 Middle Chinese 173–175, 192–193, 247–248 (also see Medieval Chinese) Middle English 2 minimal degree marker 139–140 modality deontic modality 150–151 dynamic modality 150–151, 154 epistemic modality 95, 150–151 objective modality 150 subjective modality 150 Modern Chinese 13, 24, 39, 42, 90, 171–172 Modern English 2 morpho-syntactic process/change 1, 169–170, 182–183 na` 那 12, 56–63, 69–82, 171 neutral tone 73 new information 36–38, 45, 71, 75, 78 ni phrase (Japanese) 297 nil classifier (construction) 192–193

Subject index

305

nominalization 186, 188–189, 196, 247–248 non-identifiable (referent) 74, 81 noun 5, 10, 56–60, 63–64, 67, 70–73, 79–83, 95, 123–125, 169–176, 183–197, 215–216, 221–225, 229– 230, 242, 249–250, 257, 260, 275– 276, 298 numeral (word) 63, 80–81, 191, 194

Sino-Tibetan languages 13, 215–227, 229 situational use 56, 59–60 specific (referent) 72–73, 262 source meaning 171, 176–181, 188 static yeˇ 12, 22–26, 29, 47–49 syntactic function 222 syntactic structure 215 syntagmatic change 3

Old Chinese 174, 237, 247, 263–266 (also see Archaic Chinese)

Tai (language) 215–227, 229 Taiwan Mandarin 119 theme 275–276, 294, 297, 299–300 to/ for phrase 297 tone 63, 71–73, 80–81, 257 tone sandhi 191 topic 75, 79 topic continuity 75, 79 topic chain 75 topic marker 69, 71, 74, 78, 240 tracking use 57–60, 68, 71, 77 transference (of meaning) 186, 188– 189

perfect aspect marker 12, 23–24, 33– 38, 49–50, 91, 100, 106 (see anterior) perfective aspect marker 4, 24, 38, 48– 49, 89, 92, 97–98, 104, 110–111 performative (function/use) 159 persistence 186, 188 phono-syntactic (process) 169–170, 191 physical domain 165 polysemy 3–5, 10, 178–180, 184–186 predicate 115, 222, 225, 231 qualia structure 117–118, 120, 125 reanalysis 284, 288–289 recognitional use 12, 59–60, 71, 79, 83 referent 55, 58–60, 72–75, 81, 83, 169, 242 referential (feature) 55, 60–63 reduced predicate maker 63, 70 repeater 13–15, 170, 215–217, 219, 221, 223–231, 233 resultative resultative aspect 24, 28, 33, 47–49 resultatie verb 90, 98–100, 245–246 semantic change 1–4, 10–13, 149, 154, 160–165, 170–171, 181–189, 194– 197, 255–257 semantic reanalysis 13, 182–186 sentence final particle 25, 64, 139

unidirectional 1, 124 verb 5–8, 30–33, 63–67, 69, 90, 92, 178, 194–197, 239–244, 252, 260 auxiliary verb 24, 118, 131–132, 149–151 causative verb 93, 131–132 coordinate verb 245, 257 coverb 276–282 directional verb 12, 93–95, 101, 109, 283 ditransitive verb 14, 275–276, 283– 284, 291, 300 intransitive verb 245–245 serial verb (construction) 3, 10, 14, 92, 97–100, 195, 285–286, 291– 292, 296–297 stative verb 49 transitive verb 3, 14, 276, 289–292 volitional verb 13, 161

306

Subject index

verb of acquisition 282–283, 291, 294–296 creation 283, 291, 294–296, 300 deprivation 282–283, 293 movement 283, 291, 294, 296, 300 transfer 283, 291 volition 13, 149–152, 156–157, 161– 162, 165 word string 251 word-internal element 268–269 xiaˇng 想 149, 161–163

ya`o 要 149, 152, 161, 163–165 yeˇ 也 12, 21–51 yı´ 一 12, 63–65, 80–83, 125–126, 171, 183, 191–192 yua`n 願 149, 155–157 yu` 欲 149, 157, 161–162 zhe` 這 (see proximal demonstrative) 12, 56–60, 65–82, 171