Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives [1 ed.] 9781402083204, 1402083203

Citation preview

Space in Languages of China

Dan Xu Editor

Space in Languages of China Cross-linguistic, Synchronic and Diachronic Perspectives

Editor Dan Xu INALCO/CRLAO, Paris France

ISBN 978-1-4020-8320-4

e-ISBN 978-1-4020-8321-1

Library of Congress Control Number: 2008925335 © 2008 Springer Science + Business Media B.V. No part of this work may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying, microfilming, recording or otherwise, without written permission from the Publisher, with the exception of any material supplied specifically for the purpose of being entered and executed on a computer system, for exclusive use by the purchaser of the work. Printed on acid-free paper 9 8 7 6 5 4 3 2 1 springer.com

Contents

Preface…………………………………………………….……………….. vii Introduction: How Chinese structures space………………………......……..1 Dan Xu Section A. Space: A Cross-linguistic perspective Chinese localizers: Diachrony and some typological considerations…............................................................................................ 15 Hilary Chappell and Alain Peyraube Syntax of space across Chinese dialects: Conspiring and competing principles and factors……………………………………………….………39 Danqing Liu The linguistic categorization of deictic direction in Chinese –– with reference to Japanese………………………………………..……..69 Christine Lamarre The locative words in the waxiang dialect spoken in Guzhang, Hunan………………………………………………………… 99 Yunji Wu From locative to object markers: The parallel development of two postpositions in Bai………………………………………………………. 119 Jingqi Fu and Lin Xu The role of visual space in sign language development……………………………………………….………………143 Shun-chiu Yau

v

vi

Contents

Section B. Space in Synchronic and Diachronic Chinese Asymmetry in the expression of space in Chinese –– the Chinese language meets typology………………………….…....…175 Dan Xu Two conditions and grammaticalization of the Chinese locative……………………………………………………………….…....199 Chaofen Sun Origin and evolution of the locative term hòu ‘BACK’ in Chinese………………………………………………………………… 229 Fuxiang Wu The inessive structure in archaic and medieval Chinese: An evolutionary study of inessive demonstrative uses from archaic to early modern Chinese……………………………………………………...…………......249 Qingzhi Zhu and Wenjie Chen Subject Index……………………………………………………...…..…...267 Language Index……………………………………………..…….…..…...271 Author Index……..……………………………………………..……..….. 273

Preface

This collective work began in 2004 thanks to Grant Number 03326 awarded by the Research Department of the French Government to the project entitled L’Espace et ses représentations en Asie orientale à travers divers langages. The participants are from universities and institutions in several countries, working in various domains. They all have the same strong interest: investigating ‘Space’ in languages of China. Over the past decade, this topic has been the subject of debate in many disciplines including linguistics and psychology, but there have so far been few studies of Chinese and related languages. My gratitude goes to the authors, who have supported this project and given their contributions to this book. Our collaboration over the years has been enjoyable and fruitful. We have shared our experiences and exchanged our points of view, which are not always the same. Without their indispensable help and constructive observations, this book would never have been completed. Many thanks to Craig Baker for his efficient help with editing work in English. I take responsibility for any remaining mistakes. Dan Xu Paris

vii

INTRODUCTION: HOW CHINESE STRUCTURES SPACE Dan Xu INALCO/CRLAO, France Space has long been a popular topic in linguistic research. Numerous books on the subject have been published over the past decade. However, none of these books were based on linguistic data from Chinese. The Chinese language is an “atypical” SVO language1 and deserves more attention and study. In this volume, contributors working in different specialties present and analyze the expression of space in languages of China. Not only Mandarin Chinese (the standard language) is investigated; several other dialects, as well as a minority language of China and Chinese Sign Language are studied. Cross-linguistic, synchronic and diachronic approaches are used to investigate phenomena related to space. This work does not claim to challenge or revise ongoing theoretical proposals, since the contributors are aware that problems explaining the expressions of space in Chinese have been largely neglected in past research. Even the available data is not very well described. In this book, we try to provide general linguists and those who are interested in the Chinese language with a reliable presentation and description of spatial expressions in Chinese. The papers collected here are empirical, descriptive and sometimes tentative. Our aim throughout has been to stimulate discussion rather than to offer solutions. In this book, some contributors focus on spatial structures, while others concentrate on spatial terms. In section 1 of the Introduction, the language situation in China is presented. Then, we introduce some important recent debates about the Chinese language. Finally, we give a summary of the articles which study the expression of space using different approaches. As the contributing scholars argue, Chinese shares many common features with other languages, but also presents some particular properties. 1. Language Situation in China What does “Chinese” mean? It is not easy to give a short answer. The language situation in China is very complex, not only for those who do not know this language, but also for linguists who have been working on it for many years. Generally speaking, “Chinese” refers to Mandarin2, or rather the “standard language” based on Beijing dialect, which is spoken on TV. In almost every province, educated people are bilingual; they speak a dialect and the standard language. Most people can at least understand Mandarin. The Chinese dialects are classified into ten groups3: (1) Mandarin: mainly spoken in north of the Yangtze River, and in southern provinces including Sichuan, Yunnan, Guizhou, etc. More than 662 million4 people speak Mandarin.

1 D. Xu (ed.), Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives, 1–14. © Springer Science +Business Media B.V. 2008

2

D. Xu

(2) Jin: spoken in Shanxi province and zones contiguous with this province, for instance some regions of Hebei, Inner Mongolia, Henan and Shaanxi. (45 million) (3) Wu: spoken in Shanghai, Zhejiang, etc. (69 million) (4) Min: distributed in Fujian, Taiwan, etc. (55 million) (5) Hui: attested in Huizhou, Anhui province. (3 million) (6) Gan: spoken in Jiangxi. (31 million) (7) Hakka: located in zones between Fujian, Guangdong and Jiangxi. (35 million) (8) Yue: used in Guangdong, Guangxi, Hongkong, Macao. (40 million) (9) Xiang: spoken in Hunan. (30 million) (10) Pinghua: found in Guangxi. (2 million). While people in the North understand each other when they travel across provinces, people in the South cannot communicate easily when they visit a neighboring village. Cantonese is almost as much a “foreign language” for people from Beijing as the Breton language (spoken in Bretagne of France) is for people from Paris. In other words, the mutual intelligeability among Chinese dialects is low. In the South, ancient Chinese pronunciation is better preserved. The evolution source has been in the North, since most capitals in history were located in the North. However, a common cultural and historical background, as well as the same written form, have united the different ethnic groups and dialects for millennia. Recent excavated texts from different provinces show us that as early as the Warring States period (475–221 BC), a large majority of the written forms were similar and had the same origin; only some particles had variant forms. Unfortunately, we do not know their pronunciation since Chinese characters do not directly reflect sounds. Nevertheless, researchers have attempted to reconstruct them using the Qièyùn, the first rime dictionary from 601 BC, the Shījīng (Odes) dated from around the eleventh to sixth centuries BC, and phonetic series of characters (see Karlgren, 1957, Fanggui Li, 1980, Pulleyblank, 1991, Baxter, 1992, Sagart, 1999 among others). Works based on these three data sources, especially the eminent research of the Qing Dynasty (1644–1911) scholars, clearly describe the pronunciation around the time of the Qièyùn. The reconstruction of Old Chinese (11th century BC to first century AD) began in the last century. As expected, there are still many problems and divergent points of view. The complexity of the Chinese languages is evidently due not only to the vast geographic region where they are spoken; the long time period for which data is available means that assigning different time periods to the history of Chinese is also often a subject of debate. The earliest texts, divinatory texts inscribed on bones and shells (usually called “inscriptions on bones”), can be traced back to the 14th century BC. The style of characters changed over time, but the continuity of the writing system is evident. Scholars working on phonology and syntax, including the contributors to this book, often have different points of view on the division of the history of Chinese into time periods5.

Introduction

3

Scholars generally agree that Chinese belongs to the Sino-Tibetan language family. However, comparative studies “are still at a relatively primitive level” (Norman, 1988, 13); many problems remain unclear and data is unavailable for many languages. Comparative methods used for Indo-European languages some-times cannot be used because descriptions of languages are incomplete or non-existent. Moreover, investigation of the Sino-Tibetan family is much more complex than Indo-European, because a large portion of the languages have been never investigated and have never had written forms. These non-Han language speakers have permanent contact with Han people (Chinese people) in the South and Southwest. This situation also causes complication in the dialects. Geneticists 6 propose a continuous southward movement of Han people. Southward migrations “occurred during almost all periods in the past two millennia” (Bo Wen et al. 2004, 304). “Studies on classical genetic markers and microsatellites show that the Han people, like East Asians, are divided into two genetically differentiated groups, northern Han and southern Han, separated approximately by the Yangtze river.” (302). Their genetic observation suggests that the Chinese dialects in the North may be less heterogeneous then those in the south. The vast regions of the North have had language contact with the Altaic languages: Mongolian within China, and Manchu spoken in some villages in Heilongjiang province7, while in the south contact has been with the Tibeto-Burman family in the West and Southwest, and the Miao-Yao and Tai languages in the South. This presents an intricate situation. The geneticists conclude that “the massive movement of the northern immigrants led to a change in genetic makeup in southern China, and resulted in the demographic expansion of Han people as well as their culture” (304). Tone systems are characteristic of Chinese and many other Asian languages. However “there is now considerable evidence to suggest that the various tone systems within Sino-Tibetan may not be directly cognate, i.e. that tone systems have developed independently in various branches of the family.” (DeLancey, 1987, 805). The rise of the tone system in Chinese was partially caused by the loss of voiced stops. This process was repeated in other Asian languages such as Thai and Vietnamese (Haudricourt, 1954). Tones evidently compensated for the loss of the distinctive feature of voiced stops. Old Chinese may have possessed clusters. “If the morphology in OC [Old Chinese] was wiped out, the reason seems to be that the one character–one syllable development, urged perhaps by the rise of tones, was not favorable to recording a morpheme containing more than one syllable or clusters.” (Dan Xu, 2006, 2). Today researchers know that structural resemblances do not imply genetic relationship, and vice versa. Typologically speaking, Chinese has the word order SVO while “all TB [Tibetan-Burman] languages are OV, except for Bai and the Karen languages, which are VO (and more specifically SVO).” (Dryer, 2003, 43; see also Jingqi Fu and Lin Xu in this volume). Dryer has identified “a number of characteristics that are highly atypical of VO languages” in Mandarin. In fact, there seems to be a strong correlation for VO languages to have prepositions and OV languages to have postpositions. In Mandarin, however, both prepositions and postpositions are common (see Dan Xu, 2006, Danqing Liu in this volume).

D. Xu

4

With this schema in mind, readers will understand why in this book, the topic “Space in languages of China” comprises many approaches and perspectives. 2. Different Approaches to Space in Languages of China The intent of this volume is for authors working on different domains to focus their investigation on one topic: the expression of space in various languages in China, both oral language and sign language, Mandarin and other Sinitic languages, as well as other languages of China. In this book, many dialects are examined, including Wu and Yue dialects (Liu Danqing), Waxiang of an unidentified dialect spoken in Hunan (very little research on this dialect has been done, see Yunji Wu), and Jizhou of Hebei province, which is a Mandarin-speaking region (Lamarre). The Bai language, spoken in some regions of Yunnan province, is also investigated (Jingqi Fu and Lin Xu). The linguistic affiliation of Bai has been a topic of debate. Sign language is not ignored in this study (Shun-chiu Yau), whereas previous investigations of space mainly focused on oral language. Almost the entire history of China is covered, from Old Chinese to Middle Chinese, Modern Chinese, and contemporary Mandarin (Chappell and Peyraube, Fuxiang Wu, Chaofen Sun, Qingzhi Zhu and Wenjie Chen, Lamarre, Dan Xu, etc.). If we can take the vast geographic area of China as a projection of time, we will see that the dialect varieties represent different depths in time. In other words, the different dialects form continuums corresponding to historic periods. The Wu and Xiang dialects still preserve the voiced stops, while in most other Chinese dialects these sounds have disappeared or become distinctive tones. The “entering tones,” which are in fact syllables ending in the stops -p, -t, -k, are well preserved in Yue, Min and Hakka, while in most regions they have been lost (except in some Jin dialects, which are isolated by mountains and seem more conservative for northern dialects). Current dialects present fine-grained patterns to diachronic analyses. In this book, the following topics are discussed. They are also subjects of current investigations in general linguistics: – – – –

grammaticalization typology of motion events (satellite-framed vs. verb-framed languages) adpositions (prepositions and postpositions) phonological change and its impact on syntax.

2.1. Grammaticalization Almost all authors in this book have dealt directly or indirectly with the process of grammaticalization. It consists of a lexical item becoming a grammatical item, or a less grammatical element becoming a more grammatical one. The Chinese language offers rich examples of this process, and studies on this subject have flourished for two decades. The Chinese language has always been a serial verb construction language. Almost all prepositions originated from verbs. Some conjunctions also came from verbs. For example, the preposition zài grammaticalized from an existence verb (see Danqing Liu, Chaofen Sun in this volume). Evidently, the localizers (particles following an NP and indicating location in space) in Chinese

Introduction

5

grammaticalized from a subclass of nouns, and changed from a specific and prototypical meaning of localizers to a general one through grammaticalization (see Chappell and Peyraube in this book). The locative term hòu ‘back’ in Chinese, however, came from a verb via grammaticalization (see Fuxiang Wu in this volume). In many other languages investigated by some linguists, the body part ‘back’ is mainly a source of locatives. It is clear that grammaticalization is a general phenomenon in human languages. The locatives have developed into object markers via a dative stage in Bai (See Jingqi Fu and Lin Xu). In sign languages as well, grammaticalization is reported (by Armstrong, 2002, cited by Yau in this volume). Some morphemes may currently be undergoing this process in standard Mandarin; for example jìn ‘enter’, according to Danqing Liu (see in this book), “is halfway in grammaticalization from a full verb to a spatial goal marker”. The verb zǒu can be also used as a directional ‘away’ (see Lamarre in this book). In standard Mandarin, the motion verbs ‘come’ and ‘go’ are found as main verbs, satellites (grammaticalized), and bound morphemes (see Dan Xu in this volume). 2.2. Typology of motion events According to Talmy’s (2000) framework, the conceptualization of a motion event can be realized as set of different conceptual components in human languages. Thus two main types “Verb-framed languages” (V-languages) and “Satellite-framed languages” (S-languages) are found. In V-languages, path is expressed by the main verb, while in S-languages, path is indicated by a verb-sister position (verb affixes, verb particles). With his empiric investigation, Slobin (2004) proposes a third type, the “equipollently-framed language”. In this type, path and manner are expressed by equivalent grammatical forms. Wälchli (2001, cited by Berthele, 2004, 98) refines previous theories (Tesnière, 1959, Talmy, 2000) with three models of encoding the path, i.e. Verb encoding (by the verb stem), Adnominal encoding (by prepositions, postpositions or case marking) and Adverbal encoding (by verb affixes or verb particles). Generally speaking, Chinese and Japanese confirm the two typological differences established by Talmy: Chinese behaves as an S-language, and Japanese as a V-language. Lamarre notes that this categorization cannot account satisfactorily for the expression of deictic path. These typological features do however “exert an indirect influence on the strategies available to a language to combine deictic path, nondeictic path and the manner or cause of motion in a same verb complex” (Lamarre in this volume). Dan Xu (2006) proposes that the Chinese language has undergone a typological change from a V-language to an S-language. In contemporary standard Chinese, some motion verbs cannot match the S-language pattern because they behave as main verbs when an agent is the Figure (see Talmy, 2000) and must be considered as satellites when a patient is the Figure moved by an outside force (see Dan Xu in this volume). 2.3. Adpositions As has been mentioned, prepositions in Chinese often originated as the grammaticalization of verbs. This point of view is widely adopted by the linguistic community. However, for some locative particles attached to an NP in Chinese, there

6

D. Xu

is disagreement. They can be viewed as “postpositions” (see Danqing Liu in this volume), localizers (see Chappell and Peyraube in this volume), “locative terms” (see Fuxiang Wu in this volume) or “NP enclitics” (see Chaofen Sun in this volume). Even the term “postposition” has only begun to be used in recent research of the Chinese language due to advanced typological investigations in general linguistics. Linguistic typology shows a clear-cut correlation between VO order with prepositions and OV order with postpositions. Standard Chinese is classified as a VO language, and the term “postposition” troubles some scholars. This paradox cannot be avoided if we are limited to the existing descriptions and approaches to Chinese. Personally, I think that the Chinese language should not be treated as homogeneous because of its long history and permanent contacts with other non-Han languages. The two orders VO and OV coexisted in Old Chinese (Dan Xu, 2006), and some OV vestiges are found in frozen expressions. Modern English is a VO language, though 15th century English was an OV language. In consequence, many examples of OV order remain in words such as ‘book-seller’, ‘easter-egg-hunt’ and so forth (see Givón, 1971). The same thing happens in Chinese. Though it is considered to be a VO language, in some expressions OV order is required (see Ren Zhou, 2006)8. This means that an expected pure order is difficult to find in Chinese, which is undergoing steady often “invisible” evolution. If the status of these locative particles or postpositions is in disagreement, at least scholars agree that most of them arose from nominal elements. Actually, “postposition” implies a syntactic treatment, while “enclitic” implies a morphological interpretation. In other words, the former is freer while the latter is more bound. Perhaps these observations reflect scholars’ different perceptions of the degree of the grammaticalization of these locative particles. The debate remains open. In sign languages, the “adpositions” are expressed by hand movements and facial expressions. The general tendency is that the Ground (see Talmy, 2000) precedes Figure; “locatives preceding the subject and predicates from SL [sign language] are abundant” (see Yau). 2.4. Phonological change and its impact on syntax In presenting the languages in China in section 1, it has been suggested that Old Chinese phonology might have had clusters, voiced and unvoiced distinctions, which are completely unknown in standard contemporary Mandarin. Phonological change, often arising from phonetic modification in the first place, affects the morphology and even the syntax of a language. Scholars have long noted that the Chinese language in transmitted versions shows more and more dissyllabic words starting in the Han (206 BC–220 AD). Evidently this adjustment helped the language to avoid too many homophones. This innovation has multiple consequences in morphology as well as in syntax. Let us observe a few examples relative to spatial expressions. In Old Chinese, a single verb indicated both manner and path. However, starting in the Han, a growing number of verbs could only follow another verb, and expressed path. The satellite was no longer optional. These V-V compounds are often called verb-resultative compounds. Some of them became lexicalized dissyllabic words, while others remained at the syntactic level as a main verb plus a

Introduction

7

satellite. The localizers evolved in the same way. Qingzhi Zhu and Wenjie Chen point out that in contemporary standard Chinese, the dissyllabic localizers come from monosyllabic ones. Their meaning has become more and more abstract and their function has specialized as locative markers. This morphosyntactic change corresponds to phonological needs since the Han. Chaofen Sun notices that the spatial terms that behave more like clitics form a phonological unit with an NP9. Scholars have noted that these locative terms or postpositions play a more important role in indicating space than the preposition zài, since zài can be omitted while postpositions cannot. Lamarre indicates that standard Chinese shows a strong tendency to use bipartite path verbs (nondeictic path + deictic path) in motion events. All these mentioned facts suggest that syntactic choices are often triggered by phonological changes and constraints. 3. Organization of the Book10 In Section A “Space: a Cross-linguistic Perspective”, comparative investigations are made between several Chinese dialects, standard Mandarin, and other languages. Waxiang, a dialect almost unknown to linguists, and Bai, a language that is very controversial in linguistic discussions, have also been studied. Chappell and Peyraube’s paper investigates localizers (fāngwèicí). Localizers express the relative spatial positions of objects. They can be monosyllabic or dissyllabic. Usually monosyllabic localizers follow ordinary nouns, changing them into “place words” (chùsuǒcí) as in : zhuōzi shang (table-on) ‘on the table’. This is especially true for the two localizers shàng ‘on’ and lǐ ‘in’, the versatility of the others being quite low in spoken language. Disyllabic localizers are formed by adding a suffix (usually biān, miàn or tóu) or a prefix (yǐ or zhī). Unlike monosyllabic localizers, they can be used alone as place words and can be subjects or objects, and can be combined with nouns to express position. The paper draws a general outline of the evolution of the localizer system through the different stages of the Chinese language and their use and meaning in different Sinitic languages. As Danqing Liu points out, Modern Chinese marks a spatial role for head verbs syntactically. The word order “preposition + NP + postposition” is common. When a verb or NP has a spatial meaning, the preposition and/or postposition can be absent. The rule of omission varies between dialects: postpositions are more easily omitted in Mandarin and Cantonese than in Wu dialects, while pre-verbal prepositions are more easily omitted in Wu. Spatial prepositional phrases tend to occur pre-verbally in all modern Chinese dialects, but they more often occur post-verbally in Mandarin than in Wu dialects. In Cantonese, spatial NPs without prepositions are more inclined to follow the verb than in Mandarin or Wu dialects. Lamarre’s paper examines two typologically opposite languages: Chinese, a satellite-framed language and Japanese, a verb-framed language (according to Talmy’s framework). She discusses the linguistic encoding of deictic motion in Chinese and Japanese, focusing on clauses where the verb complex expresses the manner or the cause of motion and/or non-deictic path, with deictic direction (toward or away from the speaker). She demonstrates that Japanese and Chinese, despite their different typological status, both rely heavily on deictic directionals, i.e. spatial deixis (vs.

8

D. Xu

person deixis). She points out that Chinese also frequently uses path verbs, behaving like a Verb-framed language. In her study on locative expressions in the Waxiang dialect, spoken in Western Hunan, China, Yunji Wu shows that the locative words in Waxiang do not share much in common with either Mandarin or the Xiang dialect group, the main dialects spoken in Hunan. In the Waxiang dialect, ‘mountain’ and ‘river’ are used as references for directions or locations. There are more distinctions among objects than in Mandarin and the Xiang dialects. There is a three-way, sometimes described as even a four-way system of demonstrative pronouns. There are three words for ‘up’. There is a distinction for the word ‘side’: pHiE55ta refers to a place close to the object or person referred to, while pHiE55la refers to a place closer to the speaker. Jingqi Fu and Lin Xu studied the diachronic pathways of object markers from locative markers in Bai, a Sino-Tibetan language spoken in Yunnan, China. Bai has a pair of postpositions that are used as both locative and object markers with contrastive meanings. The authors have shown the path of grammaticalization of the locative into object markers via a dative stage. The critical contrast of ‘on X’ vs. ‘around X’ has evolved into ‘central participant’ vs. ‘peripheral participant’. Crucial to this development is the notion of contact vs. absence of it. This contrast within the locative permits a split into direct vs. indirect roles, unlike in other languages where the Dative- Locative affinity is reflected with a single morpheme. Developments of the two postpostions in different dialects of Bai are also discussed. Shun-chiu Yau analyzes space with another approach. He reveals how Sign Language (SL) has taken advantage of its visual-spatial particularity to develop independently and rapidly within a very short period. The chapter also argues that there is a strong link between SL and gestures. Gestures are common to all humans, and not exclusively practiced by deaf signers. At the theoretical level, insistence on this gestural link is of utmost importance for those who are convinced that gestures once played a crucial role in the emergence of human language. Thus, SL observations and analyses are of theoretical interest not only to sign researchers, but also to those working on general linguistics. In Section B. Space in Synchronic and Diachronic Chinese, four works have investigated space in Chinese from synchronic and diachronic approaches. Dan Xu shows that contemporary Chinese is likely a satellite-framed language, while Old Chinese was a verb-framed language. Motion verbs do not fit very well into this dichotomy because some of them have kept their verbal features in the serial verb construction, in which other verbs [-motion] are grammaticalized as satellites. Asymmetry is one of the characteristics of human language. The Chinese language also presents numerous asymmetrical cases. The asymmetry in language reflects the speaker’s asymmetric perception of space. The motion verbs lái ‘come’ and qù ‘go’, the spatial terms qián ‘before, front’ and hòu ‘back’, shàng ‘above’ and xià ‘down’ are asymmetrical at both the syntactic and semantic levels. These issues are treated from a cognitive point of view. Chaofen Sun has proposed three types of Chinese locative: definite, specific and general. The grammaticalization of the Chinese locative construction is a renewal process involving two conditions: a selectional restriction and a multi-syllabic

Introduction

9

constraint that jointly define the prototypical Chinese locative. The author claims that there are neither postpositions nor circumpositions in Chinese as in Amharic. Furthermore, none of the kind of word-order changes hypothesized by Greenberg (1995) for Amharic ever happened in Chinese history. Following Himmelmann’s (2004) characterization, in which three types of expansion (host-class, syntactic, and semantic-pragmatic) are considered essential to grammaticalization, special attention is given to the expansion of the two conditions in these three domains over time. Fuxiang Wu’s paper focuses on the source model and semantic change of the locative term hòu ‘back’ in Chinese. In specific, the semantic and morphosyntactic changes taking place when hòu occupied the head position in genitive structures are studied. According to Svorou (1986, 1993) and Heine (Heine1997, Heine et al. 1991), in cross-linguistic studies, there are three source models for spatial terms: body part terms, environmental landmarks, and relational part terms. The locative term hòu in Chinese originated from a motional process. Although this does not correspond to the mainstream source models of BACK-REGION grams, it represents a typological feature in human languages rather than a language-specific characteristic. Metonymy is the mechanism in the meaning change of hòu. Qingzhi Zhu and Wenjie Chen point out in their paper that location is an important category of space. In Chinese, from the time of the oracle bone inscriptions to the present, one means used to express space or location is demonstrative pronouns. Important changes occurred over the history of Chinese demonstrative pronouns. Before Middle Chinese, there was no co-occurrence of inessive pronouns and demonstratives to indicate location. In Late Middle Chinese, a system of inessive pronouns had developed such as zhèlǐ, zhèbiān, zhèr ‘here’, nàli, nàbiān, nàr ‘there’. These words became specialized into locative demonstrative pronouns. The paper focuses on their development, and tries to answer the question of how this evolution occurred and analyze the reasons for this change. The authors of this volume present different points of view on the expression of space in language and related theoretical issues. Space is a topic that is both classical and modern, of enduring interest. We hope that with our studies of space in languages of China, more general linguists become interested in the Chinese language and better understand how Chinese structures space. Chinese presents similarities with other languages in expressing space, as well as displaying some less common behaviors which merit further investigation. These studies of space give insight into not only general linguistics but also other domains such as anthropology and psychology.

Notes 1

See Dryer, 1992. People often refer to Mandarin of the north. In fact, Mandarin has some subgroups spread in the north, but also in south provinces such as Sichuan, Yunnan, etc. In these provinces, many non Han (non Chinese) languages are spoken, for instance languages belonging to the TibetoBurman family in west and southwest, Miao-Yao and Tai in the south. 3 Cf Language Atlas of China, 1987. 2

10

D. Xu

4

The statistics are provided by the above document dated from 1987. Those seen in Wikipedia are higher for some dialects: 836 million for Mandarin, 71 million for Wu, 71 million for Yue, 60 million for Min, etc. 5 Here I have taken the periodization proposed by Dan Xu (2006), which is simplified and takes account of typological change in Chinese Syntax. Please see also the chronological dynasties of Chinese history provided in the appendix at the end of the Introduction. 6 Cf the report of Bo Wen et al. published in Nature, 2004. 7 See Jie Zhao, 1989. 8

圖書閲覽室 *閲覽圖書室 túshū yuèlǎn shì yuèlǎn túshū shì book read room read book room ‘book reading room’ 汽車修理工 *修理汽車工 qìchē xiūlǐ gōng xiūlǐ qìchē gōng car repair worker repair car worker ‘car repairer’ 9 See also Feng Shengli, 2000. 10 This section is based on abstracts provided by the authors of this volume.

Introduction

11

References Ameka, Felix K. (1995). The linguistic construction of space in Ewe. Cognitive Linguistics. 6-2/3, 139–181. Ames, R. T. & H. Rosemont, Jr. (1998). The Analects of Confucius. New York: Ballantine Books. Baxter, William H. (1992). A Handbook of Old Chinese phonology. Trends in Linguistics Studies and Monographs 64. Berlin: Mouton de Gruyter. Bellavia, Elena. (1996). The German über . In Martin Pütz and René Dirven (Eds.), The Construal of Space in Language and Thought (pp. 73–107). Berlin. New York: Mouton de Gruyter. Berthele, Raphael. (2004). The typology of motion and posture verbs: a variationist account. In B. Kortmann (Ed.), Dialectology meets Typology: Dialect Grammar from a Cross-Linguistic Perspective (pp. 93–126). Berlin: Walter de Gruyter. Bilmes, Leela. (1995). The Grammaticalization of Thai ‘Come’ and ‘Go’. Proceedings of the Twenty–first Annual Meeting of the Berkeley Linguistics Society (pp. 33–46). Berkeley: Berkeley Linguistics Society. Bybee, Joan, Perkins, Revere & Pagliuca, William. (1994). The Evolution of Grammar. Chicago and London: University of Chicago Press. Clark, Herbert H. (1973). Space, Time, Semantics, and the Child. In Timothy E. Moore (Eds.), Cognitive Development and the Acquisition of Language (pp. 27–63). Academic Press New York and London. Comrie, B. (1981). Language universals and linguistic typology. Chicago: Chicago University Press. DeLancey, Scott. (1990). Sino-Tibetan languages. In Comrie, B. (Ed.), The World’s Major Languages (pp. 797–810). Oxford University Press. Dryer, Matthew S. (1992). The Greenbergian word order correlations. Language, Vol. 68(1), 81–138. Dryer, Matthew S. (2003). Word Order in Sino-Tibetan Languages from a Typological and Geographical Perspective. In Graham Thurgood & Randy J. LaPolla (Eds.), The Sino-Tibetan Languages (pp. 43–55). Féng Shènglì 馮勝利. Hànyǔ yùnlǜ jùfǎxué 漢語韻律 法學. Shànghǎi jiàoyù chūbǎnshè 上 海教育出版社. Givón, Talmy. (1971). Historical Syntax and Synchronic Morphology: An Archaeologist’s Field Trip. Chicago Linguistic Society, 7, 394–415. Haudricourt, A. (1954). De l’origine des tons en vietnamien. Journal Asiatique, 242, 69–82. Karlgren, B. (1957). Grammata serica recensa. Bulletin of the Museum of Far Eastern Antiquities, 29, 1–332. Language Atlas of China. (1987). Hong Kong: Longman. Levinson, Stephen C. (2003). Space in Language and Cognition. Cambridge University Press. Lǐ Fāngguì 李方桂. (1980). Shànggǔ yīn yánjiū 上古音研究. Běijīng: Shāngwù yìnshūguǎn 北京: 商務印書館. Munnich, Edward et al. (2001). Spatial Language and spatial representation: a cross-linguistic comparison. Cognition, Vol. 81, 171–207. Norman, Jerry. (1988). Chinese. Cambridge University Press. Peyraube, Alain. (2003). On the history of place words and localizers in Chinese: a cognitive approach. In Yen-hui Audrey Li & Andrew Simpson (Eds.), Functional Structure(s): Form and Interpretation (pp. 180–198). London & New York: Routledge Curzon.

12

D. Xu

Pulleyblank, E. G. (1991). Lexicon of reconstructed pronunciation in Early Middle Chinese, Late Middle Chinese and Early Mandarin. Vancouver: University of British Columbia Press. Sagart, Laurent. (1999). The Roots of Old Chinese. Amsterdam/Philadelphia: John Benjamins Publishing Company. Slobin, Dan I. (2001). Language and thought online: cognitive consequences of linguistic relativity (draft). Slobin, Dan I. (2004). The many ways to search for a frog: Linguistic typology and the expression of motion events. In Strömqvist & Verhoeven (Eds.), Relating events in narrative: Vol. 2. Typological and contextual perspectives (pp. 219–257). London: Lawrence Erlbaum. Svorou, Soteria. (1993). The Grammar of Space. Amsterdam/Philadelphia: John Benjamins Publishing Company. Talmy, L. (2000). How Language Structures Space. Toward a Cognitive Semantics. Vol. 1, 177–254. MIT Press. Tesnière, Lucien. (1959). Eléments de syntaxe structurale. Paris: Klincksieck. Traugott, E. C. (1978). On the Expression of Spatio-Temporal Relations in Language. In Greenberg et al. (Eds.), Universals of Human Language (pp. 369–400). Stanford: Stanford University Press. Wen, Bo. et al. (2004). Genetic evidence supports demic diffusion of Han culture. Nature. 431, 302–305. Xu Dan. (2006). Typological change in Chinese syntax. Oxford University Press. Zhào jié 趙傑(1989). Mǎnyǔ yánjiū 滿語研究 (Study on Mandchou), Mínzú chūbǎnshè 民族 出版社. Zhōngguó shǒuyǔ 中國手語(2003). Zhōngguó lóng rén xiéhuì biān 中國聾人協會編. Huáxià chūbǎnshè 華夏出版社. Zhōu Rèn 周韌 (2006). Gòngxìng yǔ gèxìng xià de Hànyǔ dòng-bīn shì míng fùhécí yánjiū 共性與個性 的漢語動賓飾名復合詞研究. Zhōngguó yǔwén 中國語文, 4, 301–312.

Introduction Appendix 1 Chinese Chronology

Dynasty

Dates

Xia 夏 Shang 商 Western Zhou 西周 Eastern Zhou 東周 Chun Qiu 春秋 Zhan Guo (Warring States period) 戰國 Qin 秦 Western Han 西漢 Eastern Han 東漢 Three Kingdoms 國 Western Jin 西晉 Eastern Jin 東晉 Nan Bei Chao 南北朝 Sui 隋 Tang 唐 Wu Dai 五代 Northern Song 北宋 Southern Song 南宋 Yuan 元 Ming 明 Qing 清

2070–1600 BC 1600–1046 BC 1046–771 BC 770–256 BC 770–476 BC 475–221 BC 221–206 BC 206 BC–23 AD 25–220 220–280 265–316 317–420 420–581 581–618 618–907 907–960 960–1127 1127–1279 1271–1368 1368–1644 1644–1911

Source: Xīnhuá Zìdiǎn 新華字典. 2004. Běijīng: Shāngwù yìnshūguǎn 北京: 商務印書 館.

13

D. Xu

14 Appendix 2

Sinitic Languages Map taken from Wikipedia, the free encyclopedia

Harbin

Shenyang

BEIJING

JIN

Xi’an Nanjing

MANDARIN

HUI Chongqing

XIANG

Shanghai Hangzhou

WU

GAN MIN

Changsha HAKKA Guangzhou PING CANTONAIS

Hong Kong

Taipei

CHINESE LOCALIZERS: DIACHRONY AND SOME TYPOLOGICAL CONSIDERATIONS Hilary Chappell and Alain Peyraube Centre de recherches linguistiques sur l’Asie Orientale * CNRS and EHESS, Paris, France Keywords: localizers, diachrony, typology, prototypes, Sinitic languages, Chinese Abstract: Localizers (fāngwèicí 方位词) express the relative position of objects. Either monosyllabic or dissyllabic in form, they belong to a closed sub-class. The monosyllabic set comprises the following localizers: shàng 上 ‘up’, xià 下 ‘down’, qián 前 ‘front’, hòu 后 ‘back’, lǐ 里 ‘inside’, wài 外 ‘outside’, zuǒ 左 ‘left’, yòu ‘right’, dōng 东 ‘east’, xī 西 ‘west’, nán 南 ‘south’, běi 北 ‘north’, zhōng 中 ‘middle’, jiān 间 ‘in, middle’, páng 旁 ‘side’, nèi 内 ‘inside, within’. In the majority of cases, monosyllabic localizers follow ordinary nouns, changing them into place words (chùsuǒcí 处所词) as in: zhuōzi shàng (table-on) ‘on the table’. This is especially the case for the two localizers shàng ‘on’ and lǐ ‘in’, the versatility of the others being quite reduced as far as the spoken language is concerned. Disyllabic localizers are formed by adding a suffix (typically biānr 边儿, miànr 面儿 or tóu 头) or a prefix (yĭ 以 or zhī 之). Distinct from the monosyllables, they can be used alone as place words, serve as subjects or objects, or be combined with nouns to express position. Our analysis will provide the general outlines for the evolution of the system of localizers through different stages of the Chinese language (Archaic period, Medieval period, etc.), including their use and meaning in different Sinitic languages. It is shown that the evolution followed the direction of semantic change: dìngxiàngxìng 定向性 > fànxiàngxìng 泛向性, i.e. from a specific and prototype meaning for each localizer to a general one through reinterpretation. We propose that the Prototype model, rather than the Basic Meaning model, can be applied to explain this process of semantic change. Similar processes of syntactic and semantic change which can be observed in modern Sinitic languages (or Chinese dialects) are also briefly commented upon.

1. Introduction There exists in Chinese a special category of words expressing the relative position of objects. They form a closed class (or subclass) and can be monosyllabic or disyllabic. They are called ‘localizers’ (fāngwèicí 方位词). They contribute to creating a system of spatial representation in Chinese which is markedly different from those found in the Indo-European languages. Nonetheless, neither their nature, their function nor their meaning have remained constant during the history of the Chinese language. The systems of localizers also differ in Sinitic languages other than Mandarin. In this article, we set out to examine the evolution of these words through different stages of the Chinese language (Archaic Chinese, Medieval Chinese and the modern period) and in a tentative and far less detailed way, deal with certain of their uses and meanings in several Sinitic languages, distinct from Mandarin. 15 D. Xu (ed.), Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives, 15–37. © Springer Science +Business Media B.V. 2008

H. Chappell and A. Peyraube

16

2. Place Words and Localizers In order to understand the usage of Mandarin Chinese localizers, it is necessary to grasp the concept in Chinese of chùsuǒcí 处所词 ‘place words’.1 A special category of words exists in Chinese called ‘place words’ (chùsuǒcí 处所词) by Y. R. Chao (1968: 519 ff.), that can be defined as follows: place words are substantives which can be objects of verbs or prepositions of place or movement, including the verbs lái 来 ‘come to’, dào 到 ‘arrive at’, or the prepositions zài 在 ‘at’, dào 到 ‘to’, cóng 从 ‘from’, wǎng 往 ‘toward’, etc.; (see Chao, 1968: 520 ff., Peyraube, 1980, 10 ff.). Substantives which are not place words usually cannot occupy such positions. Thus, you cannot say in Chinese * dào mén qù 到门去 [to door go] ‘go to the door’, but one can say dào Zhōngguó qù 到中国去 [to China go] ‘go to China’ or dào xuéxiào qù 到学校去 [to school go] ‘go to the school’. While the terms ‘China’ or ‘school’ are place words, ‘door’ is not. More precisely, place words can be: (i) (ii)

(iii)

(iv)

(v)

place names or geographical locations, such as Zhōngguó 中国 ‘China’, Bālí 巴黎 ‘Paris’;2 nouns with an inherently locative value, i.e. nouns for places used as place names, such as xuéxiào 学 校 ‘school’, fànguǎnr 饭 馆 儿 ‘restaurant’, túshūguǎn 图书馆 ‘library’; disyllabic localizers expressing spatial deixis such as lǐtou 里头 ‘inside’, dōngbiānr 东边儿 ‘the east side’, pángbiānr 旁边儿 ‘side, beside’; see below the whole list of disyllabic localizers; common nouns followed by monosyllabic or disyllabic localizers, such as zhuōzi shàng 桌子上 [table on] ‘on the table’, fángzi bèihòu 房子背后 [house back] ‘at the back of the house’; demonstrative locative pronouns such as zhèr 这 儿 ‘here’, nàr 那 儿 ‘there’ and nǎr 哪儿 ‘where’.

If, most of the time, place words are related to verbs or prepositions of place or movement, they are not restricted to such verbs or prepositions. They can be objects of other verbs as well. For example, wǒ kànbújiàn mén bèihòu 我看不见门背后 [I cannot-see door behind] ‘I can’t see behind the door’. Localizers (fāngwèicí 方位词), that express the relative position of things, may have different functions according to whether they are monosyllabic or disyllabic. They form a closed class (or subclass). The monosyllabic localizers comprise: shàng 上 ‘up’, xià 下 ‘down’, qián 前 ‘front’, hòu 后 ‘back’, lǐ 里 ‘inside’, wài 外 ‘outside’, zuǒ 左 ‘left’, yòu ‘right’, dōng 东 ‘east’, xī 西 ‘west’, nán 南 ‘south’, běi 北 ‘north’, zhōng 中 ‘middle’, jiān 间 ‘in, middle’, páng 旁 ‘side’, nèi 内 ‘inside, within’. Disyllabic localizers are usually formed by adding a suffix (typically biānr 边 儿, miànr 面儿 or tóu 头) or a prefix (yĭ 以 or zhī 之) to the monosyllables3.

Chinese Localizers

17

Most of the time, the monosyllabic localizers follow ordinary nouns, changing them into place words (as in (iv) above). This is especially the case for the two localizers shàng and lǐ, the degree of versatility of the others being quite low in the spoken languages. They cannot be used alone, except in certain fixed phrases based largely on Classical Chinese (as in Shàng yǒu tiāntáng, xià yǒu Sū Háng 上有天堂 下有苏杭 [above there-is heaven below there-is Suzhou Hangzhou] ‘Above there is Heaven, below there are Suzhou and Hangzhou’), or as objects of the prepositions wǎng 往, xiàng 向 or cháo 朝 ‘to, toward’, as in wǎng lǐ zǒu 往里走 [to inside move] ‘move forward’; see Wen Lian (1959: 8), Y. R. Chao (1968: 525), Peyraube (1980: 31). Which categories do these place words and localizers belong to? As far as the localizers are concerned, nearly every viewpoint has been put forward. They have been considered to be adjectives (Ma Jianzhong, [1898], Mǎ shì wén tōng, chapter 3), adverbs (Lü Shuxiang, 1947, Li Jingxi and Liu Shiru, 1955), nominal suffixes (Cartier, 1972), postpositions forming a discontinuous constituent with the prepositions they frequently co-occur with (Hagège, 1975: 220 ff., Peyraube, 1980: 53 ff.),4 spatial enclitics or particles (Sun Chaofen, forthcoming), and even pronouns (Rygaloff, 1973: 143). Usually, however, they are considered as a subclass of nouns, for the good reason that it does not seem desirable to distinguish the monosyllabic localizers from the disyllabic localizers. Thus the latter, like the monosyllabic localizers, express position when they are combined with nouns, but they also can be used alone as place words, functioning as either subjects or objects. Despite this, not everyone agrees with the judgement of place words as belonging to the category of nouns. It could be a case of an autonomous category of words, which, unlike common nouns, do not have the possibility of being modified by classifier phrases of the form Numeral + Classifier, but which may, by contrast, be objects of the prepositions zài 在 ‘at’ or dào 到 ‘to’ or have an adverbial function, precisely what common nouns are able to do. This is the view of Li Chongxing (1992) and Chu Zexiang (1997, 2006) who claim that Zhu Dexi (1982: 42–45) already tended to the view that place words, and also time words and localizers were separate and autonomous categories.5 We take the following view in this article, on the model of Chu Zexiang (2006) that localizers – in the same way as place words – form a subclass of nominals, that nonetheless can be considered autonomous. However, this is not the main focus of our proposal which is to outline the historical evolution of the localizers. 3. From Archaic Chinese (11th c. BC) to Medieval Chinese (3rd c. AD) 3.1. Archaic Chinese In Archaic Chinese (11th – 2nd centuries BC.), localizers are only monosyllabic and much less frequent than they are in contemporary Chinese. The main reason why they are less used is because they are not needed to follow ordinary nouns in order to transform them into place words. In other words, there are no fundamental differences between ordinary nouns and place words in the classical language. Another characteristic of the period is that the locative prepositions (for the large

18

H. Chappell and A. Peyraube

part yú 于/於 ‘at, to, from, etc.’) are needed to introduce place words when these place words are not direct objects of locative or movement verbs. Localizers do already exist in Chinese since the Pre-Archaic period (14th – th 11 c. BC.), i.e. in the language of the oracle bone inscriptions. Zhao Cheng (1988: 269–272) makes an inventory of seven of them: dōng 東 ‘east’, nán 南 ‘south’, xī 西 ‘west’, běi 北 ‘north’, zhōng 中 ‘middle’, zuǒ 左 ‘left’ and yòu ‘right’. Shàng 上 and xià 下 are also attested in the language of the oracle bone inscriptions, but their meaning is probably respectively ‘Heaven’ and ‘Earth’, and not ‘above, on’ and ‘below, under’. These seven localizers are still used during the Archaic period, together with some others, namely shàng and xià, meaning by this stage ‘above, on’ and ‘below, under’, nèi 內 ‘inside’, wài 外 ‘outside’, and, to a lesser extent, qián 前 ‘front’ and hòu 后 ‘back’, which appeared later, though not as late as lǐ 裏, which Chu Zexiang (1996) dates back to the 4th or 5th c. AD. See also Yang Bojun and He Leshi (1992: 89–92). Apart from very rare and marginal exceptions (for instance two examples of wàimian 外面 ‘outside’ in the Mozi [but this text dated as 3rd c. B.C. displays several original phenomena not attested in other texts of the same period; see Peyraube 1988: 101, Zhang Wanqi 1998]), these localizers are always monosyllabic and, contrary to contemporary Chinese, they can be used alone to express place (chùsuǒ 處 所 ). Thus, they are place words and they can be subjects, objects or even adverbials. They are used both as (i) place words like the disyllabic localizers in contemporary Chinese (see ex. 1– 4), and as (ii) position words (or spatial deictics) following nouns (see examples above, 5–9). Examples: (1)

瞻之在前，忽焉在後 (論語: 子罕) zhān zhī zài qián hū yān zài hòu look them be-at front suddenly then be-at behind ‘Looking at them, they are in front of me, but they then suddenly appear behind.’

(2)

周公居東二年 (書經: 金滕) Zhōu gōng jū dōng èr nián Zhou prince live east two year ‘Prince Zhou has been living in the east for the last two years.’

(3)

今拜乎上 (論語: 子罕) jīn bài hū shàng now worship at above ‘Now, the practice is to worship above (the hall).’

(4)

晉侯在外十九年矣 (左傳: 僖公二十八年) Jìn hóu zài wài shí jiǔ nián yǐ Jin prince be-at outside ten nine year part. ‘The Prince of Jin has been away for nineteen years.’

Chinese Localizers

(5)

王坐於堂上 (孟子: 梁惠王上) wáng zuò yú táng shàng king sit at hall aloft ‘The king was sitting aloft in the hall.’

(6)

王立於沼上 (孟子: 梁惠王上) wáng lì yú zhǎo shàng king stand at pond on ‘The king was standing by a pond.’ 6

(7)

則是方四十里為阱於國中 (孟子: 梁惠王下) zé shì fāng sì shí lǐ wéi jǐng yú guó zhōng thus this square four ten li be pitfall at kingdom middle ‘Thus those forty square li are a pitfall in the middle of the kingdom.’

(8)

(9)

19

孟孫立於房外 左傳：定公六年 Mèng Sūn lì yú fáng wài Meng Sun stand at room outside ‘Meng Sun was standing outside the room.’ 射其左，越於車下，射其 ，斃於車中 左傳：成公二年 shè qī zuǒ yuè yú chē xià shè qī yòu bì yú chē zhōng shoot (an arrow) his left pass at chariot under shoot his right die at chariot inside ‘(He) shot (the one who was on) his left who fell down under the chariot, he shot (the one who was) on his right who died inside the chariot.’

If common nouns in Classical Chinese do not necessarily need to be followed by a localizer to be place words, as is the case for contemporary Chinese, there are nevertheless two conditions for this to be possible: (i) they must be objects of verbs of place or movement (ex. 10), and/or (ii) they must be introduced by a locative preposition, yú 于/於, or less often hū 乎, zhū 諸 or even zhī 之 (zhī can be considered sometimes as an equivalent of zhū, according to Li Chongxing 1992), as shown in the following examples (11–13). There are some exceptions, especially when the nouns used as place words are subjects (as in example 14) or in some other cases when the nouns are taken to have a locative value and are used as place names, as in (15): (10)

不之堯之子而之舜 孟子：萬章上 bù zhī Yáo zhī zǐ ér zhī Shùn negation go Yao det.-part. on but go Shun ‘(The princes) went not to the son of Yao, but to Shun.’

H. Chappell and A. Peyraube

20

(11)

子張書諸紳 論語：衛靈公 Zǐ Zhāng shū zhū shēn Zi Zhang write it+at sash ‘Zi Zhang wrote them (these words) on his sash.’

[zhū 諸 = zhī 之 + yú 於] (12)

(13)

(14)

(15)

婦人笑於房 左傳：宣公十七年 fùrén xiào yú fáng princess laugh at room ‘The princess laughed in (her) room.’ 公薨於車 左傳：桓公十八年 gōng hōng yú chē prince die at chariot ‘The prince died in the chariot.’ 塗有餓莩而不知 孟子：梁惠王上 tú yǒu è piáo ér bù zhī fā road there-be hungry body and negation know start ‘There are people dying from famine on the road and (you) don’t know how to start (issuing the stores of your granaries).’ 樹吾墓檟 左傳：哀公十一年 shù wú mù jiǎ plant my tomb catalpa ‘Plant catalpas on my tomb.’ 7

Even when place words are formed by Nouns + Localizers, it is still rare to have the preposition yú deleted. An exception is: (16)

韓厥執縶馬前 左傳：成公二年 Hán Jué zhí zhì mǎ qián Han Jue take strap horse in-front-of ‘Han Jue took a strap in front of the horses.’

Only when the verb has the third person or demonstrative pronoun zhī as its object, the preposition yú does not seem to be needed, as in: (17)

子產使校人蓄之池 孟子：萬章上 Zǐchǎn shǐ xiào rén xù zhī chí Zichan order field-officer man keep it pond ‘Zichan ordered his field-officer to keep it (the fish) on the pound.’

Chinese Localizers

21

However, as hypothesized by Li Chongxing (1992), one can assume, in these cases, that a locative preposition yú has been deleted before the noun chí ‘pond’ (in ex. 17), or even that zhī might involve the fusion: zhū = zhī ‘it’ + yú ‘to’. 3.2. Pre-Medieval Chinese In Pre-Medieval Chinese, which roughly corresponds to the Han period (206 BC. – 220 AD), the characteristics of place words and localizers have changed. These changes can be summmarized as follows: (i)

common nouns are no longer used as place words, or, at least, it is not as easy as it was in Late Archaic. Examples of place words in the Shĭ jì 史記 (1st century B.C.) which are formed by ‘Common nouns + Localizers’ and which probably could not have been used as place words without a localizer are given in examples (18–20). (ii) the locative preposition yú is no longer needed to introduce a place word which is not the object of a verb of place or movement. Examples (18–20) also reveal this trend, as well as those in (22–23). (iii) localizers become functional words, though they still express a clear and precise position of things. Moreover, localizers which are disyllabic in composition, start to be used at the end of this period. Examples: (18)

(19)

(20)

桓公與 人蔡姬戲船中 史記：齊 公世家 Huán gōng yú fūren Cāi Jī xì chuán zhōng Huang prince and spouse Cai Ji have-fun boat in ‘The prince Huang and his spouse Cai Ji had fun in a boat.’ 孔子去 適宋，與弟子習禮大樹下 史記：孔子世家 Kǒngzǐ qù Cáo shì Sòng yú dìzǐ xí lǐ dà shù xià Kongzi leave Cao go Song with disciple practice rite big tree under ‘Kongzi left Cao and went to Song to practice the rites with his disciples under a big tree.’ 出朝，則抱以適趙盾所 史記：晉世家 chū cháo zé bào yǐ shì Zhào Dùn suǒ leave court then carry-in-the-arms with go Zhao Dun place ‘(She) left the court and then, carrying (her child) in her arms, went with (him) to Zhao Dun’s place.’

In this last example, there is no localizer after the personal noun Zhao Dun, but a noun expressing a place (suǒ ‘residence’), which shows that personal names can no longer be used as place words, as they could be earlier in Classical Chinese (see ex. 10). It is obvious if one compares this example taken from the Shĭ jì with the similar one extracted from the Zuo zhuan (5th c. B.C.), where the personal name Zhao (Dun) is used as a place word:

22

(21)

(22)

H. Chappell and A. Peyraube

出朝，則抱以適趙氏 左傳：文公七年 chū cháo zé bào yǐ shì Zhào shì leave court then carry-in-the-arms with go Zhao family ‘(She) left the court and then, carrying (her child) in her arms, went with (him) to Zhao’s place.’ 8 西與秦將楊熊戰白馬 史記：高祖本記 Xī yú Qín jiāng Yáng Xióng zhàn Bái Mǎ Xi and Qin general Yang Xiong fight Bai Ma ‘Xi and the general of Qin, Yang Xiong, fought at Bai Ma.’

[In this example, the place word is a place name] (23)

殺義帝江南 史記：高祖本記 shā Yì Dì jiāng nán kill Yi Di river south ‘(He) killed Yi Di south of the river.’

[The place word is a noun followed by a localizer]9. There are still cases, of course, where both the locative preposition yú and a localizer following the noun are present, as in: (24)

種瓜於長安城東 史記：蕭相國世家 zhòng guā yú Cháng’ān chéng dōng plant melon at Chang’an city east ‘(He) planted melons on the eastern side of the Chang’an city.’

However, according to Li Chongxing (1992), in chapter 8 of the Shi jì (Gāo zǔ běn jì), there are 80 place words which are not introduced by the locative preposition yú, while only 14 are. If one compares this situation to the one prevailing in Late Archaic Chinese, there is no doubt that there has been a change. Such a change, as mentioned before, is mainly due to the fact that ordinary nouns and place words are now differentiated. As the place word is now formed by a noun with its postposed localizer, the locative preposition is not necessary. But the most important characteristic of the localizers in this period (PreMedieval) is undoubtedly the third one (see (iii) above), i.e. localizers tend to behave like real functional words. They start to play an important grammatical role, as one of their main functions is to follow nouns and transform them into place words. They still express of course a real and precise position, having what Lü Shuxiang (1984: 294) called dìngxiàngxìng 定向性 (‘precise position character’): ‘on’, ‘under’, ‘in’, ‘outside’, ‘in front of ’, ‘behind’, ‘left’, etc. But in the case of some of them (as for zhōng 中 or jiān 間 ‘in’), the meaning of a precise position is somewhat blurred. Examples showing these tendencies follow:

Chinese Localizers

(25)

(26)

(27)

23

是時桓楚亡在澤中 史記：項羽本記 shì shí Huán Chǔ wáng zài zé zhōng that time Huan Chu take-refuge (be)-at marsh in ‘At that time, Huang Chu took refuge in the marsh.’ 於空中作音樂 道行般若經 yú kōng zhōng zuò yīnyuè at air in make music ‘(They) made music in the air.’ 不復還在世間 人本欲生經 bú fù huán zài shì jiān negation again return at world in ‘(He will) not return to the world any more.’

Another important change which began in Pre-Medieval is the appearance of disyllabic localizers. They are still infrequent, but their occurrence is not so rare. Shàngtóu 上頭, for instance, is used on several occasions in Zheng Xuan’s (127–200) commentaries of the Shī jīng 詩經. There are also several examples of disyllabic localizers in the Late Han vernacular Buddhist translations of the 2nd c. AD: these are shàngtóu 上頭 in Cháng ā hán shí bàofǎ jīng 長阿含十報法經 (Taisho 13) translated by An Shigao, shàngtóu 上頭 and hòutóu 後頭, but also zuǒmiàn 左面 in Xiū xíng běnqǐ jīng 修行本起經 (Taisho 184) by Kang Mengxiang and Zhu Dali. These disyllabic localizers can be used alone as subjects or objects, as in contemporary Chinese. For example: (28)

上頭有 長阿含十報法經 shàngtóu yǒu above there-be ‘There are (some) above.’

Disyllabic localizers, however, are not yet used after nouns. Monosyllabic localizers are still preferred in the position following nouns to transform them into place words. 4. Medieval Chinese (3rd–13th c. AD) During the Six Dynasties period (220–581), i.e. in Early Medieval Chinese, there are still common nouns and monosyllabic localizers used as place words, as in Archaic Chinese (see examples 29–31), but this kind of example becomes rarer. Usually, place words are either place names (as in example 32) or ordinary nouns followed by localizers or pronouns (examples 33–36):

24

(29)

(30)

H. Chappell and A. Peyraube

載王於車而殺之 世說新語：仇隙 zài wáng yú chē ér shā zhī carry king at chariot and kill him ‘(He) carried the king in the chariot and killed him.’ 上有萬仞之高，下有不測之深 世說新語：德行 shàng yǒu wàn rèn zhī gāo xià yǒu bú cè zhī shēn above there-be ten-thousand ren det.-part. height below there-be negation fathom det.-part. depth ‘Above there is a height of ten thousand ren, below there is an unfathomable depth.’

[The monosyllabic localizers shàng and xià are used as place word subjects. For more examples of this kind in the Shì shuō xīn yǔ dated 420–444, see Zhang Zhende et al. 1995] (31)

淵自督糧在後 三國志：271 Yuān zì dū liáng zài hòu Yuan personally take-charge-of provision (be)-at rear ‘Yuan personally took charge of the provisions in the rear.’

[The localizer hòu is a place word object of the verb zài] (32)

祖崩於洛陽 三國志：294 Tàizǔ bēng yú Luòyáng Taizu pass-away at Luoyang ‘Taizu passed away at Luoyang.’

[The place word Luòyáng is a place name] (33)

(34)

(35)

於本母前宴飲 世說新語：方正 yú Běn mǔ qián yànyǐn at Ben mother in-front-of feast ‘(They) feast in front of Ben’s mother.’ 10 在船中彈琴 世說新語：言語 zài chuán zhōng tán qín (be)-at boat in play lute ‘(He) was in the boat and played the lute.’ 母王 人在壁後聽之 世說新語：文學 mǔ Wáng fūrén zài bì hòu tīng zhī mother Wang lady (be)-at wall behind listen them ‘(His) mother Lady Wang was behind the wall listening to them.’

Chinese Localizers

(36)

25

可擲著門外 世說新語：方正 kě zhì zhe mén wài can throw to door outside ‘(It) can be thrown out of the door.’

In all the above examples, the localizers are monosyllabic. However, there are now a few cases of place words formed by Nouns + Disyllabic Localizers; one can find several examples in the Buddhist translations of the period, for instance in the Chū yào jīng 出曜经 (ca. 4th c.): (37)

鼠在瓶裏頭 出矅經：637 1 shǔ zài píng lǐtóu mouse be-at bottle inside ‘The mouse is in the bottle.’

(38)

尋出門外頭 出矅經：631 1 xún chū mén wàitóu search go-out door outside ‘(He) went outside to search.’ 11

According to Chu Zexiang (2006), the locative syntagma with the highest frequency is the group composed of Noun + Localizer. In fact, for the Shì shuō xīn yǔ he identifies 199 cases of Preposition + Noun without a localizer, 114 cases of Preposition + Noun + Localizer and 261 cases of Noun + Localizer, that is, without any preposition. A comparison with the situation found in the Shĭ jì enables a measurement of these diachronic changes. It turns out that in the Shĭ jì, representtative of the earlier Pre-Medieval stage, we find 1507 cases of Preposition + Noun, 219 cases of Preposition + Noun + Localizer and 1126 cases of Noun + Localizer.12 But the most important point to note, for this period of Early Medieval, is uncontestably the capacity of certain locatives to gradually acquire the function of expressing undifferentiated localization. It was shown earlier that in the PreMedieval period, the meaning of some localizers began to be blurred (see examples 25–27 in particular). In Early Medieval Chinese, this tendency strengthens considerably. Several of the monosyllabic localizers no longer express a ‘precise position’ (the dìngxiàngxìng 定向性 of Lü Shuxiang 1984: 294) but a rather ‘vague position’ (fànxiàngxìng 泛向性). The grammatical function of the localizer then completely overrides its original semantic value. This is the second tide of the grammaticalization of position words: Noun > dìngxiàngxìng or precise localizer > fànxiàngxìng or generic localizer. The cases of shàng 上 and zhōng 中 are particularly prominent, but cases of qián 前 ‘in front of ’, 下 xià ‘under’, biān 邊 ‘on the side of ’, tóu 頭 ‘at the head of ’ etc. are also involved. See Li Chongxing (1992). Examples of this phenomenon follow:

26

(39)

(40)

H. Chappell and A. Peyraube

長文尚小，載著車中... 文若亦小，坐著膝前 世說新語：德行 Chángwén shàng xiǎo zài zhè chē zhōng ... Wénruò yì xiǎo zuò zhe xī qián Changwen still tiny carry at carriage in ... Wenruo also tiny sit at knee in-front-of ‘Changwen was still tiny and was carried in the carriage ... Wenruo, who was also tiny, was seated on (his) lap.’ 雖長大猶抱著膝上 世說新語：方正 suī zhǎng dà yǒu bào zhè xī shàng even-so grew big still hold at knee on ‘(And) even when (he) was fully grown, (he) still used to hold (him) on his lap.’

It is evident from (39) and (40) that qián 前 and shàng 上 could be used interchangeably in Early Medieval to express the same generic position. (41)

負米一斗，送著寺中 六度集經·23 下 fù mǐ yì dǒu sòng zhe sì zhōng carry-on-the-back rice one dou deliver to temple in ‘(He) carried one dou of rice on his back and delivered (it) to the temple.’

Especially zhōng is often used as an undifferentiated localizer, though it may still have its ‘precise position’ interpretation: ‘inside’ or ‘among’ or even ‘between’, as in (42) and (43): (42)

(43)

口中含嚼， 著掌中 百喻經·上 kǒu zhōng hán jiáo tù zhè zhǎng zhōng mouth in keep-in-the-mouth chew spit at palm inside ‘(He) chewed it inside his mouth and then spat it out into the palms of his hands.’ 時五百人中有一人最上智慧 生經一 shì wǔ bǎi rén zhōng yǒu yì rén zuì shàng zhìhuì this five hundred people among there-be one man most top intelligence ‘Among these five hundred people, there (must) be one who is most intelligent.’

Hence, the Late Medieval period (6th–13th c.) is mainly characterized by the use of almost all of the monosyllabic localizers to express an undifferentiated localization, i.e. a vague interpretation of position, which later became uncommon and so remains today. Wang Ying (1995) has studied the use of localizers in 1928 Tang poems included in the anthology Táng shī bié cái 唐識別裁 (dates for the Tang dynasty are 618–907). The following monosyllabic localizers are attested, together with the frequency of their occurrence: shàng 上 (319), xià 下 (167), qián 前 (178), hòu 后

Chinese Localizers

27

(29), lǐ 裏 (122), nèi 内 (19), wài 外] (157), zuǒ 左 (6), yòu (7), zhōng 中 (423), jiān 間 (110), dōng 東 (116), nán 南 (113), xī 西 (133), běi 北 (120), páng 旁 (25), zé 則 (8), biān 邊 (103), tóu 頭 (96), dĭ 底 (23). Disyllabic localizers are not yet frequent: only 3 occurrences for shàngtou 上頭, 3 for qiántou 前頭, 5 for zhōngjiān 中間 (and 1 for zhōngyāng 中央), 1 for dōngbiān 东邊 (and 1 for dōngmiàn 東面, 1 for dōngtóu 東頭), 1 for nánbiān 南邊 (1 for nánmiàn 南面 and 2 for nántóu 南頭), 1 for xībiān 西邊 (and 3 for xītóu 西頭), 1 for pángzé 旁則 and 1 for zépáng 則旁. The monosyllabic localizers are used (i) as adverbials before a verb; (ii) after a preposition forming a locative PP with it that can be used as an adverbial or as a complement; and (iii) after nouns, transforming them into place words. Situation (iii) represents by far the most common use of the localizers under the Tang, which explains why Ōta (1958) calls the localizers ‘post-auxiliary nouns’ (hòuzhù míngcí 后助 詞). In short, the situation at the end of the Tang period is not very far removed from that which is found today in contemporary Chinese. What is striking about Wang Ying’s analysis is that in noticing the fact that almost all of the monosyllabic localizers can express undifferentiated localization, if not a very vaguely defined position, he concurs with remarks of the Qing philologist Yú Yuè 俞樾 (1821–1907) given in his Qǔ yuán zá zuàn 曲園雜纂. Thus, for instance, the localizer dōng ‘east’ is sometimes used to express the meaning of wài ‘outside’, or xī ‘west’ is used for nèi ‘inside’. As a matter of fact, if different editions of certain texts are compared, it is not rare to find different localizers in different editions. In Wang Fanzhi’s poems, we find: qián ‘in front of ’ used for biān ‘at the side of ’, xià ‘below’ for nèi ‘inside’, dōng ‘east’ replaced by xī ‘west’, tóu ‘at the head of ’ replaced by shàng ‘on, above’, zhōng ‘in, middle’ replaced by biān ‘at the side of ’, lǐ ‘in, inside’ replaced by tóu ‘at the head of ’, biān ‘at the side of ’ replaced by qián ‘in front of ’, and even xià ‘under, below’ replaced by shàng ‘on, above’ or dōng ‘east’ replaced by xī ‘west’, i.e. some localizers are replaced by their antonyms! In the following examples, hǎi xītóu = hǎi nántóu (sea west = sea south) in (44), lín shàng = lín wài (forest on = forest outside) in (45), and lǐ shàng = lǐ xià (litree on = li-tree under) in (46): (44)

節 行 落盡海南頭 西頭 全唐詩 125/289 jié lŭxíng máo luò jìn hái nántóu (xītóu) bamboo-stick travel ox-tail-decoration fall entirely sea south (west) ‘(He) traveled with his bamboo stick decorated with ox-tail (so often that) the ox-tail decorations all fell into the southern (western) end of the sea.’

(45)

林外 上 九江 (全唐詩 126/293) lín wài (shàng) jiǔ jiāng píng forest beyond (on) Jiu Jiang flat ‘The Jiujiang river was flat beyond the forest.’

28

(46)

H. Chappell and A. Peyraube

紫燕櫪上 下 嘶(全唐詩 169/396) zǐ yàn lì shàng (xià) sī purple swallow li-tree on (under) cry-out ‘Purple swallows cried out on (under) the li-tree.’

This situation shows that the concrete meaning of some monosyllabic localizers has often been completely bleached through a process of grammaticalization, and that these monosyllabic localizers have become real function markers indicating a very vague position with the straightforward syntactic function of transforming the nouns to which they are attached into place words. The meaning of ‘precise position’ (dìngxiàngxìng) of the monosyllabic localizers is, at the same time, still retained. Wang Ying gives the following figures for the two most common localizers of the period, biān and wài: for biān, 73 of them express a precise position and 30 an undifferentiated localization, while for wài 108 express a precise position and 49 an undifferentiated localization. Peyraube (2003), after having discussed several models taken from Klein and Nüse (1997), has proposed to account for such a situation in the following terms: For each localizer, there is a nucleus meaning, which can be precisely described, the one corresponding to a precise position. This meaning, however, is modifiable through semantic operations, possibly motivated by textual constraints. Other interpretations can then be derived. The ‘basic meaning’ model put forward by Klein and Nüse (1997) is apparently not the right one to fit the Medieval Chinese situation. It is not likely that Chinese localizers could obtain specific and particular interpretations by adding some features to a general or basic meaning. Things seemed to have worked in the opposite way. The lexical meanings of shàng, xià, qián, hòu, etc. are to be considered specific ones, representing a typical interpretation: ‘up’, ‘down’, ‘front’, ‘back’, etc. Other uses in particular contexts are then derived through reinterpretations. If the spatial features of the relatum (or ‘ground’, ‘landmark’) (for instance ABOVE, IN CONTACT WITH, in the case of shàng) are not in accordance with the specific meaning, a reinterpretation is needed, as the speaker conceptualizes the theme and the relatum differently. This reinterpretation implies a cognitive effort. As the evolution of Chinese localizers has been from ‘precise position or specific meaning’ (dìngxiàngxìng) to ‘vague position or general meaning’ (fàngxiàngxìng), it is most felicitous for our analysis to have recourse to a prototype model. 5. Localizers in other Sinitic Languages In this final section on modern Chinese dialects, we draw some parallels with different stages in the history of Chinese, concerning diachronic change in both spatial syntax and semantics, without however making any claims about historical relationships. To some extent, we find the paradoxical situation of undifferentiated localization expressed by certain localizers in several contemporary Sinitic languages. First of all, in Mandarin, the locatives shàng and lǐ, or even xià, often express a vague localisation removed from their prototypical meaning. This is normally

Chinese Localizers

29

viewed as a derived meaning of locatives. To begin, let us consider the examples from Mandarin: (47)

(48)

她在飞机上看书 tā zài fēijī shàng kàn shū 3sg at plane on read book ‘She reads a book on the plane.’ (and not on top of the plane) 他住在乡下 tā zhù zài xiāng xià 3sg live at country below ‘He lives in the country.’ (and not under the country).

These derived metaphorical meanings exist in all languages.13 Relating to our earlier discussion in section 4 of Medieval Chinese, in the Huaxian 华县 dialect of Shaanxi (Zhongyuan Mandarin), the locative xià has the meaning of shàng ‘on’, for which two examples are given below: (49)

(50)

把书放到桌子下 bǎ shū fàng dào zhuōzī xià object-marker book put at table on ‘Put the book on the table.’ 我在楼下住哩 wǒ zài lóu xià zhù li 1sg at building above live aspect ‘I live in one of the top storeys’.14

The Jiangjin dialect ( 江 津 , Southwestern Mandarin, Sichuan) uses the postnominal localizer gāo 高 in a fairly undifferentiated manner to mean all of ‘in’, ‘on’, ‘about’, ‘up’ and ‘on top of ’. Depending on the class, shape and function of an object, it may refer to (i) the top and/or large flat surface of objects including tables, seats, sofas and beds (ii) the inside part of a container, but equally its outside surface, for certain round shaped objects with openings such as bsowls and basins (iii) the surface of thin objects including paper, textiles and sheets (iv) the top or high part of an object such as a building or mountain (v) in a derived sense, the contents or scope of an article or newspaper (See Huang Borong 1996: 113–114). The Shanghai dialect of the Wu group is similarly interesting in that it makes use of an undifferentiated locative, lA53 拉, distinct from the demonstrative pronouns

30

H. Chappell and A. Peyraube

which serve to change common names into place words. Specifically, it is used to transform possessor nouns into place nouns, while retaining the possessive meaning of the NP, as the following example shows (see Xu and Tang 1988: 414-415; Huang Borong 1996: 114). (51)

Shanghai dialect (Wu) 小毛拉父 ˛iç34mç13 lA53 vu13 (name) LOC:POSS father ‘Xiao Mao’s father’ (literally: the father at Xiao Mao’s place)

Mandarin 小毛家里的爸爸 xiăo máo jiā-li de bàba (name) family POSS father ‘Xiao Mao’s father’ (literally: the father in Xiao Mao’s family’)

The use of lA53 拉 can be compared with the use of jiā-li de ‘family POSS’ in standard Mandarin, typical in expressions denoting family and kin relationships. The use of a cognitive schema based on a locative construction to express possession is a well-known strategy (Heine, Claudi and Hünnemeyer 1991; Chappell and McGregor 1995). Compared with lA5 拉, the demonstrative i53mi13 伊面 in Shanghainese which is close in function to postnominal nàr or nàli ‘there’ cannot be used to change a possessor noun into a locative possessor: (52)

Shanghai dialect (Wu) 老太婆拉兒子 lç13t’a34bu13 lA53˜i13tsÆ elderly:lady LOC:POSS son ‘the elderly lady’s son’

Mandarin 老 婆家里的兒子 lăo tài pó jiā-li de érzi elderly:lady family POSS son ‘the elderly lady’s son’

Shanghai dialect (Wu) * 老太婆伊面兒子 *lç13t’a34bu13 i53mi 13 ˜i13tsÆ34 *elderly:lady LOC:DEM son By way of contrast, lA53 拉 cannot follow a common noun that is not part of a possessive NP, whereas i53mi13 伊面 ‘there’ can: (53)

Shanghai dialect (Wu) *到台子拉去 *tç34dE13 tsÆ34 lA53t˛’y 34 to table LOC:POSS go

到台子伊面去 tç34dE13 tsÆ34 i53mi 13t˛’y34 to table LOC go ‘go to the counter’

These examples with lA53 拉 , in particular, recall the Pre-Medieval period which saw the adoption of a postnominal localizer as a requirement with personal

Chinese Localizers

31

names when these had a locative meaning equivalent to ‘at X’s place’; see example (20) above from the Shĭ jì. With regard to the second theme of grammaticalization and semantic change discussed in the diachronic sections of this analysis, we next consider data on localizers from the variety of Sin-on Hakka (Huizhou subgroup, Southern Guangdong), in which most localizers have a disyllabic form. These can be used independently as nouns, or as adverbials, or may follow the head noun (Chappell and Lamarre 2005: 88). Recall that disyllabic localizers first began to appear in the Pre-Medieval period, becoming more and more common by the time of Late Medieval Chinese (see section 4 above). Each spatial relation has several alternative forms for which some are listed in (54): (54)

Sin-on Hakka (Huizhou subgroup, Southern Guangdong) tang3 kau1 頂高 ‘above’ kyok6 ha1 腳下 4 1 šong kau 上高 ‘above’ tai3 ha1 底下 4 2 men theu 面頭 ‘above’ nui4 tu3 ‘within’ ngoi4 men4 外面 內肚 1 4 sim nui ‘within’ ngoi4 poi4 外背 心內 1 4 ti poi ‘inside’ □背 3 1 tu li ‘inside’ 肚裡 tshai4 nui4 在內(肚) ‘inside’ (tu3)

‘below’ ‘below’ ‘outside’ ‘outside’

The following example shows that these disyllabic localizers can serve alone as head nouns: (55)

在腳下放緊 tshoi1 kyok6 ha1 fong4 kin3 at below place-DUR ‘to put it down below’ They may also directly serve as the object of a verb:

(56)

去奈裡去个裡 hi4 nai4 li1? hi4 kai4 li1 go wherego there Where are you going to?’ ‘I’m going over there.’

It appears that the strategy of Noun + Localizer is generally employed as the preferred structure, recalling the use of a similar pattern that became prevalent in the Pre-Medieval and Early Medieval Chinese periods and that has remained so up until the present. Note that in the following examples, taken from Chappell and Lamarre (2005: 95), the use of tshoi1 ‘at’ is optional, and that both monosyllabic and disyllabic localizers can occur in the postnominal position.

H. Chappell and A. Peyraube

32

(57)

(tshoi1) tsok6 šong4 (OR: tsok6 shong4) (tshoi1) then1 tang3 (tshoi1) tshui4 kai4 hat6 ha1

(在) 桌上

‘on the table’

(在) 天頂 (在) 罪个轄下

(tshoi1) wuk6 poi4 (tshoi1) ho2 pen1 (tshoi1) fong2 tu3 li1 (tshai4) mun2 ngoi4

(在) 屋背 (在) 河邊 (在) 房肚裡 (在) 門外

‘in Heaven (above)’ ‘under the domination of sin’ ‘behind the house’ ‘at the river’ ‘inside the room’ ‘outside the door’

Another extremely interesting observation which can be made about spatial deictics in this variety of Hakka concerns processes of grammaticalization and semantic change. In particular, the expression poi4 heu1 背後 meaning ‘behind’ (composed with poi4 背 ‘back’) can be combined as a kind of enclitic with other spatial expressions such as the following: (58)

šong4 poi4 ‘up there’ 上背 ha1 poi4 ‘down there’ 下背 ha1 poi4 kai4 下背个 ‘lower’

ti1 poi4 □背 ngoi4 poi4 外背

‘inside’ ‘outside’

Even the term for ‘roof ’, uses this morpheme: nga3 poi4 (tang3) 瓦背主瓦背頂 ‘tiled roof ’. This appears to suggest that the same general principles of semantic change are in evidence for Hakka that we have described for Pre-Medieval Chinese in section 3 above. That is, a concrete noun, with the core meaning of a body part, ‘back’, grammaticalizes into a function word, specifically a spatial deictic, with the precise meaning of ‘behind’ via the process of metaphor. In its wake, a second semantic change takes place involving generalization to the meaning of ‘position’, ‘side’. As for the case of Early Medieval Chinese, presented above, these two similar semantic changes can be summarized as follows: (i) fully lexical concrete noun (body part term) > specific localizer (dìngxiàngxìng) (ii) specific localizer (dìngxiàngxìng) > generic localizer (fànxiàngxìng). The first change, whereby body part nouns evolve into spatial deictics, is not at all uncommon crosslinguistically, as discussed in detail in Heine and Claudi (1986) on SPACE IS AN OBJECT with regard to similar phenomena in African languages. For example, words for ‘forehead’ or ‘breast’ may grammaticalize into the spatial deictic ‘in front of ’ while the container metaphor associated with ‘belly’ evolves into the spatial deictic ‘in’, ‘within’. An interesting question for further research would be

Chinese Localizers

33

why certain body part terms are privileged over others in undergoing the second semantic change of generalization, that is, why poi4 背 ‘back’ generalized further from the spatial meaning ‘behind’ to refer to vague location in the formation of Hakka disyllabic localizers, but not tu3 肚 ‘belly’, which stopped after its first semantic change, involving metaphor, to ‘within’. 6. Conclusion A few concluding remarks can be made from the above sketch. To be qualified as place words, common nouns or personal names do not need to be followed by a localizer in Classical Chinese. Localizers are, however, found placed after nouns, but then they always express a precise, concrete localization (dìngxiàngxìng). When they are not direct objects of locative verbs or verbs of movement, place words are usually introduced by a locative preposition. From the Pre-Medieval period, it is more and more the case that the preposition yú is no longer needed to introduce place words, while the use of localizers following ordinary nouns or personal names starts to become necessary to qualify them as place words. Disyllabic localizers also start to be used towards the end of this period. Less and less we find ordinary nouns being used as place words in the Early Medieval period. Moreover, some monosyllabic localizers start to express an undifferentiated localization, a generic position (fànxiàngxìng). Following on from this, in Late Medieval, the use of monosyllabic localizers which express an undifferentiated localization spreads. During the Tang period, it affects almost all the localizers, which can, consequently, often be used interchangeably. The evolution follows the direction: dìngxiàngxìng > fànxiàngxìng, i.e. from a specific and prototype meaning for each localizer to a general one through reinterpretation. The Prototype model, rather than the Basic Meaning model, can be applied to explain this process of semantic change, as argued in detail in Peyraube (2003). Finally, it was shown that similar processes of syntactic and semantic change can be observed in the modern Chinese dialects, an area deserving of further detailed research.

* This research has benefited from the Australian Research Council (ARC) Grant n° A00106262 and from a grant of the French Ministry of Higher Education and Research (Action Concertée Incitative n° 03326.)

34

H. Chappell and A. Peyraube

Notes 1

Unless otherwise indicated, the use of ‘Chinese’ generally refers in this analysis to ‘Mandarin Chinese’. 2 Sun Chaofen (2006, forthcoming) who calls place words “spatial nominals” has observed that all place names have to be polysyllabic. He also proposes a ‘multi-syllabic constraint’ for the NPs marked by the preposition zài ‘at’: zài requires a multi-syllabic spatial nominal. 3 For all possible combinations, including combinations of two monosyllabic localizers to form disyllabic localizers, see Peyraube 2003: 184. 4 Greenberg (1995) discusses this subject, with reference to circumpositions in other languages, noting that languages which have such circumpositions are quite rare and might thus be in a transitory stage, in the process of changing from ‘languages with prepositions’ into ‘languages with postpositions’ or changing from ‘languages with postpositions’ into ‘languages with prepositions’. 5 Zhu Dexi, however, classes these place words, time words and localizers within the nominal category (tǐcí 体词), on the same level as the numerals (shùcí 数词) or the classifiers (liàngcí 量词). 6 The English translations in 5 and 6 have been taken from Legge (1861). As indicated by the translation, the localizer shàng does not express the meaning ‘on’, but a vague location. 7 This is certainly a case of a common noun. Even though ‘tomb’ is not preceded by a preposition, nor followed by a localizer, this common noun may be considered as one with a locative value. 8 Another interesting example is the following one from the Shĭ jì, which can be compared with example (15) above taken from Zuǒ zhuàn. The remark can be made that a locative has been added to transform the noun into a place word: shù wǔ mù shàng (plant my tomb on) ‘plant (it) on my tomb’. 9 Actually there is another example in the same chapter of the Shĭ jì where the preposition yú is present: shā Yì Dì yú jiāng nán. It shows that we only have a tendency. 10 As shown by example (33), there are many locative PPs introduced by yú in Early Medieval which are preverbal, a situation which really began in the Late Han period, ca. 200 AD. See Peyraube (1994). 11 Some other examples of wàimiàn 外面 and zuǒmiàn 左面 can be found in the Xián yù jīng 賢愚經 (ca. 445). 12 In the contemporary language, Chu Zexiang (2006) counts, in a corpus composed of several novels, 83 occurrences of ‘Prep. + Noun’, 227 occurrences of ‘Prep. + Noun + Loc.’, and 367 occurrences of ‘Noun + Loc.’ If we compare these figures with those given for chapters 41 to 60 of Hóng Lóu Mèng 紅樓夢 (18th century), we can observe an increase in the quantity ‘Prep. + Noun’ without a localizer: 58 ‘Prep. + Noun’, 348 ‘Prep. + Noun + Loc.’ and 528 ‘Noun + Loc.’. It would nevertheless be taking a risk to think that the reverse movement could take place (locative phrase with a localizer and without a preposition > locative phrase with a preposition and without a localizer), by means of a cyclic movement, thus allowing us to return to the situation which characterized the language of the Archaic period, particularly the period of Classical Chinese. In a separate article, Chu Zexiang (2004) additionally observes

Chinese Localizers

35

that the localizer in a PP headed by the preposition zài can be dropped when the noun itself is preceded by a modifier (including a relative clause) as long as the NP is polysyllabic. He also observes that the longer the modifier is, the more likely it is for the enclitic to be dropped (see also Sun Chaofen (forthcoming) and Chappell and Thompson (1992) for a similar patterning regarding the presence or absence of de 的 in heavy NPs). 13 Thus, in English, as in Chinese, the usual expression is ‘on the train’, ‘on the plane’ while in French it is respectively ‘dans le train’, ‘dans l’avion’. Another interesting comparison is seen in the following French expressions and their English equivalents: ‘sur la route’ = ‘on the road’, but ‘dans la rue’ = ‘in the street’. 14 To express the meaning of ‘below’, the disyllable dǐxià has to be used; cf. Huang Borong (1996: 113–115).

36

H. Chappell and A. Peyraube

References Cartier, Alice. (1972). Les indicateurs de lieu en chinois. La linguistique 2,8. Chappell, Hilary & Sandra A. Thompson. (1992). The semantics and pragmatics of associative de in Mandarin discourse. Cahiers de Linguistique Asie Orientale 21, 199–229. Chappell, Hilary & William McGregor (Eds.). (1995). The grammar of inalienability. A typological perspective on body parts terms and the part-whole relation. (Empirical approaches to language typology 14). Berlin: Mouton de Gruyter. Chappell, Hilary & Christine Lamarre. (2005). A grammar and lexicon of Hakka: Historical materials from the Basel Mission Library. Collection des Cahiers de Linguistique – Asie Orientale 8, Paris: École des Hautes Études en Sciences Sociales. Chao, Yuen Ren. (1968). A Grammar of Spoken Chinese. Berkeley and Los Angeles: University of California Press. Chu, Zexiang. (1996). Hanyu kongjian fangwei duanyu lishi yanbian de jige tedian [Some characteristics of the historical evolution of space and locative phrases]. Gu hanyu yanjiu 1, 57–61. Chu, Zexiang. (1997). Xiandai hanyu de mingming xing chusuoci [A study of nomenclative place words in Contemporary Chinese]. Zhongguo yuwen 5, 326–335. Chu, Zexiang. (2004). Hanyu ‘zai + fangwei duanyu’ li fangweici de yinxian jizhi [The mechanisms responsible for the deletion of localizers in ‘zai + locative phrases’]. Zhongguo yuwen 2, 112–122. Chu, Zexiang. (2006). Hanyu chusuoci de cilei diwei ji qi leixingxue yiyi [The wordclassstatus of place words in Chinese and its typological significance]. Zhongguo yuwen 3, 216–224. Greenberg, Joseph H. (1995). The diachronic typological approach to language. In M. Shibatani and T. Bynon (Eds.), Approaches to Language Typology (pp. 145–166). Oxford: Clarendon Press. Hagège, Claude. (1975). Le problème linguistique des prépositions et la solution chinoise. Louvain: Peeters. Heine, Bernd and Ulrike Claudi. (1986). On the metaphorical base of grammar. Studies in Language, 10(2), 297–335. Heine, Bernd, Ulrike Claudi, & Friederike Hünnemeyer. (1991). Grammaticalization: A Conceptual Framework. Chicago: University of Chicago Press. Huang, Borong (Ed.). (1996). Hanyu fangyan yufa leibian [Grammatical Sketch of Chinese Dialects]. Qingdao: Qingdao chubanshe. Klein, Wolfgang, & Ralf Nüse. (1997). La complexité du simple: l’expression de la spatialité dans le langage humain. In Langage et cognition spatiale (pp. 1–23). Paris: Masson. Legge, James. 1861. The Chinese classics. Taipei: Southern Material Center, Inc., 1983 (reprinted). Li, Chongxing. (1992). Chusuoci fazhan lishi de chubu kaocha [A preliminary study on the historical development of place words]. In Jindai hanyu yanjiu, eds. Hu Zhu’an, Yang Naisi and Jiang Shaoyu, 24–63. Beijing: Shangwu yinshuguan. Li, Jingxi & Liu, Shiru. (1955). Zhongguo yufa jiaocai [Materials of Chinese Grammar]. Beijing: Wushi nian chubanshe. Lü, Shuxiang. (1947). Zhongguo wenfa yaolüe [Precise of Chinese Grammar]. Shanghai: Shangwu yinshuguan. Lü, Shuxiang. (1984). Hanyu yufa lunwen ji [Selected readings on Chinese Grammar]. Beijing: Shangwi yinshuguan.

Chinese Localizers

37

Ma, Jianzhong. (1898). Ma shi wen tong [Basic principles for writing clearly and coherently by Mister Ma]. Beijing: Shangwu yinshuguan, 1983 (reprinted). Ōta, Tatsuo. (1958). Chungokogu Rekishi Bunpo [Historical Grammar of Chinese]. Tokyo: Konan Shoin. Peyraube, Alain. (1980). Les constructions locatives en chinois moderne. Paris: Langages croisés. Peyraube, Alain. (1988). Syntaxe diachronique du chinois – Evolution des constructions datives du XIVème siècle av. J.-C. au XVIIIème siècle. Paris: Collège de France. Peyraube, Alain. (1994). On the history of Chinese locative prepositions. Zhongguo jingnei yuyan ji yuyanxue 2, 361–387. Peyraube, Alain. (2003). On the history of place words and localizers in Chinese: a cognitiveapproach. In Yen-hui, Audrey Li & Andrew Simpson (Eds.), Functional Structure(s): Form and Interpretation (pp. 180–198). London & New York: Routledge Curzon. Rygaloff, Alexis. (1973). Grammaire élémentaire du chinois. Paris: PUF. Sun, Chaofen. (May 30–31, 2006). Historical geography, phonology, word order and the Chinese locatives. International Conference celebrating the publication of the complete works of Li Fang-kuei and the publication of the Bulletin of Chinese Linguistics. Academia Sinica, Nankang, Taiwan. Sun, Chaofen. Spatial mapping and grammaticalization in Chinese (forthcoming). Wang, Ying. (1995). Tang shi fangweici shiyong qingkuang kaocha [Research on the use oflocalizers under the Tang dynasty]. Lü Shuxiang xiansheng jiushi huadan jinian wen ji, (pp. 207–213). Beijing: Shangwu yinshuguan. Wen, Lian. (1959). Chusuo, shijian he fangwei [Place, Time and Position]. Shanghai: Jiaoyu chubanshe. Xu, Baohua & Tang, Zhenzhu. (1988). Shanghai Shiqu Fangyanzhi [A Record of the Dialect of Shanghai City]. Hong Kong: Joint Publishers. Yang, Bojun & He, Leshi. (1992). Gu hanyu yufa ji qi fazhan [A Grammar of Ancient Chinese and its evolution]. Beijing: Yuwen chubanshe. Zhang, Wanqi. (1998). Liangci mei de chansheng ji qi lishi yanbian [Appearance and historical evolution of the classifier ‘mei’]. Zhongguo yuwen 3, 208–217. Zhang, Zhende et al. (1995). Shi shuo xin yu yuyan yanjiu [Linguistic Research on the Shi shuo xin yu]. Sichuan: Bashui shushe. Zhao, Cheng. (1988). Jiaguwen jianming cidian [Concise dictionary of oracle-bone inscriptions]. Beijing: Zhonghua shuju. Zhu, Dexi. (1982). Yufa jiangyi [Lessons on Grammar]. Beijing: Shangwu yinshuguan.

SYNTAX OF SPACE ACROSS CHINESE DIALECTS: CONSPIRING AND COMPETING PRINCIPLES AND FACTORS Danqing LIU Chinese Academy of Social Sciences Keywords: spatial expressions, syntactic commonality and diversity, Chinese dialects, universal principles, typology Abstract: This chapter examines the commonality and diversity of syntactic structures of spatial expressions among several major Chinese dialects, including Mandarin, Wu dialects (in particular Shanghainese) and Cantonese, in the framework of typology. It claims that what underlies the apparent dialectal diversity in spatial expressions is the interaction among several factors, which include universal principles such as the Relator Principle and the Principle of Economy, common mechanisms such as grammaticalization and word order harmony, and language-specific rules such as the prosodic-syntactic rule of stress assignment. The conspiring and competing relationships among these factors help shape the syntax of space across Chinese dialects.

1. Introduction: Universality and Diversity of Spatial Expressions Across Languages Spatial roles, including location, spatial source and goal (grammatically locative, ablative, elative or allative cases), are among the most basic roles that a language often appeals to in a sentence. It is well-known that a typical news report should contain information concerning the Six Whs, namely Who, When, Where, What (to do), (to) Whom/What, and Why. In fact, these six Whs represent the crucial information which a sentence should provide in human languages. It is thus not surprising that almost all of these Whs have their corresponding syntactic functions in a clause or sentence: subject for who, predicative verb for what (to do), object for (to) whom/what, temporal adverbial for when, spatial adverbial for where, and causal phrase/clause for why. In this sense, expressing spatial roles is a universal task of human grammars. A spatial ‘role’ is not just a semantic category, but it also performs a syntactic function with a thematic role in a clause. It is not the case that an expression which is spatial in meaning must correspond to a spatial role in a clause syntactically. For instance, while Melbourne is a place name, hence, a spatial expression semanticcally, in the sentence ‘Melbourne plays a big role in the economy and education of Australia’, Melbourne does not take up the spatial role but a sort of agentive role, and functions as the subject of the sentence syntactically. It is only in sentences like ‘We met each other in Melbourne’ that Melbourne will take up the spatial role. Moreover, the syntactic counterparts of the Six Whs are not of equal importance in argument structure and clausal syntax. For instance, the constituents answering Who and (to) Whom/What typically occur in the subject and the object positions of a

39 D. Xu (ed.), Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives, 39–67. © Springer Science +Business Media B.V. 2008

D. Liu

40

clause respectively, with both being direct arguments projected by the verbal head, as assumed by the formal linguists. The temporal and spatial constituents, corresponding to When and Where respectively, usually occur as adverbials (adverbs, adpositional phrases, or adverbial clauses), hence indirect thematic roles. In current syntactic theories, spatial adverbials can take up either argument position in a sentence, such as in ‘He put the book in his study’, or adjunct position, such as in ‘He is reading in his study’. Since they are adverbials, there are always morphological or syntactic means to mark their status as adverbials in human languages. All languages are equipped with some devices to mark spatial expressions, but the nature of these devices may vary. In languages with morphological cases, spatial cases usually belong to one of the basic cases. In Latin, a typical inflectional language, ablative cases are assigned for most nouns and locative cases for some placedenoting nouns. In Uyghur, a typical agglutinative language spoken in Xinjiang, China, there are three spatial cases marked by suffixes to nouns. They are -ta for location, -¥a for locative goal and -din for locative source. Relevant examples are given below (cited from Litip Tohti 2001: 226–227). (1)

a. Bala karwat-ta uxlawatidu child bed-Loc slept ‘The child slept on the bed’ b. Män bazar-¥a bardim I market-Goal went ‘I went to the market’ c. Biz šinja0-din kälduq we Xinjiang-Source came ‘We came from Xinjiang’.

In some languages, spatial roles are linked by adpositions (prepositions/postpositions), which are syntactically more independent and analytical than morphological case-marking devices. For example, in English there are no spatial cases but only prepositions as space markers, such as in, at, on, under, beneath, above, to, from and complex prepositions like in front of, on top of. In the Yi language spoken in Bijie, Guizhou, China, an analytical Tibeto-Burman language, spatial constituents are marked by postpositions, as in “Ýi21x–21 ko33 džo55 ” (lit. pool-in-at), “¥o13ka33 k*ue33 k*–21”(lit. south from north to) (data from Ding 1993: 276, 280). In some languages, a preposition co-occurs with a postposition to signal a spatial expression, as in German, Ewe, Amharic and Pashto. (2)

German: von hier ab (Zhang C.: 356) from here from ‘from here on’

Syntax of Space Across Chinese Dialects

(3)

Ewe: le xn megbé (Heine et al. 1991: 140–141) at house behind ‘behind the house’

(4)

Amharic: bä-bet wast (Greenberg 1995) in house interior ‘in(side) the house’

(5)

Pashto: pa kor ks e (Greenberg 1995) in house interior ‘in(side) the house’.

41

A combined form of a preposition and a postposition is referred to as circumposition. Moreover, a combination of a morphological case and an adposition can also be found in some languages. For example, in Russian, although nouns can be marked with locative case, an NP in locative case alone is not enough to encode a spatial expression in a clause and a preposition is required. The boundary between morphological case markers and adpositions is not always clear-cut. In Japanese, for example, spatial markers such as -de ‘at, in, on’, -e ‘in, into, to’ and -ni ‘at, in, to’ are called (locative) ‘case markers’ by some grammarians and ‘postpositions’ by others. Such a situation exists more readily in postpositional languages, since postpositions are often less free and more dependent than prepositions and thus behave more like case affixes. Even in languages whose spatial expressions are usually marked by grammatical means, there are still cases in which an NP serves directly as a syntactic spatial expression based on its semantic meaning without grammatical marking. This is considered as a syntactic spatial expression because it occurs in an adverbial position without occupying any slot for direct arguments. Under such a case, word order might still work if it is not a case of free word order. This is often the case in Chinese. The way-phrase in English sometimes serves this function as well, as in ‘Let’s meet the way out’. Contemporary typological studies make a crucial distinction between headmarking and dependent-marking (cf. Nichols 1986). If a dependent-head relationship is marked on the head, it is an instance of head-marking; and if it is marked on the dependent, it is an instance of dependent-marking. For example, in English, one can say ‘the man’s house’, with the genitive suffix on the dependent ‘the man’. In Hungarian, by contrast, a different morphological strategy is adopted, as shown in (6) (6)

az ember hàz-a the man house-3Sg ‘the man’s house’ (cited from Nichols 1986: 57)

While the possessive noun ember is left unmarked morphologically, the head noun hàz is suffixed with a, the marker for a noun modified by a third person singular

42

D. Liu

genitive. In addition, since adpositions are also markers for dependent relations, in the English phrase ‘the house of the man’, the preposition of is also a dependentmarker, since it takes the genitive phrase ‘the man’ as its immediate constituent. The parameter of dependent-marking vs. head-marking is also relevant to spatial expressions. It is observed that some categories like person, number and gender prefer head-marking, while adjuncts or peripheral/oblique NPs denoting location, direction, instrument, manner, etc. strongly favor dependent-marking in the form of case markers or adpositions. (cf. Nichols 1986: 78–79, Song 2001: 200). Yet, as will be seen in this chapter, some types of head-marking do occur in spatial categories in Chinese. The framework underlying the above analysis follows a typological tradition reflected in many language descriptions. From the point of view of Talmy’s typology of spatial expressions (cf. Lamarre’s chapter in this volume), this framework may fit the Satellite-framed languages more than Verb-framed languages. In Verb-framed languages, clauses expressing spatial relations take the form as the English clause ‘John entered the room’ instead of that of ‘John came in the room’. In the former clause, with the ‘path verb’ to enter serving as the predicative verb, the spatial expression the room functions as the direct object of the verb instead of an adverbial. In this chapter, however, we will still follow the traditional typological framework, in view of the following three points. Firstly, Chinese, especially Modern Chinese, is basically a Satellite-framed language, and the traditional model can serve to describe and analyze its syntax of space well. Secondly, the focus of Talmy’s insightful framework is how verbs conceptualize spatial meanings while the main objective of this chapter is to examine how the spatial NPs in a clause are expressed syntactically. The traditional framework seems to serve our objective better. Thirdly, Talmy’s model focuses on sentences denoting spatial movement and existence, while the traditional model aims to cover more phenomena of typological significance, which allows us to integrate morphosyntactic facts regarding spatial expressions into a more global typological view of the languages and dialects in question. Readers who are interested in Chinese facts under Talmy’s typological framework may refer to Lamarre’s chapter in this volume. In the following, I will begin my analysis by examining the typological characteristics of spatial expressions in Chinese across dialects. Data are mainly taken from Standard Mandarin, Beijing Mandarin and a number of Wu dialects, particularly Shanghainese Wu. 2. Commonality in Spatial Expressions Among Chinese Dialects Chinese dialects have much in common in the ways they express spatial relations. The followings are some main universals among dialects in this respect. 2.1. Lacking morphological spatial cases, such as locative, ablative, elative and allative cases Chinese as a whole is a ‘caseless’ language. All the meanings and functions indicated by morphological cases in other languages are expressed in this language by analytical means or simply left unmarked [1]. This caseless state also holds true in

Syntax of Space Across Chinese Dialects

43

spatial category. The existence of morphological spatial cases is not reported in any varieties of Chinese. Below are some examples. For the abbreviations used in this chapter, see the Abbreviations. (7)

出于幽谷，迁于乔木。 诗经 chū yú yōu gǔ, qiān yú qiáo mù. (Shi Jing) go-out at deep valley move at tall tree ‘(The birds) went out from a deep valley and moved to a tall tree’

(8)

我在教室里看书。 wǒ zài jiàoshì-lǐ kàn-shù. I at classroom-inside read-book ‘I am/was reading books in the classroom’

(9)

他跳进了池塘。 tā tiào-jìn-le chítáng. 3Sg jump-enter-Asp pool ‘He jumped into the pool.’

(10) 侬住辣上海口伐？ nung zy leq Zanghe va? 2Sg live at Shanghai Ptc ‘Do you live in Shanghai?’ (11) 佢行去学校。 kheui hang heui hok’hau 3Sg walk go school ‘He walked to school.’ In (7), both clauses use the preposition 于 yú ‘at, to, from’ to introduce the source or goal in the postverbal position, while in (8), a preposition (grammaticalized from a verb giving existential meaning) and a postposition (from a locative relational noun) are used to mark the spatial adverbial preceding the verb. In this volume, postpositions are also called ‘localizers’ by Chappell and Peyraube, and ‘locatives’ by Sun. (7) and (8) exemplify the typical patterns of spatial expressions in Archaic and Modern Chinese respectively. In (9), a directional verb, or ‘path verb’, 进 jìn ‘enter/into’ is inserted between the verb and the spatial expression (as goal). The word jìn is halfway in grammaticalization from a full verb to a spatial goal marker. The way Shanghainese uses to introduce spatial constituents, as in (10), is similar to that in Archaic Chinese which signals a locative role following the verb with a preposition. In (11), a semi-grammaticalized deictic motion verb (‘path verb’) heui ‘go/to’ is used to signal the goal role in Cantonese. The Mandarin or Shanghainese counterpart of this verb, being a full verb instead, cannot function in this way. Although various means are adopted by different varieties of Chinese to link spatial

44

D. Liu

constituents, no morphological cases are involved. Prepositions, postpositions, and directional verbs used as markers of spatial roles in the above examples are either function words or semi-grammaticalized words. 2.2. Spatial components with zero marking in a clause NPs without spatial marking can also play a spatial role in a clause under certain conditions. Word order then becomes the only means of conveying spatial meaning in these cases. Yet there are differences among dialects as to where these unmarked spatial components are placed and what kind of NPs can function in this way. Consider the examples below. (12) 他住(在)干面胡同 Tā zhù (zài) Gānmiàn hútong 3Sg lives (at) Ganmian Lane ‘He lives in Ganmian Hutong’

Compare: 伊住*(辣)干面胡同) i zy laq Koemie udung he lives at Ganmian Lane ‘He lives in Ganmian Hutong’

(13) 老王今朝(到)杭州去。 Lo-uong jingtso (to) ongtseu qi Lao-Wang today (to) Hangzhou go ‘Lao-Wang will go to Hangzhou today.’ (14) 老王今朝杭州来。 Lo-uong jingtso (to/zung) ongtseu le Lao-Wang today (to/from) Hangzhou come ‘Lao-Wang will come to/from Hangzhou today.’ (15) 佢今日嚟/去深圳。 Kheoi kanyat lai/heui Santsan. 3Sg today come/go Shenzhen ‘S/he will come/go to Shenzhen today.’ For some verbs taking an inherent spatial argument such as those denoting ‘come, go, fly, live, sit, put’, spatial expressions can follow the verb without a preposition in Mandarin as well as in many other varieties of Chinese, though prepositions like 在 zài can also be inserted between the verb and the spatial expression. In Shanghainese and most Wu dialects, however, the preposition in such cases can hardly be omitted (cf. Shi 1999, section 2.2.3, Liu 2003: 232, 276). Interestingly, Shanghainese and many Wu dialects can (sometimes must) have the spatial expression preceding the verb and the preposition can then be omitted, leading to ambiguity, as shown in (14). When the spatial preposition-free expression precedes the verb 去 qi ‘go’, it denotes the goal (destination), and the omitted preposition must be 到 to ‘to’, as in (13). If the spatial expression precedes the verb 来 le ‘come’, its interpretation will depend on the relation between the place denoted

Syntax of Space Across Chinese Dialects

45

by the expression and the location of speech. If the speech location and the denoted place are identical, the spatial expression will indicate the goal, and the omitted preposition should be 到 to ‘to’. If the utterance occurs in a place other than the place denoted by the spatial expression, the expression will indicate the source (the starting location), and the omitted preposition should then be 从 zung ‘from’, hence the ambiguity in (14). If the verb is a directional compound V + Vd with le/qi ‘come/go’ as Vd (directional verb), the preposition-free spatial expression will invariably be interpreted as source regardless of its speech location, e.g. 杭州带来 Ongtseu ta-le (Hangzhou carry-come: carry here from Hangzhou) and 杭州送去 Ongtseu sung-qi (Hangzhou send-go: send there from Hangzhou). The above features are summarized in the following table. Table 1: Interpretation of Spatial NPs with zero marking in Shanghainese PATTERN ROLE OMITTED PREPOSITION NP + 去 qi ‘go’

goal

到 to ‘to, towards’

NP + 来 le ‘come’

goal (speech place= NP)

从 zung ‘from’

source (speech place ≠NP)

到 to ‘to, towards’

NP + V + 来 le/去 qi

source

从 zung ‘from’

Finally, in Cantonese, the verb 嚟 or 去 lai/heui ‘come/go’ usually selects a goal role which follows the verb directly without appealing to a preposition, as in (15). This pattern has gradually become an acceptable or even canonical pattern in Standard Mandarin since the 1950’s, though colloquial Beijing Mandarin still prefers the pattern of ‘preposition + NP + 来/去 lai/qu ‘come/go’’. In many Wu dialects such as the Suzhou and Shaoxing dialects, this Cantonese pattern is ungrammatical, though in contemporary Shanghainese, this pattern has become marginally acceptable due to the influence of Standard Mandarin. We will come back to this issue in the later part of this chapter with a typological explanation. 2.3. Intrinsic spatial verbs As related to the phenomena discussed in Section 2.2, certain intrinsic spatial verbs in every variety of Chinese can assign spatial roles to NPs with no spatial marking. I termed these verbs as ‘spatial-role-assigning verbs’ in Liu (2002a). These verbs usually denote static existence (being), movement (come, go, walk, run, fly, swim…), placement of things (put, hang, carry…) and so forth, with the exact semantic range of the lexicon in question varying among dialects. The most typical and prevalent example of such verbs might be 在 zai ‘be at’ in Mandarin and its equivalents in other dialects. Examples of such verbs in Standard Mandarin are given below.

D. Liu

46

(16)

(17)

(18)

(19)

(20)

(21)

他在上海。 tā zài Shànghǎi. 3Sg be(-at) Shanghai ‘He/She is in Shanghai’ 我明天到香港。 wǒ míngtiān dào Xiānggǎng. I tomorrow arrive(-in) Hong-Kong ‘I will arrive in Hong Kong tomorrow’ 请走人行道。 qǐng zǒu rénxíngdào. please walk(-on) sidewalk ‘Please walk on the sidewalk’ 大门朝南。 dàmén cháo nán. main-gate face south. ‘The main gate faces south’ 这封信寄广州。 zhè-fēng xìn jì Guǎngzhōu. This-Cls letter mail Guangzhou. ‘This letter will be mailed to Guangzhou’ 郑明去年调省政府了。 Zhèng Míng qùnián diào shěngzhèngfǔ le. Zheng Ming last-year change-job(-to) provincial-government Ptc. ‘Zheng Ming changed his job to the provincial government’.

Some of these verbs play not only a special role in the synchronic syntax of Chinese, but also an active role in the diachronic process in which they constitute an important grammaticalizational source for spatial prepositions (cf. Liu 2002a). Some examples are 在 zài (to exist at, to be located at > at, in, on), 到 dào (arrive at/in > toward, to, till) and 朝 cháo (to be toward > toward). Regarding role-assignment, we should mention another group of verbs which are directional verbs such as 进 jìn ‘enter’, 出 chū ‘go out’, 过 guò ‘go over, pass by, go through’ in Mandarin. Their functions will be discussed in the later part of this chapter. 2.4. Circumpositions for spatial meanings A spatial circumposition typically consists of a preposition denoting spatial relations like ‘at’, ‘to’, ‘from’ and so on, and a postposition or locative relational noun denoting spatial locations like ‘inside’, ‘above’, ‘below’, ‘on’. In Mandarin, for

Syntax of Space Across Chinese Dialects

47

instance, we have zài…lǐ (在…里 at…inside ≈ in), cóng…wài (从… 外 from outside), dào…shàng (到…上 to…on ≈ onto) and so on[2]. 2.5. The spatial component preceding the verb as the usual order It is common in modern Chinese dialects to have the spatial component preceding the verb, especially in cases where a preposition or circumposition is employed (PP + V), despite the fact that Archaic Chinese (cf. Peyraube 1994, Sun 1996, Zhang Ch. 2002, among many others) and most VO languages (cf. Dryer 1992) have the reverse order as their basic order, i.e. ‘V + PP’. The following examples with the same interpretation are from Archaic Chinese, Standard Mandarin, Shanghainese and Cantonese. (21) a. 晋人及秦人战于令狐。( 春秋 文公 7 年) Jìnrén jí Qínrén zhàn yú Lìnghú (Chunqiu, Wengong Year 7) Jin-people and Qin-people fight at Linghu ‘Jin People and Qin People fought in Linghu’ b. 晋人和秦人在令狐打 。 Jìnrén hé Qínrén zài lìnghu dǎzhàng Jin-people and Qin-people at Linghu fight c. 晋人脱秦人辣令狐打 。 Jingning teq Djingning laq Ling’u tangtsang Jin-people and Qin-people at Linghu fight d. 晋人同秦人喺令狐打 。 Tseonyan tung Tsheonyan hai Ling’wu ta’tseong Jin-people and Qin-people at Linghu fight. As shown in the above examples, the prepositional phrase for ‘in Linghu’ follows the verb for ‘fight’ in Archaic Chinese while it precedes the verb in all the three modern dialects. 2.6. Summary Chinese as a whole is an analytic language without morphological case, and there are thus no spatial expressions marked with case morphology. The basic syntactic way of indicating a spatial role is to employ prepositions (usually derived from verbs), postpositions (usually derived from locative nouns) or a combination of these two categories, which can be termed as circumpositions. When the verb, the noun, or both of them are of inherent spatial meaning, static or dynamic, the grammatical marking might and sometimes must be omitted, though the range of such verbs and nouns varies among different dialects. Modern dialects of Chinese unanimously have the order of PP + VP as the basic word order for spatial expressions headed by a preposition. This order is almost the only counterexample to a widely-attested typological observation, which shows that all VO languages except Chinese have the order of VP + PP (cf. Dryer 1992, 2003). In this respect,

D. Liu

48

Archaic Chinese is comparable to most VO languages rather than to Modern Chinese. 3． ．Dialectal Diversity in Terms of Spatial Expression: Marking Types and the Rule Governing the Use of Space Markers Although Chinese dialects share many obvious commonalities in terms of syntactic realization of spatial constituents as described above, they also behave differently from one another in many aspects. 3.1. Ellipsis of locative postpositions In principle, a preposition in Modern Chinese cannot take an ‘ordinary’ NP as its argument, such as *zài zhuōzi (lit. at table). Its argument should be either an NP with locative meaning in itself (e.g., zài Běijīng, ‘at/in Beijing’) or a phrase headed by a noun-derived postposition (e.g., zài zhuōzi-shàng, lit. ‘at table on’). However, the exact range of nouns with locative meaning differs greatly from dialect to dialect. Place names are treated as inherent locative nouns in all dialects, and the postposition is not allowed to occur after a place name, e.g., zài Běijīng (*lǐ), ‘at Beijing (inside)’. The diversity is found in common nouns. In Standard Mandarin or Beijing Mandarin, a fairly large number of nouns are treated as both entity nouns and locative nouns. For these nouns, a postposition is optional in a prepositional phrase. If one puts a postposition after the noun, the noun is treated as an entity noun, e.g., 在学校里 zài xuéxiào-lǐ (lit. ‘at school inside’). One can also omit this postposition, allowing the preposition to govern the noun directly, e.g., 在学校 zài xuéxiào. In the latter case, xuéxiào ‘school’ is treated in the same way in which a place name is treated. This group of nouns covers many items like those denoting ‘ground, building, store, house, hall, office, room, government, company, post-office, park, garden, city, lake, river, sea’ (there are observations that forms with or without postpositions might have slight semantic or pragmatic differences, c.f. Chu 2003, etc.). In contrary, in Shanghainese and many Wu dialects, such nouns are all treated as entity nouns, and they must therefore be followed by a locative postposition after a spatial preposition. Only place names are treated as spatial nouns and can be governed by a spatial preposition. Compare the two sentences below. (22)

a. 在学校(里) zài xuéxiào(-lǐ) at school(inside) b. 辣学堂*(里) laq uqdong*(-li) at school-inside

/ / / / / /

到操场(上) dào cāochǎng(-shàng) to playground-on 到操场*(上) to tshozang*(-long) to playground -on.

The difference is not only in semantic range, but would also lead to syntactic consequences. In the case of common nouns, the occurrence of a postposition is only optional syntactically in a spatial prepositional phrase in Standard Mandarin where

Syntax of Space Across Chinese Dialects

49

it is obligatory in Wu dialects. Wu dialects allow the absence of spatial postpositions only in proper nouns which denote place names. The more active role of postpositions in Wu dialects is typologically harmonic with the stronger verb-final tendency in Wu (cf. Liu 2001, 2003)[3]. 3.2. The ‘ellipsis’ of spatial prepositions: Postposition phrases/locative NP or ircumposition phrases A canonical and full-fledged spatial expression in Chinese contains three parts which can be represented in the following form. (23)

preposition + NP + postposition

In certain conditions, the preposition, the postposition or both can be omitted, resulting in a partial-fledged expression in various dialects. We have already come across cases involving the ellipsis of postpositions in Sections 2.3, 3.2 and those involving the absence of both in Section 2.2. In this section, we will look at cases which involve the ellipsis of prepositions in spatial expressions. A sharp contrast lies between Wu dialects and Beijing Mandarin regarding their omission of the preposition. In Beijing Mandarin, the postverbal spatial expression often occurs without prepositions, while in Wu dialects it is the preverbal spatial expressions that can occur, in a relatively freer manner, without prepositions. Let us consider the postverbal position first. (24) a. 你睡(在/de)小床上吧。 nǐ shuì zài/de xiǎochuáng-shàng ba 2Sg sleep at small-bed–on Ptc ‘You sleep on the the small bed.’ b. 侬困*(辣)小床上吧。 nung khun laq xiozong-long ba. 2Sg sleep at small-bed-on Ptc (25) a. 行李搁(在/de)房间里了。 xíngli gē (zài/de) fángjiān-li le. luggage put at bedroom-inside Ptc ‘The luggage has been put in the bedroom’ b. 行李摆*(辣)房间里勒。 angli pa *(laq) vongke-li leq luggage put at bedroom inside Ptc. The pair of examples in (24) involves an intransitive verb while the pair in (25), a transitive verb. In both pairs, the behavior of the prepositions is consistent within each dialect. That is, the prepositions are optional and are often dropped in Beijing Mandarin while they are obligatory in Wu dialects. In fact, due to recent

50

D. Liu

influence from Mandarin, the omission of prepositions in these cases may be marginally acceptable for young Shanghainese speakers, but it is unacceptable for speakers of other Wu dialects near Shanghai, such as the dialects of Suzhou and Wuxi, that are generally more conservative than Shanghainese. The ellipsis of the postverbal preposition poses a challenge to syntactic analyses in Mandarin. Many Chinese scholars treat these preposition-free postverbal spatial expressions in Mandarin as objects (referred to as ‘locative objects’, cf. Fan X. et al. 1998: 49, Shi 1999). What underlies this account is the traditional claim that locative postpositional phrases called 方位短语 fāngwèi duǎnyǔ (locative phrases) in Chinese should be nominal, and what we term as postpositions can be categorized as a subclass of nouns. According to our postposition analysis, the spatial NP and the spatial postposition are distinct in their syntactic status. Their syntactic variation is also demonstrated in the postverbal position. Compare the two sentences below. (26) a. 他睡(*在)/(了)小床。 tā shuì (*zài)/(-le) xiǎochuáng 3Sg sleep(at)/(-Asp) small-bed ‘He slept on the small bed. / He used the small bed as his place for sleeping’ b. 他睡(在)/(*了)小床上。 tā shuì (zài)(*-le) xiǎochuáng -shang. 3Sg sleep(at)(*-Asp) small-bed-on ‘He slept on the small bed.’ As shown in (26a), when an NP without postposition follows the verb, an aspect marker can be inserted between the verb and the NP, as in other VO constructions, but the insertion of a preposition is not allowed. In contrary, when a postpositional phrase (PostP) follows the verb, a preposition can be inserted but not an aspect marker, as in (26b). It is a general constraint in Mandarin that verbs followed by a preposition cannot take any aspect markers, unless the aspect marker follows the preposition (V + Pre + Asp + NP). Sentence (26) clearly illustrates that even if a preposition does not overtly occur before a postpositional phrase, the postverbal spatial expression still observes the rule of prepositional phrases in Mandarin. They do not allow any postverbal aspect markers to occur. In this sense, spatial expressions headed by a postposition should not be considered as real objects. They are adverbial in nature. On the other hand, an NP following the verb, even if it has some spatial meaning, is a real object (locative or instrumental), which explains why it rules out the insertion of a preposition[5]. Now let us turn to the preverbal position. Interestingly, it demonstrates a reverse rule: the omission of spatial prepositions is hardly grammatical in Beijing Mandarin but is much more easily accepted in Wu dialects. Consider the following sentences. (27) a. 你??(在)门口椅子上坐一会儿。 nǐ ??(zài) ménkǒu yǐzi-shàng zuò yi-huír.

Syntax of Space Across Chinese Dialects

b.

51

you ??(at) gate chair on sit one-while ‘You can sit in the chair at the gate.’ 侬(辣)门口头椅子高头坐脱歇。 nung (laq) mengkheudeu itsy-kodeu zu-theq xiq. you (at) gate chair-on sit-Asp while.

(28) a. 他*(到)教室里去了。 tā *(dào) jiàoshì lǐ qù le. he *(to) classroom-inside go Ptc ‘He went to the classroom.’ b. 伊(到)教室里去勒。 i (to) jioseq-li qi leq. he (to) classroom-inside go Ptc. The preverbal spatial preposition is almost obligatory in Beijing Mandarin but optional in Shanghainese and many other Wu dialects. Preverbal spatial expressions without prepositions have been a common or even unmarked phenomenon in Wu dialects. Liu (2003: 212–214) has cited many examples in this pattern from early Wu texts (published from 1802 to 1902, based on the Suzhou dialect). In fact, some Wu dialects are even farther from Beijing Mandarin in this aspect. For example, the optional prepositions in (27b) and (28b) are not allowed to occur at all in the Shaoxing dialect. The spatial expression solely relies on the postposition as a spatial marker (cf. Liu 2003 Ch. 13, particularly examples (12–13), p. 260). The omission of prepositions in different positions has different implications in syntactic typology. Adpositions, including prepositions and postpositions, belong to a kind of ‘relator’, which links one syntactic constituent with another. According to Simon Dik’s Relator Principles (cf. Dik 1997: 406–408, Rijkhoff 2002: 295), relators have their preferred positions (a) at the periphery of the relatum with which they form one constituent (if they do so), and (b) in between the two relata. Since both prepositions and postpositions observe Relator Principle (a), no further discussion will be needed regarding this principle. We will thus focus on (b). When a spatial expression precedes the verb, only the postposition can occur in between the two relata (NP and V). When a spatial expression follows the verb, only the preposition can occur in between the two relata (V and NP). In Wu dialects, there is a unified rule governing the occurrence and absence of prepositions, which is as follows: Prepositions located between the verb and the spatial expression must occur overtly This rule, based on the Relator Principle (b), only allows the ellipsis of the preposition in preverbal position. In contrast, Beijing Mandarin allows the postverbal preposition to be absent, violating Relator Principle (b). The omission of a preposition here might be attributed to the principle of economy, because there is informational redundancy in using a circumposition to mark a spatial expression.

52

D. Liu

This account will imply that the principle of economy is ranked higher than the Relator Principle in Mandarin. However, this explanation seems to be problematic when attention is drawn to the preverbal position. When a prepositional phrase precedes the verb, the preposition is not in between the two relata but is basically not omissible in Beijing Mandarin. Here the principle of economy does not work. At the preverbal position, it is Wu dialects that observe the principle of economy. Thus, by employing the principles of relators and of economy, we can adequately explain the facts in Wu dialects but would fail to explain the facts in Beijing Mandarin. The problem lies in that there is a factor of prosodic syntax that determines the working domain of economy. In Mandarin, as well as in many Chinese dialects, a preposition remains a genuine preposition only in the preverbal position. In the position immediately following the verb, the preposition will undergo a reanalysis and be adjoined to the preceding verb instead of the following NP (cf. Feng 2005 [1997]: 93–97, and his citation of Li 1990). Feng considers that this reanalysis is driven by a prosodic rule, namely the stress-assignment rule in Mandarin Chinese. This explains why when an aspect marker occurs, it must follow the preposition instead of the verb. This reanalysis might turn a preposition into an enclitic by depriving it of its prosodic and syntactic autonomy. A typical example of such weakening process is the postverbal clitic de, which is believed to be a weak form of zài ‘at’, dào ‘to’ or both[4]. In fact there are similar forms with the same meaning and function in dialects of Nanjing ([.tε], Liu 1995) and Nanchang ([.tεt], Xiong 1994), and hence, their exact etymological source is subject to further research. In fact, postverbal prepositions have the nature of head-marking means, unlike the dependent-marking of preverbal prepositions. It is now clear that in Beijing Mandarin a preposition is subject to ellipsis only when it is reanalyzed as an element or an enclitic adjoined to the verb. In other words, ellipsis caused by the principle of economy is ranked lower than reanalysis in Beijing Mandarin, which according to Feng (2005) is driven by prosody. In contrary, the Relator Principle (b) is ranked higher than the principle of economy and prosodicallymotivated reanalysis thus plays no role in the ellipsis of prepositions in Wu dialects. 3.3. The ‘preposition-like’ usage of directional verbs As shown by sentences (9) and (11) in Section 2.1, directional verbs can function as some kind of markers for spatial expressions in various Chinese dialects. As markers for spatial expressions, these verbs are semi-grammaticalized elements. Some Mandarin directional verbs have already been discussed in Talmy’s framework of V-language and S-language in Lamarre’s chapter in this volume. What concerns us here is the diversity among dialects in the use of directional verbs as spatial markers. Directional verbs can be divided into two groups. One group consists of locative directional verbs denoting directions based on objective locations. They are 进 jìn ‘enter, in’, 出 chū ‘out of’, 上 shàng ‘up’, 下 xià ‘down’, 过 guò ‘passing, past, over’ and 开 kāi ‘leaving, separate’ in Standard/Beijing Mandarin. The other group consists of deictic directional verbs denoting directions based on the speaker’s location and perspective. They are 来 lái ‘come’ and 去 qù ‘go’. A locative

Syntax of Space Across Chinese Dialects

53

directional verb and a deictic directional verb can be combined into a compound in the order of “locative + deictic”, such as 进来 jìn-lái (come in, literally ‘entercome’), but these compounds are not our issue. 3.3.1. Locative directional verbs A locative directional verb following another verb has one of the two functions, either as an adverb-like element, or as a preposition-like element (cf. Fan J. 1963). As an adverb-like element, it will not change the argument structure, but as a preposition-like element it helps introduce a spatial expression to the verb as its spatial role, as shown by the following examples. (29)

(30)

拿出钱包 ná chū qiánbāo (~ná qiánbāo) take out wallet ‘to take the wallet out of a pocket or a bag’ 跑出大门 pǎo chū dàmén (~ *pǎo dàmén) run out gate ‘to run out of a gate’.

In (29), ná qiánbāo ‘take wallet’ forms a VO construction, which is independent of the occurrence of the directional verb chū ‘out’. Thus chū is adverbial in nature and has no linking function. In (30), chū introduces the spatial expression to the argument structure. Without chū, pǎo dàmén cannot be a grammatical unit. In this sense, chū in (30) plays a preposition-like function. Since the directional verb following a full verb is syntactically combined with the preceding verb rather than the following NP, it is not a genuine preposition. To put it in another way, a preposition that is added onto the spatial NP belongs to a dependent-marking device, while a directional verb added onto the full verb is a device for head-marking. The usage as exemplified in (9) and (30) exists in many but not all dialects of Chinese. In Wu dialects, where the adverbial usage of postverbal directional verbs sounds very natural, their preposition-like usage will be syntactically marginal or unacceptable. When we tried to elicit the Wu counterparts of sentences in the form of ‘V(full verb) + Vd + NP’ from our informants, they gave sentences in the form of ‘Vf + Prep + NP + Postp’ instead. There is a distinction not only between a directional verb and a preposition, but also between an NP and a PostP, because a spatial preposition must govern a postpositional phrase instead of an NP in Chinese, especially in Wu dialects (cf. Section 3.1). In some cases, the informants confirmed that the sentences in the form of ‘Vf + Vd + NP’ would be unacceptable in their mother tongue. Below are two Mandarin sentences used in our questionnaire, by which 12 dialects throughout the Wu region were investigated. No speaker used the Mandarin pattern to give relevant sentences in their own dialects. For the sake of convenience, only Shanghainese examples will be provided to contrast with the Mandarin examples.

54

D. Liu

(31) a. 小偷躲 进了卫生间。 xiǎotōu duǒ jìn-le wèishēngjiān. thief hide enter/into-Asp bathroom ‘The thief hid himself in the bathroom.’ b. 小偷呀辣卫生间里。 xiotheu ia laq uesengke-li. thief hide at bathroom-inside (32) a. 有几个观众跳上了舞台。 yǒu jǐ-ge guànzhòng tiào shàng-le wǔtái have several-Cl audience jump up/on-Asp stage ‘A few from the audience jumped onto the stage.’ b. 有几个观众跳到舞台高头。 yeu ji-eq goetsung thio to ude-kodeu have several-Cl audience jump to stage-on. For some informants in Wu dialects, the corresponding pattern of these Mandarin examples is totally unacceptable. For others, sentences in this pattern are acceptable only for one or two directional verbs (mostly 进 jìn ‘enter’). As to directional verbs like 上 shàng ‘up’ and 下 xià ‘down’, no informant gave sentences in the pattern of the Mandarin sentence in (32a). In short, the pattern in which the directional verb serves to introduce the spatial expression is very common in Standard Mandarin and maybe in most varieties of Mandarin, but is marginal or even unacceptable in the syntax of Wu dialects. Although directional verbs in Wu dialects can hardly introduce spatial NPs in the position following full verbs, as full verbs they can freely take spatial NPs as their arguments, such as 进卫生间 jing uesengke ‘enter the bathroom’ and 上舞台 zong ude ‘get onto the stage’ in Shanghainese. Therefore the problem is not whether the directional verb can take an NP as its spatial role or not, but whether it can help another full verb to take an NP as its spatial role or not, with the directional verbs being a sort of function word (preposition-like words, but head-markers). The latter function is a result of the grammaticalization of the former function. The process has not yet taken place in Wu dialects. 3.3.2. Deictic directional verbs (come/go) Deictic directional verbs are those for ‘come’ and ‘go’. We have seen in Section 2.3.1 that the locative directional verbs can serve to introduce a spatial NP for the preceding full verb in Standard Mandarin but can hardly do so in Wu dialects. How about deictic directional verbs then, in terms of such a function? In Cantonese, 嚟 lai ‘come’ and 去 heui ‘go’ can freely take the following NP as their spatial object (denoting the goal), and the two verbs can also freely introduce spatial NPs for the preceding full verbs, functioning as spatial headmarkers. Consider the sentences below.

Syntax of Space Across Chinese Dialects

(33)

(34)

(35)

55

我哋行去 去沙田 ngotei hang heui Sathin. we walk go/to Shatin ‘We will walk to Shatin’ 啲文件带咗嚟公司喇。 ti mankin taai tso lai kungsi la. those document bring Asp. come/to company Ptc ‘Those documents have been brought to the company (here).’ 你攞去边度？ nei lo heui pintou? you carry go/to where ‘Where are you carrying (this) to?’

The Cantonese pattern demonstrated above does not exist in Standard/Beijing Mandarin and Wu dialects. The verbs denoting ‘come/go’ in those dialects cannot occur between a full verb (either intransitive motion verbs like hang (walk’ in (33) or transitive action verbs like taai (carry’ in (34)) and a spatial NP. The typological diversity here is understandable. As mentioned before, the preposition-like function of directional verbs is a result of grammaticalization from their full verb usage. In Standard Mandarin, the wide use of lai/qu (come/go’ taking a spatial (directional) object is a rather recent development (and still a novel usage in Beijing speech), contrary to the situation in Cantonese, whose usage has long been common and basic. Grammaticalization is a diachronic process and will take some time to go through. There is not sufficient time for deictic directional verbs in Standard Mandarin to derive a preposition-like function. In Wu dialects, the use of a deictic directional verb taking a spatial object is inhibited. Without this usage as a precondition, the grammaticalization from full directional verbs into spatial headmarkers can hardly take place. That is why there is no preposition-like function of words denoting ‘come/go’ in Wu dialects. To an extent, the diversity among dialects in terms of the preposition-like function is related to their typological characteristics in terms of the ordering of the spatial components in the clause. We will discuss this issue in Section 4 below. 4． ．Typological Diversity Among Dialects Regarding the Word Order of Spatial Components From the perspective of word order typology, many features are shared by most or all dialects of Chinese. The most significant ones are as follows: 1) SVO as main order in a clause; 2) co-existence of prepositions, postpositions (especially for spatial constituents) and the combination of both, namely circumpositions; and 3) preference for the preverbal position for adverbials, including spatial adverbials. On the other hand, there are still some crucial variations among dialects in the ordering of spatial expressions.

56

D. Liu

4.1. The ordering of the nominal spatial expression as relative to the verb For some verbs with intrinsic spatial meaning, the NP itself is enough to serve as their spatial role and no space marking is necessary in Chinese. The typical members of these verbs include pure motion verbs (deictic and locative directional verbs), and static existence verbs. The diversity mainly lies in the deictic directional verbs. The order of the spatial argument relative to locative directional verbs or existence verbs is generally the same across dialects except for certain special ordering regarding existence verbs in a few dialects. Deictic directional verbs denoting ‘come/go’ are the purest motion verbs that take a spatial component as their basic argument. In some dialects of Chinese, the spatial component (usually the goal, sometimes the source) which combines with such verbs does not require any space marking. Spatial components without space marking are referred to as ‘spatial NPs’. The linear position of spatial NPs, however, varies with dialects. Spatial NPs always precede the deictic verbs in Shanghainese and other northern Wu dialects (if not all Wu dialects), but follow them in Cantonese. Consider the Shanghainese examples below, with the first two already cited in Section 1.3. (36)

(37)

(38)

老王今朝(到)杭州去。 Lo-uong jingtso (to) Ongtseu qi Lao-Wang today (to) Hangzhou go ‘Lao-Wang will go to Hangzhou today.’ 老王今朝(到/从)杭州来。 Lo-uong jingtso (to/zung) Ongtseu le. Lao-Wang today (to/from) Hangzhou come ‘Lao-Wang will come to/from Hangzhou today.’ *老王今朝来/去杭州。 *Lo-uong jingtso le/qi Ongtseu Lao-Wang today come/go Hangzhou ‘Lao-Wang come/go to Hangzhou today’ .

When the place name Ongtseu precedes le ‘come’ or qi ‘go’, the preposition for goal or source is optional and is more often omitted, at least for older Shanghainese speakers and the speakers of other Northern Wu dialects. Then, the place name as a spatial NP plays the spatial role of the verb. Such a spatial NP cannot follow the verb, as shown in (38). In contrast, such a spatial NP always follows the verb in Cantonese. Therefore, the Cantonese counterparts of the above Shanghainese examples would become unacceptable if a spatial NP preceded the verb, as shown below. (39)

*王生今日杭州去。 *Wong-Sang kanyat Ongtsau heui.

Syntax of Space Across Chinese Dialects

57

Mr. Wang today Hangzhou go ‘Mr. Wang will go to Hangzhou today.’ (40)

(41)

*王生今日杭州嚟。 *Wong-Sang kanyat Ongtsau lai. Mr. Wang today Hangzhou come ‘Mr. Wang will come to/from Hangzhou.’ 王生今日嚟/去杭州。 Wong-Sang kanyat lai/heui Ongtsau. Mr. Wang today come/go Hangzhou ‘Mr. Wang will come/go to Hangzhou.’

The dialectal distinction shown in the above examples is not merely a wordorder phenomenon. Chinese grammarians usually treat the Wu pattern of ‘spatial NP + V’ as an adverbial-V construction whereas the Cantonese pattern of ‘V + spatial NP’ is treated as a VO construction. Such an analysis is supported by syntactic and pragmatic evidence given below. 1) In Wu dialects, a preposition can be added before the spatial NP (see the bracketed elements in (36) and (37)), while in Cantonese, the pattern of ‘V + spatial NP’ allows no prepositions to be inserted. This means that the spatial NP in the Wu pattern is an indirect role to the verb and is adverbial in nature, while it is a direct argument of the verb and functions as an object in Cantonese. 2) The Shanghainese spatial NP preceding the verb is different from the preverbal spatial topic. Syntactically, a spatial NP must occur immediately before the verb, while the spatial topic can be separated from the verb by other elements. Semantically, spatial NPs can serve as the information focus, answering questions like ‘where are you going’ or ‘where did you come from’, while a spatial topic must carry given or presupposed information, as shown below: (42) 侬明朝有啥事体？ — 明朝 杭州 *明朝 去。 a: nung mingtso ieu sa zythi? b: (mingtso) Ongtseu (*mingtso) qi 2Sg tomorrow have what thing (tomorrow) Hangzhou (*tomorrow) go What will you do tomorrow? (I will) go to Hangzhou (tomorrow) (43) 侬杭州啥辰光去？ — 杭州 明朝 *杭州 去。 a: nung Ongtseu sa zengkuong qi? b: (Ongtseu) mingtso (*Ongtseu) qi you Hangzhou what time go? (Hangzhou) tomorrow (*Hangzhou) go When will you go to Hangzhou? (I will) go (to Hangzhou) tomorrow In (42), the place name Ongtseu bears information focus and must immediately precede the verb. In (43), Ongtseu conveys the presupposed information and serves as a subtopic (following the subject). And thus it must precede the temporal adverbial mingtso ‘tomorrow’. What is exhibited here indicates that spatial NPs, which can serve as information focus, are normal spatial adverbials, and they are

D. Liu

58

sharply different from spatial topics that must bear given or presupposed information. They occupy different syntactic slots in the clause. Now let us examine the situations in Standard and Beijing Mandarin. The pattern of ‘来 lái/去 qù (come/go) +spatial NP’ has become a kind of canonical pattern in Standard Mandarin, as mentioned before. This seems to suggest that Standard Mandarin and Cantonese are alike in this respect. Traditionally, however, this is not the pattern of Beijing Mandarin. In traditional Beijing Mandarin, the spatial argument of lái/qù ‘come/go’ is usually introduced by a preposition or semigramaticalized directional verb such as 到 dào ‘to’, 往/望 wàng ‘to, toward’, 上 shàng ‘go up to’. The spatial expression governed by such a function word generally precedes the verb and in some cases follows the verb. In modern works written by Beijing writers, we can hardly find the pattern now shared by Cantonese and Standard Mandarin (see detailed investigation in Liu 2000). Consider the following Beijing Mandarin examples. (44)

(45)

(46)

他明年到/上北京来。 tā míngnián dào/shàng Bēijīng lái. he next-year to/up Beijing come ‘He will come to Beijing next year.’ 汽车到/上/往西单去了。 qīchē dào/shàng/wàng Xīdān qù le. automobile to/up/toward Xidan go Asp-Ptc ‘The automobile went to Xidan.’ 他来到了北京。 tā lái dào le Bèijīng. he come to Asp Beijing. ‘He came to/arrived in Beijing.’

In terms of word order, Beijing Mandarin is closer to Wu dialects than to Cantonese with the spatial argument to these verbs mostly preceding the verb. On the other hand, interestingly, since Standard Mandarin has fully accepted the pattern of ‘lái/qu (come/go) + spatial NP’, it is closer to Cantonese than to Wu dialects in this parameter. It is certainly easy for Beijing Mandarin to accept the Standard Mandarin pattern, but the original Beijing Mandarin pattern is still preferred by native Beijing Mandarin speakers, according to my preliminary observations. As for locative directive verbs such as 进 jìn ‘enter’, 出 chū ‘be out of’, 上 shàng ‘get onto’, 下 xià ‘get down’ and 过 guò ‘be over, be past’ in Standard Mandarin, dialects show more commonality than diversity. When they are used as full verbs with a spatial argument, they are almost unanimously followed by a spatial NP. Finally, we come to existence verbs. The purest existence verbs are 在 zài ‘be at’ in Mandarin and its dialectal equivalents . In most dialects, including most Wu

Syntax of Space Across Chinese Dialects

59

dialects, we have the order of ‘zài + spatial NP’. However, at least in some Wu dialects, the reverse order is also a natural or even preferred pattern. Notice the 绍兴 Shaoxing examples below (cited from Liu 2003: 267): (47)

a. 伊来亨图书馆里。 ~ b. 伊图书馆里来亨。 [Ûi leha0 duspku—-li] ~ [Ûi duspku—-li leha0] he be-at library-inside he library-inside be-at ‘He is in the library.’

Strictly speaking, dusp ku—-li in the above example is not a spatial NP but a postposition phrase. In fact, if one replaces it with a place name like Nanjing, the situation will be the same. The order of (47b) above is rather rare among Chinese dialects, but in Shaoxing, this order is favored, as compared with the order of (47a). In the 宁波 Ningbo dialect which is close to Shaoxing, when being asked the dialectal counterpart for the Mandarin pattern of ‘zài + spatial NP’, the informant may come up with another pattern, a pattern which has the spatial expression preceding the existence verb. Compare (48) with (49) which share the same interpretation. (48)

有个陌生人在房间里。 yǒu ge mòshēngrén zài fángjiān-li. have Cls stranger be-at room-inside ‘There is a stranger in the room’

(49) 有一个生头人到 屋落里头来该。 [i; i+km sadœ;…i0 tn olo-lidœ; leke] have one-Cls stranger to room-inside be-at The dialectal diversity regarding the order of spatial NPs described in this section conforms well to the characteristics of the dialects in question with respect to word order typology (see discussions in Liu 2000, 2001). Cantonese, with stronger verb-middle features, instantiates a more typical SVO type than most other Chinese varieties, and its spatial expressions follow the verb in most cases. Wu dialects are not a typical SVO type, with more verb-final patterns and a richer and more active postposition system due to their stronger topic-prominent nature. In the Wu region, the dialects around Shaoxing and Ningbo exhibit more characteristics of verb-final languages. Accordingly, Wu dialects, particularly the Shaoxing and Ningbo dialects, have more spatial expressions preceding the verb. Mandarin stands somewhere in between Cantonese and Wu dialects in terms of word order type. The order of spatial expressions is in line with this overall model. 4.2. The ordering of adpositional phrases Lamarre (2003) notes an interesting fact about the ordering of adpositional phrases. A spatial prepositional phrase (PP) in a static situation such as ‘live

60

D. Liu

somewhere’ must precede the verb in the dialect of Jizhou, located in Hebei province near Beijing, and in many other northern dialects, whereas it can follow the verb in Standard Mandarin. Wu dialects are similar to Standard Mandarin in this aspect, according to our investigation. Lamarre’s observation is very important. However, if one considers the more general situation regarding the ordering of spatial adpositional phrases, Wu dialects will be found to have more constraints on postverbal PPs than those in Standard Mandarin and other northern dialects. A number of spatial prepositions that can precede the verb can also follow the verb in Standard/Beijing Mandarin, such as 在 zài ‘at’, 往 wàng ‘towards, to’ and 向 xiàng ‘in the direction of’. In the postverbal position, the form de (present only in colloquial Beijing Mandarin), considered to be a weakened form of 在 zài ‘at’, 到 dà o ‘to’ or both, can also be found and it is no longer a genuine preposition due to its binding to the preceding verb. In formal Standard Chinese, more spatial prepositions are found to follow the verb, such as 于 yú ‘at, to, from’ and 自 zì ‘from, since’. In Wu dialects, however, only very few spatial prepositions can follow the verb. In Shanghainese, only laq ‘at’ and its compound forms laqla, laqhe ‘at here/there’ can follow the verb. In fact, we do have prepositions like 望 mong (=Mandarin wàng) and 向 xiang in Shanghainese, but they can only precede the verb. Therefore, the chances for spatial prepositions to follow the verb are slimmer in Shanghainese than in Mandarin. In some dialects of Wu, no spatial prepositions that precede the verb can follow the verb as well. There actually exist some postverbal spatial markers, but they are enclitics of the verb instead of prepositions governing the following NP. Etymologically they can hardly be related to the preverbal spatial prepositions, and thus are inherent head-marking means. For example, in the 金华 Jinhua dialect, the 313 535 corresponding prepositions for Mandarin zài ‘at’ include 来 [l' ] and 隑[I' ], which, with their governed spatial NPs, can only precede the verb, but the postverbal marker for spatial expression is [dŒ], which is in neutral tone. Consider the following examples. (50)

(51)

老师来/隑黑板上写字。 [l#usp l'/I' hŒp#-Ýi#0 sia zp] teacher at blackboard-on write character ‘The teacher is writing on the blackboard.’ 首饰都放特俺革里。 [siusŒ tu f#0 dŒ a kŒli] jewelry all put at I here ‘The jewelry is all kept in my possession.’

The phrase [l' hŒp#-Ýi#0] or [I' hŒp#-Ýi#0] (lit. ‘at blackboard-on’) in (50) is a free prepositional phrase that can occur in other syntactic positions, while [dŒ a kŒli] in (51) is not a unity, and can never occur in positions other than the postverbal one. Therefore no spatial preposition can follow the verb in the Jinhua

Syntax of Space Across Chinese Dialects

61

dialect. The preference for spatial adposition phrases in Wu dialects to precede the verb is harmonic with its stronger verb-final tendency, as mentioned in Section 3.1. In Wu dialects, even prepositions which can follow the verbs have fewer chances to follow the verb. For instance, we have zǒu zài dàlù-shàng (lit. ‘walk at big road-on’) and piāo zài bànkōng-zhōng (lit. ‘waft at the middle-sky-in’) in Mandarin, whereas their corresponding adpositional phrases in Wu dialects like Shaoxing and Ningbo tend to precede the verb. We mentioned ‘adposition’ instead of ‘preposition’ because when a spatial expression precedes the verb in Wu dialects, it often goes without a preposition, and the postposition then becomes the only overt marker for spatial role. This phenomenon conforms to the greater prominence of postpositions in Wu dialects. In addition, as mentioned before, for an adpositional phrase located in the preverbal position, only the postposition stands in the preferred relator position, i.e. between the two relata (cf. Dik 1997: 407, Rijkhoff 2002: 295). When in the postverbal position, only the preposition is in the preferred position. A preverbal adpositional phrase without preposition observes both the Relator Principle and the Principle of Economy. It seems that Cantonese, being a more typical SVO dialect, does not have more postverbal spatial adpositions than Wu dialects. Recall, however, it is often the case that spatial verbs will take preposition-free spatial NPs as their postverbal objects. This is on a par with its stronger SVO nature. In sum, Wu dialects are closer to the type PP + VP than Beijing/Standard Mandarin, where more VP + PP patterns exist in the latter. Cantonese does not have a relatively larger number of postverbal prepositions than Wu dialects and Mandarin, and spatial verbs generally take preposition-free spatial NPs as their objects. Differences among these dialects regarding their orderings of spatial expressions conform to the overall situations of these dialects in word order typology. 5． ．Concluding Remarks: Conspiring and Competing Factors on the Syntax of Space Space is a basic category that human languages appeal to morphosyntactic means to encode the relevant information. As an analytic language with no morphological cases for spatial roles, Chinese basically relies on syntactic means, namely word order and function words or semi-grammaticalized words, to mark spatial roles for their verbal heads. A number of verb-derived prepositions and noun-derived postpositions are used to signal spatial relations. A full-fledged spatial expression in Chinese is in the form of ‘preposition + NP + postposition’. Under certain conditions, prepositions, postpositions or both can be absent, resulting in a reduced form of spatial expressions, namely NP + postposition (more often), or preposition + NP. When the verb, the noun, or particularly both the verb and the noun, bears intrinsic spatial meaning, an NP alone can function as a spatial constituent. Both prepositions and postpositions are crucial means for dependent-marking, and syntactically they are combined with the spatial argument or adjunct of the verb. However, in many modern Chinese dialects, when a preposition occurs postverbally, it will undergo reanalysis and be adjoined to the preceding verbs rather than

62

D. Liu

following the postverbal NP. This process is triggered by prosodic factors, according to Feng (2005), and the preposition thus becomes a head-marking means in this position. A weakened form of the postverbal preposition can lose all its properties as a preposition and become a real head-marking verb suffix, such as de ‘at, to’ in Beijing Mandarin. In addition, a directional verb can be adjoined to the preceding head verb and help to introduce the following NP as the spatial role (goal) to the head verb in many dialects of Chinese. In such a case, directional verbs as semigrammaticalized space markers are a kind of head markers. In addition to postverbal prepositions and their weakened forms and directional verbs, there are also postverbal aspect markers that can function like prepositions to introduce spatial constituents in many dialects, though they are not discussed in details in this chapter. Compared with postverbal prepositions and directional verbs, they are genuine function words and perform the function of head-marking. Examples of such markers include 阿[a/ia/ua/ka/0a] (at, to; perfective and durative marker) in the Taixing dialect, Jianghuai Mandarin (cf. Li R. 1957), 倒[tno] (at, to; durative marker) in the Nanjing dialect (cf. Liu 1995), 哒[ta21] (at, to; durative marker) in the Changsha dialect (cf. Wu 2001) and so on. Xu (1994) has noted the close relationship between spatial prepositions and aspect markers in Mandarin, but she claims that the postverbal prepositions that can be weakened and deleted are just aspect markers and no longer prepositions. If we adopt the notion of head-marking, however, we do not have to deny the space-marking function of aspect particles. A marker attached to the head performing the function of linking a dependent could be a natural phenomenon in the typology of marking direction. The generalization and discussion above reveal an important fact: there is a variety of head-marking means that are active in encoding spatial role in Chinese. This fact is noteworthy if we recall the claim of some typologists that adjuncts or peripheral/oblique NPs – those denoting location, instrument, etc. – strongly favor dependent-marking in the form of case markers or adpositions. (cf. Nichols 1986). ‘Spatial adpositional phrase + V’ is the basic pattern shared by all modern Chinese dialects, though the unmarked pattern for Classic Chinese is ‘V+adpositional phrase’. In this parameter, the classic pattern rather than the modern one follows the order shared by an overwhelming majority of the VO languages in the world (cf. Dryer 1992, 2003). Besides the commonality, there is also typological diversity among Chinese dialects in terms of spatial expressions, a major issue in this chapter. In Chinese, a preposition usually governs a postpositional phrase instead of an NP, unless the NP is an inherent spatial NP like a place name. In Beijing Mandarin, many nouns have double statuses. As a normal noun, such a noun can be followed by a postposition to co-occur with a preposition whereas as a spatial noun it can be governed directly by a preposition without the postposition. Cantonese is similar to Mandarin in this respect. By contrast, in Wu dialects, only place names belong to spatial nouns. All of the other nouns must be followed by postpositions before being governed by prepositions. In this respect, Wu dialects seem to be the farthest from

Syntax of Space Across Chinese Dialects

63

Classic Chinese and most prepositional languages typologically, because a spatial preposition in Classic Chinese and languages like English can freely govern an NP. Beijing Mandarin tends to have the spatial preposition omitted in the postverbal position, due to the prosodic rule that makes the postverbal preposition a weakened and unstable clitic. In contrast, Wu dialects tend to have the preposition omitted in the preverbal position and must maintain its overt occurrence in postverbal position, a preferred position for preposition as a relator, according to the Relator Principle (b) (cf. Dik 1997: 407). The ellipsis can be motivated by the Principle of Economy in both dialects. In addition, the case of Wu dialects exhibits the force of the Relator Principle, while the Beijing Mandarin case exhibits the force of prosodic syntax (cf. Feng 2005). Locative directional verbs (denoting ‘enter, be out, be up, be down…’) can introduce an NP as a spatial role for the preceding full verb in Cantonese and Standard/Beijing Mandarin, but they can hardly do so in Wu dialects. With regard to this function, directional verbs behave like prepositions (but head-marking) and are semi-grammaticalized in nature. Deictic directional verbs (denoting ‘come/go’) only have this function in Cantonese but not in Standard/Beijing Mandarin and Wu dialects. This distinction reflects the various degrees of grammaticalization of directional verbs, which in turn is due to the different functions of directional verbs as full verbs. Both locative and deictic directional verbs as predicative full verbs can be followed by spatial expressions as objects in Cantonese. That forms the basis underlying their gramaticalization from full verbs to space markers. In Beijing Mandarin, taking spatial objects has long been a basic function of locative directional verbs, but had not become a regular function of deictic verbs till a few decades ago. That accounts for the fact that only the former group has obtained the function as space markers. In Wu dialects, the pattern of ‘locative directional verb + spatial object’ is grammatical but is much less frequently used than that of ‘preposition + NP + postposition + verb’, and the pattern of ‘deictic directional verb + spatial object’ is ungrammatical (deictic verbs are intransitive in Wu). That explains why few directional verbs have developed the function of space markers in Wu. As previously mentioned, spatial prepositional phrases, like other prepositional phrases, usually occur preverbally in all modern Chinese dialects. However, for a few prepositions, the spatial prepositional phrase (typically denoting goal) can follow the verb. The prepositions in the latter order are more numerous in Beijing Mandarin than in Wu dialects. This variation is in line with the general typological fact that Wu dialects are less typical as a SVO-preposition dialect than Beijing Mandarin and show more characteristics of verb-final and postpositional languages. Furthermore, in the Wu region, the closer a dialect is to the verb-final and postpositional type (as in the cases with the Wu dialects of Shaoxing and Ningbo), the more rarely it employs postverbal prepositional phrases (cf. Liu 2003). Spatial prepositional phrases following the verb in Cantonese are not so common, but there are many more cases in Cantonese than in other dialects that a spatial NP follows the verb without a preposition, functioning as a direct object. This is in line with the

D. Liu

64

general fact of word order typology in Cantonese, which is the closest to a typical SVO language among Chinese dialects. It appears that typological harmony in terms of word order plays a significant role in encoding spatial roles in Chinese dialects. So far we have seen several factors that dominate the syntactic behavior of spatial roles in Chinese dialects. They are the Relator Principle, the Principle of Economy, prosodic syntax, the degree of grammaticalization and its basis, word order harmony. In some cases, different factors play conspiring roles in shaping a syntactic pattern. For instance, in the Wu pattern of ‘(omitted preposition+) NP + postposition + V’, the Principle of Economy causes the omission of the preposition; the Relator Principle keeps the overt occurrence of the postposition, and the word order harmony with its stronger verb-final tendency leads the spatial expression to precede the verb. In other cases, different factors may compete with one another, and which factor ends up winning in relevant cases varies with dialects. For example, in Beijing Mandarin, the prosodic-syntactic factor makes the postverbal prepositions become enclitics, which in turn can be omitted due to the Principle of Economy. In Wu dialects, the Relator Principle plays a bigger role, causing the maintenance of the preposition between the verb and the spatial NP, despite the fact that prepositions in non-relator positions are more often omitted in Wu dialects than in Beijing Mandarin. A more systematic generalization of how different factors interact with one another in the syntax of space will be subject to further research based on richer dialectal data in the future. * The author thanks Prof. Dan XU (Paris), Prof. David Xing CHENG (New York), Prof. HU Jianhua (Beijing), Dr. SHEN Yuan (Shanghai), Dr. Peppina LEE (Hong Kong), Dr. TANG Zhengda (Beijing) and Dr. CHEN Yujie (Hangzhou) for their various kinds of constructive help with the preparation and revision of this chapter.

Abbreviations



2Sg 3Sg NP Vd Loc Asp Ptc Cls

Archaic Chinese Standard Mandarin (spoken) Beijing Mandarin (identical with Standard Mandarin in most but not all aspects) Shanghainese, a Wu dialect spoken in Shanghai. Cantonese, a Yue dialect spoken in Guangzhou, Shenzhen, Hong Kong and Macau. second person singular third person singular nominal phrase directional verb locative adposition (preposition or postposition) aspect marker sentence-final particle classifier

Syntax of Space Across Chinese Dialects

65

Notes 1. It is observed that several Chinese dialects seem to demonstrate some features similar to morphological cases. In these dialects, some person pronouns can be thought of as having genitive case marking forms, such as nga ‘my’ as opposed to ngai ‘I/me’ in some Hakka dialects. Some pronouns tend to function as subjects instead of objects while others go the reverse way, as in some Gan, Hui and Wu dialects. Their real nature awaits in-depth study. 2. For detailed discussions about preposition and postposition, see Hagège’s work as cited by Xu (1990), and Liu (2003). For circumpositions, see Liu (2002b, 2003). 3. As for dialects other than Standard Mandarin and Wu, little attention has been paid to the rules governing the absence/ occurrence of postpositions, so it is hard to make a crossdialectal generalization on this issue. According to my personal impression, many dialects like Jianghuai Mandarin or Xiang dialects maybe stand in between Standard Mandarin and Wu dialects in this respect, and Yue (Cantonese) dialects may be as free as, or even freer than Mandarin in absence of postpositions. 4. Chao (1979[1968]: 178, 333) regards de as a mixture of zài ‘at’ and dào ‘to’. 5. The ‘spatial object’ without postposition can be better interpreted as a kind of patient or manner than as spatial role, though it really contains some spatial meaning.。See a detailed analysis in Ren (2000, Chapter 5).

66

D. Liu

References Chao, Y. R. 赵元任. (1979). [1968]. A Grammar of Spoken Chinese (Translated into Chinese by Lü Shuxiang). Beijing: The Commercial Press. (Original English edition, 1968). Chu, Zhexiang 储泽祥. (2004). Hanyu ‘zai+fangwei duanyu’ li fangweici de yinxian jizhi. [The mechanism of the presence or absence of the locative words in ‘zai + locative phrase’ in Chinese]. Zhongguo Yuwen, 2. Ding, Chunshou 椿寿. (1993). Yiyu Tonglun [A Comprehensive Introduction to the Yi Language]. Guiyang: Guizhou Minzu Press. Dik, S. C. (1997). The Theory of Functional Grammar. Part 1: The Structure of the Clause. ed. By Kees Hengeveld, Second, revised version. Berlin New York: Mouton de Gruyter. Dryer, M. S. (1992). The Greenbergian word order correlations. Language, 68(1), 43–80. Dryer, M.S. (2003). Word order in Sino-Tibetan Languages from a typological and geographical perspective. In G. Thurgood and R. LaPolla (Eds.), Sino-Tibetan Languages. Richmond: Curzon Press. Fan, Jiyan 范 继 淹 . (1963). Dongci he quxiangxing houzhi chengfen de jiegou fenxi [Structural analysis on the verb and the following directional constituent]. Zhongguo Yuwen, 2. Fan, Xiao 范晓 (Ed.) (1998). Hanyu de Juzi Leixing [Sentence Types in Chinese]. Taiyuan: Shuhai Press. Feng, Shengli 冯胜利. (2005). [1997]. Hanyu de Yunlyu, Cifa yu Jufa [Interaction between Morphology, Syntax and Prosody in Chinese]. Beijing: Peking University Press. Greenberg, J. H. (1995). The diachronic typological approach to language. In Shibatani & Bynon (Eds.), Approaches to Language Typology. Oxford: Clarendon Press. Heine, B., Claudi, U., & Hünnemeyer, F. (1991). Grammaticalization: A Conceptual Framework. Chicago: University of Chicago Press. Lamarre, C. (2003). Cong Hebei Jizhou fangyan dui Xiandai Hanyu ‘V zai+chusuo’ geshi de zaitantao [The construction of ‘V zai+location’ in Modern Chinese, a revisit from the Hebei Jizhou dialect.]. In Dai Z. (Ed.), Hanyu Fangyan Yufa Yanjiu he Tansuo [Grammatical Research and Inquiry of Chinese Dialects]. Harbin: Heilongjiang People Press. Li, Renjian 李人鉴. (1957). Taixing Fangyan-zhong dongci-de houfu chengfen [Enclitics to verbs in the Taixing dialect]. Zhongguo Yuwen, 5. Litip, Tohti. 力提甫·托乎提. (2001). A Study on Generative Syntax of Uyghur and other Altaic Languages (in Chinese). Beijing: Minzu Press. Liu, Danqing 刘丹青. (1995). Nanjing Fangyan Cidian [A Dictionary of the Nanjing Dialect] (as a volume of Great Dictionary of Modern Chinese Dialects). Nanjing: Jiangsu Education Press. Liu, Danqing 刘丹青. (2000). Yueyu jufa de leixing tedian [Typological characteristics of Cantonese syntax]. Hong Kong: Asian and Pacific Journal of Language in Education, 3(2). Liu, Danqing 刘 丹 青 . (2001). Hanyu fangyan yuxu leixing-de bijiao [A typological Comparison of word order among Chinese dialects]. Japan: Study of Contemporary Chinese 2. Liu, Danqing 刘丹青. (2002a). Fuyuan shici yu yufahua (Role-assigning content words and grammaticalization). In Pan Wuyun (Ed.), Dongfang Yuyan yu wenhua (Language and Culture in the East). Shanghai: Oriental Publishing Center. Liu, Danqing 刘丹青. (2002b). Hanyu-li de kuangshi jieci [Circumpositions in Chinese]. Contemporary Linguistics, 4.

Syntax of Space Across Chinese Dialects

67

Liu, Danqing 刘丹青.(2003).Yuxu Leixingxue yu Jieci Lilun [Word Order Typology and a Theory of Adposition]. Beijing: Commercial Press. Nichols, Johanna. (1986). Head-marking and dependent-marking grammar. Language, 62(1), 56–119. Peyraube, Alain. (1994). On the history of Chinese locative prepositions. Taipei: Languages and Linguistics in China 中国境内语言及语言学 , 2. Ren, Ying 任鹰 (2000). Xiandai Hanyu Feishoushi Binyu Ju Yanjiu [Studies in the Nonpatient Object in Contemporary Chinese]. Beijing: Social Sciences Academic Press. Rijkhoff, Jan. (2002). The Noun Phrase. Oxford: Oxford University Press. Shi, Youwei 史有 . (1999). Chusuo binyu chubu diaocha (A preliminary investigation of locative objects). Retrieved From www.usc.edu/dept/LAS/ealc/chinling/articles/ shi2.pdf / Song, Jae Jung. (2001). Linguistic Typology: Morphology and Syntax. Longman. Sun, Chaofen. (1996). Word-Order Change and Grammaticalization in the History of Chinese. Stanford: Stanford University Press. Wu, Yunji. (2001). The development of locative markers in the Changsha dialect. In Hilary Chappell (Ed.), Sinic Grammar: Synchronic and Diachronic Perspectives. Oxford: Oxford Press. Xiong, Zhenghui 熊正辉. (1994). Nanchaing Fangyan Cidian (A Dictionary of the Nachaing Dialect (as a volume of Great Dictionary of Modern Chinese Dialects). Nanjing: Jiangsu Education Press. Xu, Dan 徐 丹 . (1990). Pingjie Jieci Wenti ji Hanyu-de Jiejue Fangfa [Review of ‘Le Probleme Linguistique de Prepositions et la Solution Chinoise’ by Claude Hagège]. Zhongguo Yuwen [Chinese Language], 6. Xu, Dan 徐丹. (1994). Guanyu Hanyu-li ‘dongci + X + didianci’ de juxing [About the pattern of ‘verb + X + locative word’ in Chinese]. Zhongguo Yuwen [Chinese Language], 3. Zhang, Caiyao 张才尧. (1994). Shiyong Deyu Yufa [A Practical German Grammar]. Beijing: Waiyu Jiaoxue yu Yanjiu [Foreign Language Teaching and Research] Press. Zhang, Cheng 张赪. (2002). Hanyu Jieci Cizu Cixu de Lishi Yanbian (Historical Change of the Word Order of Preposition Phrases in Chinese). Beijing: Beijing Language and Culture Univerisity Press.

THE LINGUISTIC CATEGORIZATION OF DEICTIC DIRECTION IN CHINESE —— WITH REFERENCE TO JAPANESE —— Christine Lamarre University of Tokyo Keywords: spatial deixis, subjectivity, motion events, directionals Abstract: This paper discusses the linguistic categorization of deictic direction in Mandarin Chinese, with reference to Japanese. It focuses on the following question: to what extent should the prevalent bimorphemic (nondeictic + deictic) structure of Chinese directionals be linked to its typological features as a satellite-framed language. We know from other satellite-framed languages such as English, Hungarian, and Russian, that this feature is not necessarily directly connected to satellite-framed patterns. Furthermore, verb-framed languages like Korean or Japanese also frequently combine deictic and non deictic paths in the same verb complex. Our conclusion is that the typological behavior of a language from the viewpoint of Talmy’s typology (Verb-framed vs. Satellite-framed) has no direct bearing on its choice of bipartite encoding of paths (deictic together with nondeictic). However, these typological features likely exert an indirect influence on the strategies used by a language to combine deictic paths, nondeictic paths, and the manner or cause of motion in a same verb complex, and ultimately also on the frequency of using deictic paths. We conclude that in order to understand why the overt linguistic expression of deictic direction is prevalent in some languages and not in others, we must take into account at least two different levels of linguistic organization, one pertaining to the syntactic devices expressing causativity and change of state, the other pertaining to the semantic devices through which languages express subjectivity.

0. Introduction We discuss here the linguistic categorization of deictic direction in Mandarin Chinese, with reference to Japanese, and focus on the following question: to what extent should the prevalent bimorphemic (nondeictic + deictic) structure of Chinese directionals be linked to its typological features as a satellite-framed language. This paper is organized as follows. First our methodology and basic concepts are introduced in Section 1. Section 2 sketches the linguistic characteristics of Chinese from the viewpoint of Talmy’s framework: Chinese behaves in many ways like a satellite-framed language, which often encodes the path of motion in a verbal prefix or a verb particle while expressing the manner or cause of motion in the verb root. However, it also frequently uses path verbs, thus behaving like a verb-framed language, just like Japanese and Korean. In Section 3, we describe Chinese path directionals (or ‘satellites’), which provide two separate slots for deictic and nondeictic paths when path information is added to other semantic components of a motion event (manner, cause, etc.), as in 搬出来 bān-chu-lai ‘carry out (toward speaker)’. In Section 4, we compare Chinese to a geographically close but typologically distant language: Japanese. Section 5 compares briefly Chinese with French and English. 69 D. Xu (ed.), Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives, 69–97. © Springer Science +Business Media B.V. 2008

70

C. Lamarre

1. The Path of Motion and its Linguistic Categorization 1.1. Verbal, adverbal and adnominal encoding The term ‘linguistic categorization’ used in the title of this paper determines the way various languages code paths in a spatial motion event, and sometimes categorize it into a distinct grammatical category. When they express motion events, languages often encode the path of motion into several distinct categories. Berthele (2004) distinguishes the following types of path encoding: i. ii.

verbal encoding: uses path verbs; adnominal encoding: combines the locative noun phrase (the Ground in Talmy’s typology) with adpositions (prepositions or postpositions); iii. adverbal encoding: makes use of verb particles and verb affixes. Type i is seen in French or Japanese, which typically use verbs to express core paths notions like ‘come’, ‘exit’, ‘descend’, ‘cross’ etc. These languages are called ‘verb-framed languages’ in Talmy’s typology (we call them V-languages hereafter). Type iii corresponds to Talmy’s satellite-encoding: satellites ‘relate to the verb root as a dependent to a head’ (Talmy 2000:102); examples include English verb praticles, Russian prefixes, and Hungarian preverbs. Languages where type iii is prevalent are called ‘satellite-framed languages’ (hereafter S-languages). Type i. and type iii often are, to a large extent, mutually exclusive, at least in ‘dominant and typical patterns’. Chinese, however, where both types are frequently used, challenges this typology (see sections 2 and 3 for details). Languages using verbs to code core path meanings sometimes have path verbs expressing both autonomous and caused motion, as is the case with French sortir expressing both ‘exit’ and ‘make exit’, and Japanese deru and dasu expressing, respectively, autonomous and caused motion ‘out’ (infix -s- is causativizing in Japanese). Others like Thai or Chinese lack path verbs expressing caused motion, and so have to combine them in serial verb constructions or in verbal compounds to obtain causative motion meaning. Type ii often combines with the two other types, as is seen in languages like French (for instance in se précipiter dans la maison ‘to rush in the house’) which mainly encodes path into verbs, and in languages like Hungarian (ház-ba [house-into] ‘into the house’), which typically encodes path into preverbs. In Chinese, a language with a split type of encoding that allows the use of Type i and Type iii, adnominal encoding is observed too, ex. 从屋里 cóng wū-li [from room-inside] ‘from inside the room’, which includes both a preposition ‘from’ and a postposition ‘inside of’. As deictic path is expressed much more often through satellites or verbs than through prepositional phrases (hereafter PPs) in the languages we investigated, we focus here on types i and iii, the two categories which do not depend on the overt expression of the Ground Noun in the clause. We sometimes use the English adverbs ‘hither’ and ‘thither’ to translate directionals marking, respectively, motion towards and away from the speaker. However we use English verbs ‘come’ and ‘go’ to gloss Chinese deictic directionals in the examples, to show their close connection with deictic verbs. In the discussion, we refer to deictic verbs or directionals marking the

The Linguistic Categorization of Deictic Direction in Chinese

71

motion towards the speaker as venitive, and to verbs or directionals marking a motion away from the speaker as andative verbs or directionals. 1.2. S-languages and V-languages, deictic and nondeictic directionals Talmy (1985, 2000) grouped languages according to the way they package the semantic components of a motion event into various linguistic forms. The path categorization parameter is the term most relevant to our present discussion. Path in a motion event, as defined by Talmy (2000:25), is the path followed by the Figure object with respect to the ground object. We give here special significance to whether the syntactic slot used for deictic paths is distinct from the slot designed for nondeictic paths, i.e. to a possibly bipartite categorization of paths in the typical expression of a motion event. A closer look at languages which grammaticalize path into a closed-class category (Slanguages) shows that some of them, like Slavic languages (Russian, Polish) and Hungarian, can affix only one prefix (or satellite) to the verb root. This is not the case in Chinese, which belongs to the group of languages allowing several path satellites to be affixed to a verb root. These path satellites, often called ‘directionals’ or ‘directional complements’ in the literature, may be apportioned among various subsets, usually appearing in distinct syntactic slots, which correspond to various semantic and cognitive categories. Craig (1993: 24) noticed for Jakaltek (a Mayan language spoken in the Chiapas and in Guatemala) that deictic directionals were assigned the most external slot when several directionals were fixed to the verb root: (1)

a. ten-ik-toj

vs. b.

ten-il-tij

hold-in-thither

hold-out-hither

‘push into away from speaker’

‘pull out toward speaker’

German and Chinese also converge in assigning the deictic satellite the most external position, as illustrated in examples (2) and (3): (2)

(3)

a. hin-aus-werfen

b. her-aus-tragen

thither-out-throw

hither-out-carry

‘throw out away from speaker’

‘carry out toward speaker’

a. 扔出去

b. 搬出来

rēng-chu-qu

bān-chu-lai

throw-exit-thither ‘throw out away from speaker’

carry-enter-hither ‘carry in toward speaker’

In Lolovoli (an Oceanic language spoken in N.E. Ambae), deictic directionals and geographic directionals (‘motion across/on the level’; ‘motion up/landward’; ‘motion down/seaward’) are combined to form a set of nine directional verbs. These directionals combine with manner verbs to describe the direction of movement that is concurrent with the action expressed by the verb, and form verb complexes such as rororo hamai ‘creep up toward the speaker’ or soi hivo ‘fall down away from

72

C. Lamarre

the speaker’. The three deictic path elements are: ‘away (from deictic centre)’; ‘towards deictic centre’; ‘towards addressee, or past/future deictic centre’ (see Ross, 2004, Hyslop, 2001: 199–229; 292; 295–98). Let us now look in detail at the Chinese twofold encoding of spatial paths, deictic and nondeictic. Chinese directionals are usually called ‘directional complements’, as they follow a verb to indicate the direction of the action, and share many features with the verbal items following a verb to indicate the result of the action (the ‘resultative complements’). Interestingly, the combinatory possibilities of nondeictic and deictic directionals are the formal criteria generally adopted in grammar books to define the grammatical category of path directionals. Table 1 lists the main path satellites (or directionals), to show how deictic and nondeictic items combine. Symbol ∅ indicates zero. Pd stands for ‘deictic path’, Pnd for ‘nondeictic path’, and NPLOC for the Locative or Ground Noun Phrase (the reference of the motion). There are 9 core members of the grammatical category of path satellites (usually called ‘directional complements’), on which everyone agrees, 7 nondeictic and 2 deictic. Table 1. Standard Mandarin simple and bipartite path satellites Pnd Pd zero ∅ lai hither qu thither zǒu away

zero ∅ ∅

来 lai 去 qu 走 zǒu

up shang*

up qi*

down off xia 下 xia

in jin

out chu

back hui

over, past through guo

to dao**

上 shang

起 qi

进 jin

出 chu

回 hui

过 guo

到 dao

上来 shanglai 上去 shangqu /

起来 qilai

下来 xialai 下去 xiaqu

/

/

进来 jinlai 进去 jinqu /

出来 chulai 出去 chuqu /

回来 huilai 回去 huiqu /

过来 guolai 过去 guoqu /

到…来 dao…lai 到…去 dao…qu /

*-qi is source-oriented, used for instance in 站起来 zhànqilai ‘stand up’, whereas -shang is goal-oriented, used for instance in 爬上去 páshangqu ‘climb up (a tree)’. ** -dao ‘to’ cannot combine directly with deictic directionals, and must followed by a locative phrase, like in 扔到河里去 rēngdao héli qu ‘throw into the river’. As path verbs and path satellites are written with the same characters, Table 1 may be used to describe simple and bipartite path verbs as well. However, we must pay attention to the following points: when a path verb is used after a manner or cause verb in a verb complex, i.e. when it functions as a path satellite (or directional), it will be unstressed and lose their tonal value, as for instance in 寄回来 jì-hui-lai [send-back-hither] ‘send back toward the speaker’ (the two last syllables do not bear any tonal marks, they are atonal syllables). When they function as path verbs, in the case of a bipartite item, the nondeictic component will bear stress, and the deictic element will be unstressed, as in 回来 huí-lai [return-hither] ‘come back’. Deictic motion verbs lái and qù recover their original tonal value when used alone, as in 去 qù ‘go!’ We discuss path verbs in Section 2, and path satellites in Section 3.

The Linguistic Categorization of Deictic Direction in Chinese

73

1.3. More on Table 1: path verbs and path satellites Table 1 includes three items which behave less prototypically than the core items corresponding to English verb particles ‘up’, ‘down’, ‘in’, ‘out’, ‘over/through’ and ‘back’. In this section we comment on the irregular item -qi ‘up’, and on two more controversial items, -dao ‘to’ and -zŏu ‘away’, as well as on other path verbs which do not function as satellites and are not included in Table 1. The source-oriented path verb qǐ ‘rise’ In Standard Mandarin, the deictic opposition is neutralized for the source-oriented path verb qǐ ‘rise’, which can only combine with the venitive directional -lai, probably for cognitive reasons (note that this is not necessarily the case for other dialects). For instance, a group of children looking at a kite rising up into the sky (and going away from them, the deictic center) will use the path verb qǐ ‘rise’ with the directional -lai and exclaim: 起来了， 起来了! qǐ-lai-le, qǐ-lai-le ! (lit. ‘it’s coming up’). This asymmetry is observed when -qi appears after co-event verbs and functions as a satellite too. The status of 到 -dao ‘to’ The verb 到 dào means ‘to arrive’. In a postverbal position it introduces a reached goal and means ‘to’, its status is a controversial issue. Only -dao is unable to combine directly with the deictic directionals -lai et -qu. In Standard Mandarin it must be followed by a locative NP, like in păo-dao-xuéxiào-lai [run-to-school-come] ‘run to school toward the speaker’. Patterns like *păo-dao-lai [run-to-come] being ungrammatical. Some linguists argue that it should therefore be excluded from the repertory of path directionals and be treated as a postverbal preposition. Others, for instance Liu Y. (1998), consider that it shares most of its syntactic features with the nondeictic path verbs/directionals, and should be included in the list. We find the latter analysis more convincing and therefore included -dao in Table 1. A source-oriented andative directional We noted before the possibility for deictic and nondeictic path directionals to combine is often taken as a defining feature of the category. This explains why almost no Chinese grammar acknowledges the existence of a third deictic motion verb, 走 -zŏu ‘go away’, nor is it included in the list of directional complements. As a source-oriented andative directional it means ‘away’ and is frequently used. It cannot combine with nondeictic directionals, though, and is thus generally considered as a ‘resultative complement’, not as a ‘directional complement’. The opposition observed in northern Chinese between source-oriented vs. goal-oriented andatives directionals -qu ‘thither’ and -zou ‘away’ is quite recent, mainly a 20th cen. development (see Lamarre and Liu 2001 and Lamarre 2006 for a more detailed discussion). Note that the andative directionals are less ‘essentially deictic’ than the venitive one, like in many other languages (see François 2003). Other Path verbs We mentioned that Table 1 could be used as a list of simple and bipartite path verbs as well. But actually this would list only a small part of the path verbs used in the language, i.e. only those which function as satellites, and can appear after verbs expressing the manner or cause of motion. Path verbs like 升 shēng ‘rise’, 沉 chén ‘sink’, 钻 zuān ‘make one’s way into’, tuì ‘move back, return’ or 穿 chuān ‘pass through’ are semantically very similar to 上 shàng ‘ascend’, xià ‘descend’, 进 jìn ‘enter’, 回 huí ‘move back’ or 过 guò ‘pass through’, but did not

C. Lamarre

74

grammaticalize into path satellites (or ‘directional complements’, as they are generally called in the literature). Even a path verb such as 掉 diào ‘fall’, which can function as a satellite after a manner-of-motion or a cause-of-motion verb is not generally included in the list of path satellites because it cannot combine with the deictic directionals -lai and –qu. Example (4) compares 掉 diào ‘fall’ to another semantically similar path verb xià ‘descend’. Both mean as a directional ‘off’ when they function as a satellite, (4a) and (4b) can be used to describe the motion of a hat blown off by the wind for instance. 掉 -diào cannot combine with deictic directionals, as is shown in (4c). Conversely, it frequently enters the paradigm of manner-of-motion verbs, to combine with path satellite -xia, as shown in (4d). (4)

a. 吹 下 来 chuī-xia-lai

b. 吹 掉 chuī-diào

blow-descend-come

blow-fall

‘blow off’

‘blow off’

c. * 吹 掉 来 * chuī -diao-lai

d. 掉 下 来 diào-xia-lai

fall-descend-come

blow-fall-come

‘blow off’

‘fall down hither’

Talmy (2000: 53–57) distinguishes between various components of the path: the deictic component, the vector component, and the conformation components. Table 1 shows that Chinese categorization makes a clear-cut distinction between deictic and nondeictic components, which are assigned different syntactic slots. The distribution of the other components is much more complex, and we will not discuss it further. We adopt here the binary opposition ‘deictic’ versus ‘non deictic’, which proves sufficient for the issues dealt with in this paper. 1.4. Deictic path is special Deictic Path is a special kind of path. Talmy mentions that: ...unlike Spanish, Korean can represent both Path components concurrently in nonagentive sentences [...] Thus Korean is a characteristically Path verb type of language, but it structurally distinguishes the Deictic component from the Conformation component of Path and accords it higher priority when both components are present. (Talmy 2000: 57) We will see that Japanese behaves very similarly to Korean: for autonomous motion, it frequently expresses deictic and nondeictic motion together, like in deteike ‘go out!’ or oritekita ‘[she/he] came down’. Furthermore, Slobin (1997:30) noted that S-languages do not necessarily work along the same pattern: Path satellites in German encode directionality, as in English. But in addition, they encode deictic viewpoint by means of the particles hin ‘thither’ and her ‘hither’, which can combine with the directional particles. The range of possibilities is thus considerably greater than in English, allowing for

The Linguistic Categorization of Deictic Direction in Chinese

75

expression of VIEWPOINT PERSPECTIVE along with directionality of motion. (Slobin 1997: 30) In other terms, it appears that the way a language encodes deictic paths cannot be entirely predicted by the way it encodes (nondeictic) paths in general. Evidence of the specificity of deictic paths can be found in the fact that many typical S-languages, which are supposed to encode path into satellites rather than into verbs, do have deictic motion verbs and use them frequently. For instance, English typically uses satellites for nondeictic paths, but uses deictic verbs come and go instead of satellites (there are no verb particles conveying deictic direction, the adverbs hither and thither are now obsolete). This allows deictic and nondeictic paths to be combined in the same verb phrase, ex. come in. Similarly, in Hungarian, a typical S-language, none of the core path meanings (‘out, in, up, down’ etc.) can be encoded by a verb, except deictic paths. Hungarian uses two deictic motion verbs megy ‘to go’ and jön ‘to come’ (note that megy is sometimes translated as ‘walk’, and is less deictically marked than jön). Like English, Hungarian also frequently combines these deictic verbs with path satellites (in Hungarian, preverbs), as in for instance be-jön [in-come-3SG-PRE] ‘she/he comes in’. Hungarian also has two deictic path preverbs ‘to here/to there’. German has deictic path satellites (see ex. 2 above), but actually frequently uses deictic verbs, in spite of the fact that the andative item gehen still often retains its manner meaning of ‘going by foot’. In contrast with this, Russian and Polish have to encode both deictic and nondeictic path in prefixes. As they lack synthetic deictic motion verbs like come, they will combine a manner-of-motion verb (‘move on foot’, ‘move by vehicle’, ‘fly’ etc.) and a prefix expressing a deictic (or quasi-deictic) path ‘toward the speaker’, ‘away from the speaker’ instead. These structural differences make it easy to express both deictic and nondeictic paths in the same verb phrase in English, German or Hungarian (come out, go down etc.), whereas in Russian or Polish, this proves to be quite difficult, just as it is in French or in Spanish, two V-languages. Let us now turn to Chinese. 2. Deictic Motion in Chinese: The Verbal Encoding. 2.1 General facts about the encoding of paths in Chinese Chinese frequently uses path verbs, like French. However, like German or Jakaltek, it can encode deictic paths, nondeictic paths, and a co-event in the same verb complex (see ex. 3 above). It appears thus to be an exception to the general tendency for languages to choose between type i) of verbal encoding) and type iii of adverbal encoding. We argued in Lamarre (2003, 2007a) that it is indeed better to treat Chinese as a ‘split’ type, rather as an ‘equipollent’ type like Thai (see Slobin 2004), for the following reasons: 1) Chinese does not randomly use verbal or adverbal encoding of paths, the V-language type of encoding is actually available only in the case of autonomous motion events; in caused motion events, paths will necessarily make use of adverbal encoding. Example (5a) shows the bipartite path verb jìnlai [enter-come] ‘come in’ used to answer when someone knocks at the door. Example (5b) may be used in a

C. Lamarre

76

narrative to describe someone running into the room that holds the deictic center (or the hero in a narrative). In both cases the Figure (the moving entity) is a human being controlling his / her motion, which allows the speaker to choose between a V-language type of encoding (a path verb, type a.) and an S-language type of encoding (type b). In example (6), however, the figure is an inanimate object which relies on an external force to be moved out of a pocket or a bag, a typical instance of a caused motion event. Only an S-language type of encoding, i.e. a co-event verb ‘take’ followed by the path satellite ‘out’ (derived from the path verb ‘exit’) will be available in the latter case in Chinese (6b). Verbal encoding is ruled out (6a is not grammatical). (5)

(6)

a. 进来! jìn-lai

b. 小王 跑进来 说:... Xiǎo-Wáng pǎo-jin-lai shuō

enter-come

Young-Wang run-enter-come say

‘Come in !’

‘Wang run in [toward speaker] and said:….’

a. *把 证件 出来 *Bǎ zhèngjiàn chū-lai OM

ID-card exit-come

b. 把 证件 拿 出 来! Bǎ zhèngjiàn ná-chu-lai OM

ID-card take-exit-come

‘Take out your identity card’ This contrasts with French and Japanese, two V-languages, which can use causative path verbs sortir and dasu ‘make come /go out’ (cf. French: sortez vos papiers [exitCAUS-IMP your papers], Japanese: mibunshoo wo dase [papers-ACC exitCAUS-IMP]. Thus, we can state that, in Chinese, verbal encoding is available only for autonomous motion events, whereas adverbal encoding (S-language type of strategy) is the only type of encoding valid both for autonomous and caused motion events. According to Talmy (2000: 64–66), similar ‘split patterns’ also exist in Tzeltal, a Mayan language, and in Emai, a Niger-Congo language. 2) Chinese is likely to have been a V-language in its early stages and shows a drift towards an adverbal type of encoding (see Talmy 2000: 118, Lamarre 2003, and also Peyraube 2006 and D. Xu 2006: 146–188 for a discussion of historical data). This reminds us of the way an S-language like Latin evolved into Romance V-languages, leaving some fossilized patterns in French like path-encoding prefixes (see Kopecka 2004 and 2006). It is therefore more enlightening to treat Chinese as a language exhibiting split encoding, to grasp its historical and dialectal variation, rather than to classify it into a third category of ‘equipollent languages’. In the case of autonomous motion events, when both patterns are available, functional and stylistic factors appear to weigh heavily on the strategy used to make the choice. If we compare dialogues taken from TV series and literary texts including descriptions, we find that V-language patterns like (5a) are overwhelmingly prevalent in dialogues (more than 9/10), whereas the proportion of V-language patterns and S-language patterns is much closer to 50 /50 in literary corpora. Literary texts use more co-event verbs for descriptive purposes. We will see in Section 4 that the distinction between autonomous and caused motion is also

The Linguistic Categorization of Deictic Direction in Chinese

77

crucial to understand the differences between Chinese and Japanese when it comes to expressing deictic direction. 2.2. Path verbs tend in Standard Chinese to be bimorphemic We stated above that when the figure is moving on its own and has control over its motion (what we call ‘autonomous motion events’), Chinese may encode the path either in the main verb, like a V-language, or in a path satellite (or directional), like an S-language. Mandarin Chinese path verbs frequently take the form of a bipartite verb where a nondeictic element combines with a deictic one. Simple nondeictic path verbs may also be used when they are followed by a locative noun phrase indicating the goal, route, or sometimes also the source of the motion. This constraint can be understood as follows: both the locative NP and the deictic directional function as a kind of reference to the path of motion, an objective one for the former, a subjective one in case of the latter. Table 2 shows the proportion of clauses with and without deictic directionals for four nondeictic path verbs, ‘exit’, ‘enter’, ‘cross, go/come over’ and ‘return’, in a spoken corpus (7.5 hours of dialogues taken from the TV series Married for ten years). In this table and hereafter, Vco stands for ‘co-event verb’, the verb expressing the manner or the cause of the motion (the term ‘co-event verb’ is borrowed from Talmy 2000). Pd stands for ‘deictic path’ and Pnd for ‘nondeictic path’. NPLOC stands for the Locative (or Ground) NP used as the reference for the motion. Table 2. Four nondeictic path verbs with/without deictic directionals (dialogues) Followed or not by a locative NP With or without deictic directionals -lai/-qu ‘enter’ 进 jìn ‘exit’ 出 chū ‘cross, come/go over’ 过 guò ‘return’ 回 huí total

yes

no

total

without

with

without

with

8 2 0 24 34

0 0 0 4 4

4 0 0 8 12

23 43 38 160 264

35 45 38 196 314

Table 2 shows the prevalence of the bipartite pattern [nondeictic path + deictic path], especially when no locative NP follows (264/ 276, i.e. more than 95%). If we look into the details of the data, we can observe that the relatively low usage of deictic directionals for ‘enter’ and ‘return’ is due to the high frequency with which quasi-lexicalized compounds are used, where the locative NP is the object of the path verb, like 回家 huí-jiā [return-home] ‘return home’ (which accounts for 21 out of the 24 occurrences of ‘return + NPLOC) or 进屋 jìn-wū ‘enter the room’. Jìn ‘enter’ do not require a deictic directional in the disyllabic compound qǐngjìn ‘please get in’. Huí ‘return’ may be used alone, not guò ‘cross, go /come over’. Literary corpora show the same tendency. This can be seen as reflecting a lexicalization pattern, path verbs tending to be lexicalized into disyllabic items which include a kind of objective or subjective

78

C. Lamarre

reference as their second syllable . We will see in section 3 that this holds for path directionals too. 2.3. Synthetic or analytic deictic motion verbs It is well known that deictic paths may be lexicalized into the verb when combined with other meanings, for instance in the transitive English verbs bring and take (see Fillmore 1997: 82–102 and Talmy 2000: 161). French equivalents apporter/ emporter and amener/emmener show the non-dominant (and already fossilized) pattern of path prefixes added to a verb root (see Kopecka 2006). These verbs incorporate the manner or motion and some kind of deictic information. Chinese equivalents 拿来 ná-lai [take-come], 拿去 ná-qu [take-go] and 拿走 ná-zŏu [take-away], as well as Japanese equivalents mottekuru [take-te-come] / motteiku [take-te-go], will also be ‘analytic’, i.e. bimorphemic (see Uehara 2006), they can be decomposed into a deictic motion element at least partly similar to deictic verbs ‘come’ (lái and kuru) and ‘go’ (qù and iku), and a co-event verb (ná and motsu) meaning ‘take/hold in the hand’. In Chinese, ‘basic’ deictic motion verbs lái and qù also show an ‘analytic’ variant. Northern and central Chinese dialects often use an analytic type of encoding to express the deictic motion to a goal, by combining the semantically bleached path verb shàng ‘ascend’ or dào ‘arrive’ followed by the goal NP, then by the deictic directionals -lai/qu. In this case these verbs loose their original meaning, shàng no longer conveys any motion upwards, and dào includes the motion before the ‘arrival’ instead of profiling the arrival itself, both just mean ‘move to’. For instance, our language consultant from Jizhou (Hebei province, a northern dialect close to Standard Mandarin) was reluctant to use the synthetic pattern, and preferred shàng to dào as the neutral verb of motion in the analytic pattern. In Heyang and Yongshou, two Mandarin dialects spoken in Shaanxi, the synthetic encoding pattern is not used either but dào is preferred in the analytic pattern (see Tang and Lamarre, in print). However, more conservative southern Sinitic languages such as Cantonese and Taiwanese use synthetic motion verbs cognate to Mandarin lái and qù, followed by the goal locative NP instead (the analytic pattern is not accepted, see D. Liu 2000 and Liu’s contribution in this volume for a discussion on Cantonese vs. Beijing and Shanghai). Standard Mandarin, the national koine, uses both northern (analytic) and southern (synthetic) patterns indifferently, and uses both shàng and dào: (7)

你 上 哪儿 去 了主 下午 找 不 到 你。 Nǐ shàng nǎr qu le, xiàwǔ zhǎo-bu-dào nǐ。 2SG move:to where go PFV afternoon look-NEG-ACH 2SG

‘Where have you been? This afternoon I couldn’t find you.’ [Ten years, 3] (8)

咱们 现在 去 哪儿 ? [Ten years, 8] Zánmen xiànzài qù nǎr 1PL now go where

‘Where are we going now?’ (asked by the driver to his boss)

The Linguistic Categorization of Deictic Direction in Chinese

79

The analytic variant of the ‘basic deictic verbs’ ‘come’ and ‘go’ found in Northern Chinese reminds us of that used in Longgu (an Oceanic, Austronesian language spoken on Guadalcanal, Solomon Islands), as described in Wilkins and Hill (1995). Wilkins and Hill (1995) challenged the general opinion that deictic motion verbs ‘come’ and ‘go’ are lexical universals which manifest a universal deictic opposition. In Longgu, ‘come’ and ‘go’ are expressed through the combination of a deictically neutral motion verb la ‘travel, move along a path’, and of directional particles -mai ‘hither’ and -hou ‘away from here’. Chinese, more than Japanese, provides evidence to support Wilkins and Hill’s point of view. We consider the ‘analytic pattern’ described above for Northern and central dialects (as well as Standard Mandarin where it coexists with the synthetic pattern) as reflecting the same bimorphemic pattern of lexicalization we described in the preceding section 2.2. In the case of deictic motion verbs, the first morpheme or ‘nondeictic path’ gets bleached and does not express a concrete nondeictic path any longer This allows deictic and nondeictic path verbs to be unified into a bimorphemic structure and may be seen as some kind of leveling (see Lamarre 2007b). The opposition between a source-oriented 走 zǒu ‘walk, go away, leave’, vs. a goal-oriented andative qù observed in Northern Chinese is another piece of evidence in favor of Wilkins and Hill’s conclusion: deictic motion verbs ‘come’ and ‘go’ are not basic, and their lexical semantics vary to the extent that no ‘universal’ can be safely assumed. In Chinese, the venitive verb lái has been extremely stable, but the andative verb shows huge variation both in history and in dialects. One parameter of this variation is that of source-oriented vs. goal-oriented andative motion . 3. Deictic Direction in Chinese: The Adverbal Encoding 3.1. Prototypical Path Satellites are bimorphemic Table 1 listed both simple and bipartite deictic and nondeictic path directionals, which suggests that speakers are free to express only deictic, or only nondeictic paths. However, we showed in section 2.2 and 2.3 that there is a strong tendency for path verbs to be bimorphemic, i.e. to be composed of a nondeictic and a deictic element. This tendency exists when path verbs function as satellites (or directional) too. Data show that although clauses where the nondeictic directional is used alone (like [V co + Pnd + zero]) or clauses where the deictic directional is used without the nondeictic one (like [Vco + zero + Pd]) are not prohibited; the prototypical path satellite, like the prototypical path verb, is in Chinese bipartite, and combines a nondeictic and a deictic element, the verb phrase taking the form [Vco + Pnd + Pd]. We describe constraints on monomorphemic path satellites in the next sections. 3.2. Nondeictic path directionals tend to be followed by deictic directionals The tightest constraint is observed for combinations of the type [Vco+Pnd]. Previous studies (Ju 1992, Y. Liu 1998:36–8, Qi 1998:180) noted that unless the verb is followed by a locative NP (or by another NP like the patient of the verb, the Figure), path directionals are usually bipartite, the deictic direction being overtly expressed. Y. Liu (1998:36) gives the following example as ungrammatical:

80

C. Lamarre

(9) *下课 了， 孩子 们 跑出。 xiàkè le, háizi-men pǎo-chu class:end PFV child-PLUR run-exit

‘classes are over, the children run out’ In order to become grammatical, the verb compound ending the sentence should be extended with a deictic directional like -lai (if the speaker looks at the children from outside the classroom), i.e. pǎo-chu-lai [run-exit-come] ‘run out toward the deictic centre’. If a noun phrase appears after the verb complex, this constraint disappears. Table 3 gives the relative frequency of use of non-deictic path directionals (Pnd) with and without deictic path directionals (Pd) in a sample of 7.5 hours of TV dialogues (128 clauses). We chose clauses which include a combination of a co-event verb and a path directional not followed by any NP. Data show that, in most of the cases, the use of a non-deictic directional also implies the expression of deictic direction. In other terms, pattern (2), [Vco-e PndPd], which uses a bimorphematic path satellite, is the most frequent. The proportion of autonomous motion and caused motion clauses is given in the brackets (autonomous / caused) . Table 3. Correlation in the use of non-deictic and deictic directionals (dialogues) Pnd pattern

(1) VcoPnd ∅ (2) VcoPndPd (3) Vco∅Pd total

Pnd =∅

起 -qi ‘up’ *

进 -jin ‘in’

出 -chu ‘out’

上 -shang ‘up’*

/

/

/

/

6 (2/4) /

18 (7/11) /

2 (2/0) 1 (1/0) /

7 (3/4) /

10 (9/1) 7 (3/4) /

6

18

3

7

17

31 (3/28) 31

下 -xia ‘down’

过 -guo ‘across over’ /

回 -hui ‘back’

18 (5/13) /

28 (5/23) /

18

28

/

total

12 (11/1) 85 (26/59) 31 (3/28) 128 (40/88)

* Both 起 -qi and *上-shang mean ‘up’ but the former is used for goal-oriented motion, the latter source-oriented motion (see Table 1). Only two non-deictic directional items (goal-oriented ‘up’ and ‘down’) out of seven happen to combine with a co-event verb without a deictic directional (pattern [Vco Pnd]). They are used after verbs of change of posture and placement (‘sit’, ‘kneel’, ‘squat’, ‘put down [your bag]’) where the deictic opposition is less likely to operate for quite understandable semantic reasons. Other directionals, ‘in’, ‘out’, ‘across /over’, and the source-oriented ‘up’, demand the use of a following deictic directional (pattern [ Vco Pnd Pd]). Thus in this sample of dialogues, 88% of the clauses including a nondeictic directional (85 out of 97) also included a deictic one, if the syntactic environment required it (no NP after the verb complex). We now know which directionals are less likely to take overt markers of deictic direction: those expressing vertical paths and where the human body does not undergo any change of position. The 31 clauses that do not include nondeictic directionals ([VcoPd] patterns) will be discussed in the next section. Note that Table 3 only lists directionals used in their spatial meaning .

The Linguistic Categorization of Deictic Direction in Chinese

81

These findings confirm an observation made by Y. Liu (1998: 38): she looked at clauses that included a co-event verb followed by -shang ‘up’ (goal-oriented) in her (written) corpus, and found that in 97% of the 365 clauses where -shang was not followed by deictic directional -lai or -qu, the verb was followed by a locative or a Figure NP. On the other hand, most of the clauses using the bipartite path directional ( pattern [V+Pnd+Pd]) did not include any postverbal NP, and the verbal complex appeared in the clause-final position. The expression of deictic motion can be considered as one of the linguistic devices which express subjectivity. It is thus not surprising that dialogues should include more deictic directionals than literary descriptive texts or ‘objective’ news reports. Let’s now compare a spoken corpus (the same TV series) to a literary corpus (which includes some dialogues too). In spoken Standard Mandarin, Figure NPs rarely appear after the verb complex, in other terms, one of the conditions that allow deictic directionals to be omitted is rarely met. Locative NPs other than goal NPs rarely appear after the verb complex too. Table 4 shows that the opposite is true in literary corpora: in many clauses describing a motion event, the locative NP or the Figure NP appears after the verb, making it easier to leave deictic information covert, which fits with the ‘objectivity’ of written texts. The two corpora compared here are dialogues of the TV series Married for ten years (7.5 hours) and Chen Rong’s novel At middle age. We looked at the 7 main nondeictic path directionals (-dao is excluded, like for Table 3). Three patterns were taken into accounts: In pattern (A) no locative or Figure NP appears after the verb complex. This is an environment which usually requires the expression of deictic direction (the verb complex is clause-final), both in written and in spoken style. In pattern (B) the nondeictic path satellite is followed by a locative NP. Pattern (C) is restricted to caused motion sentences or presentative sentences, where the nondeictic path satellite is followed by the Figure NP. (B) and (C) are patterns which are said to allow the omission of the deictic directional. Table 4. [Vco + Pnd] compounds with and without postverbal NPs, and frequency of the expression of deictic path, in a contemporary novel and in TV dialogues (% of clauses with deictic path per pattern) pattern: ( position of the locative or /and Figure NP) TV dialogues % of expressed deictic path literary corpus % of expressed deictic path

(A) NPLOC and/or NPfig covert or preverbal 97 85 (88%) 102 81 (80%)

(B) Vco Pnd +NPLOC

(C) Vco Pnd + NPFIG

4 1 (25%) 51 5 (10%)

2 2 (100%) 69 27 (40%)

For Pattern (A), in the literary corpus, -xia ‘down’ accounted for 20 out of 21 occurrences of a nondeictic path directional without -lai or -qu, like in the dialogue corpus (see Table 3 above). As for Pattern (B) and (C), Table 4 shows that they are actually rare in dialogues, where the Figure NP usually appears before the verb when overtly expressed. Locative NPs mainly occur after -dao in dialogues, not often after the other path verbs. We mentioned earlier that the goal marker -dao requires to be followed by a goal locative NP, and this can account for some specific

82

C. Lamarre

behavior concerning its lexicalization into a bimorphemic path verb. Thus Table 4 shows that even if we take into account these patterns where NPs appear after the verb complex, the general tendency shown in Table 3 is not altered for spoken data. Let us give examples of some of these patterns. Example (10) illustrates the most frequent pattern in the case when the figure is overtly expressed (pattern A): it is put before the verb with the help of the object marker (OM) bǎ: (10)

你 给 妈 把 电视 给 搬 过 去 吧。 [Ten years, 6] Nǐ gěi mā bǎ diànshì gěi bān-guo-qu ba 2SG BEN Mum OM TV:set AGT carry-cross-go PRT

‘carry the TV set to Mum’s place’ In example (11) the Figure is covert but easily inferred from the situation (Pattern A): the speaker refers to a carpet she brought back from Australia and sold to the hearers. The locative NP expressing the source of the motion is expressed through a preverbal PP, showing the pattern [Preposition+NPLOC+Vco-e + Pnd + Pd]. The co-event verb is a dummy verb (see Talmy 2000: 284–5). (11)

我 呢主 从 那边 弄 回 来 也 挺 不 容易 的。[Ten years, 2] wǒ ne cóng nèibian nòng-hui-lai yě tǐng bù róngyì de 1SG PRT from there get-return-come also very NEG easy PRT

‘I had quite a lot of trouble to bring it back from there’ The following examples illustrate pattern (B), [Vco-e+Pnd+NPLOC], where a locative NP appears after the verb and deictic direction may be omitted. The deictic directional is added in example (12), not in example (13), a descriptive sentence in a narrative. (12)

能从火车站一直走回滨河路去。 [Ten years, 8] néng cóng huǒchēzhàn yìzhí zǒu-hui-Hébīnlù-qu can from station all: the: way walk-return-Hebin-street-go

‘We could walk from the station all the way back to Hebin Street’ (when we were young and in love) (13)

三个大汉 进了手术室主… [Middle age, 11] sān ge dà-hàn chōng-jìn-le shǒushù-shì,… 3 CL big-fellow rush-enter-PFV operating-room

‘Three burly fellows rushed into the operating room,…’ Example (14) illustrates pattern (C), [Vco-e+Pnd+NPFIG], where the figure of a caused motion event appears after the verb complex. Patterns (B) and (C) are widespread in the written style, but actually quite rare in dialogues. (15) 她叹了口气，从抽屉里拿出点零钱说:… [Middle age, 8] tā tàn-le kǒu qì cóng chōuti-li ná-chu diǎn lingqián shuō 3SG heave-PRF CLF sigh, from drawer-inside take-exit some small.change say

‘She sighed, took a few coins out of the drawer, and said…’

The Linguistic Categorization of Deictic Direction in Chinese

83

3.3. Deictic directionals tend to be preceded by nondeictic directionals As for constraints on [Vco+Pd] (i.e. pattern (3) in Table 3, where the nondeictic path element does not appear), our data show that, in dialogues, this pattern is mainly used to encode caused motion events (the proportion was of 3 autonomous motion clauses vs. 28 caused motion clauses in Table 3). In autonomous motion events, where the co-event verb expresses the manner of motion, andative directional –qu is rare. Lamarre 2005 argues that manner-of-motion verbs like ‘run’ or ‘walk’ combined with the andative verb qù depart in many ways from typical directional compounds, and notes that such patterns are typical of the written style (see also Liu Y. 1998:36, and Xu Dan’s contribution in this volume). This is confirmed by our dialogue data, where the 3 autonomous motion clauses illustrating pattern (3) all use the venitive directional -lái. On the other hand, in the novel At middle age, 35 of the 50 clauses following this pattern describe autonomous motion events, and 13 out of these clauses used the andative directional -qù, all were descriptive sentences in a narrative. Examples (16) and (17) illustrate the heterogeneity of pattern (3) [Vco+Pd] when it comes to aspectual features and textual function. Example (16) is an imperative sentence taken from a dialogue, the speaker complains about the taste of the fish she was served and orders the waitress to call the proprietor of the restaurant, the verb-directional compound is causative. Example (17) is taken from a narrative, the PP expressing direction ‘towards the ward’ proves that the clause is atelic in spite of the use of a directional. (16)

把 你们 老板 叫 来。 [Ten years, 2] bǎ nǐmen lǎobǎn jiào-lai OM

2PL boss call-come

‘call your boss [so that he comes here]’ (17) 赵天辉两手插在白大褂的衣兜里，一边同孙逸民谈着，一边向病房走去。 [Middle age, 6] Zhào Tiānhuī liǎng-shǒu chā-zài bái dàguà de yīdōu-li, yìbiān tóng Sūn Yìmín tán-zhe yìbiān xiàng bìngfáng zǒu-qù. Zhao Tianhui two-hands insert-at white coat GEN pocket-inside, SIM with Sun Yimin talk-DUR SIM toward ward walk-go ‘Zhao Tianhui, both hands thrust into the pockets of his white coat, was walking toward the ward while talking with Sun Yimin.’ If we look at non-standard varieties of Chinese, we can observe that there are several dialects which do not allow deictic directionals to combine directly with the co-event verb (pattern (3) [Vco Pd] in Table 3), and use dummy elements instead to fill up the slot of the non-deictic directional. For instance, Mandarin 拿来 ná-lai [take-ø-come] ‘bring’ becomes [take-dummy Pnd –come], with a dummy [.t´/] in

84

C. Lamarre

Shanxi and in Shanghai, a dummy [.lç] in Hebei, or a dummy [.tç] in Shaanxi (see Lamarre and Liu 2001, Lamarre 2002, Tang and Lamarre [in print]). We interpret this as an indirect consequence of the bimorphemic pattern of lexicalization of path verbs. In Chinese, path verbs, when they function as the resultative predicate in a resultative compound, i.e. when they function as satellites after another verb, tend to have a bounding function. In the case of bimorphemic path verbs, only the first item (the nondeictic one) carries out this syntactic function, leaving to the deictic item a mere ‘directional’ function. In Standard Mandarin, this is reflected by constraints on the andative directional -qu ‘thither’, and by the development of another andative directional, -zou ‘away’, semantically more adapted to its bounding function. But some dialects developed a semantically bleached resultative form, used when there is no need to express a concrete nondeictic path. It enters the slot of the nondeictic path resultative and achieves this bounding role. 3.4. Bimorphemic path satellites and Mandarin’s typological characteristics Let’s go back to our initial research question: is this tendency to express deictic path whenever nondeictic path is expressed related to the S-language encoding observed in Chinese? The answer then is apparently ‘no’: data (Table 2) show that even when Chinese behaves like a V-language and expresses autonomous motion events through path verbs, like in 回来 huí-lai! [return-come] ‘come back!’, the typical pattern is a twofold encoding [nondeictic + deictic path]. However, if we look at historical data, we can see that bipartite path verbs are far from being frequent in Classical Chinese. Cui (2005), in a comprehensive study on the evolution of Chinese path verbs from Archaic to Medieval Chinese, showed that the use bipartite path verbs of the type [nondeictic path+deictic path] is not a feature of Archaic Chinese (also called ‘Classical Chinese’). It increases only from the Medieval Period (after the 3rd cen.), following the same evolution as the resultative constructions. To make a comparison between Classical Chinese and Modern Chinese, we surveyed samples of various Chinese texts written in Classical Chinese (Classical Chinese survived the Archaic Period as a written language), and looked at their translations into modern Chinese. Table 5 presents data from two texts of which several translations into Modern Chinese are available, A New Account of Tales of the World (Shishuo Xinyu, 5th cen), and 3 stories taken from Pu Songling’ s Strange tales of Liaozhai (end of 17th cen.). We chose stories related to everyday-life events, preferably with dialogues where deictic motion was likely to be expressed, and checked several translations, to avoid any personal bias. The data given in Table 5 are restricted to the three high-frequency path verbs ‘enter’, ‘exit’, and ‘return’. Even the most conservative translators, who avoided colloquial expressions, regularly added deictic directionals in the 51 cases when the original text used only nondeictic path verbs (the modern Chinese sentences taken into account here are those where the verb complex is not followed by any locative or Figure NP, to be consistent with the rule at work in modern Chinese). Table 5 details the deictic directional added in the modern translation, -lai ‘hither’ or -qu ‘thither’. Please note that the lexical item ‘exit’ remained stable over the period in question, whereas the lexemes for ‘enter’ and ‘return’ underwent lexical replacement (the item at the right

The Linguistic Categorization of Deictic Direction in Chinese

85

of the arrow is the modern lexeme). In the case of ‘return’, several lexemes coexisted in Classical Chinese, only huí is used now as directional complement in Standard Chinese . We can thus conclude that even if we cannot find any real evidence for a causal link between the bipartite structure and the S-language tendency, the typological change towards an S-language probably occurred hand-in-hand with the evolution of path verbs towards a bipartite structure [nondeictic+deictic]. Besides, this development is observed only for a small subset of the nondeictic path verbs used in Chinese, precisely those which express the core path schematic meanings and grammaticalize into path satellites. Other path verbs (for instance those evoked in 1.3) do not show this lexicalization pattern. Table 5. Deictic direction added to nondeictic path verbs in modern translations of classical texts

Shishuo Modern Chinese

入 rù Æ 进 jìn ‘enter’ rù 2 jìn-lai 1 jìn-qu 1

出 chū ‘exit’ chū 8 chū-lai 7 chū-qu 1

Liaozhai Modern Chinese

rù 8 jìn-lai 1 jìn-qu 7

chū 9 chū-lai 7 chū-qu 2

还 huán, 归 guī, 反/返 fǎn Æ 回 huí ‘return’ fǎn 4 huán 8 guī 2 huí-lai 8 huí-qu 6 guī 8

fǎn 2

total

24

27

huí-lai 6 huí-qu 4

Another factor to be taken into account is the tendency for monosyllabic words to develop into disyllabic words widely attested in the history of Chinese lexicon. To sum up Section 3, directionals in Standard Mandarin tend to exhibit a bipartite structure [nondeictic+deictic], which manifests itself by various constraints weighing on patterns using non-deictic directionals without deictic ones, or, at a lesser extent, deictic directionals without nondeictic ones. These constraints are especially strict in the colloquial style, whereas the written language is less rigid. All this points to a high categorization (a high degree of grammaticalization) of path directionals: when a directional is used, it must be taken from a closed-class category of less than 10 elements, and likely takes the shape of a twofold directional [Dnd+Dd]. Standard Mandarin’s looser rules are probably the result of koineization, and maybe also reflect the fact that as a written language it retains fossilized structures from older layers of the language, and maybe even artificial structures coined more recently. 4. The Coding of Paths in Japanese 4.1. Japanese complex predicates and the expression of paths Japanese is a V-language which differs from V-languages like French or Spanish in its wide use of verb complexes. The behavior noted by Talmy (2000: 57) for Korean (see our quotation in 1.4 above) basically applies to Japanese too:

86

C. Lamarre

characteristically, Japanese is a Path verb type of language, but structurally it distinguishes the Deictic component from the Conformation component of Path. Let us give below a more detailed account of these various verbs complexes. The first type of complex predicate is obtained through the combination of two verbal elements at the lexical level, to create compound verbs like kake-agar-u [run-ascend-NPAST] ‘run up’, suberi-de-ru [slip-exit-NPAST] ‘slip out’, and oshi-kom-u [push-enterCAUS] ‘throw in’. (See Matsumoto 1996: 237–81 for an account in English of the semantic relationship between the two components). The first two examples denote autonomous motion, the third example caused motion. Note that this compounding process is not fully productive, and many combinations of manner/cause and path possible or frequent in Chinese or English are left out: kake-agar-u ‘run up’ can exist but not aruki-agar-u ‘walk up’ (see Matsumoto 1996:277). This type of compound verb is not of direct interest to us here, because such verbs usually only combine manner or cause with nondeictic paths, not with deictic paths (*kake-ik-u/*hashiri-ik-u [run-go-NPAST] ‘go running’, *komi-ik-u /*hairi-ik-u [enter-go-NPAST] ‘go in’). Note also that some path and manner verbs have special verb roots used only in verb compounds. Thus in the following pairs: -kom-u vs. hair-u ‘enter’, kake- vs. hashir-u ‘run’, the former are bound forms, the latter are free forms. The other type of verb complex, involving te-linkage, occurs at the syntactic level, and is called by Shibatani (2003) the ‘conjunctive compound construction’. All verb complexes involving a deictic path component fall into this type. Here is a list of the possible combinations involving deictic direction: a) [nondeictic path verb expressing autonomous motion+ te + deictic path verb]: de-te-it-ta (exit-te-go-PAST) ‘go out’; hait-te-kur-u (enter-te-comeNPAST) ‘come in’; modot-te-kur-u (return- te - come- NPAST) ‘come back’; agat-te-ik-u (ascend-te-go-NPAST) ‘go up’ b) [manner verb + te + deictic path verb]: hashit-te-ik-u [run-te-go-NPAST] ‘go running’; oyoi-de-ki-ta [swim-te-come-PAST] ‘(I) came swimming’ c) [compound verb (co-event verb + nondeictic path) + te + deictic path verb]: kake-kon-de-ku-ru [run-enter-te-come-NPAST] ‘run in toward the speaker’ d) [co-event verb + te + nondeictic path verb + te + deictic path verb]: hasit-te- modot-te-kuru [run-te-return-come-te-NPAST] ‘run back toward the speaker’ Matsumoto (1996) calls type b) a ‘participial complex motion predicate’, where the co-event verb expresses the manner of motion (‘run’) or the accompanying event of motion (‘carry’) (see also Tanaka and Matsumoto 1997). 4.2. The expression of deictic paths in autonomous and caused motion events As was noted by Shibatani (2003), although nondeictic motion may be used without deictic direction being expressed (for example in kanojo-wa heya-kara de-ta [she-TOP room-from exit-PAST] ‘she got out of the room’) in casual dialogue, the speaker often expresses both types of path as in pattern a), (heya kara de-te-kita [room-from exit-come-PAST] ‘came out from the room’), otherwise the sentence is felt as lacking an important information: the speaker’s location when the event occurred. The same has been noted for Korean, where this tendency is said to be even stronger than in Japanese (see Fukami 1999).

The Linguistic Categorization of Deictic Direction in Chinese

87

Japanese is known for its various devices expressing subjectivity, and spatial deixis is just one of the ways it is expressed, together with the verbs of ‘giving’ and ‘receiving’; person deixis is underdeveloped. Verbs of coming and going -iku and -kuru, in their basic directional – spatial – use, behave as the head of these complex predicates, bearing the marks of tense, passivation etc. (see Matsumoto 1996, note that Shibatani 2003 argues that even in type b) the manner verb and the deictic motion verb carry the same semantic weight). Two points are to be noted here, which bear some importance in understanding the difference with Chinese. First, the deictic verb, the last item of the verb complex, shares its subject with the preceding verbs, and may only express the motion of the subject of the first verb. Second, in each of the four cases, a) to d), the interpretation of -kuru or -iku as the deictic direction of a motion event is triggered by the semantics of the first predicate. That is to say, whenever the semantics of the first predicate blocks this interpretation of a ‘simultaneous motion’, when for instance the first predicate is an action verb like ‘eat’ or ‘play’, the semantic relationship between the predicates is interpreted as being sequential, ex. tabe-te-ik-u [eat-te-come-NPAST]‘eat and then go’ or ason-de-ku-ru [play-te-come- NPAST]‘play (somewhere) before coming’ etc. e) In complex predicates where the first verb is a causative path verb, the deictic direction is interpreted as the direction of the motion of the agent, with a sequential reading, for instance dashi-te-ik-u [exitCAUS-te-go-NPAST] means ‘take something out and go (somewhere)’ and not ‘take something out away from the speaker’. f) When the first verb expresses an action like ‘take’ or ‘carry’, this difference is blurred by the fact that the agent and the patient move conjointly: hakon-de-ku-ru [carry-te-come- NPAST] means ‘to carry and come’, but as the carrier and the carried object move conjointly, the situations in which such verb complexes are used is roughly similar to the situations wherein Chinese uses 搬来 bān-lai [carry-come] (see example (21) below). The semantic mismatch becomes obvious for verbs like ‘hit’, ‘throw’, ‘kick’ or ‘send (a letter)’, in which the agent does not move together with the patient. For instance okut-te-ik-u [send-te-go- NPAST] can only mean ‘send (a letter etc.) and then go (somewhere)’, and never ‘send a letter away from the speaker’, whereas in Chinese 寄去 jìqu [send-go] precisely means ‘send a letter away from the speaker’. g) As a result, when Chinese combines a causative co-event verb with both nondeictic and deictic paths, there can be no Japanese equivalent. ‘Take out (your purse from your bag)’ will be expressed in Chinese with the verbal complex 拿出来 ná-chu-lai [take-exit-come], but the parallel verb complex in Japanese tori-dashi-teku-ru [take-exitCAUS-te-come- NPAST] will imply that the agent moves towards the speaker after having taken out the purse from the bag (see ex. 23 below). Let’s first look at a few sentences expressing autonomous motion, where the encoding of deictic motion coincides in Chinese and in Japanese. Pattern a) (ex. 18) is similar to a Chinese nondeictic path verb followed by a deictic path element: (18) [Jap.] hait-te-ik-u enter-te-go-NPAST

[Ch.] jìn-qu

de-te-ku-ru

modot-te-ku-ru

exit-te-come- NPAST

return-te-come- NPAST

chū-lai

huí-lai

88

C. Lamarre enter-go

exit-come

return-come

‘go in’

‘come out’

‘come back’

Patterns b), c) and d) have Chinese equivalents too, see examples (19) and (20): (19) [Jap.] 流

=

行く nagare-te-ik-u

[Ch.] 流去 liú-qù (written style)

flow-te-go-NPAST

flow-go

‘go flowing’

‘[water] flows away toward some place’

(20) [Jap.] 駆 け 込 or 走 っ 入 っ 来る 来た hashit-te-hait kake-kon-de-k u-ru -te-ku-rurun-enter-te-co me-NPAST

=

[Ch.] 跑进来 pǎo-jin-lai

run-te-enter-tecome-NPAST

run-enter come

‘come in running’ In the case of caused motion, however, Japanese and Chinese widely diverge. This is not obvious at first sight, because in everyday life, quite a few caused motion events where the agent pushes and carries things imply the joint motion of the agent (the subject) and the object patient of the cause verb, as in example (21) below. Chinese ná and Japanese motsu mean ‘take/hold in the hand’: (21) [Ch.] 你 把 东西 拿来 ! nǐ bǎ dōngxi ná-lai

＝

[Jap.] 荷物 持っ ！ nimotsu-o mot-te-ki-te

2S OM thing take-come

luggage-ACC take-te-come-IMP

‘Bring over the suitcase [to here]!’ However, as in Japanese the head of the verb complex is the final verb ‘come/go’, the deictic motion expresses the motion of the subject. When the subject (the agent) and the object (the figure) move in different directions, like in examples (22) and (23) where only the patient is ‘packed in’ or ‘taken out’, Japanese cannot express deictic direction any longer, or if it does, the meaning of the clause is changed. (22)

把 书 装 进 箱子里 bǎ shū zhuāng-jìn xiāngzi-li

~ 把 书 装 进去 ~ bǎ shū zhuāng-jin-qu

OM

book pack-enter box-inside

OM

book pack-enter-go

cf. Jap. hon o hako-ni tsume-kom-u

book-ACC box-at pack-enterCAUS- NPAST

(＊把 书 装 进) (*bǎ shū zhuāng-jin) OM

book pack-enter

(# tsume-kon-de-ik-u [put-enterCAUS-te-go-NPAST] would mean ‘pack the books in before leaving’)

The Linguistic Categorization of Deictic Direction in Chinese

(23)

你 把 护照 拿 出 来 Nǐ bǎ hùzhào ná-chu-lai

?? 你 把 护照 拿 出 ?? nǐ bǎ hùzhào ná-chu

2SG OM passport take-exit-come cf. Jap. pasupooto-wo dashi-nasai

2SG OM passport take-exit passport-ACC exitCAUS-IMP

89

(# dashitekinasai [exitCAUS-te-come-IMP] would require the owner to go and fetch the passport, take it out, then come back) The Chinese sentence is not acceptable without the deictic element -lai if not followed by the ground NP or the Figure NP, whereas the Japanese translation would have to delete this deictic element. In the situation illustrated by (23), the subject of the sentence is ordered by a policeman to take out (from the pocket of her bag) her passport. The Japanese complex predicate with -ki-nasai ‘come’ will be accepted only in another situation, for instance if the owner of the passport moves to another room to remove her passport from a suitcase, and then come back. Example (24) is taken from the Chinese novel Rickshaw (by Lao She). Japanese deletes the original deictic information, and translates the motion with its causative path verbs dasu and ireru (corresponding to the non-causative verbs deru and hairu). Interestingly, French and English translations delete this deictic information too: (24) [Ch.] 他 把 车 拉 出 去 又 拉 进来 了, ……[Rickshaw chap. 1] tā bǎ chē lā-chu-qu yòu lā-jin-lai le 3S OM rickshaw pull-exit-go again pull-enter-come PFV

[Jap.] Kare-wa kuruma-wo dashi-tari ire-tari… 3S -TOP rickshaw-ACC exitCAUS-ALT enterCAUS-ALT

[Fr.] Il sortit le pousse du garage et le rentra. [Eng.] he pulled the rickshaw back and forth through the gateway... Example (25) is taken from a compendium of Chinese directional complements providing Japanese translations (Hou et al. 2001). Needless to say that these ‘complements’ are troublesome for the Japanese learners of Chinese. Japanese kaesu is the causative form of kaeru ‘return’. (25)

[Ch.]把 球 扔 了 回 去 bǎ qiú rēng-le-hui-qu OM

ball throw-PFV-return-go

[Jap.]ボ ル 投げ返した booru-o nage-kaeshi-ta ball-ACC throw-returnCAUS-PAST

‘threw back the ball (away from the speaker)’ 4.3. Specific uses of deictic motion verbs in Japanese linked to subjectification In the precedent section, we emphasized the pervasive use of deictic directionals in Chinese caused motion clauses, whereas Japanese has to delete this information. However, Japanese does express deictic motion in cases where Chinese

90

C. Lamarre

does not. This section presents two linguistic patterns in which the tendency to express any motion as being related to the location of the agent triggers some verbal compounds with a sequential reading [do something and then go/come] into developing a derived meaning: many activities of everyday life will be expressed as related to the deictic motion of the agent. For instance, someone leaving work in the evening will suggest her colleagues to have a drink ‘on their way home’, that is ‘to drink and then go’ non-de-ik-u [drink-te-go-NPAST]. The closest Chinese equivalent would be sentences like 吃了饭再走 chī-le fàn zài zǒu [eat-PFV rice and.only.then leave] ‘eat something before you leave’, i.e. a complex sentence including a conjunction to link the sequential actions. In most of the cases, however, the Chinese sentence used in equivalent situations will not involve any expression of motion. We refer the reader to Uehara 2006 for a detailed discussion on deixis and subjectivity in Japanese. Another case of grammaticalization (and subjectification) of deictic motion verbs is the use of verbs followed by -kuru ‘come’ to express that event in which the action denoted by the first predicate directly affects the speaker (or someone he/she has empathy with). Such use is restricted to the venitive deictic verb. Shibatani (2003) calls it an ‘inverse marker’, and links it to voice phenomena. In this case, -kuru is not interpreted as a sequential motion even when it appears after causative motion verbs like ‘throw the ball’ booru o nageru. In the following example (26a), borrowed from Shibatani (2003), nageru ‘throw’ and kuru ‘come’ combine to express the deictic direction of the ball (patient), apparently just as in Chinese (26b): (26a) 建 ぼくにボ ル 投げ 来た Ken-ga boku-ni booru-o nage-te-ki-ta

(26b) 把 球 扔 (过) 来 bǎ qiú rēng-(guo-)lai

Ken-NOM 1S-to ball-ACC throw-te-come-PAST

OM

‘Ken threw me the ball’

‘throw me the ball’

ball throw (over) come

However, such constructions can be accounted for as an extension of the spatial deictic direction, to express affectedness. This explains why they lack a corresponding pattern with the andative verb (example 27a is also taken from Shibatani 2003). Subjectivity can only be expressed by the deictic center ‘I’: (27a) *Boku-ga Hanako-ni booru-o nage-te it-ta. I-NOM Hanako-to ball-ACC throw-te-go-PAST

intended meaning: ‘I threw the ball toward Hanako.’ In Chinese, on the opposite, the andative directionals -qu can combine with ‘throw’ to express that the ball is sent away from the deictic center’, as in (26b): the similarity of (26a) and (26b) was deceitful. (27b) 把 球 扔 (过) 去 bǎ qiú rēng-(guo-)qu OM

ball throw (over) come

‘throw him/her the ball’

The Linguistic Categorization of Deictic Direction in Chinese

91

5. Chinese Expresses Deictic Direction More Often than French or English French, like other Romance languages, is known for its pervasive use of personal pronouns, and this widespread use of personal deixis is probably one factor allowing it to deal so loosely with spatial deixis. As we showed in Section 3, one of the factors explaining the pervasive expression of deictic direction in Chinese is the lexicalization pattern of path verbs and path satellites: path verbs are most of the time bimorphemic, and deictic direction gets readily expressed in sentences where nondeictic path is marked. We showed that this tendency was stronger in some syntactic configurations (when no postverbal NP appears in the clause). The following sentences are taken from the famous Chinese novel Rickshaw (1937) and its French and English translations. Let us look first at causative motion events including the nondeictic path satellite ‘exit’ -chu. Neither French nor English expresses the direction ‘toward the speaker’ as Chinese does when someone takes out something from one’s pocket or a bag etc. English translates the co-event verb as ‘take out’, French as sortir ‘take out’. (28)

祥子 没 去 端 碗， 先 把 钱 掏 了出 来：… (chap. 4) Xiángzi méi qù duān wǎn, xiān bǎ qián tāo-le-chu-lai... Xiangzi NEG go hold bowl first OM money pull-PFV-exit-come

[Fr.] Avant de toucher à la nourriture, Siang-tse sortit une liasse de billets … [Eng.] Hsiang Tsu did not take the rice bowl. He took out his money instead. Similar examples can be seen throughout the novel. This tendency is not limited to clauses that include a co-event verb, and indeed holds in cases where the nondeictic path element functions as the main verb too. Example (28) includes two tokens, one in the narrative, and the other in a dialogue part. In the former, the viewpoint is that of the hero (Xiangzi). English, thanks to its combination of deictic motion verbs and verb particles expressing nondeictic path, has no trouble in translating it as go/come in. French, on the other hand, can only choose between deictic and nondeictic motion, and regularly chooses the latter entrer/sortir ‘enter/exit’ throughout the novel: (29)

他 决定 进去 见 她 […] 大着 胆 叫了 声 刘 姑娘。——“进来 ！” [Rickshaw, Chapter 6] tā juédìng jìn-qu jiàn tā [...] dà-zhe dǎn jiào-le shēng Liú-gūniang. ——Jìn-lai 3SG decide enter-go see 3SG big-DUR guts call-PFV CL Liu-miss. —— enter-come

[Fr.]

…il se decida à entrer chez Tigresse,… [...] il appela : — Mademoiselle Tigresse ! — Entre. [Eng.] ...until he decided to go in and see her. [...] and then very bravely called her name. “Come in!”

Our last example illustrates the way Chinese adds deictic information in translations from English, another S-language. It is taken from Harry Potter (vol. 2 The Chamber of Secrets). Note that in this case it would be difficult for English to translate the Chinese text without omitting the deictic information.

92

(30)

C. Lamarre

…and Aunt Petunia’s hand appeared, pushing a bowl of canned soup into the room. [The Chamber of Secrets, Chapter 2, p. 22) [Ch.] 从 洞口 推 进 来 一碗 罐头 汤。 cóng dòngkǒu tuī-jin-lai yì wǎn guàntóu tāng from hole push-enter-come one bowl canned soup

[Fr.]

.... poussant à l’intérieur de la chambre un bol rempli de soupe en boîte.

These examples, together those given in Section 4, show that the overt expression of deictic direction is to some extent constrained by various language-specific factors. 6. Conclusion and Perspectives for Further Research Our analysis of Chinese and Japanese data confirmed Talmy’s and Slobin’s remarks based on other languages: the way a language encodes deictic paths cannot be totally predicted by the way it encodes nondeictic paths. However, we suggest that this slightly disappointing statement should be, to some extent, offset by the following observations: 1) The encoding of deictic motion is not totally free of the constraints brought on by the typological features of the language as an S-language or a V-language. Talmy (2000: 213–288) provided various cross-linguistic evidence to show that in an S-language, satellites function to express change of state as well. We argue that the V-language characteristics of Japanese blocked the marking of deictic viewpoint in caused motion events, and favored the semantic extension of venitive directionals into ‘subjectivity markers’ expressing affectedness (Section 4). Chinese path satellites, on the other hand, function in many ways like the postverbal devices used to express change of state and causation in non-motion events (often called ‘resultative complements’). The discrepancy between Chinese and Japanese observed here for causative motion contrasts with their similarity in encoding autonomous motion, and is adequately accounted for by Talmy’s framework. 2) On the other hand, deictic path is special: whatever the language, its special link to the linguistic expression of subjectivity motivates its different categorization. Languages which assign spatial deixis an important role in the expression of subjectivity will thus often find a way to express both types of paths, nondeictic and deictic, in the same clause. Chinese and Japanese have been characterized as being ‘lococentric’ (Paris 1992, Rygaloff 1977). Paris (1992, 1997) noted that in Chinese spatial deixis was prevalent in the deictic triad ‘I/now/here’, person being secondary . This factor obviously works at a very different level in a language from the more clearly syntactic factors related to causation and change-of-state, and we should be aware that both types of mechanisms are likely to influence the way deictic path gets categorized in a language. Some languages like French or Russian show a very low amount of interest in the linguistic expression of deictic direction, in spite of their belonging to different types (French is a V-language, Russian an S-language). We will end this short study by a last observation and suggest some directions for further research. We focused quite narrowly on Standard Mandarin deictic directionals -lai and -qu, and we saw that some Chinese dialects have developed a highly constrained bipartite category of path satellites: the overt expression of

The Linguistic Categorization of Deictic Direction in Chinese

93

nondeictic path very often entails the expression of deictic direction as well. Although cross-linguistic data (Russian, Hungarian for S-languages, Japanese and Korean for V-languages) prove that this bipartite expression of paths cannot be explained merely as the result of the typological change Chinese underwent in moving from a V-language to an S-language, we still believe that in the case of Chinese the assignment of nondeictic and deictic direction to separate slots has something to do with the development of directional satellites (directional complements). Although this claim definitely needs to be substantiated, our opinion is that the bipartite nature of Chinese directionals results from a compromise between their semantics and their syntactic position. Semantically, a deictic directional is not necessarily ‘bounding’, especially the andative item -qu. By combining with more bounding elements (nondeictic path satellites), deictic directionals are somehow ‘liberated’ from the syntactic tasks of satellites, and so play a less constrained role. List of the abbreviations used in the glosses 1SG, 2PL etc.: 1rst person singular, 2nd person plural; ACH: achievement marker; ACC: accusative marker; AGT: agentivity marker ; ALT: alternative form (Jap., marks alternating events); BEN: preposition marking benefactive; CL: classifier; DUR: Ch. durative suffix –zhe; GEN: genitive marker; IMP: imperative form; NOM: nominative marker (Jap.); NPAST: nonpast; OM: Ch. object marker bǎ; PAST: past; PFV: perfective suffix -le and sentence particle le which marks a change of state; PRE: present tense; PRT: sentence-final particle; SIM: adverb marking two simultaneous actions ; TOP: topic marker; VERBCAUS: (Jap.) the causative form of the verb. List of the abbreviations used in the text (by order of apparition) S-language: satellite-framed language; V-language: verb-framed language, PPs : Prepositional Phrases; NP: Noun Phrase; Pd : deictic path; Pnd : nondeictic path; Vco-e: co-event Verb; NPFIG: the Figure NP, i.e. the noun phrase expressing the moving entity; NPLOC : Locative or Ground NP; Ch.: Chinese; Fr. : French; Jap. : Japanese.

Notes See Chao (1968:458) for a table showing the correspondence between German separable prefixes and Chinese directional complements. The development of -zou ‘away’ as a source-oriented andative satellite can be seen as a strategy designed to solve the contradiction between the semantics of an andative marker with no perfectivizing effect on the verb it follows and the bounding function required from path satellites (path satellites function as the resultative predicate in a resultative compound, see Lamarre 2006 for more arguments). A comparison with languages using deictic satellites in languages in which motion path satellites are linked to boundedness (aspectual perfectivity), Hungarian and Russian, provide some hints about the ways in which a language can combine the role of perfectivization and the semantics of deictic direction. These two languages also use a preverb meaning ‘away’, u- in Russian and el- in Hungarian. We thank here Yulia Koloskova for sharing with us her intuitions on Russian, see also Koga et al. (2006). Chao 1968:458–467 is an exception on this matter: he treats both 掉 diào ‘fall, out, off’ and -zŏu ‘away’ as a type of directional complements which does not combine with deictic items. Mandarin Chinese bipartite path verbs like 进去 jìnqu [enter-go] ‘go in’ are usually analyzed as Verb-Result compounds of the same syntactic nature as directional compounds (formed by a co-event verb and a path directional) like 扔进去 rēngjinqu [throw-enter-go]. Both types of compounds can be split by the insertion of a negation to express impossibility,

94

C. Lamarre

ex. 进不去 jìn-bu-qù [enter-NEG-go] ‘cannot go in’ and 扔不进去 rēngbujìnqù [throwNEG -enter-go] ‘cannot throw in’. However, it must be noted that some dialects do not allow such insertions in the former case (See Tang and Lamarre, in print). Besides, verbs like 进去 jìnqu ‘go in’ do not exhibit some other features characterizing resultative compounds, for instance they can be negated by bù and not only by méi. We believe that these path verbs are better analyzed as lexicalized bimorphemic verbs rather than resultative compounds. These analytic forms are often given another treatment in many grammar books, i.e. are considered as the combination of the preposition dào and a deictic motion verb. This treatment is problematic, though: dào would be the only preposition that can combine only with the two deictic verbs ‘come’ and ‘go’. The reader may consult Xu Dan’s contribution in this volume for a further discussion of the historical evolution of Chinese deictic motion verbs. As for dialectal variation, in some northwestern dialects, zŏu ‘go away’ can be followed by a goal locative NP and as a path satellite may even combine with nondeictic path satellites. Data about - dao are not included in this table, nor in Table 4. It is very tricky to establish a distinction between the different uses of directionals. To avoid being to arbitrary, we made use of Y. Liu 1998’s criteria. She distinguishes between the spatial, fulfillment, and aspectual uses of directionals. Although her criteria are not ideal, she gives very complete lists of each use, which are easy to refer to. Note that her ‘spatial’ uses include some metaphoric uses too, that we also included, so that our definition of a motion event is quite broad (for instance we include money transactions). Table 4 shows the proportion of clauses where a deictic path is expressed when a nondeictic path is also expressed, and thus does not include constructions without nondeictic path directionals like Tā jìle yì bāo cháyè laile ‘he sent a pack of tea [hither]’ (Married ten Years). Table 4 takes account of both autonomous and caused motion events, for the 7 nondeictic core path directionals (-dao excluded). For 12 tokens out of 51, the modern translation also added a manner verb like ‘walk’ (the most frequent addition), ‘run’, and ‘flow’ before the path elements. That is, in modern Chinese a S-language or adverbal type of encoding [co-event verb + path satellite] was used to translate Classical Chinese path verbs. This reflects the shift from a V-language to a S-language that occurred in Chinese. «[Le] chinois traite la personne comme un lieu (+ humain): personne et lieu sont coextensifs. » (Paris 1992 : 174)

The Linguistic Categorization of Deictic Direction in Chinese

95

References for Corpora Luòtuo xiángzi, by Lao She (Le pousse-pousse, tr. by F. Cheng & A. Cheng, Picquier, 1990; Rickshaw, tr. by J. M. James, U. of Hawaii Press, 1979; Rakuda Shooji, tr. by T. Nakayama, Hakuteisha, 1991.) Abbreviated as [Rickshaw]. Rén dào zhōngnián [At middle age] by Chen Rong: 人到中年 (First published in 1980) Harry Potter: The Chamber of Secrets. (By J.K. Rowlings, 2000, Bloomsbury).French tr.: J.F. Ménard, (Gallimard Jeunesse 2000), Chinese tr.: Mǎ Àixīn (Hālì Pōtè yǔ mìshì , Renmin Wenxue Chubanshe, 2000). Abbreviated as [Middle age]. Shishuoxinyu: translated by Wang Jianshe, Shehui Kexue Chubanshe 2004 世说新语 选译新注 北京:社会科学文献出版社、王建设 译注 2004 Strange tales of Liaozhai : “Nie Xiaoqian” (book 1), “A laughing girl named Yingning” (book 2), and “A girl called Rouge” (book 10). The modern Chinese translation used here was published in Taipei, Guojia Chubanshe, translation by He Changren, in 2004: 聊齋誌異 白話本 蒲松齡著，臺北：國家出版社 2004, 何昌仁譯 Jiéhūn shí nián [Married ten years] (VCD, directed by Gao Xixi, Beijing Dianshi Yishu Zhongxin Chubanshe). Abbreviated as [Ten years]. We are indebted to the 21st cen. COE project ‘Evolutionary Cognitive Sciences in Komaba’ (the University of Tokyo), which funded data-gathering and coding for a database using Chinese TV drama series.

References Arakawa, K. 荒 清 秀 (1994). 買ッテクル 买来 Kattekuru to măilai [Kattekuru and măilai ]ん 愛知大学外国語研究紀要 Aichi Daigaku Gaikokugo Kenkyuu Kiyoo, 18, 71–81. Berthele, R. (2004). The typology of motion and posture verbs: a variationist account. In B. Kortmann (ed.), Dialectology meets Typology: Dialect Grammar from a CrossLinguistic Perspective (93–126). Berlin: Walter de Gruyter. Chao, Y. R. (1968). A Grammar of Spoken Chinese, Berkeley: University of California Press. Craig, C. (1993). Jakaltek directionals: their meaning and discourse function. Languages of the World, 2, 23–36. (Lincom Europa). Cuī, D. 崔 达 (2005). 中古汉语位移动词研究 Zhōnggǔ Hǎnyǔ wèiyí dòngcí yánjiū [Studies on Medieval Chinese motion verbs]. 合 肥 Hefei: 安 徽 大 学 出 版 社 Anhui University Press. Fillmore, C. (1997). Lectures on Deixis. Stanford University: CSLI Publications. François, A. (2003). Of Men, Hills, and Winds: Space Directionals in Mwotlap, Oceanic Linguistics, 42-2, 407–437. Fukami, K. 深見 兼孝 (1999). 日本語の いく くる 朝鮮語の a/ə gada a/ə oda Nihongo no teiru/tekuru to Choosengo no a/gada-a/oda [Japanese teiku-teiru and Korean a/ gada-a/ oda], 広島大学教育学部紀要 Hiroshima Daigaku Kyooikugakubu Kiyoo, [Bulletin of the Faculty of Education, Hiroshima University]. Part 2, Vol. 38, 47–52. Hóu, J. 侯 精一 et al. (2001). 中国語補語例釈 日文版 Zhōngguóyǔ bǔyǔ lìshì – Rìwénbǎn. [A compendium of Chinese complements: Japanese edition]. 北京 Beijing: 商务印书馆 Commercial Press. Hyslop, C. (2001). The Lolovoli dialect of the North-East Ambae Language, Vanuatu. Pacific Linguistics 515. Camberra: The Australian National University. Jū, H. 居 红 (1992). 汉语趋向动词及动趋短语的语义和语法特点 Hànyǔ qūxiàng dòngcí jí dòngqū duǎnyǔ de yǔyì hé yǔfǎ tèdiǎn [Semantic and Syntactic

96

C. Lamarre

Characteristics of Chinese Directional Verbs and Verb Phrases], 世界汉语教学 Shìjiè Hànyǔ Jiàoxué [Chinese Teaching in the World], 4, 276–282. Koga, T., Koloskova Y. et al. (2006). The linguistic encoding of motion events in English, Russian and German ⎯ a contrastive study with Japanese ⎯ Paper read at the 132nd meeting of the Linguistic Society of Japan, the University of Tokyo (17–18 July 2006). Kopecka, A. (2006). The semantic structure of motion verbs in French : typological perspectives. In M. Hickmann and S. Robert (eds.), Space in Languages: Linguistic Systems and Cognitive Categories (pp. 83–101). Amsterdam/Philadelphia: John Benjamins. Lamarre C. 柯理思 (2002). 汉语方言里连接趋向成分的形式 Hànyǔ fāngyánli liánjiē qūxiàng chéngfèn de xíngshì [An Investigation of the various markers inserted between verbs and directionals in Han dialects], 中国语文研究 Zhōngguó Yǔwén Yánjiū [Studies in Chinese Linguistics] 1, 26–44. (Hong Kong). Lamarre, C. 柯理思 (2003). 汉语空间位移 件的语言表达 – 兼论述趋式的几个问题 Hànyǔ Kōngjiān wèiyí shìjiàn de yǔyán biǎodá [The linguistic encoding of motion events in Chinese], 现代中国语研 Xiàndài Zhōngguóyǔ Yánjiū [Contemporary Research in Modern Chinese], 5, 1–18 (Kyoto). Lamarre, C. 柯 理思 (2005). 讨论一个非典型的述趋式 “走去”类组合 Tǎolùn yí ge fēi diǎnxíng de shùqūshì ‘zǒuqù’ lèi zǔhé [On a non-prototypical Verb-Directional construction: zŏuqù and similar compounds]. In J. Sh０n, Wú and Hóng (eds.), 语 法 语法研究( ) (pp. 53–68). 商务印书馆. Lamarre, C. (2006). 动词后置成分 走 的语法 Dòngcí hòuzhì chéngfèn zǒu de yǔfǎhuà [The grammaticalization of postverbal directional –zou ‘away’]. In N. Ogoshi (ed.), Contrastive Studies of Grammaticalization and Categorization in East Asian Languages— from particulars to universals, Project Report for Grant-in aid for scientific research n. 14310221 (2002-05), (pp. 104–116). University of Tokyo, Japan. Lamarre, C. (2007a). The Linguistic Encoding of Motion Events in Chinese: With Reference to Cross-dialectal Variation. In C. Lamarre and T. Ohori (eds.), Typological Studies of the Linguistic Expression of Motion Events, Volume 1: Perspectives from East and Southeast Asia (pp. 3–33). Tokyo: Center for Evolutionary Cognitive Sciences at the University of Tokyo (21st century COE Program). Lamarre, C. (2007b). 试探北方官话的指示位移动词 上/到+处所词+来/去 格式 Shìtàn Běifāng Guānhuà de zhǐshì wèiyí dòngcí yǔ “shàng/dào + chùsuǒcí + lai/qu” géshì [Northern Chinese synthetic and analytic deictic motion verbs]. Paper read at the 4th Conference on Mandarin Dialects, Ankang (Shaanxi), Oct. 2007. Lamarre, C., & Liu, S. 柯 理思、刘 淑学. (2001). 河北冀 方言 拿 了走 一类的 格式 Héběi Jìzhōu fāngyán nábuliǎozǒu yílèi de géshì [Patterns like nábuliǎozǒu in the Jizhou dialect of Hebei]. 中国语文 Zhōngguó Yǔwén, 5, 428–438. Liu, D. 劉 丹青 (2000). 粵語句法的類型學特點 (Yuèyǔ jùfǎ de lèixíngxué tèdiǎn) [The typological characteristics of Cantonese Syntax]. 亞太語文教育學報 (Yà-Tài yǔwén jiàoyù Xuébào) [Asia Pacific Journal of Language in Education], 3-2, 1–29 (Hong Kong). Liu, Y. (ed.) 刘月华 主编 (1998). 趋向补语通释 Qūxiàng bǔyǔ tōngshì [Directional complements: a compendium]. 北京 Beijing: 北京语言文 大学出版社 Běijīng Yǔyán Wénhuà Dàxué.Matsumoto, Y. (1996). Complex Predicates in Japanese:

The Linguistic Categorization of Deictic Direction in Chinese

97

A Syntactic and Semantic Study of the Notion ‘Word’. Stanford: CSLI Publications and Tokyo: Kurosio Publishers. Matsumoto, Y. (2003). Typologies of lexicalization patterns and event integration: Clarifications and reformulations. In S. Chiba et al. (eds.), Empirical and theoretical Investigations into Language: A Festschrift for Masaru Kajita (pp. 403–418). Tokyo: Kaitakusha. Paris, M.-C. (1992). Démonstratifs et personne en chinois standard. In M-A. Morel and L. Danon-Boileau (eds.), La deixis (pp. 166–175). Paris: Presses Universitaires de France. Paris, M.-C. (1997). L’espace énonciatif en chinois moderne: mais où la triade est-elle passée? In C. Fuchs and S. Robert (eds.), Diversité des langues et représentations cognitives (pp. 93-105). Ophrys. (English translation : John Benjamins 1999) Peyraube, Alain. (2006). Motion Events in Chinese: A diachronic study of directional complements. In Hickmann and Robert (eds.), Space in Languages : Linguistic Systems and Cognitive Categories (pp. 121–135). Amsterdam/Philadelphia: John Benjamins. Qi, H. 齐沪扬. (1998). 现代汉语空间问题研究 Xiàndài Hànyǔ kōngjiān wèntí yánjiū [Studies on spatial issues in Modern Chinese], 上海 Shanghai : 学林出版社 Xuélín Chūbǎnshè. Ross, M. (2004). The Grammaticization of directional verbs in Oceanic languages. In I. Bril and F. Ozanne-Rivierre (eds.), Complex Predicates in Oceanic Languages — Studies in the Dynamics of Binding and Boundness (pp. 297–329). Berlin: Mouton de Gruyter. Rygaloff, A. (1977). Existence, possession, présence (être et avoir), Cahiers de Linguistique Asie Orientale, 1, 7–16. Shibatani, M. (2003). Directional Verbs in Japanese. In E. Shay & U. Seibert (eds.), Motion, direction and location in languages (pp. 258–86). Amsterdam: John Benjamins. Slobin, D. (1997). The Universal, the Typological, and the Particular in Acquisition. In D. Slobin (ed.) The Crosslinguistic Study of Language Acquisition. Vol. 5: Expanding the Contexts (pp. 1–39). Mahwah / London: Lawrence Erlbaum. Slobin, D. (2004). The many ways to search for a frog: Linguistic typology and the expression of motion events. In Strömqvist & Verhoeven (eds.) Relating events in narrative: Vol. 2. Typological and contextual perspectives (pp. 219–257). London: Lawrence Erlbaum. Talmy, L. (1985). Lexicalization patterns: Semantic structure in lexical form. In T. Shopen (ed.) Language typology and syntactic description, Vol. III: Grammatical categories and the lexicon (pp. 57–149). Cambridge: Cambridge University Press. Talmy, L. (2000). Toward a Cognitive Semantics – Vol. II: Typology and process in Concept Structuring. Cambridge (Ma) : MIT Press. Tanaka, S. & Matsumoto, Y. (1997). 空間 移動の表現 Kuukan to Idoo no Hyoogen [Linguistic Expressions of Space and Motion]. 東京 Tokyo: 研究社 Kenkyusha. Tang, Zhengda. & Lamarre, C. (in print). A contrastive study of the linguistic encoding of motion events in Standard Chinese and in the Guanzhong dialect of Mandarin (Shaanxi) , Bulletin of Chinese Linguistics, 2.1. Uehara, S. (2006). Toward a typology of linguistic subjectivity: A cognitive and crosslinguistic approach to grammaticalized deixis. In A. Athanasiadou et al. (eds.), Subjectification: Various paths to subjectivity (pp. 75–117). Berlin: Mouton de Gruyter. Wilkins, D., & Hill, D. (1995). When “go” means “come”: Questioning the basicness of basic motion verbs. Cognitive Linguistics, 6-2/3, 209–259. Xu, Dan. (2006). Typological change in Chinese Syntax. Oxford: Oxford University Press.

THE LOCATIVE WORDS IN THE WAXIANG DIALECT SPOKEN IN GUZHANG, HUNAN1 Yunji Wu Asia Institute The University of Melbourne Keywords: Locative words, frames of reference (FoR), the Waxiang dialect, Guzhang, western Hunan Abstract: There has been a great deal of thought about spatial frames of reference (FoR) in psychology, neurocognition, linguistics and elsewhere (Asifa Majid et al., 2004 : 108). For example, Levinson examined the literature of the FoR (2003, Chapter 1) and claimed that in languages, there are three frames of references. They are absolute, intrinsic and relative (2003: 53). However, this type of study has not attracted enough attention to Chinese Linguistics although some studies on space on Chinese have been carried out (e.g., McCullough, 1993 and Peyraube, 2006). Which FoR or FoRs Chinese belong to is a big topic since there are many varieties of Chinese. This paper is an initial study on the expression of locative words2 of one of the varieties of China: the Waxiang dialect.

The Waxiang dialect is a dying language spoken by possibly 300,000 people (according to local estimates) in western Hunan province, China. This paper does not intent to identify which FoR(s) the Waxiang expressions belong to, but intent to provide detailed data for further study on FoR. It will describe the expressions of the locative words from different views, e.g., from language, from the environment and culture point of views. It will show that the locative words in the Waxiang dialect are much complex than either Mandarin or the Xiang dialect group, the main dialects spoken in Hunan. The following six subtitles will be found in this paper: (1) (2) (3) (4) (5) (6)

The expression of the four directions The expression of ‘top’ and ‘bottom’ The expression of ‘inside’ and ‘outside’ The expression of ‘front’, ‘back’, ‘left’ and ‘right’ The expression of distances Summary.

1. The Expression of the Four Directions There are words for directions in the Waxiang dialect corresponding with Mandarin such as 东[tau55] ‘east’, 南[noN55] ‘south’, 西[˛i55] ‘west’ and 北[pF25] ‘north’. However they are not commonly used to indicate directions and location in everyday life. Waxiang-speaking communities are located in the mountainous areas. The villages are usually along rivers. Therefore, when directions or locations are mentioned, mountains and rivers are used as the references. The mountains are treated 99 D. Xu (ed.), Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives, 99–117. © Springer Science +Business Media B.V. 2008

100

Y. Wu

as latitude and the rivers as longitude. Latitude is divided into 高头[kau55 ta] [top + head] ‘top; north’, 当中[təN55tiau55] [to be + middle] ‘middle’ and ∂ 3 头 [tə 41ta] [? + head] ‘bottom; south’. Longitude is divided into ∂头[tsI25ta] [? + head] ‘upper reaches; west’ 当中[təN55tiau55] [to be – middle] ‘middle’ and ∂头[o25ta] [? + head] ‘lower reaches; east’. The following illustrates how the system of directions is expressed in the Waxiang-speaking area. The Zhang family, which has three siblings, lives on the top of the mountain and the Wang family lives below them. The river runs from west to east, as shown below: Zhang Elder brother

Zhang Second brother

Zhang Younger brother

Upper reaches Wang family

Fig. 1.1

Lower reaches

The following expressions can be heard: (1)

张家坐到高头那一村。 tiN55 koN55 t˛iE13 tau25 kau55 ta13 oN25 i33 t˛Hyai55 Zhang family sit COMP4 top that CL house ‘The Zhang family is located at the top of the mountain.’

(2)

张家大哥哥坐到∂头那条室。 tiN55 koN55 lu13ku55ku55 t˛iE25 tau25 tsI25 ta oN25 tiau25 t˛i41 Zhang family elder brother to sit COMP upper reaches that CL house ‘The eldest brother’s house is located to the left.’

(3)

渠∂哥哥坐到当中那条室。 zF13 ˜iaN13 ku55ku55 t˛iE25 tau25 təN55tiau55 oN25 tiau25 t˛i41 3nd Sg5 small brother to sit COMP middle that CL house ‘The second brother’s house is located in the middle.’

The Waxiang Dialect Spoken in Guzhang, Hunan

(4)

渠 哥哥坐到∂头那条室。 zF13 so55 ku55ku55 t˛iE25 tau25 o25 ta oN25 tiau25 t˛i41 3nd third brother to sit COMP lower reaches that CL house ‘The youngest brother’s house is located to the right.’

(5)

王家坐到张家的∂头的。 ƒ´N13 koN55 t˛iE25 tau25 tiN55koN55 o25 ta33 ti Wang family to sit COMP Zhang family lower reaches MOD ‘The Wang family is located below the Zhang family.’

101

In Guzhang city, the river runs from south to north. If two families are located as per the map below, the Wang family can say: 张家坐到我头的: [tiN55 koN55 t˛iE13 ∂ tau25 u25 o25 ta33 ti] “The Zhang family is located down from our house” and the Zhang family can say: 王家座到我∂头的[´N13 koN55 t˛iE25 tau25 u25 tsI25 ta ti] “The Wang family is located up from our house’.

Lower reaches Zhang family

Wang family Upper reaches

Fig. 1.2 However, in Mandarin, as well as in the Xiang dialects, if two families are located as per map above, the Wang family will say: 张家住在我家的北面/ 面： zhāng jiā zhù zài wǒ jiā de běi miàn / shàng miàn. “The Zhang family is located north to my house/up from our house” and the Zhang family will say: 王家住在我 家的南面/ 面 wáng jiā zhù zài wǒ jiā de nán miàn/xià miàn. “The Wang family is located south to my house/down from our house”. [o25 ta] and [tsI25 ta] can also be used to indicate the locations of objects. People will judge whether an object is [o25 ta] ‘at bottom’ or at [tsI25 ta] ‘at top’ depending on the course of the river. 2. The Expression of ‘Top’ and ‘Bottom’ Three words can be used to express ‘top’. They are: 头 [ta13] [head], 高头 55 [kau ta] [high – head], and 皮头[fF13 ta] [skin – head]. Two words can be used to describe ‘bottom’. They are: ∂头 [tə 41ta] [? + head] and ∂头[ku41ta] [? + head].

Y. Wu

102

2.1. Words for ‘top’ Table 2.1 below summarizes the distribution of the three words which combine with some commonly used nouns. The symbol ‘+’ indicates that the word can be combined with a noun, while ‘–’ means that the word cannot be combined with a noun. All the corresponding words in Mandarin can combine with the word shàng, although the word shàng ‘top’ may have different indications. For example, 桌子 zhuō zi shàng refers to the top of a desk while 手巾 shǒu jīn shàng refers to anywhere of a towel. The situation is more complex in the Waxiang dialect. There are three words which can means ‘top’. The table shows that there are different patterns of the combinations when the three combine with different nouns.

Table 2.1. Nouns plus words for ‘top’

台 ‘table’ 子 ‘bench’ 床板 ‘bed’ 车 ‘car’ 桥 ‘bridge’ 盖 ‘wok’ 花板 ‘ceiling’ 地 ‘earth’ 水 ‘water’ 口 ‘mouth’ 肚子 ‘belly’ 楼 ‘building’ 山 ‘mountain’ 墙 ‘wall’ 门 ‘door’ 钩子 ‘hook’ 脸 ‘face’ 树 ‘tree’ ‘sky’ 屁股 ‘bottom’ 背 ‘back’ 身 ‘body’ 边 ‘side’ 柜子 ‘cupboard’ 瓶子 ‘bottle’

13

ta paN25tsI

kHuai33tsən13 tsHo55 t˛iau13 tsçN55 kua33 tHai55xua55pən25

tHF25 tsu25 kHa25 təu25 la55 sai55 t˛iE)13 mai55 ka55t˛i miE33pau55 tsa33 tHai55 pHi55ku33 pF33 sai55 piE33 kuei25 pin13tsa

ta13 + + + + + + + + + + + + + + + + + + + + + + + – –

kau55ta + + + + + + + + + + + + + + + + + + + + + – – + +

fF13ta + + + + + + + + + + + + – – – – – – – – – – – + +

(Continued)

The Waxiang Dialect Spoken in Guzhang, Hunan

灶 ‘oven’ 碗 ‘bowl’ 灯 ‘lamp’ 头发 ‘hair’ 手 ‘hand’ 手巾 ‘towel’ 窗户 ‘window’ 钉子 ‘nail’ 书 ‘book’ 信 ‘letter’ 纸 ‘paper’ 梳子 ‘comb’ 耳朵 ‘ear’ 鼻子 ‘nose’  ‘foot’ 路 ‘road’ 街 ‘street’

tsau33 əN25 taN55 lau25 mau55

˛i´ 25 po55tsa

lin13 mai55tsa toN55tsa

t˛iə 55 ˛iE33 tsF25 sF55 tsa ˜iaN25tuo41 pi55kaN55

ku41 sau13 ka55

– – – – – – – – – – – – – – – – –

+ + + + + + + + + + + + + + + – –

103

+ + + + – – – – – – – – – – – – –

All of the three words indicate the upper side of an object but 皮头 [fF 13 ta13] emphasizes the surface of the top. 高头 [kau55 ta13] is more flexible and can be combined with most objects, while 头[ta13] seems to have more limitations. The following examples present the difference between 皮 头 [fF13ta] and 高 头 [kau55ta]/[ta13], although in some combinations the words are interchangeable. (6)

(7)

车高头放得有一筐萝卜。 tsHo55 kau55ta fəN41 tF41 va25 i13 lu55 su55 la55 pHF vehicle top to put COMP to have one CL radishes ‘There is a basket of radishes in the car.’

车皮头放得有一筐萝卜。 tsHo55 fF13ta fəN41 tF41 va25 i13 lu55su55 la55 pHF vehicle top to put COMP to have one CL radishes ‘There is a basket of radishes on the top of the car.’

If there is a basket of radishes in the vehicle, it is more common to use the word ‘inside’ than the word [kau55ta] ‘on the top’. (8)

棉花衣高头有把眼。 pa41 Nai25 miE)55 xua55 i55 kau55ta va25 coat top to have CL hole ‘There is a hole in the coat.’

104

(9)

Y. Wu

棉花衣皮头有把眼。 pa41 Nai25 miE)55 xua55 i55 fF13ta va25 coat top to have CL hole ‘There is a hole in the coat.’

Example (8) refers to the hole found anywhere on the coat while (9) refers the hole outside of the coat only. (10) 灯高头有把。 tən25 kau55ta va25 pa41 mu41ti lamp top to have CL fly ‘There is a fly on the lamp.’

(11)

灯皮头有把。 tən25 fF13ta va25 pa41 mu41ti lamp top to have CL fly ‘There is a fly on the lamp.’

Example (10) refers the fly anywhere on the bulb and lampshade and (11) refers the fly on the top of the bulb or lampshade. (12)

(13)

嘴高头有把泡。 kHa25 kau55ta va25 pa41 pau55 mouth top to have CL blister ‘There is a blister on my lip (anywhere on the lips).’

嘴皮头有把泡。 kHa25 fF13 ta va25 pa41 pau55 mouth top to have CL blister ‘There is a blister on my lip (on the upper lip only).’

Example (12) refers a blister anywhere on the lips while (13) refers a blister on the upper lip only. (14)

(15)

8

楼高头 人。 la55 kau55ta pa13 əN13 floor top no person ‘There is no one on the second floor.’

8

楼皮头 人。 la55 fF13 ta pa13 əN13 floor top no person ‘There is no one on the roof (or the top floor).’

The Waxiang Dialect Spoken in Guzhang, Hunan

(16)

(17)

105

那条碗高头雕得有花。 oN33 tiau13 əN25 kau55ta tiau55 tF41 va25 xua55 that CL bowl top to curve COMP to have flower ‘There are curved flowers on the bowl (either inside or outside the bowl).’

那条碗皮头雕得有花。 oN33 tiau13 əN25 fF13ta tiau55 tF41 va25 xua55 that CL bowl top to curve COMP to have flower ‘There are curved flowers on the bowl (outside of the bowl only).’

Some nouns are not usually combined with fF13 ta, for example ‘hat’, although this is possible in some contexts:

(18)

帽皮头 加把帽。 mau25 fF13 ta xai13 ko55 pa41 mau25 hat top still to add CL hat ‘There is a hat on the top of that hat.’

The above distinction between 皮头[fF13ta] and 高头[kau55ta]/[ta13] has not found in either Mandarin or other dialects spoken in Hunan. Further research needs to be carried out on the words 高头 [kau55 ta13] and 头[ta13]. They may stand in contrast to each other in some way. Please keep in mind that some nouns such as ‘road’ and ‘street’ cannot combine with ‘up’ or ‘top’ like Mandarin does. We will found that these two combine with the words ‘inside’ and ‘outside’ due to the environment situation. 2.2. The expression of ‘bottom’ Both 头 [tə 41 ta] and 头 [ku41 ta] can express ‘down’ in the Waxiang dialect.

Table 2.2: Nouns plus words ‘bottom’

台 ‘table’ 子 ‘bench’ 车 ‘car’ 门 ‘door’ 桥 ‘bridge’ 楼 ‘floor’

13

ta poN55 tsI

tsHo33 mai55 t˛iau13 la55

tə + + + + + +

41

ta

ku41ta + + + + + +

(Continued)

Y. Wu

106

床 ‘bed’ 碗 ‘bowl’ 缸缸 ‘cup’ 灯 ‘lamp’ 木 ‘tree’ 阳 ‘sun’ 鼻公 ‘nose’ 耳朵 ‘ear’ 嘴 ‘mouth’ 眼珠 ‘eye’ 山 ‘mountain’  ‘foot’ 水 ‘water’

kHuai25 dz´n13 ´N25 kaN55 kaN taN55 tsa33 əN33 ta pi55 kaN55 ˜iaN25tuo41 kHai25 Nai25t˛y55 sai55 ku25 tsu25

+ + + + + + + + + + + + +

+ + + + + + + + + + + – –

Except for the last two, all of the nouns can combine with both ∂ 头 [tə 41 ta] and  头 [ku41 ta] . General speaking, they are interchangeable. However, since [ku41ta] means ‘foot’ in the Waxiang dialect, people may make a distinction when a specificlocation needs to be pointed out. (19)

笔在台头。 pa41 tsHF33 ta13 tə 41ta pen to be at table down ‘The pen is under the table (anywhere under the table).’

(20)

笔在台头。 pa41 tsHF33 ta13 ku41ta pen to be at table down ‘The pen is under the table (only near the legs of the table).’

3. The Expression of ‘Inside’ and ‘Outside’ There are three words for the expression ‘inside’. They are:  [la55] [?], [lin41ta] ∑头 [? + head] and  [nuei13la55] [? + ?]. There is only one word for the expression ‘outside’: 尾 [t´n41 mai25 ] [？ + tail]. 3.1. The expression of ‘inside’ The three words for inside, [la55], [lin41ta] and [nuei13la55], are interchangeable at times and not at others. See the following combinations:

The Waxiang Dialect Spoken in Guzhang, Hunan

107

Table 1. 3. Nouns plus words ‘inside’ 河 ‘river’ 箱子 ‘box’ 手 ‘hand’ 口 ‘month’ 口袋 ‘bag’ 缸缸 ‘cup’   ‘basket’ 鼻公 ‘nose’ 耳朵 ‘ear’ 肚子 ‘belly’ 碗 ‘bowl’ 柜子 ‘cupboard’ 路 ‘road’ 村 ‘village’ 城 ‘city’ 镇 ‘town’ 街 ‘street’ 心 ‘heart’ 履 ‘shoe’ 衣 ‘clothes’ 房间 ‘room’ 被絮 ‘quilt’

55

kHa ˛in55tsa ˛i´ 25 kHa25 taN55 kaN55 kaN pHau55 tsaN13 pi55 kaN55 ˜iaN25tuo41 təu25 ´N25 kuei25 sau55 t˛Hyai55 tsI13 tsaN55 ka55 ˛ie55 li25 i55 t˛i25 fa13 su55

la55 + + + + + + + + + + + + + + + + + + – – – –

lin41ta + + + + + + + + + + + + + + – – – – + + + +

nuei13la55 + + + + + + + + + + + + – – – – – – + + + +

It can be seen from the table that there are different patterns of combinations. However, most nouns can combine with all three words. Sometimes the three words are interchangeable and at others they are not. See the examples below: (21)

(22)

他手里有一分铜。 zF25 ˛i´ 25 la55 va25 i13 f´55 dau13 3nd Sg hand inside to have one CL coin ‘There is a one cense coin in his hand (held by his two fingers and visible).’

他手里有一分铜。 i13 f´55 dau13 zF25 ˛i´ 25 lin41la55 /nuei13la55 va25 3nd Sg hand inside to have one CL coin ‘There is one cense coin in his hand (held in his hand and not visible).’

108

(23)

(24)

Y. Wu

他口里有块糖。 zF25 kHai25 la55 va25 kHua13 l´N13 3nd Sg month inside to have CL sweet ‘There is a sweet in his mouth (visible).’ 他口里有块糖。 zF25 kHai25 lin41la55 /nuei13la55 va25 kHua13 l´N13 3nd Sg month inside to have CL sweet ‘There is a sweet in his mouth (invisible).’

Only the above examples which have contrasts between [la55], [lin41ta] and [nuei13la55] are found so far and more research needs to be carried out on the differences of these three words. 3.2. The expression of ‘outside’ There is only one word for outside. It is [t´n41 mai25]. Here are some examples: (25)

你跟箱子罩一块布。 ˜i25 kai55 ˛in55tsa tən41mai25 tsau33 kHua13 pu22. 2nd SG disposal M box outside to cover CL clothes ‘Cover the box with clothes.’

(26)

门有人。 mai55 təu41mai25 va25 ´N55 door outside to have person ‘There is someone outside.’

(27)

缸缸不干净。 kaN55 kaN tən41mai55 pa13 kH´N55 t˛Hi55 cap outside not clean ‘The outside of the cup is not clean.’

It is mentioned in the above section, the words ‘road’ or ‘street’ cannot combine with the word ‘top’ as Mandarin does, but it can combine with the words ‘inside’ and ‘outside’. In the Waxiang speaking area, you can often hear parents warn their children when they go out: (28)

行要行头边，莫行走外面尾边。 ƒçN13 sau25 iau33 ƒçN13 lin41ta13 pHi55，mo25 ƒçN13 tən41mai25 pHi55. To walk road to want to walk inside side don’t walk outside side ‘(You) should walk along inside of the road not outside of the road.’

The Waxiang Dialect Spoken in Guzhang, Hunan

109

As mentioned before, Guzhang is located in the mountainous areas and the common situation is that one side of the road is against the mountain and the other side is beside abyss. [lin41ta13] ‘inside (of the road)’ indicates the side against the mountain while [tən41mai25] ‘outside (of the road)’ indicates the side beside to abyss.

4. The Expression of ‘Front’, ‘Back’, ‘Left’ , ‘Right’ and ‘Side’ The word for ‘front’ is 头 [ta13pi25], for ‘back’ 尾头[mai25ta], for ‘left’左边 [tsu pHi55], for ‘right’ is 右边[za55pHi55]; for ‘side’ the words are 边里[pHiE55la]/边头 [pHiE55ta]. The usages for the first four are straightforward but some discussion need to be given to the last two: 边 [pHiE55la] and 边头[pHiE55ta]. 边 [pHiE55la] and 边头[pHiE55ta] are sometimes interchangeable; at other times there is a distinction between them. 边头[pHiE55ta] refers to the side closest the speaker and 边 [pHiE55la] refers to the side away from the speaker. For example, if someone comes back from outside with wet shoes, he will say that he went to the 溪 边头 [kHa55pHiE55ta] ‘bank of the stream’. If he went for a walk along the bank, he will say that he went to the 溪边[kHa55pHiE55la] ‘along the stream’. If you ask someone to bring a chair close to you, you will use 边头[pHiE55ta], while if you ask someone to put a chair by the door, for example, then 边[pHiE55la] will be used (see examples 29–30). 55

(29)

你跟子放到我边头来。 ˜i25 kai55 pəN55tsI fəN33 tau u25 pHiE55ta zF13 2nd Sg disposal M bench to put to 1st Sg side to come ‘Take the bench beside me.’

(30)

你跟子放到那边去。 ˜i25 kai55 pəN55tsI fəN33 tau çN33 pHiE55la kHF25 2nd Sg disposal M bench to put to that side to go ‘Take the bench the side.’

5. The Expression of the Distances In Chinese there is a distinction between the two-way system and the three-way system for using demonstrative pronouns to indicate distances. Most dialects such as Mandarin has a two-way system in which zhè ‘here’ is distinguished from 那 nà ‘there’. Some dialects have a three-way system (Ogawa, Kanju, 1981; Lü and Jiang, 1985 and 1990; Bo, Hong 1999; Yunji Wu, 2002 among others). The Waxiang dialect is one of those that have a three-way (or even four-way) system. It has the following two sets of terms:

110

Y. Wu

This here

This there

That here

That there

ai33

oN25

oN25pHi33

oN25 oN25pHi33

ai33

oN25

oN25

oN25 oN25pHi33

In the second set, the [oN25] for ‘that’ needs to be stressed when it is pronounced. (31)

是我的缸缸。 ai33 tsHF33 u25 ti kaN55 kaN. this to be 1st sg SP cup ‘This is my cup.’

(32)

是余哥哥的缸缸。

oN25 tsHF33 a13 ku55 ku ti kaN55kaN that to be 1st sg elder brother SP cup ‘That is my elder brother’s cup.’

(33)

是余的缸缸。 oN25pHi33/oN25 stressed tsHF33 a13 xa25 zi25 ti kaN55kaN that to be 1st sg younger brother SP cup ‘That (further) is my younger brother’s cup.’

(34)

是余妹妹的缸缸。 oN25 oN25pHi33 tsHF33 a13 mei25mei ti kaN55 kaN that to be 1st sg younger sister SP cup ‘That (further) is my younger brother’s cup.’

If there are three objects, [ai33/oN25/oN25pHi33] is used. But if something else stands between the second and the third objects, [ai33/oN25/oN25 oN25pHi33] is used, as shown below: (35)

是我的室。

ai33 tsHF33 u25 ti55 t˛i41 this to be 1st sg SP house ‘This is my house.’

(36)

是余哥哥的室。 oN25 tsHF33 a13 ku55 ku ti55 t˛i41 this to be 1st sg elder brother SP house ‘ That is my elder brother’s house.’

The Waxiang Dialect Spoken in Guzhang, Hunan

(37)

111

是余的室。 oN25 oN25 pHi33 tsHF33 a13 xa25 zi25 ti55 t˛i41 this to be 1st sg elder brother SP house ‘That is my younger brother’s house.’

In the above section, the words for the different directions and locations in the Waxiang dialect have been introduced. In the following section, I will provide two cases when local people talk about directions and/or locations. Case 1: I drew two maps for my informant and asked him to tell me the directions for the No.1 high school and the Wang family. The No.1 high school is assumed to be in the county town and the Wang family to be located at the top of a mountain. Below are the maps and his descriptions. No.1 high school | | | | | | _________| ___________|tailor’s shop | | | | The passer-by and the speaker

Fig. 5.1 Passer-by: Excuse me, where is the No.1 high school? Speaker: Go ahead. Turn right when you see the intersection. Turn left when you see a tailor’s shop. Then you will be there (then you can see the school). (38)

你往头 行，到十字路口往 边行，到裁缝铺 里再往左边行，就到 边的。 ˜i25 uan41 ta13pi25 ƒəN13，tau25 sI13 tsI25 lu25 kHəu41 uan41 2nd Sg toward in front to walk to arrive intersection toward

za55pHi55 ƒçN13，tau33 zF13pau55pu33 çN33 li，tsai25 uan41 tsu55 right side to walk to arrive tailor’s shop there then towards left

pHi55 ƒçN13，t˛iə 25 tau33 piE55 ti. side to walk then to arrive the place ASP

Y. Wu

112

Wang family | | |________ | | | | The passer-by and the speaker

Fig. 5.2 Passer-by: Where does the Wang family live? Speaker: You follow this direction and go up. Don’t turn right when you see a branch road. Go up and you will arrive at the Wang family’s house. (39)

你就跟  行，乘高 ，到高头分 那里，莫走 手那边，就往高头 行，就到王家室 的。 ˜i25 t˛iə 25 kai13 ai55 li ƒçN13， tsHən25 kau55 kHF55，tau33 kau55 ta 2nd Sg then follow this side to walk to go top to go to arrive top

fai55sau25 çN25li，mo25 ƒçN13 za55 ˛iə 25 çN25li，t˛iə 25 ƒçN13 branch road there don’t to walk right hand there then to walk

uan41 kau55ta ƒçN13，t˛iə 25 tau33 əN13kçN55 t˛i25 ta13 ti. toward top to walk then to arrive Wang family house top ASP Case 2: In Guzhang, the direction in which a house faces depends on the way the mountains run. No matter which direction a house faces, people consider that the right-hand

困房[kHuai41fəN13] Addition bedroom

倒房[tau33 fəN13] Addition bedroom

困房[kHuai41fəN13] Main bedroom

困 房 / 床 板 通室[toN55t˛i41] 床板[dzən13pən25] 41 13 13 25 [kHuai fəN ]/[dzən pən ] Room for family activities Fireplace Addition bedroom or fireplace

Fig. 5.3

The Waxiang Dialect Spoken in Guzhang, Hunan

113

side when you are facing the house is east. People believe east is venerable. Therefore the main bedroom and the fireplace are located on the east side of a house. Here is the common plan of a Waxiang house: The size of the house depends on the number in the family and the financial situation. The simplest house will have a middle and a right-hand side only. Some houses also have two storerooms. There are no fixed places for toilet and kitchen. They are normally situated outside the house. In some houses, the kitchen is in the room with the fireplace. The following conversation regarding locations occurred between my informant and me in January, 2005: (40)

火床一般在什么位置

？Where is the fireplace located?

火床安在东方那间房的左边。 dz´n13 p´n25 oN55 tau25 tau55 oN25 kHau55 t˛i41 tsu55pHi55 fire place to locate to East side left side ‘The fireplace is located on the left side of the room in the east.’

(41)

老人坐在什么地方？Where do the old people sit? 老人坐到当头边。 laN25 ko55 t˛iE13 tau25 t´n55 ta25 pHi55 Old people to sit to top nearer side ‘Old people sit on top nearer to the fire place.’

(42)

客人坐在什么地方？Where do the guests sit? 客人坐到尾头两边。 kHo41´N55 t˛iE13 tau25 t´ 41 mai33 iau55 lin41 ta tso25 pHi55 guests to sit on to and two side ‘The guests sit on either side (of the fire place).’

(43)

小孩坐在什么地方？Where do the children sit? 岁数轻的坐到头。 tsua33 sa33 t˛Hi55 ti t˛iE13 tau25 fu13li55tən41ta. age young SP to sit in front of the fireplace. ‘The kids sit right in front of the fireplace.’

Y. Wu

114

(44)

在卧室里，床安在什么地方？‘Where is the bed placed in the bedroom?’ 困床开到房 要顺到室开。 kHuai25 tsən13 kHa55 tau25 fəN13 la iau33 yE33 tau25 t˛i41 kHa55 bedroom to locate to room to need to to follow to the house to locate ‘The bed has to follow the direction of the house.’

(45)

妻睡觉的方向和位置是怎么样的？‘How do the couple sleep in bed?’

哥要困到 边，姑娘哥要困到头边。 a13 səN33 ko55 iau33 kHuai33 tau25 t´n41mai33 pHi55，ku55 ˜iaN13 ko55-33 husband need to sleep to outside wife iau33 kHuai33 lin41 ta41pHi55. need to sleep inside ‘The husband sleeps on the outside of the bed and the wife sleeps on the inside of the bed (in earlier days, husband and wife slept on different sides of the bed).’ 6. Summary In summary, at least twenty-three words can be used for directions in the Waxiang dialect. They are: 东 南 西 北

  边 边 头 高头 皮头 头 头

 头  尾

[tau55] [noN55] [˛i55] [pF25] [ai33] [oN25] [oN25pHi55-33] [oN25 oN25pHi33] [ta13] [kau55 ta] [fF13 ta] [tə 41ta] [ku41ta] [la55] [lin41ta] [nuei13la55] [t´n41 mai25]

‘east’ ‘south’ ‘west’ ‘north’ ‘here’ ‘there’ ‘there (further)’ ‘there (further) ’ ‘top’ ‘top’ ‘top’ ‘bottom’ ‘bottom’ ‘inside’ ‘inside’ ‘inside’ ‘outside’

The Waxiang Dialect Spoken in Guzhang, Hunan

头 尾头 左边 右边 边头

[ta13pi25] [mai25ta] [tsu55pHi55] [za55pHi55] [pHiE55ta]

边

[pHiE55la]

115

‘front’ ‘back’ ‘left’ ‘right’ ‘side’ near the speaker orthe object ‘side’ (away from the speaker or the object).

The two main distinguishing features of the expressions for directions and location in comparision with Mandarin and the Xiang dialects, in the Waxiang dialect are as follows: i) ii)

Mountains and rivers are used as references in referring to directions or locations. There are more distinctions among objects than in Mandarin and the Xiang dialects. There is a three-way, even four-way system of the demonstrative pronouns. There are three words for ‘up’, and the expression has at least ‘top’ and ‘above top/higher than top’. There is a distinction for the word ‘side’: [pHiE 55ta] refers to a place close to the referred object or person, while [pHiE 55la] refers to a place further to the speaker.

It must be pointed out that among the twenty-three words, there are about a third words which lexical sources are unknown. It will bring difficulty when the FoR of the Waxiang dialects to be identified. Must more study need to be carried out in this area. Abbreviations ASP : COMP : CL : Sg : MOD : SP : M:

Aspectual marker complement classifier singular modal particle structural particle marker

116

Y. Wu

Notes 1

This project is supported by University of Melbourne and CRLAO, Paris. I would like to thank Prof. Dan Xu and Prof. Christine Lamarre for their helpful comments on this paper. This paper is based on my fieldwork in Guzhang in January and September in 2005. My informant is Mr. Guangxun Xiang who was born in 1935 and had four year’s education. 2 Locative words in this paper refer to the nouns for directions, such as south or north, nouns for locations such as in front or back, right or left, as well as demonstrative pronouns as such there and here. 3 A box  stands for a character which lexicon source is unknown. 4 Abbreviations can be found at the end of this paper.

The Waxiang Dialect Spoken in Guzhang, Hunan

117

References: Bao, Houxing, & Wu Yunji. (1985). Yuánlíng Xiānghuà jìlüè [A survey of the Waxiang dialect spoken in Yuanling]. Journal of hunan normal university, 40–60. Heine, Bernd & Tania Kuteva. (2002). Word lexicon of grammaticalization. London: Cambridge University Press. Hong, Bo. (1999). Jiānguŏjí-hàntáiyŭ zhuīzhǐ [Essays on the Chinese and Thai Languages]. TianJing: Nankai Daxue Chubanshe. Levinson, S.C. (2003). Space in language and cognition: explorations in cognitive diversity. London: Cambridge University Press. Lü, Shuxiang, & Jiang Lansheng. (1985). Jìndài hànyŭ zhishi dàicí [The Demonstrative Pronouns in Early Modern Chinese]. Shanghai: Xuelin Chubanshe. Lü, Shuxiang, & Jiang Lansheng. (1990). Zhǐshì dàicí de èrfēnfă hé sānfēnfă [Two-way and three-way demonstrative pronouns]. Zhongguo yuwen, 6, 401–405. Majid, Asifa., Bowerman, Melissa., Kita, Sotaro., Haun, Daniel B.M., & Levinson, Stephen C. (2004). Can language restructure cognition? The case of space. TRENDS in Cognitive sciences, 8(3), 108–114. McCullough, K. E. (1993). Spatial information and cohesion in the gesticulation of English and Chinese speakers. Paper presented to the American Psychological Society, Chicago. Ogawa, Kanju [Xiaochuan Huanshu]. (1981). Sūzhōu fāngyán de zhǐshì dàicí [The demonstrative pronouns in the Suzhou dialect]. Fangyan, 4, 287–288. Peyraube, Alain. (2006). Motion events in Chinese – A diachronic study of directional complements. In Maya Hickmann & Stephane Robert (Eds.), Space in Languages: Linguistics systems and cognitive categories (pp. 121–135). Amsterdam; Amsterdam; Philadelphia: J Benjamins. Van der Zee, E. & Jon Slack. (2003). Representing direction in language and space. Oxford University Press. Wu, Yunji. (2002). The old layer of the personal and demonstrative pronoun system and its development in Hunan. Journal of Chinese Linguistics, 30(2), 310–342. Yang, Wei. (1999). Yuánlíng Xiānghuà Yánjiū [Research on the Yuanling Waxiang Dialect]. Changsha: Hunan Jiaoyu Chubanshe.

FROM LOCATIVE TO OBJECT MARKERS: THE PARALLEL DEVELOPMENT OF TWO POSTPOSITIONS IN BAI1 Jingqi Fu St. Mary’s College of Maryland

Lin Xu Chinese Academy of Social Sciences

Keywords: Bai language, postposition, locative, dative-locative affinity, object marker Abstract: While locative-dative affinity is not rare cross-linguistically, Bai offers an unique case of specific connection through two of its postpositions that have grammaticalized from locative to object markers, in contrast with each other. We argue that the locative contrast between ‘on X’ vs. ‘near X’ is basis for the distinction between ‘direct object’ and ‘indirect object’. Dialectal data and other Tibeto-Burman languages are brought to bear on the issues of grammatical object markers derived from locatives and in particular the critical notion of ‘contact’.

1. Introduction The Bai language is caseless except for the genitive vs. default distinction in its pronominal system. However, there exist two object markers that may be thought of as accusative vs. dative or direct vs. indirect object markers. The same two morphemes also serve as locative postpositions with distinct meanings. Cross-linguistically, it is not unusual for Dative to be derived from Locative, as was the case in old English, Latin and IE (Pasicki, 1998). What is rare is for two locative postpositions to develop into object markers in a parallel fashion. Bai therefore offers a glimpse into how different locative meanings can develop into dative and accusative markers. We will explore the pathways from the locative to object markers in the parallel development of both postpositions. Bai is an ethnic minority language spoken by 1.6 million people mainly in the Dali region of southwest China. It is considered a Sino-Tibetan language, but due to the long standing contact with Chinese, more than 60% of its vocabulary consists of Chinese loans and its grammar exhibits both Chinese and Lolo-Burmese features. Consequently, its genetic affiliation within the family has been subject to debate, having been classified as either a Tibeto-Burman or a sinitic language. The current study does not bear directly on the issue of the genetic affiliation of the Bai language as it presents something rather unique. The three Bai dialects studied are Jianchuan (Central), Dali (Southern) and LEME (Northern, formerly Bijiang). The LEME and Dali data is mostly based on our 2004 fieldwork and the rest is based on the standard sources and our own judgments and previous fieldwork. The examples that follow are from Jianchuan unless otherwise mentioned. The two morphemes in question are -no33 and -v§55. They are postpositions, attaching to nouns or noun phrases. Either can serve as a locative marker or object marker; this distinction was illustrated in Xu and Zhao (1984). Being a locative marker, X-no33 means ‘on X’ whereas X-v§55 means ‘around X’, as shown in (1). 119 D. Xu (ed.), Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives, 119–141. © Springer Science +Business Media B.V. 2008

120

(1)

J. Fu and L. Xu

a. ku21to42 -no33 ts 33 ji>21k55 bridge on there is person ‘There is a person on the bridge.’

ji>21 CL

b. ku21to42 -v§55 `ts 33 ji>21k55 ji>21 bridge near there is person CL ‘There is a person near the bridge.’

(Xu and Zhao 1984: 56)

The examples in (2) show the objective use. The same verb su#44 ‘speak’ with an object can either mean ‘criticize him’, when the pronoun has an -no33 or ‘talk to him’ when the pronoun has an -v§55. (2)

a. o31 su#44 #31khu#33 > tsi33 -no33 1-SG speak Proper Name ‘I criticize/scold Little Doggie.’ b. o31 su#44 #31khu#>33tsi33 -v§55 1-SG speak Proper Name ‘I speak to Little Doggie (about something).’ (Xu and Zhao 1984 : 55)

The postposition -no33 seems to mark the direct object and -v§55 seems to mark the indirect object. We will see that the distribution of object markers is more complex than the distinction between direct and indirect object. The paper will explore the connection between objective and locative uses in three Bai dialects. Section 2 will introduce the locative use of the postpositions and Section 3 the objective use. Section 4 presents a dialect comparison and Section 5 the conclusion. 2. Locative Markers The distinction between -no33and -v§55 as locative markers is quite clear. The contrast in (3) and (4) provides further illustration, with -no33 referring to the top surface of the table or mountain and -v§55 the area around the table or foot of the mountain. (3)

a. m#55 ts 33 tsi55tsi33 tso>55-no33 3-PL at table CL ‘They play on the table.’ b. m#55 ts 33 tsi55tsi33 tso>55 -v§55 3-PL at table CL ‘ They play near the table.’

w#42se>42 play w#42se>42 play

From Locative to Object Markers

(4)

121

o31 21 sv>42 -no33 , mo31 21 sv>42 ko55 -v§55 1-SG go mountain 3-SG go mountain-foot ‘I go up in the mountain and he goes to the foot of the mountain.’

In (3) it can also be shown that the host for the postposition is an NP because there is an intervening classifier. Both postpositions can be part of a compound, with a discernable meaning distinction. (5) lists compounds with -no33 and (6) compounds with -v§55. (5)

a. t 21po21 no33 head ‘on the head’

b. tse>55 no33 body ‘on X’s body/with X’

c. * 33 no33 back ‘in the back’

(6)

a. me21 v§55 door ‘at the door’

b. ti31 v§55 field ‘in the field’

c. tu44 v§55 mouth ‘on/around the mouth’

d. t 21 v§55 front ‘in front/before’ The activities associated with a noun in (5) as a location are most likely to occur on top of this location, e.g. on the head or on the body. Likewise, nouns in (6), as location, denote a reference point around which the action takes place, e.g. at the door or in the field. The minimal contrast between ‘front’ (6d) and ‘back’ (5c) is more idiomatic. Even so, one can think of ‘front’ as an area and ‘back’ as a location with a surface (back of the body), hence the distinction of ‘on vs. around’. The idiomatic usage of these postpositions shows that the locative meaning is older and more original than the grammatical markers. In some cases, such as nouns expressing cardinal directions, both postpositions can be used as shown in (7). (7)

a. tv>55 fv§55 {no33/v§55} east ‘east’

b. se>55 fv§55 {no33/v§55} west ‘west’

Rather than saying that the two postpositions are interchangeable, we show in (8)-(9) that context will determine the appropriate postposition. That is, the NP-no33 in (8) refers to a location on which an action takes place (e.g. living at), whereas the NPv§55 refers to a location from which or toward which an action or event is directed, be it walking from, toward, or looking toward. (8)

mo31 kv§42ke21 3-SG live at ‘He lives in the east.’

tv>55 fv§55 no33 east

J. Fu and L. Xu

122

(9)

a.

mo31 s#35 tv>55 fv§55 v§55 3-SG from east ‘He comes from the east.’

* 35 come

b.

mo31 pe44ts21 tv>55 fv§55 v§55 3-SG walk-toward east ‘He walks toward the east.’

c.

mo31 h#>55 ts21 3-SG look toward ‘He looks towards east.’

tv>55 fv§55 v§55 east

The basic distinction of ‘on vs. around’ is therefore reflected in compounds as well as phrases. Besides the ‘on vs. around’ distinction, it is important to note that -no33 implies a surface contact. In this sense it contrasts with to>33no33 which means ‘above’ without contact. The examples (a) and (b) in (10) show this contrast: (10) a. m 55 t 21po21 kho33 -no33 3-GEN head CL ‘There is a mosquito on his head.’ b. m 55 t 21po21 kho33 to>33 -no33 3-GEN head CL above ‘There is a mosquito over his head.’

ts 33 have

mo44 tsi33 t 21 mosquito CL

ts 33 have

mo44 tsi33 t 21 mosquito CL

By extension, an object in contact with the action is likely to be marked with –no33. Consider (11). (11) a. th44 m 55 -no33 kick 3-SG-GEN ‘to kick him a few times’

k35 ko35 several CL

b. th44 mo31 k35 ko35 kick 3-SG several CL ‘to kick at him a few times’ Note that when a pronoun is followed by a postposition, it automatically takes on genitive form. For example, mo31+-no33Æ m 55 -no33. Both sentences in (11) contain the predicate ‘to kick him’. The object ‘him’ with -no33 receives the kicks, but without -no33 ‘him’ may or may not receive the kicks. So the NP-no33 implies a contact with the action, in this case the recipient of the kicks.

From Locative to Object Markers

123

In contrast, -v§55 is associated with a location not necessarily in contact with the action/object. The examples in (12) and (13) illustrate the general location use. In (12) the hand is waving in the general direction of ‘him’. In (13) sitting closer to a person, may or may not involve physical contact. (12) o31 s>55 1-SG wave ‘I wave to him.’

-v§55 m 55 3-SG-GEN

(13) o31 kv§42 ji>44 1-SG sit into ‘May I sit closer to you?’

-v§55 th44 n 55 tsi31t#42? 2-SG-GEN a bit OK (Xu and Zhao 1984:89)

s 33 hand

While X-no33 refers to a location X that is in contact with the action, be it the destination or the object, X-v§55 refers to a location X around which action takes place. In summary, as locative postpositions, -no33 and -v§55 have a clear distinction in both compounds and sentences. While X-no33 means ‘on X’, with contact, X-v§55 has the meaning of ‘around X’. 3. Object Markers Besides the locative NP, -no33 and -v§55 can also attach to an object NP. Here again the two have a distinction. As shown in (2) above, NP with -no33 corresponds more or less to a direct object while the NP with -v§55 to an indirect object. We will show that the distribution of the postpositions is more complex than a simple direct vs. indirect object distinction can explain, and we will discuss how the various uses of the object markers can be related to the original locative meanings. Section 3.1 will be devoted to the indirect object of the Dative construction; 3.2 to other double object constructions; 3.3 to the single object of a transitive verb and 3.4 to the object of a prepositional phrase. 3.1. Double object constructions We start with dative constructions because they allow a clear distinction between -no33 and -v§55. The distribution of the two postpositions depends on the role of the NP host in the sentence. Generally, indirect objects in dative constructions can be marked by either postposition. If the referent of the indirect object is the recipient of the exchange, -no33 is used; on the other hand if the referent of the indirect object is the source of the exchange, then -v§55 is used. In (14) and (15) the pair of verbs has opposite meanings: sell vs. buy; lose vs. win. The indirect objects of sell and lose, the recipients of the transfers in (a), are marked with -no33, whereas those of buy and win, the sources of the transfers in (b), are marked with - v§ 55.

124

J. Fu and L. Xu

(14) a. o31 k 33-zi31 {m 55 - no33/*m 55 -v§55} 1-SG sell-give 3-SG-GEN 3-SG-GEN ‘I sold him four acres of land.’

ti31 sv§55 field 4-acres

b. o31 m#42 -t 44 {*m 55 - no33/m 55 -v§55} 1-SG buy -reach 3-SG-GEN 3-SG-GEN ‘I bought four acres of land from him.’

ti31 sv§55 field 4-acres

(15) a. o31 tshB55-zi31 {m 55 - no33/*m 55 -v§55} tse21 ko>33 1-SG lose give 3-SG-GEN 3-SG-GEN money two ‘I have lost two dollars to him.’

kho55 dollar

b. o31 t 44-t 44 {*m 55 - no33/m 55 -v§55} tse21 ko>33 kho55 1-SG win reach 3-SG-GEN 3-SG-GEN money two dollar ‘I have won two dollars from him.’ Clearly the distinction between -no33 and -v§55 is not explainable in terms of direct vs. indirect object as the contrast in (2) might have suggested, since the same indirect object can take either postposition. To further illustrate, when an indirect object may be ambiguous between recipient and source roles, the choice of postposition can disambiguate. In (16), the verb p21 ‘return’ can mean ‘return something to the owner’ or ‘have the borrower return something’. The postposition -no33 is associated with the recipient, the original lender, while -v§55 with source, the one returning the book. (16) a.. mo31 p21  55 3-SG return 1-SG-GEN ‘ He returned a book to me.’

- no33

b. mo31 p21  55 -v§55 3-SG return 1-SG-GEN ‘ He had me return a book (of his).’

sv§55 book

tshu#44 CL

sv§55 tshu#44 book CL (Xu and Zhao 1984: 55)

Likewise the verb pho55 ‘to rent’ can either mean ‘rent to someone’ or ‘rent from someone’, depending on the resultative ending on the verb. The indirect object with -no33 expresses the receiver (the tenant), and the one with -v§55 expresses the source (in this case the land owner). (17) a. mo31 pho55 zi31  55 3-SG rent give 1-SG-GEN ‘He loaned me four acres of land.’

-no33

b. o31 pho55 (t 44) m 55 1-SG rent obtain 3-SG-GEN ‘I have rented four acres of land from him.’

ti31 sv§55 field 4-acre -v§55

ti31 sv§55 field 4-acre

From Locative to Object Markers

125

Similarly, the referent of the indirect object can be ambiguous based on the choice of postposition without the verb changing meaning. The predicate th#>55 li33 in (18) means ‘to accept gifts/dowries’. But the indirect object with -no33 refers to the person for whom the gift is intended (the ultimate recipient) and the indirect object with -v§55 refers to the person who offers the gift (the source). The recipient here is the intended one, the goal; it is not the subject of ‘accepting’, but the person who ends up having the gift. (18) a. n 55 pu35 th#>55 n 55 2-SG-GEN uncle accept 2-SG-GEN ‘Your uncle accepted the gift/dowry for you.’

-no33

li33 gift/dowry

b. n 55 pu35 th#>55 n 55 -v§55 li33 2-SG-GEN uncle accept 2-SG-GEN gift/dowry ‘Your uncle accepted the gift from you.’ (Xu and Zhao1984 : 55) In the same vein, compound verbs with the meaning of GIVE/TO are compatible with objects taking -no33. A list of some of them is given in (19) and examples are given in (20)–(21). (19

ne44 pi55 k 33 zi31 v55> pi55> ne44 thi31 p 21 pi55 55 pi55 su#44 zi31 su#44 pi55 k#>55 zi31

(20)

#>55 -zi31 #31khu#>33tsi33 find -to Proper Name ‘Find Little Doggie a playmate.’

-no33

(21)

k#>55 zi31  55 - no33 teach give 1-SG-GEN ‘Teach me a song.’

khv§44 tsh55> song CL

take bring sell to write to take away(to) bring come call for say to say to teach to t55 buddy

ji>21 CL

In summary, in a dative construction, the object-no33 marks the recipient /goal and the object-v§55 the source. It is not hard to see how the recipient /goal vs. source

126

J. Fu and L. Xu

meanings relate to locative contrast of ‘on’ vs. ‘around’, since the recipient will receive the object after the transfer, the object is ‘on’ the recipient, but the source is the location from which the object moves; the object is therefore not ‘on’ it after transfer, it is in the vicinity of the transfer. However there is more to it than the recipient vs. source contrast, as we will see below. The contrast just outlined only reflects the basic meaning distinction of the two postpositions. Consider the two sentences in (22), each containing a recipient. (22) a. o31 liu55-zi31 m 55 - no33 1-SG leave-give 3-SG-GEN ‘I left a bag with him (by handing it over to him).’

n 21 me21 bag CL

b. o 31 liu55tsh 55 m 55 - v§55 n 21 me21 1-SG leave-finish 3-SG-GEN bag CL ‘I left a bag for him (for someone to give to him later)’.

(a) and (b) are identical except for the verb ending and the choice of postposition. With -no33 the transfer of the ‘bag’ is to the recipient’s hands; with -v§55, the bag is left for the intended recipient. Even though -no33 could express an intended recipient/goal when contrasted with the source -v§55 (cf. 15), when both refer to recipients, -no33 refers to the immediate recipient and -v§55 to the distant recipient. Indeed examples (23) and (24) show that the distant recipient, the goal not yet reached, is expressed by -v§55. The intended recipient, the person to whom a bride is promised in (a) or the persons to whom a will is left in (b) are waiting to receive the object. (23)

mo31 u44  55 -v§55 m 55 jv>55thi33 3-SG promise-marriage 1-SG-GEN 3-SG-GEN younger sister ‘He promised me his younger sister’s hand in marriage.’

(24)

mo31 tshi55 tsi33 su#>55 -v§55 s>21 -v§42 3-SG leave descendant will ‘He left a will for his descendants’.

In light of (22)–(24), the recipient vs. source contrast is not sufficient, as both postpositions can serve as recipients. The basic distinction seems to be close vs. further or central vs. peripheral participant. There is a sense of relative proximity of the participant to verb action. Between a source and a recipient, a recipient is closer to the action; but between two recipients, one may be closer than the other. It is the closest participant that gets marked with -no33. The original locative contrast of ‘on’ vs. ‘around’ is carried over here: the close participant is at the location whereas the far participant is around the location. The recipient vs. source contrast is simply another instance of central vs. peripheral participant.

From Locative to Object Markers

127

3.2. Other double object constructions Let us consider other double object constructions where the notion of transfer may or may not apply. The source vs. recipient distinction can be apparent as shown by the contrast in (25). (25) a. o31 j>42 t#55t#55 -no33 1-SG carry on the back uncle ‘I put something on my uncle’s back.’ b. o31 j>42 t#55t#55 -v§55 1-SG carry on the back uncle ‘I carry something on my back for my uncle.’ (Xu and Zhao 1984: 55) The object ‘uncle’ is the recipient with -no33: the person whom the load goes on. And the ‘uncle’ is the source with -v§55: the person who the load is taken from. The same can be said of (26); the pronoun ‘him’ in ‘hitting him with the stone’ is in a sense the recipient of the stone hitting since it is the location the stone will end up. (26)

#31khu#33 > tsi33 t>44 m#55- no33 Proper Name hit 3-PL ‘Little doggie hit them with a stone.’

tso42khui55 stone (Xu and Zhao 1984: 69)

However, other constructions are less obviously analyzable in terms of recipient vs. source distinction. Consider the contrast between (a) and (b) in (27). (27) a. m#55 44  55 -no33 3-PL insult 1-SG-GEN ‘They call me old bull.’ b. mo31 55 3-SG call ‘He calls me # kh>.’

ku33  21po55 old bull

 55 - v§55 #31kh>33 1-SG-GEN Proper Name (Xu and Zhao 1984: 69)

Both verbs ‘insult’ and ‘call’ are similar in meaning, to call somebody something. Both have similar argument structure, taking double objects. The first object, the one marked with a postposition, is a recipient in the calling act in both, so the distinction recipient vs. source would not apply here. The person being insulted is more affected by insulting than the person being called a neutral name. Could it be that the object more affected by the action is more likely to be marked with -no33? We are evidently moving away from the locative meanings, and will answer this question in the next section on single object constructions. We may conclude for now that for double object constructions, the object (usually an indirect object) that is the central participant is marked with -no33 whereas the one that is peripheral participant is marked -v§55. The central participant

J. Fu and L. Xu

128

can be a recipient when contrasted with a source or close recipient when contrasted with a further recipient. The connection can be established between the central vs. peripheral contrast with the locative contrast of ‘on’ vs. ‘near/around’ of the two postpositions. The central participant refers to the person who the object of the action ends up on. The peripheral participant (source or remote recipient) is in the vicinity of the object, further away from the action than a recipient or a closer recipient. The range of usages of the two postpositions goes beyond the recipient vs. source contrast, as we have seen in examples (27). The extended meaning indicates that the two postpositions have evolved from their original locative meanings to fully grammatical ones. We consider the recipient vs. source stage as an intermediate stage in the pathway from locative to object markers. Since the dative constructions involve a transfer of something from one location to another, or from one party to another, the indirect object in a dative structure would correspond to a location for the object, the source before the transfer or the recipient after the transfer, therefore a locative meaning is apparent. The next stage would be one where the locative meaning is less apparent, if present at all. That will be the topic of the next section on single object constructions. 3.3. Single Object In double object constructions, the -no33/-v§55 distinction is rather clear, with 33 no marking the recipient and -v§55 the source. For the single object of a transitive verb, this distinction is not applicable. Rather, there is a shift from the locative orientation. Recall the contrast in (2) repeated here as (28), in which the object with –no33 functions as the direct object and the object with -v§55 as the indirect object. (28)= (2) a.

b.

o31 su#44 #31khu#33>tsi33 -no33. 1-SG speak Proper Name ‘I criticize/scold Little Doggie’. o31 su#44 #31khu#>33tsi33 -v§55. I speak Proper Name ‘I speak to Little Doggie’(about something). (Xu and Zhao1984 : 55)

The recipient/source distinction is not relevant here, as the object with -no33 is not a recipient of an exchange nor is the object with -v§55 the source. Something else determines the choice of the object’s postposition. In general, the single object of most verbs can be marked by -no33. A list is given in (29) with examples to illustrate in (30)–(31).

129

From Locative to Object Markers

(29)

Verbs with object-no33 t 33 ‘to wait’ kh#> 21‘plan’ t>44 ‘to hit’ i> 31 ‘to believe’ 44 31 55 42 t ‘to win’ i hu#> ‘to like’ k ‘to detest’ k>44 ‘to scare’ 55 pi ‘to force’ x44tsu44 ‘to scold’ ko21 ‘to love’

(30)

mo31 i31hu#>>55 3-SG like ‘He likes them.’

m#55- no33 3-PL

(31)

l33 lo42 mo33, o31 EXCL 1-SG ‘Oh, what should I do!’

jo>44 want

kh#> 21 #55s>31 -no33 j>33 plan what EXCL (Xu and Zhao 1984: 59)

However for a smaller number of verbs, the object is marked v§55. If the object refers to the source of the action, like the person to learn from in (32), then v§55 must be used. (32)

no31 * 42 m 55 -v§55 2-SG learn 3-SG-GEN ‘Don’t learn from him.’

#31 mi#55 NEG

In addition to the source, the addressee of an utterance is marked with -v§55 as in the following examples with the verbs ‘ask’, ‘tell’ and ‘talk to’. (33)

 55 ku55 pi44 1-SG-GEN aunt ask ‘Who is my aunt going to ask?’

# 31to21 who

(34)

j#44 khv21 su#44 ku55 go home talk to aunt ‘go home and tell your auntie’

-v§55

(35)

 55 t21 ph#>31 1-SG-GEN suffering talk ‘I am telling you about my suffering’.

-v§55?

n 55 2-SG

-v§55

Both source and addressee play a peripheral role in the action, while the object of a transitive verb (29) is directly involved in the action or event, be it theme or experiencer. The peripheral reading of the object-v§55 can be further illustrated by an apparent exception. The object of the verb x#55 ‘to look, to look after’ is sometimes marked with -v§55 as in (36) (b) and (c).

J. Fu and L. Xu

130

(36) a. x#55 m 55 look 3-SG-GEN ‘Don’t look at him.’

-no33

#31 mi#55 NEG

b. x#55 ji>55 -v§55 k21 look 2-SG-RESP a bit ‘look after you (take care of your things, watch you do things).’ c. #21 tu21, x#55 m 55 EXCL look 3-SG-GEN ‘Let him do what he wants!’

-v§ 55

21 u#55 go PRT (Xu and Zhao 1984: 58)

This is surprising because the object of ‘to look’ is the theme and therefore is expected to be marked with –no33. However in (b) and (c) the object -v§55 is not the Theme of the verb x#55 in a strict sense. In (b) looking after someone does not involve just that person but things related to that person. In (c), the true object of looking is the event of someone going his way. In contrast, the object-no33 in (a) serves as the sole theme of the action of looking. In other words, object-v§55 bears a loose connection with the verb, by playing a peripheral role referring to anything that is not the theme of the verb. In a sense, given two participants, the one closely involved in the action/event denoted by the predicate is marked –no33 and the one further involved is marked -v§55. So in (37), ‘combing Grandma’s hair’, ‘hair’ is the theme, the close participant. ‘grandma’ is indirectly involved in the action, therefore it is marked with -v§55. (37)

mo31 sv§55 #31ne44 - v§55 3-SG comb grandma ‘He combed grandma’s hair.’

t 21po21 hair (Xu and Zhao 1984: 54)

Switching between the two postpositions will also yield the result that the -v§55 marked object bears a more remote role in the action. In (38), for example, the object of the verb pu33 ‘bring’, when marked with -v§55 is no longer the theme but rather the indirect beneficiary. (38) a. no33 k#44 m 55 2-SG PREP 3-SG-GEN ‘Bring/carry him [over] here.’

-no33

pu33 carry

b. no33 pu33  55 2-SG carry 1-SG-GEN ‘Carry (him) for me.’

-v§55

k21 a bit

* 35 come

From Locative to Object Markers

131

What we can conclude from the distribution of the two postpositions in a single object construction is this. On the one hand, it follows the same patterns as with double object constructions: an -no33 marked object bears a role central to the action/event of the verb, say Theme or Experiencer, while a -v§55 marked object has a role further away from the action/event, say source or addressee. On the other hand the distribution of the object revealed something more. The so-called central vs. peripheral role is no longer definable in locative terms, as the theme of verbs such as ‘like’ is not a location, unlike a recipient. There is a shift from location to objective role. Central vs. peripheral location has become central vs. peripheral participants. The recipient vs. source contrast represents both location and objective role; this is how it can serve as an intermediate stage between locative and object markers. Through the different stages, what remains constant is the contrast between central vs. peripheral, the contrast due to ‘on’ vs. ‘around’ of the two locative postpositions. As object markers, the postpositions moved away from their locative meanings and gained a measure of status as grammatical words. They are not purely characterizable in terms of direct and indirect object markers yet, since either can appear as the single object of a transitive verb (‘look’ in (36)) or as the ‘indirect object’ in a double object construction, with a meaning difference. Given this contrast of central vs. peripheral participant, we can reinterpret the contrast observed in (27) where the object of ‘insult’ is marked with –no33 and that of ‘call’ is marked with -v§55. The object of ‘insult’, affected by insulting, is the target of the action; while the object of ‘call’ in ‘call a person something’ remains addressee, and plays a peripheral role, like other addressees. It turns out that this sentence type is one of the few where both –no33 and -v§55 can appear without changing the meaning of the sentence. It is exactly this type that has an ambiguous role for the human object, either as an addressee or person directly involved (calling someone a name). 3.4. Object of preposition Objects of prepositions, especially pronominal objects, can also take a postposition. Either postposition can occur as locative or object marker, with the same general distinction described above. Some prepositions select a postposition, some can occur with either, with a difference in meaning. The examples in (39) and (40) illustrate the locative use. The preposition s#35 ‘from’ is compatible with either postposition. (39)=(9a)

mo31 s#35 tv>55 fv§55 -v§55 3-SG from east ‘He comes from the east.’

* 35 come

(40) lo21tsho>55 kv>55 no33 t>55 mi21so44, ji>k>55 s#35 m 55t 21-no33 21 Mekong river on pull rope bridge people from 3-SG-GEN above go ‘On the Mekong River was built a rope bridge, people go across the river on it.’ (Xu and Zhao 1984: 45)

132

J. Fu and L. Xu

The -v§55 in the first sentence marks the source location, the east where the person comes from. The -no33 in the second sentence marks the location, the rope bridge on which the crossing is done. As both are objects of the preposition ‘from’, the postposition alone makes the meaning distinct: at the source location vs. on top of the location. In this regard, recall that the source in a dative construction always takes -v§55 like in (41). (41)

mo31 s#35  55 -v§55 3-SG from 1-SG-GEN ‘He took (something) away from me.’

ne44 ts 21 take away (Xu and Zhao 1984 : 45)

In both (41) and (39) the source location contrasts with another one, the current location. The source place contrasts with the current location. In (40) the focus is ‘on the rope bridge’, no other location is implied. Therefore we can say that source locations can be viewed as further than the current one. The surface location has no33 whereas the far one has -v§55. As object markers, the postpositions occur predictably within prepositional phrases. The most common prepositions in such combination are: k#44 and jo>42, which serve to introduce the object without a specific meaning of their own. They are comparable to Mandarin ba (把) in the so-called disposal construction. Since their lexical meaning is bleached, they do not restrict the use of the postposition; instead it is the predicate that determines which one to use. The verbs in (42) take NP-no33 as their prepositional object. They have the meaning of ‘depend on’, ‘call someone to come’ and ‘hold someone by the hand’. Either preposition can be used. (42

a. ku33 l#42 jo>44 k#44 m#55 -no33 kh#44 old PRT will 3-PL rely ‘When I get old, I will depend on them (to support me).’ b. jo>42 m 55 -no33 55 ts 31 3-SG call –come ‘Go and bring him.’ c. k#44 #31mo33 -no33 khe>55 * 35 mother hold-by-hand -come ‘Bring mother by the hand.’

In contrast, the object of the preposition in (43) are NP-v§55. Just like their counterparts without a preposition, the addressees are -v§55 marked.

From Locative to Object Markers

133

(43) a. s21 k#44 #31mo33 -v§55 su#44 try mother speak ‘try to speak to mother’ (44) b. thi31t#42 21 k#44 m 55 -v§55 only go 3-SG-GEN ‘(I) had no choice but to call him.’

55 call

In other words, as object of a prepositional phrase, the choice of postposition follows the same pattern as the object of a VP, while an addressee has -v§55, others have -no33. So far postpositions freely occur within a prepositional phrase, as locative and object marker. Regardless of the preposition, -no33 and -v§55 reflect their basic meaning distinctions: central vs. peripheral participant. We now turn to prepositions that do select the postposition for their object. The prepositions like p 31 ‘for’, fv33 ‘especially for’, and t#42 ‘with’ take -no33, as shown in (45)–(46). (45)̀̀

o31 li55 p 31 n 55 {-no33/* -v§55} 1-SG also specially for 2-SG-GEN ‘I am also happy for you’.

(46)

 55 mo33 fv33  jv>§33thi33 {-no33/* -v§55} 1-SG-GEN mother for 1-SG-GEN younger sister ‘My mother comes especially for my sister’.

(47)

t#42 m 55{ -no33/* -v§55 } w#42se>42 with 3-SG play ‘play with him’

i31xu#55 happy * 35 come

Other prepositions such as the comparative li 21 can only occur with -v§55. (48)

li 21 m 55 {* -no33/-v§55 } COM 3-SG ‘Cannot surpass him’

tu#42 Neg. catch

Not all comparative structures require -v§55, the pattern with the Chinese loanword pi31 biš 比 as shown in (49) is without any postposition. This shows that whatever requires the presence of -v§55 in (47) is not applicable in (49). Since (49) is a construction based on a Chinese loan word, it might be more recent. (49)

o55 pi31 mo31 1-SG COMP 3-SG ‘I am six years older than him.’

to42 big

fv§44 su#44 six year (Xu and Zhao 1984 : 83)

134

J. Fu and L. Xu

Some other prepositions cannot take any postposition, as illustrated by ko55‘with’: (50

no31 ko55 {mo31 /* m 55 -no33 /*m 55 -v§55 } w#42se>42 2-SG with 3-SG 3-SG-GEN 3-SG-GEN play ‘You go and play with him.’

Notice that ko55 in (50) is identical in meaning to t#44 in (47), both meaning ‘with’. Yet the choice of postposition is different. Some idiosyncrasies might be involved here, perhaps having to do with the original lexical distinction between the two prepositions. Indeed, ko55 is more of a conjunction linking two parallel elements, as in (51), whereas t#44 was once a verb with the meaning of ‘follow’, as shown in (52). (51)

#31khu#33 > tsi33 ko55 #31kh>33 Proper Name Proper Name ‘Little Doggie and #31kh>33 are together.’

(52)

h35i33 t#44 n 55 -no33 death follow 2-SG-GEN ‘(I) will follow you in death and in life.’

ts 33 a31pho55 be a group

If t#44 was originally a transitive verb, its object taking -no33 could be explained because objects of transitive verbs are marked with -no33. The obligatoriness of the object marker may be due to the meaning of ‘follow’, implying a contact between a person and his follower. The pre-determined choice of postposition by some prepositions gives us some hint of the older meaning of the postpositions and prepositions. The benefactive and the comparative with li 21 require an -v§55. This shows that this postposition is already used in extensions of its locative meaning ‘around’. They are typically meanings associated with the Dative: goal and general point of reference. Old English Dative exemplified both (Pasicki 1998). The distribution of the two postpositions within prepositional phrases is the same as with verbs. As a locative, -no33 marks the surface location while -v§55 marks the vicinity, e.g. the general area or source location or some general point of reference such as the standard for comparison. As an object marker, -no33 is associated with NPs whose referent is directly affected by the verb action, while v§55 refers to the source and goal. 4. Dialect Comparison A preliminary examination suggests that the same usages of both postpositions are found in the three Bai dialects (Jianchuan, Dali and LEME). That is, the two postpositions serve as both locative and object markers with the same range of meaning distinctions, even some of the idiosyncratic usages. There are however interesting differences.

From Locative to Object Markers

135

In the LEME dialect, -v§55 or m 55 acts as a general location marker. In (53) the noun ‘school’ is the object of ‘go’ and ‘come’ and of the preposition ‘at’. In all three cases it can be followed by either -v§55 or the postposition -he>55 ‘in’, but cannot be followed by -no33. Since the noun refers to a location either away from the speaker or where the speaker is, or where the action takes place, it no longer exclusively corresponds to a source or location around the action, the kind of meaning reserved for -v§55 in the ‘on’ vs. ‘around’ distinction. It is clear that -v§55 is used to indicate a location, regardless of its proximity to the action. (53) a. j#44 * 31sv§55h31 go school ‘go to school’

{-v§55 /-he>55 /*-no33} -in

b. tchu#44 * 31sv§55h31 come school ‘come to school’

{-v§55 /-he>55 /*-no33} -in

c. kv42 * 31 sv§55h31 at school ‘to eat at school’.

{-v§55 /-he>55 /*-no33 } -in

ji55be33 eat

Corresponding sentences in other dialects can only have -h 21 ‘inside’, rather than the equivalents of –no33 and -v§55. The examples in (54) come from Jianchuan, but their Dali equivalents are essentially the same. They contrast with LEME examples in (53) in terms of the postposition of the locative noun. (54) a. 21 u55 t#> go school ‘go to school’

{*-v§55 /-h 21 /*-no33 } in

b. * 35 u55 t#> {*-v§55 /-h 21 /*-no33 } come school in ‘come to school.’ c. ts 33 u55 t#> {*-v§55 /-h 21 /*-no33 } j 44tsh>44 at school in eat ‘to eat at school’. Interestingly, in the same dialect, LEME, -v§55 is found to serve as an aspectual marker, that of durative. In (55) -v§55 is suffixed to the verbs ‘climb’ and ‘stand’ with the durative aspect meaning.

J. Fu and L. Xu

136

(55) a. ɖi 31 ɖi 31 -no33 m#55-v§55 ni31 q35 tree -LOC climb person ‘A person was climbing a tree.’ b. ɖi 31 ɖi 31 #55-no33 kh 33-v§55 tree side stand ‘Beside the tree stands a person.’

#31 i31 one CL #31 i31 one CL

ni31 q35 person

It is not uncommon for a locative marker to become a durative marker. The Mandarin durative marker zhe is argued in Mei (1988) to be derived from a locative preposition. Perhaps the locative expression would first become a general locative marker and then an aspect marker. That is why this extension of -v§55 is found in the one dialect where it refers to a general location. The lexicalization of the two postpositions in the three dialects also points to the special status of -v§55/-mu55 in LEME. Table 1 shows words in citation forms of directions, based on the word lists from Xu and Zhao (1984). Table 1. Some direction nouns in the three dialects Meanings front/before Back Beside Above Below Inside Outside

Jianchuan t 21 v§55 * 33 no33 po21 no33 to33 > fv§33 no33 *33 fv§33 no33 kh 31 no33 u#44 no33

Dali t 21 v§55 * 33 t 55keŒ44pieŒ44 tou33 po21 eŒ 33mi42no33 kh 31mi42no33 u#44 mi31no33

LEME t 21 * 33 po21 qo55no33 do>33 mu55 po55ji33mu55 po55x >33 po55eŒ42no33

We can see that while in all three dialects some nouns take –no33, -v§55/mu55 is mostly found in LEME, with only one example in Jianchuan and Dali. This gives a hint as to the status of -v§55 in LEME. Its wider distribution shows that it is a locative devoid of meaning. Together with its general locative use and derived aspectual marker use it has gained the status of a general locative morpheme, beyond its meaning of ‘near/around’. Another difference among the three dialects is related to the extension of the locative meaning. In LEME, -no33 and -v§55 can correspond to the presence or absence of the referent of the object to the situation of utterance. Consider (56): (56) a. no42 v§55 bo55 -no33 2-SG call 3-SG ‘Call him (he is over there).’

From Locative to Object Markers

137

no42 v§55 bo55 -v§55 2-SG call 3-SG ‘Call him (he is not present).’

b.

This distinction, present vs. absent of the referent of the object, can be traced to the original distinction of ‘on’ and ‘around’ of the locative meaning. The person who is present will be at the location whereas the person who is absent is not at the location, only near or around it. In Dali dialect, a similar but not identical extension can be observed. An object with -no33 has the effect of making the event present or future whereas an object with -v§55 is associated with a past event. The contrast in (57) illustrates this point. (57) a. o42 su#44 bo55 -no33. 1-SG speak 3-SG ‘I will speak to him.’ b.

o42 su#44 bo55 -m 55. 1-SG speak 3-SG ‘I have spoken to him.’

Here again the locative trace is detectable. No33 as a locative means ‘on’ or at the current location, its association with a current/future event is understandable. M 55 (= -v§55 ) means ‘around’ or further from the action; since the past is further from the present, its effect to indicate a past event is explainable. We have seen several different paths for locative markers. While in all three dialects, locative develop into object markers via a recipient vs. source stage, some have extended the locative contrast to express the present vs. absent or present vs. past contrasts and some have picked one of the two postpositions as the general locative marker and have turned it into an aspectual marker. The common path shared by all three, from locative to object markers, may either be a universal path that each dialect independently follows or a common feature present in Bai before the dialect split. 5. Conclusion We have presented the usages of two postpositions in Bai, as locatives, recipient vs. source and finally as object markers. We have argued that locatives are on their way or have developed into object markers via a dative stage. Bai is not unusual to link object and locative. Many languages do, in the form of Dative and Locative cases, for example. What is rare is that Bai has the pairing of two locative morphemes that gives rise to two corresponding object markers. Because of the parallel development of the two morphemes in contrast, we were offered a rare opportunity to understand how the grammaticalization of locative into object marker works, particularly in Bai as opposed to other languages where the object-locative affinity involves only one morpheme.

138

J. Fu and L. Xu

In languages with morphological case, the case that is associated with and derived from the locative case is Dative. IE, Latin and Old English are just a few such languages (Pasciki 1998). Dative is known to exhibit a range of meanings. Often cited are dative of place, goal, approach, comparison, possessor, experiencer, interest and sympathy. These meanings are subsumed under ‘general reference point’ in Pasciki, tracing back to locative. Similarly, in languages with no clear morphological case, affinity between object and locative can also be observed. In a study of grammatical postpositions/suffixes in Lolo-Burmese languages, Dai (1990:92–92) compares the postpositions in 19 languages of the Sino-Tibetan family, including Bai. Three of these have identical morphemes as locative and object postpositions: tha33 for Lahu, a33 for Jinuo, la (du, tu) for Tibetan. To these three we may add Bai (-no33 and - v 55 § ) and Hani (a33). Only two out of 19 languages have the same postposition for subject and locative: Zhuoka’s k33 and Naxi’s n 33. In these two, the same morpheme also marks the source. Even though some of these forms may be cognate, especially Jinuo and Hani’s, some are independent from each other, as argued by Dai himself. A clear affinity can be established between object and locative. The locative relates more to indirect object than direct object, both in form and in usages. In languages distinguishing direct and indirect object, like Hani, the indirect object shares the form of locative (Dai 1990 : 88). In languages where there is no formal distinction between direct and indirect object, the locative postposition is used more or less after the NP that would correspond to an indirect object. In both Lahu and Tibetan, according to the standard sources (Jin Peng 1983; DeLancey 2003; Matisoff 2003), the locative postposition is used after NPs that expresses the addressees, the recipient and goal. In Tibetan, Delancey (2003: 275) further notes that the distribution of the locative after the object is strictly governed by the lexical meaning of the verbs, and that objects of verbs describing surface contact must have the locative marker. This state of affairs is true outside of Tibeto-Burman. To cite an unrelated language, Pashto, an Iranian branch of the IE language family, also exhibits the same affinity (Skalmowski 1996). In that language a directional postposition ta appears after Dative-like nouns to indicate the place, the addressee, the goal, the time, etc. In short, locative is linked to dative, or to dative-like nouns (indirect object). If there is uniformity in the dative and locative affinity across languages regardless of the morphological case, how is this affinity reflected in Bai with a pair of postpositions? How did the original locative meanings explain the objective uses? On the surface, the -no33 vs. -v§55 contrast looks much like an Accusative vs. Dative or direct vs. indirect object contrast. -No33 marked NPs are mostly Accusative-like by being the theme or direct object, or the benefactive; and -v§55 marked NPs are dative-like by expressing goal, addressee, comparative, source, etc.

From Locative to Object Markers

139

On closer inspection, the -no33 vs. -v§55 contrast cannot be explained in terms of an Accusative vs. Dative contrast. One of the central functions of -no33 is to mark the recipient in a Dative construction. The recipient is associated with Dative crosslinguistically, as the goal/recipient is prototypically Dative. The answer lies in the original locative contrast and the path the two postpositions took from locative to object marker. Because of the original contrast of ‘on’ vs. ‘around’, the pathways of the two postpositions are natural and easy to trace. ‘On’ and ‘around’, in the dative stage, translate into the recipient, the location on which the transferred object lands, and the source, the location from where the object moves. At this stage recipient and source have both a locative meaning and a participant meaning. And because of the relative proximity built into the contrast, -no33 vs. -v§55 represents closer location/participant vs. further location/participant. Beyond Dative constructions, the notion of participant takes over while still maintaining the contrast of central vs. peripheral. This leads to what looks like direct vs. indirect contrast (or Accusative vs. Dative). Because of the critical contrast of ‘on’ vs. ‘around’, the two postpositions developed in such a way that they defined themselves in opposition with each other rather than in absolute terms. The same postposition -no33 can therefore mark both recipient in a Dative construction and direct object, since in both cases it stands in opposition to -v§55 by marking something closer or more central to the action/event. In other words, the pathways of object markers from locative are related to the original meaning of the locatives. Since in the languages where the Dative case or Dative-like NP is derived from Locative, the range of meanings and usages are more or less comparable, accountable perhaps under ‘general reference point’, where we assume that the original locative must be of a general type. In most situations, as is the case with Bai’s -no33 vs. -v§55 we can no longer find the original meaning of the locatives. Where the original meaning is still detectable, the derived usages are quite telling. In Pashto for instance the object marker ta is called a directional postposition (Skalmowski 1996 : 399) that is said to be derived from ‘toward you’. Because of this meaning of moving toward, we think it is not random that the nine functions cited by the author of this postposition did not include source. Source is of course the opposite of motion towards somewhere. In Bai, there is no direction built into ‘on’ vs ‘around’, therefore source is compatible with ‘around’. The locative/object markers in the two languages mentioned above, Lahu and Tibetan, involve the notion of contact. Lahu’s tha33 is derived from ‘ascend’, ‘above’ and means ‘upper surface’, ‘top part’ (Matisoff 2003: 215). In Tibetan the original meaning of the locative la (du, tu) is unclear to us, but the restriction that it marks the argument of a ‘surface contact’ verb puts Tibetan on par with Lahu and Bai in this matter, since contact is invoked. The notion of contact is acknowledged by scholars working in grammaticalization of spatial expressions. Svorou (1993) proposes a universal path to derive spatial expressions roughly as follows.

140

J. Fu and L. Xu

Table 2. Svorou’s Evolutional path of Spatial grams from nouns Animal body part Human body part Environmental landmark

Æ

relational part of object

Æ

location adjacent to object

Æ

location in the region of object part

Crucial to the path is the stage relational part of object followed by location adjacent to object. The tha33 in Lahu exemplifies these two stages, by evolving from ‘top part’ to ‘top surface’. Wu (this volume) argued for pathways of different locative postpositions in Chinese. Of four such postpositions (shang 上 ‘above’, xia 下 ‘below’, qian 前 ‘front’, hou 后’back’), he proposed these two crucial stages for shang 上 ‘above’, while others do not have the ‘adjacent’ stage. The notion of contact may be necessary for the locative to develop into a direct object marker, as the direct object, especially the affected one, is in contact with the action. It is probably not random that -no33 in Bai and tha33 in Lahu are locatives that became object markers, nor is it an accident that surface contact verbs require la-marked objects in Tibetan. In this regard, languages like Bai, where two object markers simultaneously developed from locatives, serve as a good case study of how objects are connected with the locative.

Notes 1

This work has benefited from a grant of the French Ministry of Research and of the Faculty Development grant of St. Mary’s College of Maryland. We are grateful to the other articipants of the project and to our informants, Mr. Hu Yinan, Li Jianfeng, Zhang Xilu, and Hua Sibo. All errors are of course our own.

From Locative to Object Markers

141

References Dai Qingxia. (1990). Yimianyu de jiegouzhuci [The grammatical markers in tibeto-burman languages]. In Zangmian yuzu yuyan yanjiu [Research on tibeto-burman languages], (pp. 83–97). Beijing: Minzu. Jin Peng. (1983). Zangyu jianzhi [Outline of the tibetan grammar]. Beijing: Minzu. DeLancey, Scott. (2003). Lhasa tibetan. In Thurgood, Graham & Randy LaPolla (Eds.), The sino-tibetan languages (pp. 270–288). London New York: Routledge. Matisoff, James. (2003). Lahu. In Thurgood, Graham & Randy LaPolla (Eds.), The sinotibetan languages (pp. 208–221). London New York: Routledge. Mei Zulin. (1988). Hanyu fanyanli xuci “zhe” zi sanzhong yongfa laiyuan [The origin of the three usages of the grammatical word “zhe” in Chinese]. Zhongguo yuyan xuebao, 3, 193–216. Pasicki, Adam. (1998). Meaning of the dative case in Old English. In Van Langendonck W. & W. Van Belle, (Eds.), The dative, vol. 2: Theoretical and contrastive studies: (pp. 113–142). Amsterdam Philadelphia: John Benjamins Publishing Company. Skalmowski, Wojciech. (1996). Dative counterpart in Pashto. In Van Belle W. & W. Van Langendonck (Eds.), The dative, vol. 1: Descriptive studies: (pp. 395–406). Amsterdam Philadelphia: John Benjamins Publishing Company. Svorou Soteria. (1993). The grammar of space. Amsterdam Philadelphia: John Benjamins Publishing Company. Thurgood, Graham & Randy LaPolla. (2003). The sino-tibetan languages. London & New York: Routledge. Wu Fuxiang. (2006). Hanyu fangsuociyu de yufahua [Grammatcalization of locative expressions in Chinese] (this volume). Xu Lin & ZhaoYansun. (1984). Baiyu jianzhi [Outline of Bai grammar]. Beijing: Minzu.

THE ROLE OF VISUAL SPACE IN SIGN LANGUAGE DEVELOPMENT Shun-chiu YAU Centre de recherches linguistiques sur l’Asie Orientale * CNRS and EHESS, Paris, France Key words: sign language, cognitive-perceptual constraints, gestural creation, element ordering universals, syntactic changes Abstract: The fact that we are all, hearing or deaf, seeing creatures, and thus subject to cognitive-perceptual constraints in our communication, SL observations and analyses are therefor of theoretical interest not only to sign researchers, but also to those working on general linguistics as well.

1. Introduction Although gestures of some form were probably used when humans began to communicate with each other, the known history of the earliest sign languages (SL) of the deaf, for example Spanish or French SL, is less than three centuries old. Gestures created by deaf individuals are like wild grass, they grow during the lifetime of their creators, then disappear when their creators pass away. Gestures have to be invented anew. These gestures only develop into a language system when the deaf have the opportunity to form a small community, for instance, children grouped together in a school for the deaf. Such an opportunity for deaf youth was first created in Europe in the seventeenth century.1 It is gener ally believed that the history of oral languages (OL) goes back about 100 000 years (in Gordon Hewes’s rough estimation (1974) for example), thus the existence of SL in comparison to OL is like a newborn child compared to an adult. From this point of view, it is fascinating to find that SL has developed a rich syntax within such a short period. Certain complex structures absent in SL but present in OL might give some people the impression that SL is still not fully developed. In any case, it is difficult to define linguistic maturity. As Wittgenstein writes in his Philosophical Investigations: “... how many houses or streets does it take before a town begins to be a town?” In the same way, we may ask how many syntactic structures a SL should possess before it is considered a fully developed language. It is therefore preferable to say that, apart from those sign systems created in isolation which only have rudiments of syntax (see below), conventional SLs used in sizable deaf communities today have already developed syntactic and lexical mechanisms to meet the demand of their communities, in particular to handle situations resulting from ever-increasing contact with their respective ambient languages. Considering this, we should avoid dividing conventional SLs into “developed” and “developing” groups, a classification based on the notion of “progress” as used in geo-economics, to distinguish “developed countries” from “developing countries”.2 The notion of “progress”, in general, is irrelevant to language, and it would be hard to find qualitative or quantitative criteria for such an assessment. 143 D. Xu (ed.), Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspective, 143–174. © Springer Science +Business Media B.V. 2008

144

S. Yau

The term “visual space” in the title may need a brief explanation. Space in SL is not simply mental, the kind shared by OL (see Fauconnier 1985). The aim of this paper is to reveal how manual SL takes advantage of its visual-spatial particularity to develop independently and rapidly within a very short period, ranging from two or three centuries for the conventional SL to a lifetime for an individually created system (see Yau 1992). By adding “visual” to the term “space” in the title, I want to emphasize how SL “exploits the medium of vision and space in a systematic way” (Kyle and Woll 1985). Moreover, with a signing space provided, the visual medium is one of the two characterizing features of SL, the other being the manual mode. This visual aspect constitutes the most contrasting feature with the auditive aspect in speech. The last point in this introduction concerns the scope of this paper in SL research. In my early studies on this visual-manual medium, I maintained a strong link between SL and gestures, not only because SL can be labeled as gestural by nature, but because gestures are common to all humans, and not exclusively practiced by deaf signers. At the theoretical level, insistence on this gestural link is of utmost importance for those, myself included, who are convinced that gestures once played a crucial role in the emergence of human languages.3 In the section on element ordering universals in the present paper, though the data were obtained principally from SL observations, their significance goes beyond the domain of SL, and will shed light on some language universals as well. As far as element ordering is concerned, it is remarkable to find a similar preference for the order SOV, natural sign order, in the history of IE languages according to Lehmann (1972), or in child language with verb-medial word order as the target (see for example Slobin 1966; Parker 1970; Yau 1979).4 The same phenomenon is even witnessed in primates. According to what has been reported, the chimpanzee Washoe first used the sign order SOV, and it was only at a later stage that she was trained to adopt the signed American English sign order of SVO. (Linden 1974)5 My explanation for the phenomenon is that we are all, hearing or deaf, Washoe included, seeing creatures, and thus subject to cognitive-perceptual constraints in the organization of our utterances or gestural sequences. From this point of view, SL observations and analyses are of theoretical interest not only to sign researchers, but to general linguists as well. (See section 5: Element orderings) 2. The Two Essential Gestural Devices of SL: Iconicity and Simultaneity The two essential devices, iconicity and simultaneity, often work hand in hand in the formation of individual signs or gestural sequences. Simultaneity is not unknown in OL. For example, producing a tonal variation while pronouncing a vowel is a simultaneous realization. But in SL, simultaneity is generally exploited.6 With the help of a range of bodily visual articulators, notably the two hands, components of a single sign can be performed at the same time. Thus the sign TAPE-RECORDER in BSL (British Sign Language) represented by its two reels can be imitated simultaneously.7 Examples of simultaneous lexical fusions are also to be found in SL. The sign (one) MONTH in CSL (Chinese Sign Language) is made with two I-hands (i.e., fists

The Role of Visual Space in Sign Language Development

145

with only the index fingers extended), one in front of the other, the front I-hand, pointing up, palm forward; the rear I-hand (standing for “one”), pointing sideways, palm downward; the index of the side-pointing I-hand slides down the back of the index of the up-pointing I-hand (see Yau 1977b: 83). It is by changing the side pointing index that stands for “one” to “two”, “three” and so forth that the respective signs for TWO MONTHS, THREE MONTHS etc., are created without modifying their “monosyllabic” status. Frishberg and Gough (1973) offer the case of WEEK, a similar example in ASL. “The sign for TWO-WEEKS is considered a single word. It is made with the motion, location and orientation of the sign WEEK, but the hand configuration changes to the number TWO.”8 It can be changed to three and so forth to obtain the respective signs for THREE WEEKS, FOUR WEEKS etc. The same “with the signs HOUR and MONTH then, the numerals from ONE to NINE (except 5) are incorporated freely.” This numeral incorporation process turns out to be quite extensive in conventional SLs such as CSL and ASL.9 Frishberg and Gough (1973) further point out that “time words (in ASL) can be inflected by processes of incorporation and compounding ... These two processes are really part of the same process, namely condensing the maximum amount of semantic information into the smallest possible articulatory time, using separate parameters of the articulation to express simultaneously semantically distinct material.” On the syntactical level, the role of simultaneity is much more significant than in lexical formation, and this significance will be witnessed in sign sequences discussed in this paper. As expected, this effect of simultaneity is achieved with the help of a range of bodily visual articulators, notably the hands and the upper part of the body. Here is a simple example for illustration: an event demanding the presence of two separate entities such as “someone entering a house” can be realized simultaneously, with one hand standing for HOUSE (partially represented by the roof ) while the other acts as a human being walking into it (i.e., under the roof ). In the case of an event where Entity A (for example a dog) executes an action on Entity B (a cat, such as in ‘A dog chases a cat’), all three constituents involved in the event can be maintained visually in the signing space, with the action embedded in the hand representing Entity A directed towards Entity B kept located in space by the other hand. In a more complex linguistic situation, for instance, if the example just given is supposed to be made in “direct speech”, a signer can perform the same sequence, while at the same time turning his or her body slightly to one side, thus signaling that it is not the signer who is “talking”, but is only reporting someone’s speech.10 Of all the properties of SL, iconicity is probably best known to the general public as its hallmark. But this impression of its iconicity is only partly true. Iconicity is basically a launching device for SL, mainly to get the lexical and syntactical creation mechanism started.11 In the case of lexicon, once an iconic sign becomes established, it is subject to modification, and with time, its iconicity will be eroded, sometimes to the extent that if its iconic origin is not promptly documented, its etymology, though relatively recent, will remain unknown. Thus, what was once a transparent sign would become translucent, and even further weaken to opaque.12 Moreover, we should not have too many illusions about iconic hints in deciphering

146

S. Yau

signs. Occasionally, the iconic origin of a sign can be guessed in its citation form, but once it is enrolled in discourse by a fluent signer, it would be out of the question for a non-signer to guess its meaning. Space nevertheless provides SL with possibilities to iconically reflect certain aspects of the situation during a discourse. There are verbs which require a modification of their location according to the height of the addressee in the context. For example, according to Liddell (1995), the sign ASK-QUESTION in ASL (American Sign Language) “is directed toward the chin of the person described as being asked, but the sign GIVE is directed toward the recipient’s chest. Each individual indicating verb has a specified height on the body toward which it will be directed, ranging from abdomen level to forehead level”. According to recent observations, I find the same constraint is also present in CSL. When signing, or imagining signing, to a tall person like the basketball player Yao Ming, gestures must be raised to an appropriate height. Furthermore, I was also informed that in CSL, when referring to someone who is an important personality, for example the Party leader, the President or the Prime Minister, it is also necessary to raise the height of the gestures so as to show respect even if they are all physically short!13 But it is iconicity at the syntactic level that is my main concern. With the use of both hands in signing, an iconic schema for a basic linguistic situation involving two entities, namely S, O, and V (N, N’, R in my symbols), can be set up, and this schema is normally comprehensible to an interlocutor, even though the identification of the S, O, and V sometimes requires a shared knowledge of the cultural background between the signer and the addressee. I therefore consider the use of this iconic device as crucial in improvising the basic syntactic structure of the medium, just as its role in creating basic signs. I will return to this point in section 5 below. Despite the important contribution of iconicity to this visuo-manual medium, many studies published in the past decades have tried to minimize its role in SL.14 Several sign linguists provide their testimony, for example, Armstrong (2002) quotes Wilcox (1996) as saying “Although the existence of iconicity was noted, it is arguable that most linguists were more interested in explaining it away than in treating it as fundamental.” Or, as Boyes-Braem’s comments in her paper (1986): “Sometimes people who want to disparage SL will call it an ‘iconic’ language and intend to imply by this a language which is only capable of referring to ideas which are ‘easy to image’, ‘concrete’ or simply put, ‘primitive’.”15 3. The Range of Nonmanual Features in SL Although SL is primarily known as manual, there are a wide range of supporting nonmanual features to enrich its expressiveness. These features can be divided into two categories according to the parts of the body where they are produced: facial expressions and body movements.16 In SL, we often observe facial expressions such as raised brows, eye gaze, eye blink, or squinted eyes, etc. Many of these features are shared by OLs which also use them to express various moods, such as raised brows or eye blink in questioning, expressing doubts or surprise; knitting brows for worry or thoughtfulness; shaking the head together with an appropriate facial expression for negation.17 These

The Role of Visual Space in Sign Language Development

147

facial expressions are common to both deaf and hearing people belonging to the same cultural tradition, and perhaps to all humans. However, what distinguishes the use of some of these visual signals in SL from those in OL is their high frequency of occurrence and the specific syntactic functions assigned to them in some cases in SL. The question of these being assigned specific functions will be dealt with when the second category of nonmanual features, body movements, are involved (see section 8 “Mechanism in conventional SL). Let’s take the facial feature accompaniment of the sign RELIEVED in ASL for example. Liddell (2004: 13) describes this sign’s formation as follows: “To properly produce RELIEVED, not only must the hand move as illustrated, but the lips must be rounded and pursed throughout the sign. In addition, the signer blows out a puff of air as the hand moves downward. The form of the nonmanual signal associated with RELIEVED may be related to the common English expression, ‘phew!’.”18 A similar facial expression together with a relaxing bodily movement signaling “relief ” is also observed in CSL, and perhaps among Chinese speakers as well when they utter this word emphatically in Chinese.19 According to Poulin et al. (1995), in signed sentences involving referential shift in LSQ (Langue des Signes Québécoise), “the breaking of eye contact with the addressee is the most consistent change to indicate that the signer has entered a referential shift. However, the breaking of eye contact is not sufficient alone to specify referential shifting. It must be accompanied with a change in facial expression and/or body position.” Likewise, in DSL (Danish Sign Language), according to Engberg-Pedersen (1993: 102–3), the signer only has to change gaze direction and/or change their facial expression to shift to “reported speech”. There are some other data about facial features in DSL. Again according to Engberg-Pedersen (1993: 23; 42), eye gaze is employed for directing the addressee’s attention to a reference point; squinted eyes serve as an appeal to check the addressee’s understanding of a reference; and eye blink is one of the most frequent indicators of a major constituent boundary.20 Jouison makes the following claim in a highly technical paper (1986): “Although facial expression and gaze direction do improve the analysis of signed discourse, if paid close attention to, only the observation of synchronized or coordinated body movements can complete the picture.” He further adds that “body movements not only constitute the basis of the structural coherence of FSL (French Sign Language) sentences but also determine the order of the constituents by their influence upon the levels of articulation. ... The levels of articulation are hierarchically organized, and include the sign parameters, shape, location, movement and orientation, next their morphological features, and finally the body movements at the highest, syntagmatic level.”21 In sum, I think many of the nonmanual phenomenon in SL should be considered as basically visual rather than visuo-spatial.22 All the nonmanual features mentioned so far are located on the signer’s body, from the trunk to the head. These features, though produced on the signer’s body, do not really require any extra signing space to perform. If we consider these features as spatial, should we also consider facial expressions that accompany speech as spatial features too? The

148

S. Yau

nonmanual devices in SL thus principally take advantage of the visual aspect rather than its spatial setting. 4. The Salient Visuo-Spatial Parameter in SL: Hand Movement Hand movement makes major contributions to the functioning of SL at both the lexical and the syntactic levels. At the lexical level, hand movement is the key parameter in fostering what I call lexical branching or ramification, that is, deriving a series of cognate lexical items solely by means of modulating a root sign, especially its movement. SLs are rich in this kind of branching. The pairs PUBLIC, gōng in Chinese, and CONFISCATE chōnggōng, and the pair MAN rén and PEOPLE rénmín in CSL are the two cases in point. They are called character-signs in CSL, and the forms of these two pairs are: PUBLIC: a COIN-hand (a fist with its thumb and index finger forming a circle) placed inside the arch of the other hand in an L shape (fist with thumb and index finger widely stretched, palm side facing downward), imitating the character gōng; CONFISCATE: the character-sign PUBLIC gōng with the hand of the small “circle” pivoting downward, signifying forfeited (coin or money) to the public revenue. MAN (rén as in Zhōngguó-rén, “the Chinese”) is an imitation of the character rén. The form of the sign is: both index fingers pointing up obliquely (all other fingers closed), the tip of one against the second joint of the other.23 When this character-sign is made with a horizontal circular movement, it signifies “there are people all around” or “people, the masses”. (Yau 1987) Klima and Bellugi (1979: 31) also provide us with an interesting example from ASL, showing how cognate signs are derived by modulating the movement of the root form. The sign in question is YEAR: two fists, both thumb sides upward, one on top of the other, the one on top rotates once around the one below, and back to its original position, suggesting the earth revolving around the sun once a year. To obtain the sign EVERY YEAR, the active hand (that is the fist on top in YEAR) “brushes forward on the base hand repeatedly.” To modulate the sign YEAR to mean FOR YEARS AND YEARS: “the active hand moves above the base hand in a circle.” The paper by Frishberg and Gough (1973) is in fact the first study to claim that modulation of sign movement is a very productive lexical creation process, and it furnishes us with several illustrations. “The sign A-FEW-WEEKS-PASS-BY repeats the movement for WEEK four times, on each repetition uses a different finger (from index to pinky of the base hand) to serve as the location or base of the sign.” Frishberg and Gough mention two kinds of movement reduplication, the fast type and the slow type. Fast reduplication expresses habitual and slow reduplication indicates continuous action. For example, the sign WEEK, if the movement is changed “by repeating that same movement several times quickly we have also changed the meaning to WEEKLY. In the same way MONTH changes to

The Role of Visual Space in Sign Language Development

149

MONTHLY, DAY to DAYTIMES, that is, happening habitually in the daytime, TOMORROW to EVERY-DAY, YEAR to ANNUALLY.” On the other hand, “slow reduplication indicates continuous action. It has been defined not only by the relative slowness of the repeated movement but also by the wide circular path taken between iterations. WEEK plus slow reduplication yields FOR-WEEKS-ANDWEEKS, MONTH becomes FOR-MONTHS-AND-MONTHS, and so on.” Likewise, for expressing durative, for example, ALL-DAY, ALL-WEEK, ALL-MONTH, ALL-YEARS, the root signs have only to be “made once, but particularly slow.” Furthermore, diachronic sign modifications are mainly carried out through movement in combination with the location parameter. The old form of BEIJING in CSL was once made in the following manner: H-hand (index and middle fingers extended), side-pointing, palm inward, with indicating finger tips draw a diagonal line across the chest from the opposite shoulder down to the waist. It is an allusion to the sash worn by state officials at the beginning of last century soon after the fall of the Qing dynasty.24 But now its movement is horizontal, and is shortened to a line of a few cm drawn with the tips of the index and middle finger on the chest slightly under the shoulder. In cases like the sign BEIJING, there is only modification in form but not in meaning. Klima and Bellugi (1979: 71) also propose some old forms from ASL, showing changes in movement or the location of movement. The sign BORING recorded in 1918 has the forefinger pressing against the end of the nose, and the head bending forward. “Today the forefinger touches the end of the nose, and a turn of the wrist has been substituted for the head movement.” Another example is THINKING. “In Old FSL, an index finger moves in a small circle near the forehead and the head moved in a circular motion. Today the index finger moves in a small circle near the forehead, the head itself remaining still.” In studying the difference between some current ASL signs and their previous forms, Frishberg (1975) finds cases which previously required facial expression, environmental contact, or body movement, now only require movements of the hands. I think she has made an important point by viewing this shift in articulators as a historical tendency in SL to concentrate lexical information in the hands of individual signs in their citation forms. This tendency is interpreted by her as part of a general shift in ASL from iconicity to arbitrariness in signs. The following example is typical: “COMPARE is one of the few signs for which we have information from Old FSL, as well as from Long and Stokoe, et al. The original form of this sign had two flat hands facing the signer, separated. The eyes moved from one to the other, and then the hands moved together, eyes focused on both at once. Long describes an intermediate stage in which the hands have begun moving ‘inward and up before you side by side as if looking at them and comparing palms.’ The modern form simply rocks the two hands, either in alternation (which can be related to the older eye movements) or in unison – an arbitrary, but symmetrical change.” Despite this prominent role in sign modification, hand movement was considered in the nineteen sixties and seventies only as an equal among the several formational parameters with no particular distinction. During that period, for Stokoe

150

S. Yau

and many other sign researchers, all parameters were basically attributed with the same status in sign formation. This traditional viewpoint was revised with the publication of Liddell’s papers in the early eighties. Liddell (1984) proposed that signs consist of phonological segments divided into movements and holds. Each movement (M) or hold (H) segment contains handshape, location, orientation, and nonmanual features.25 He further adds: “This paper examines the nature of the sublexical structure of signs in American Sign Language ... the currently accepted view of a sign as a simultaneous bundle of primes – handshape, movement, location and orientation – is insufficient in several ways. Timing data are presented which show that the majority of signs are segmentable into movements and holds. ... This framework allows an interesting parallel with spoken language phonology: both divide the sequential units which make up the words of the language into major types of units. In spoken language, the units are consonants and vowels; in sign language, they are holds and movements.” Around that period, according to Boyes-Braem (1986), other researchers were also actively working on this particular parameter of SL. Grosjean (1981) found that “one actually perceives the sign parameter by parameter, beginning with the handshape, and location. However, it isn’t until one perceives the movement, that one can identify the sign.”26 In an experiment conducted by Poizner et al. (1981) it is shown that “signers were able to identify signs by their movements alone.”27 As hand movement has its intrinsic directionality, consequently, at the syntactic level, this parameter is the principal device for establishing relationships between entities set up during a discourse, and thus allows SL to exploit the visual space to the maximum. For instance, the movement of a verb of action helps distribute the syntactic roles of the nominals as N agent, N’ patient or Nloc in a basic sequence in SL. I leave this important issue to section 6 for further discussion. In sum, hand movement at the lexical level in SL is comparable to the vibration of the larynx in speech. Equally important is that without this movement parameter at the syntactic level, SL would lose all of its dynamic, and be deprived of its means of establishing syntactical relationships between entities in space. 5. Element Orderings in SL: Graphical Presentation and Theoretical Implication Several scholars in the past had faithfully recorded the basic sign order performed by deaf persons before their gestures became conventional. In the nineteenth century, Rémi Valade, the brilliant teacher at the Institution Impériale des Sourds-Muets de Paris, made a vivid description of a consistent natural sign order by deaf people in his Etudes sur la lexicologie et la grammaire du langage naturel des signes: “ ‘The cat watches the mouse’. If we ask a deaf-mute to translate this sentence by signs, we will see that he places the mouse and the cat at a small distance from each other, then he himself substitutes for the cat and makes the sign ‘watch’ in the direction of the mouse. The order of succession will then be: ‘Mouse cat watches’.” (My translation) Around the same period but sometimes after Valade’s publication in 1854, the British anthropologist Edmund Tylor also observed a similar natural order in deaf people’s signing: object - subject - action.28

The Role of Visual Space in Sign Language Development

151

In the late nineteen seventies, I followed the steps of these pioneers and studied the basic element ordering in gestural sequences provided by several isolated deaf signers, and compared them with those in conventional SLs such as CSL, JSL (Japanese Sign Language), ASL and FSL (Yau 1977a, 1979 and 1992). My findings confirm the natural sign order observed by Valade and Tylor.29 The latest study on sign order performed by an isolated deaf community was conducted collectively by Sandler et al. in 2005.30 Their report contains a careful linguistic description of Al-Sayyid Bedouin Sign Language created spontaneously about 70 years ago, without any apparent external influence in a stable existing community in the Negev region in Israel. Sandler et al. employ a balanced methodology which includes prosodic criteria and the consultant’s translation in their analysis of its syntactic structure. Here are two passages from their report: “A systematic way of marking the grammatical relations among those elements has appeared in the form of highly regular word order”, and “the vast majority of sentences in our data were unambiguous, and straightforwardly (S) O–V.” “These systematic structures cannot be attributed to influence from other languages, because the particular word orders that appear in Al-Sayyid Bedouin Sign Language differ from those found both in the ambient spoken languages in the community and in the other sign language found predominantly in the surrounding area.” Despite this series of evidence, it seems that, for one reason or another, many others are still not convinced of the validity of this natural element ordering. One of the reasons, as Wilbur (1979) suggests, is that “it is well known that the interpretations of data have been widely different among several research reports. This has led to what became termed ‘the sign order controversy’ “. Loncke et al. (1986) suppose that “a main problem is the question if one should rely on theoretical frameworks aiming respectively at syntactic (e.g., SVO, SOV, ...) or at semantic structure (e.g., Topic – Comment, Given – New), while pragmatically inspired analytical methods may seem feasible too.” (Loncke et al. 1986) 31 As far as the rudimentary syntactic development is concerned, I think Armstrong et al. (1995: 20) have asked the right question: “the crucial problem is to find how sentence forming came about in the evolutionary history of our species. How might syntax have begun?” Later in their book, they add: “The importance of the early use of visible actions to signify and communicate is that they have, by their very nature, not only the potential to represent things or events, but they also have both the elements and the order – the structure of syntax – built into them.” (p. 181) Their viewpoint can be illustrated with this example they provided (p. 100): GIRL INDEXa, BOY INDEXb, aHITb ‘The girl hit the boy.’ They then remark that “It is also possible to create an English sentence with exactly the same structure, including the spatial indexing (pointing):” “The girl was here (points to the left). The boy was there (points to the right). She hit him.” (p. 102)32 The use of SOV sign order is reported in SLs in different cultural and linguistic areas. For SLs in Asia, Nakanishi (1994) observes that in JSL (Japanese Sign Language), the preferred sign order is SOV, but OSV also appears frequently.33

152

S. Yau

Chang et al. (2005) confirm that TSL (Taiwan Sign Language) has SOV as its dominant order, as in: “(My) brother likes dogs”, which is signed as DOG BROTHER LIKE; and likewise in various other CSL dialects.34 Today in Europe, although many SLs are claimed to have SVO as their main sign order, SOV order is nevertheless often observable. While discussing the ways of encoding roles in BSL (British Sign Language), Kyle and Woll (1985: 141) give an example of the same pattern as the one above by Armstrong et al. (1995): “A sentence such as WOMANHIT-MAN may be signed in the following way. First, MAN is signed with the left hand. Then the index finger of the left hand is held upright (person classifier) in fourth person location. As the left hand is held stationary, the right hand signs WOMAN. Then the right hand, in a fist, is brought towards the left.”35 A very convincing case supporting SOV as the natural sign order comes from a recent study on LSA (Argentine Sign Language) conducted by Masone et al. (2004). “The deaf community in Argentina comprises a nonliterate group with few contacts with written or spoken Spanish; thus the main contact situation is deaf/deaf, and LSA is used within a Spanish speaking community. As we know, the word order of Spanish is SVO.” Their data show that “the basic, unmarked sign order with plain verbs in LSA is SOV for a sentence with a transitive predicate, and a change in sign order produces a different, unacceptable, and ungrammatical sentence”. Their examples are: DOG INSECT FOLLOW “The dog follows the insect.”; or BOY GLASS WINDOW THERE BREAK (perf.) “The boy broke the glass of the window.” They conclude that the analysis of their large corpus shows that “LSA has a canonical sign order, SOV”. Thus it can be claimed that in present day conventional SLs, there is not a single one which does not allow to a certain extent the use of SOV order. Such a peculiar universal linguistic situation cannot be imagined in OL, and can only be explained by the particular modality of SL. 6. The Two Registers of the Basic Element Ordering and Language Universals The modality of SL is spatiotemporal. Its ordering, therefore, ought to be represented on two registers, the temporal and the spatial. Truly speaking, it would cause distortion if the traditional schema such as the one by Greenberg (1963) for oral languages is adopted for SL representation without discrimination. When a deaf person signs a basic declarative sequence like MOUSE BITE DOG, they pose the sign DOG (or MOUSE) on one side and the sign MOUSE (or DOG) on the other, and then perform the action of the relator R (my symbol for a transitive, senorimotoric, intrinsic-directional verb) in between the two N (nominal entities), or more exactly the two points previously marked in the signing space by the two N. The N from which the action R is launched will be assigned as N agent (S), whereas the N which the R attains will be labeled as N’ patient (O). By order of appearance of these three constituents, we obtain NN’R (SOV). In certain SLs, for example ASL or FSL (according to my deaf informants of these two SLs in the nineteen seventies), there is a tendency to sign N’ before N, and therefore the order will be N’NR (OSV) on the temporal register. Anyway, in these SLs, the N and the N’ precede the R (V). This temporal bipartition is important in that N and N’ must be

The Role of Visual Space in Sign Language Development

153

executed temporally before R (V) so that the orientation of the action of R (V) can be determined. As a result, the grammatical functions of the two N will be assigned, i.e., which is to be N agent and which is to be N’ patient. However, in terms of their spatial positioning and from a static point of view, the R (V) is in fact performed between the two points previously marked by the two N. Therefore on the spatial register, the three signs, for example performed by a right-handed Chinese deaf, will be presented as NRN’ (SVO) or N’RN (OVS) on the spatial register from the observer’s viewpoint (if by a left-handed Chinese deaf, these spatial orders will be reversed).36 What is significant is that, in contrast with the temporal register, the R (V) on the spatial register, whether performed by a righthanded or left-handed deaf signer, using one of the SLs mentioned here, always lies in between the two N and that the N (agent) will be marked by the fact of its being the starting point of the R (V), and N’ (patient) by its being the terminal point of the R:

DOG (N) BITE (R) > CAT (N’) “A dog is biting a cat.”

DOG (N’) < BITE (R) CAT (N) “A cat is biting a dog.”

Graphic representation of the 3 elements, N, N’ and R on the 2 registers, T (temporal) and S (spatial)

(as for example in CSL)

(as for example in ASL as recorded in 1975)

I must admit that though this observation was made independently, the credit should first go to Valade. He first took note of this important and subtle characteristic in FSL in his book published in 1854, in which he wrote: “‘The cat watches the mouse’. ... If we ask a deaf-mute to translate ... The order of succession will be: ‘Mouse cat watches’. But if we consider the order in which the signs are posited, we will know that this is exactly the word order in the French sentence ‘cat watches mouse’” (My translation)37 Now I argue that the two orders, temporal (N and N’ executed before R) and spatial (R positioned between N and N’), are governed by cognito-perceptual constraints, and that they are universal among all SLs for this type of sequence before modifications due to conventional influences. I hope that these proposed constraints

154

S. Yau

will provide this hypothesis with some explanatory effect. See Yau 1977a, 1979, and 1992. Furthermore, I postulate that there is a link between the oral and the gestural systems. OL in the course of its existence might also be subject to the same extralinguistic constraints known to the basic sign order of the deaf. To my knowledge there is no evidence that there is any difference between the visual perception of a deaf and a hearing person towards an image, for instance, when they look at pictures showing a dog biting a cat. The existence of NRN’ (SVO) side by side with NN’R (SOV) as dominant orders in OL can reasonably be interpreted as the natural outcome of the application of these constraints. OLs are linear by nature; the vocal realization of the elements in various word orders, including NRN’ (SVO), is basically temporal. However, from a static point of view, NRN’ (SVO) in OL can be regarded as an order modeled on the positioning of the elements in space in gestural languages, and NN’R (SOV) in both oral and gestural languages as following the temporal order of appearance of the elements in the development of a visual event.38 Thus, paradoxically, when we say that the basic order of an SL is SOV, in fact, that is its order on the temporal scale. If we want to be coherent with the nature of this medium, we should say the basic order of this SL is SVO (spatial). On the other hand, when we say the word order in spoken Chinese, English or French is SVO, we are talking about an order which mirrors the ordering of the three elements of a gestural sequence on the spatial register. In view of these observations among deaf signers of different cultural areas and in different periods, and supported by orderings obtained in child language in drawing and in composition of pictographic writing39, I conclude that the element orderings NN’R or N’NR (temporal register) or NRN’ (spatial register) should be considered as linguistic universals. 7. A General Shift of Sign Order in ASL and European SLs, and Locatives as the Last Stronghold of Visuo-Spatial Constraint Fischer (1975) states that “What seems to have changed in ASL, in the last hundred years is, most importantly, that the object, which previously preceded the verb, now follows it ... the postposition is now generally a preposition, the adjective can precede or follow the noun, where before it could only follow, and the adverb precedes an adjective (as in VERY HUNGRY, where 100 years ago it was HUNGRY VERY).”40 She considers that “All of these changes can be viewed as occurring in one direction, namely, toward a closer approximation to, if not English grammar, at least English word order.” Finally she suggests that “there has been a great deal of pressure from English to conform to the syntax of the spoken language of the community. At the same time, however, there are countervailing pressures arising from the fact that ASL is a sign language ... These two pressures have interacted to produce a language which is still different from English, but conforms more to it than it used to.”41 Since then several researchers working on ASL and European SLs have agreed that SVO is the dominant order in their respective SLs. However despite this shift claimed by these researchers, there is at least one syntactic structure that resists following suit. Signed sequences with an N-locative

The Role of Visual Space in Sign Language Development

155

still appear before the N-agentive and the verb in a signed sequence on the temporal register. For example, Liddell (1980) who strongly argues for SVO as ASL dominant order admits that locatives have a normal order of OSV, for example, the sentence “The cat is (sitting) on the fence” will be signed as FENCE CAT V-CL ‘sit on’. Before we continue discussing this question, I would like to draw a distinction between the use of the conventional locatives and the context-dependent locatives in SL. By conventional locatives, I refer to those locatives indicating the cardinal points, or other directional notions such as “right”, “left”, “front”, “back”, “above”, “below”, etc. They consist of a closed list, with relatively defined positions. These conventional locatives also include those points designating “present”, “future”, and “past” along the imaginary deictic timeline.42 The context-dependent locatives, on the other hand, constitute an open list, and their positions in the signing space are entirely free.43 Our discussion here is more concerned with these context-dependent locatives, and henceforth they will be referred to simply as locatives. Locatives are syntactic constituents which are most akin to space in SL, and they are particularly sensitive to visuo-spatial constraints. Subsequently, their order of appearance in a signed sequence is highly restricted. Even in SLs such as ASL or ISL (Italian Sign Language; see Volterra et al. 1984) where the basic element ordering has apparently been subject to radical modifications, the order of appearance of the locatives in a sequence remains the same as in SLs with SOV as basic order.44 Their positional stability is explained by the strength of the cognitively based visuo-spatial constraints latent in the locative constituents, rejecting any violation of the locative’s natural positioning.45 Examples with locatives preceding the subjects and predicates from SL are abundant. I will only choose a few for illustration. In HKSL (Hong Kong Sign Language), “A man enters a house”, “A child rolls across a bed”, “A man squats on the box” or “A bird perches on the tree”, the respective locatives “house”, “bed” , “box” and “tree” all appear at the very beginning of the sequences. (Tang 2003) Likewise, in TSL, “The child ran into the house”, the locative “house” is at the head of the sequence. (Chang et al. 2005) In ASL, I find the same phenomenon, as in this example given by Baker et al. (1980): “The sequence ‘Five pencils in a row on the floor’ is rendered as ‘FLOOR, PENCIL, FIVE’, with the locative ‘FLOOR’ in the first position.” 46 Recent SL studies have adopted the terms ground and figure to represent the two nominal entities locatively related, and it is universally admitted that there is a general tendency for ground to precede figure. For example, Emmorey (1996) remarks: “The signers almost invariably expressed the ground first, and then locate the figure with respect to the ground object. This ordering may be an effect of the visual-spatial modality of sign language.” See also Tang 2003, and many others. In order to cover a wide range of coordinated pairs like figure and ground the order of which is governed by the cognito-perceptual constraints, in Yau (1983) I propose an overall paradigm of coordination: operand – operator, specified – specifier, base – appendix, focus – peripheral, topic – comment, support – supported, container – content, location – event, orientation – movement, cause – effect, means – action,

156

S. Yau

action - result, etc. When applied to natural gestural sequences, the first of the pair normally appears before its counterpart. Within this paradigm of coordination, there are the rules of mobility and size. McIntire (1982) in her long article on locatives presents a vivid and convincing description of her findings on this point: “Notice that the ordering is the mirror image of the English phrase that was used as the elicitation sentence. Although in English we have several choices as to the ordering of such phrases (e.g., the bowl with the apples in it on the table; the table with the bowl of apples on it; the apples on the table in the bowl, etc.) the ordering that the nominals are mentioned in (TABLE BOWL APPLE) is the only acceptable order for this consultant. Such a restriction on acceptability gives support to Yau’s notion that strings of nominals will follow a descending order of size and mobility.” McIntire’s example demonstrates that no matter how strong the pressure of spoken English is, it does not affect the natural order of the locatives in ASL. 8. Mechanisms in Conventional SL for Further Syntactic Development Towards the end of the nineteen seventies, there was a hot debate on the existence of subordinate structures in ASL between Liddell and Thompson, together with Coulter’s cooler but equally insightful participation. Following the chronology of the publication of articles related to this debate, it seems that Liddell (1977) first describes an intonation contour, and claims that it signals relative clauses in ASL. In the same year Thompson (1977) challenges Liddell’s claim. While admitting Liddell’s paper convincingly demonstrates the existence of a previous unnoticed contour as well as its relevance to questions of ASL grammar, he insists that this contour “does not contain any convincing evidence that relative clauses exist as a syntactic device in ASL, or that this contour serves to mark them.” At the same time, Coulter joins in with his article in 1977. Liddell then defends his position in 1978, arguing that “without these nonmanual features, these structures (with the exception of those with the relative conjunction THATa) would be indistinguishable from corresponding ASL sentences.”47 Here are two of Liddell’s classical examples: ________________________________r [REMEMBER CAT DOG BITE]S ]NP RUN AWAY “(Remember) the cat the dog bit, (it) ran away.” _______________________________r RECENTLY DOG CHASE CAT COME HOME (The dog that recently chased the cat came home.) (The cat that the dog recently chased came home.)48 Thompson’s criticism also touches on another sensitive point in Liddell’s claim, the source of his data: “Thus most of the hard evidence comes from a much more artificial source: an informant’s translations into ASL of English containing relative clauses” and “none occurred in discourse.” In answering this interrogation, Liddell writes: “Sentences containing ASL relative clauses had been elicited by asking native signers to translate into ASL a list of English sentences containing

The Role of Visual Space in Sign Language Development

157

restrictive clauses. To minimize the possibility that the translation might be somehow biased by the fact that the English sentences contained relative clauses, the informants were not asked for on-the-spot translations but were given a few days.”49 Thompson probably has reasons to doubt the validity of these elicited data. Baker et al. (1980: 163) report that “[the] restrictive relative clause does not seem to be used very often by ASL signers and, in fact, appears only once in all the examples collected for their text.”50 They also admit that “without a much larger corpus and some method for identifying potential examples that is not totally subjective, it is probably wise to refrain from speculation.” As Liddell (1976) himself points out, many of the sequences he treats as relative clauses are sometimes introduced at the beginning of the sequence by verbs such as REMEMBER, KNOW, and SEE, and relative clauses in ASL seem to pose questions to the addressees; Coulter (1977) therefore asks “Whether or not these verbs are present in the relative clause, the clause ‘invites’ the listener to identify a particular NP described by the clause, about which some comment will then be made.” He suspects, I think with reason, “that ASL relative clauses are historically derived from topic constructions, and that this is what accounts for the raised eyebrows, the ordering constraint, and the ‘question’ or ‘topic’ meaning which is associated with relative clauses.”51 I will finish this section with a few comments of my own. My impression is that the so-called restrictive relative clause in ASL is still in the course of its evolution. There is no doubt that some native signers in ASL are attempting to devise a structure to represent what is expressed in English restrictive relative clauses. At the moment, however, this structure as described by Liddell relies too much on facial expressions, and probably for this reason, sometimes results in the creation of too many semantic ambiguities. In a fairly straightforward sequence given in Liddell (1978), the author tells us that there are three possible interpretations: _______________________________________r [CAT WATCH [DOG EAT HAMBURGHER] ] ‘ME’ BUY “I bought the cat that watched the dog eat the hamburger.” “I bought the dog that the cat watched eat the hamburger.” “I bought the hamburger that the cat watched the dog eat.” (Liddell 1978) Three possible readings in a relative simple sequence with no anaphora or other complicating factors is astonishing. I suspect that this is due to its exclusive use of facial or nonmanual features for marking the clausal constituents. This amplitude of ambiguity raises the question of whether facial features are over-exploited at the expense of manual signs, consequently overloading the visual system. It might be expected that the function of these facial features would later be substituted by segmental markers, thus allowing a better equilibrium of the manual medium, and with it, the number of ambiguities perhaps could be reduced to a reasonable level, and henceforth interpretation of the restrictive relative clause would also be less dependent on contextual hints. In a word, I am convinced that ASL is now in search

158

S. Yau

of devices to express its subordinate structures, but the task is not yet accomplished, pending segmental markers to fill the gaps.52 My conviction is reinforced by the cases of grammaticalization reported in SL. For instance, Armstrong (2002: 168–69) says that “Wilcox has shown that the ASL sign GIVE has become an ASL discourse marker through metaphorical extension and a path of lexicalization – GIVE-concede, meaning, roughly, ‘I give up, you win the argument.’“ More significant is the conclusion drawn by Frishberg (1975) regarding shifts in articulators. She finds, as I have already cited in section 4, that “signs (in ASL) that previously required facial expression, environmental contact, or body movement, now only require movements of the hands.” I think this historical tendency of shifting nonmanual articulators to manual ones will happen sooner or later to consolidate the restrictive relative structure in ASL. In the early nineteen seventies, ASL researchers adopted a double strategy in the defense of their visuo-manually based medium as a fully legitimate language. As far as syntax is concerned, some of them were interested in looking for certain complex syntactic structures that until then were still considered to be non-existent in ASL. Therefore, the issue of this debate, as Thompson (1977) points out, is basically “the issue of polysentential coordination versus relativisation” There might be two different motives behind this research topic. Liddell was right to reach out for new scopes of analysis, and Thompson also had the right to argue that SL can be well satisfied with a polysentential coordination which is equally efficient. Recently, there has also been an attempt to prove that there is passive voice in ASL. ASL has predominantly been described by several authors as having only active voice, and Stokoe et al. (1965) were among the first to propose that passive constructions are not found in ASL. Isenhath (1990) reiterates this viewpoint and concludes that no English-type passive structure is found in ASL: “ASL verbs do not have voice” (1990: 39); Armstrong et al. (1995) share the same conclusion. Saeed and Leeson (1999) also deny the existence of such a construction in Irish SL. Despite these incessant confirmations of its inexistence, a few researchers, mainly Wilbur (1987), Kegl (1990) and more recently Janzen et al. (2001) maintain that a passive form may exist in SL. 9. The Case of the Syntactical Comparative Marker in CSL Spoken languages on the whole have already been well equipped since historical times with syntactical structures; for instance most of the structures of present day Chinese find their origins in the oracle-bone inscriptions of the 13th century B.C., though certain grammatical categories were developed later to attain a more satisfactory performance, as in the case of the passive voice, to provide an alternative to the Chinese passive bèi-structure which inherits a potential “victimlike” connotation. It is therefore fairly normal that SLs with only two or three centuries of history behind them manifest some syntactic gaps in their systems when compared with OL, or gaps only recently bridged. The first case that comes to my mind is the comparative schema in CSL.

The Role of Visual Space in Sign Language Development

159

In the course of the preparation of this paper, I was surprised to learn that the comparative structure in SL has rarely been studied. For instance, in reference books on ASL syntactic structures like Baker et al. (1980) American Sign Language, a Teacher’s Resource Text on Grammar and Culture or Liddell (2004) Grammar, Gesture, and Meaning in American Sign Language, or Moody (1983) Introduction à l’histoire et à la grammaire de la langue des signes on FSL, there is no mention at all of this structure. When I consulted colleagues, their replies confirmed the general absence of its treatment in SL studies.53 As a rule, only the following comparative schema is observed in sign systems created by isolated deaf individuals: A TALL ( on the one side), B SHORT (on the other side) for “A is taller than B”, or likewise A BEATIFUL, B NO (or UGLY) for “A is more beautiful than B”, or A CHILD MANY, B MANY NO “A has more children than B.” They can only compare by means of antonyms in juxtaposition, and they fail to bring out the semantic difference between A is more beautiful than B and A is beautiful, B is not beautiful or ugly (B is not necessarily ugly); or between A has more children than B and A has many children and B has few (yet the number of B’s children might still be considered as many). They sometimes have to rely on the exact numbers of children of A and B, such as A has 5 children and B has 3. Surely there is no point to solicit more complicated comparative schema like She is more intelligent than beautiful. Although these isolated deaf signers have the cognitive capacity to understand this kind of complex situation (for example when I stage a “scene” with concrete objects for them to compare), they fail to invent a comparative marker for constructing a schema. An episode in the Amerindian reservation also proves that my isolated deaf informant understood the kind of comparison I was looking for: before I left the reservation, I taught my deaf Cree informant the CSL comparative schema, minutes later he used it in front of me.54 In the mid nineteen seventies, I recorded two different comparative sequences in China: (1) in signed Chinese obtained in Beijing, and the other (2) signed in the Shanghai dialect of CSL: (1) WRITE D (for particle de) COMPETITION YOU GOOD. “I write better than you.” (literal translation of the Chinese sentence: wǒ xiě de bǐ nǐ hǎo.) (2) WRITE COMPARATIVE MARKER (pointing to the person in question) 55 Though the comparative marker in (2) is not found in the latest version of Zhōngguó shǒuyǔ “The Chinese Sign Language” (2003) with 5587 lexical entries, it has been generally used by deaf adults all over China well before 1949. The schema with this native comparative marker in its complete form is: N - verb(adv)/adj - N’ - bǐ ‘than” (+ directionality). Thus the sentence A is taller than B in Chinese will be rendered in this CSL schema as “A TALL (gāo) (on one side), B comparative marker bǐ “than” on the other side, with the bǐ -hand pointing from A towards B.56 Though the derivational history of this comparative marker has yet to be traced, and that it was invented under the stimulation of spoken or written Chinese, it

160

S. Yau

nevertheless shows that it was created internally through a process of grammaticalization. According to my knowledge, this comparative marker in CSL is the first case of its kind to be internally derived, without morphologically or syntactically copying its counterpart in the ambient spoken and written language. Despite the fact that these grammaticalizations are rare in SL, their presence is very significant because they illustrate how SL relies on internal sources to adapt to its increasingly demanding linguistic environment. 10. Concluding Remarks SL offers those working on OL a rare chance to witness the evolution of a linguistic medium of completely different modality. It also shows how SL has rapidly developed into a full-fledged system in just a few centuries by exploiting signing space and various visual articulators. Though generally subject to strong pressure from its ambient spoken counterpart, it evolves according to its own mechanisms, improvises its own devices, and does not follow the path of the creole type. It retains the prominent characteristics of its visuo-spatial modality the whole time. The aim of this paper is to demonstrate how SL takes the maximum advantage of its spatial, manual and bodily features for its syntactic development. Readers interested in its lexical aspect are referred to other related studies, for example, Klima-Bellugi 1979, Yau 1992. In section 7, I analyze the general shift of sign order in ASL and European SLs, which is generally considered to be the result of the strong pressure of the SVO word order in their respective ambient spoken languages. Within my theoretical framework, such a shift implies a change from SOV on the temporal register to SVO on the spatial order. I think that under similar pressure, CSL will sooner or later follow the same trend. On the other hand, the change in sign order observed in Western SLs in the nineteen sixties should entail a reanalysis of their entire syntactic data. Yet, to my knowledge, no such studies have so far been conceived of. Finally, I insist that both deaf and hearing people are seeing creatures. From that point of view, their languages, though of different modalities, are all subject to cognito-visual constraints. For this reason, I argue that there is a natural and common element ordering among communicative performance, in gesture, speech (child language in particular), pictographic as well as ideographic writing (such as ancient Chinese writing), and even drawings by people without any artistic training. It is this cognitive basis that will eventually offer a more satisfactory explanation to this universal observed in various modalities of expression. In the present case, it shows that studies on SL are not alien to general linguistics which is at present still dominated by research on OL.

The Role of Visual Space in Sign Language Development

161

Notes 1

However, this statement was only true in the past when conventional SL was still nonexistent. Now, the situation has changed. For instance, the signs created and used by pupils of the Hong Kong School for the Deaf were condemned after the School was converted to an ordinary secondary school a few years ago. A Nicaraguan SL which was created spontaneously by the deaf pupils at a deaf school founded in 1977, is expected to meet the same fate, and sooner or later, it will disappear and be replaced by the local conventional SL, according to researchers’ reports. (See for example, Kegl J. et al. 1999.) 2 Humbolt (1836a [Tr. 1971]: 199) has “Less Developed Linguistic Structure” for one of its chapter titles (chapter 20). In Humbolt (1836b [Tr. 1988]: 217) we read: “Now that this structure is better in one than another, is more excellent in Sanskrit than in Chinese, and in Greek than in Arabic, could hardly be kdisputed by any impartial scholar.” Time has changed and I do not think any modern linguist would adopt this attitude or make such deprecatory remarks against particular languages on our planet. 3 My early research on SL was basically marked by this conviction. See Yau 1977a; 1979. Here I would like to quote a recent remark by the neuroscientists Ritzzolatti et al. (1988) on human language genesis in relation to gestures. I should say that my position on this question, in the past as at present, is very close to theirs (Yau 1992). Ritzzolatti et al. make the following statement: “An object or event described gesturally (such as, large object – large gesture of the arms, and small object – tiny opening of the fingers) could now be accompanied by vocalization. If identical sounds were constantly used to indicate identical elements (such as, large object – large opening of the mouth, vowel ‘a’, and small object – the tiny opening of the mouth, vowel ‘i’), a primitive vocabulary of meaningful sounds could start to develop ... Manual gesture progressively lost their importance, whereas by contrast, vocalization acquired autonomy, until the relation between gestural and vocal communication inverted and gesture became purely an accessory factor to sound communication. At this point speech took off.” 4 Bautista (1983) has investigated the development of word order of children speaking Tagalog, which is a verb-initial language. In her summary, she writes: “Data obtained from 107 Tagalog-speaking children between ages of 2;2 and 4;6 support Yau’s claim of a developmental progression from enumerative description to full relational description with the children progressing from a stage of simply enumerating nominals (N) to a stage of relating these nominals (N) using the adult word orders of RNN’, RN’N and NRN’, N’RN. However, the stage of NNR description (posited by the Yau hypothesis) appeared very short-lived; the data did not provide conclusive support for the claim that NN-initial is the natural word order in child language.” I think that the fact that “the stage of NNR description appeared very short-lived” is due to the strong pressure of the joint concurrence of the four adult word orders, and the stage of NNR although very short-lived is nevertheless significant.

162

5

S. Yau

“At the beginning of the test period, Washoe put subject before object, although she put both before the verb; at the end she was using traditional subject-verb-object word order”. The relatively causal tone in which Linden, a scientific reporter, related this observation makes the description of this critical event all the more faithful and convincing. (Linden 1974: 103) 6 Lane (1977) seems to prefer the term “parallel” to “simultaneity”. He suggests that spoken language is essentially sequential while sign is predominantly parallel because of the simultaneity of the visual medium (quoted by Kyle & Woll 1985: 30). 7 However, “the sign TAPE-RECORDER has a different movement from the mime of tape reels; the movements of left and right hands are mirror images of each other in the sign, while in the mime they move in a parallel way. The sign can be used to refer to cassette recorders, where the reels are not visible, and to video-recorders.” (Kyle and Woll 1985: 116–7) This departure from a more faithful image of the object is to avoid violating the so called symmetry constraint which operates on it. According to Kyle and Woll, “The symmetry constraint states that hands should move symmetrically.” Examples of this kind are also found in CSL. In the seventies, I have criticized some of the signs jointly published by the Ministry of Internal Affairs and the Ministry of Education in Beijing in 1961. For example, the sign kē xué “science” is performed with two hands in different handshapes but making similar rotating movements. Such gestural formation is against the natural law in signing. The examples quoted here illustrate that the iconic device is also subject to the internal constraints of SL. 8 It should be noted that in American or European SL studies, terms are often directly borrowed from speech to label their equivalents in SL, for example, “word” in this quotation means “sign”. 9 See also the sequential feature within simultaneity in SL in the next section on Hand Movement. 10 See Poulin et al. (1995), Engberg-Pedersen (1993: 102–3) and others cited in the section 3 that follows. 11 For syntactic creation, see section 5 below. 12 The three terms “transparent”, “translucent”, and “opaque” are borrowed from Frishberg (1975); see also Klima and Bellugi 1979. The gradual change of signs mostly takes place in their location, such as from mouth to cheek or chin, from lower arm to hand, or from temple or upper face, etc., to the chest front. On the whole, the general tendency is from peripheric to the centre of the space in front of the signer. See Kyle and Woll 1985: 121; Woodward 1976. Here are some examples from my corpus showing how once transparent signs might become opaque. The formation of the old sign for GUANGZHOU (Canton) was iconic, but has been considered as an opaque sign of unknown origin. Fortunately I was informed by an elder Cantonese deaf signer that this sign which is made with the edges of the hands (palms upward) against the hips, in fact, indicates the length of the traditional short jacket worn by the Cantonese. This gesture was then used to represent the name sign of the city. See Yau and He 1987. The sign created for her daughter Louise by Martha Pettikwi, the deaf Cree woman who has invented her own sign system, was in fact her sign for CHUBBY, but it was

The Role of Visual Space in Sign Language Development

163

so displaced from its original position that anyone foreign to her system could not guess its meaning. See Yau 1990. 13 It is very unlikely that this constraint will be respected when a deaf person signs directly in CSL to someone politically important but physical short. 14 Curiously, as of the eighties, some linguists working on OL tend to emphasize the significance of iconicity in speech, see, for example, articles in Haiman (ed.) 1985. 15 Along with iconicity, pantomimic elements are also confronting the same fate. One might notice that deaf signers often improvise pantomimic passages in their narration so as to increase its vividness. These pantomimic elements turn out to be difficult to handle within the methodologies available. Hence there are attempts to get around the problem by eliminating them from the SL proper. For example, when Boyes-Braem (1986) writes that “the concept of a restricted set of syllable forms might also aid in the recognition of pantomime elements as being different from lexical signs – the pantomimic sequences not being based on syllabic forms”, her message can be understood as reporting that there is such an attempt in ASL. 16 Research on nonmanual features is particularly active in the United States and in Europe, and in contrast, practically nothing has been done in CSL. 17 Usually, signers use head movements to indicate negation. But Mr. Carpenter, the isolated Cree deaf signer whom I interviewed in 1977 in Attawapiskat in Northern Canada, used an extremely ideosyncratic device for this purpose, the lateral movements of the tongue. See Yau 1992. On the other hand, I have never come across the use of squinted eyes in CSL, TSL (Taiwan Sign Language), HKSL (Hong Kong Sign Language), or among the SLs created in isolation by deaf Amerindians. 18 Liddell (2004: 14) remarks that “Signs requiring specific nonmanual signals comprise only a small percentage of the ASL lexicon. ... however, nonmanual signals are more frequently present as parts of morphologically complex signs as well as in syntactic constructions.” 19 Moreover, “some action verbs can take a modification which is expressed by a mouth pattern that consists of slightly extended and protruding lips. The meaning is that the action takes place as expected, without any problems.” (Engberg-Pedersen 1993: 23; 42) I think this mouth expression is also shared by hearing people, in particular the French, with the connotation of “nothing extraordinary” or “banal”. 20 Engberg-Pedersen notes that receivers would nod or look puzzled in response to squinted eyes. She also observes that movements of the head or the body may also occur within a clause or a major clause constituent. According to my classification, body movements should belong to the second category of nonmanual expressions which I will discuss immediately. 21 The following paragraph is taken from a very succint summary of Jouison’s paper by Boyes-Braem (1986): “He (Jouison) reports that a lateral rocking movement accompanying different signs puts these signs into a coordinate relationship with each other (‘and’, ‘but’ etc.). A to-and-fro rocking movement signals a kind of subordination relationship, indicating condition or cause. The movement he reports, is the sole marker of this relationship - without the rhythmic rocking of the trunk, the signs would be just a list of citation signs.” BoyesBraem also makes an interesting comment on Jouison’s findings: “Or do other production factors influence which channel is used for this kind of marking? For example the fact that

164

S. Yau

one is standing or sitting while signing, or that one is signing to a very large audience versus one person, or that one is using a more formal versus an intimate type of discourse.” See also the case of subordinate structures in ASL in section 8 below. 22 As Elisabeth Engberg-Pedersen points out to me, gaze direction or body/head orientation should be considered as visuo-spatial. 23 This sign cannot be used as a classifier to embed an action. Only the sign HUMAN BEING made with a Y-handshape (fist with its thumb pointing upward, and little pointing downard) can be used to carry out an action. 24 It is expected that a modification of a root sign’s movement would entail automatically a slight change in its location. As for the date of this old form, unfortunately, there is no exact record. It should probably be around the turn of the 19th century. See Yau 1977b: 89 under the item “Peking”, i.e., Beijing. 25 His examples such as CHICAGO in ASL can be found in his recent publication Liddell (2004: 12). 26 This observation is in line with my appreciation of a verb of action labelled as R (relator) in SL, and consequently reinforces my analysis of the function of this R in establishing the relationship between N and N’ in a basic sign sequence. See also section 5 below. 27 The experimentors attached small point lights to the head shoulders, elbows, wrist and index finger to a signer hidden in the dark. However I relate their claim here with reservations. I suspect that the point lights may nevertheless leave some hints about the locations and the basic handshapes of the signs to be guessed. 28 For Valade, Tylor, Keep, and others on sign orders, see Yau 1992: 32–33. Kyle and Woll (1985: 47–8; 55–6) also give long quotations from Tylor (1874: 23–27). 29 In my first report on this question (Yau 1977a), I have already defined this basic sign order as the element ordering in a declarative gestural sequence containing an N (agent), an N’ (patient) and an R (a transitive, senori-motoric, intrinsic-directional verb such as bit, hit, catch, take, etc.) Though N will precede N’ or vice versa, the crucial point is that R will be performed after the two N. I have to limit my conclusions to sequences with this kind of verbs, because in sequences with verbs having a resultative complement, like build (a house) or inflate (a ballon), the orders are not necessarily subject to the above stated constraint. The constituents in these two cases, for example, will appear in the following manner: N (agent) BUILD HOUSE, and N (agent) INFLATE BALLOON. This order of appearance in fact follows the spatio-temporal development of the respective visual events. 30 I would like to thank Professor Carol Padden, Department of communication, University of California at San Diego, one of the joint authors of the reports, for sending me several papers related to their research on Al-Sayyid Bedouin Sign Language. 31 McIntire (1982) sums up the discussion on sign ordering in the late nineteen seventies as follows: “In recent linguistic literature the question of constituent ordering has been discussed by many, but five authors represent major stands on the question. Fischer (1975) has claimed that ASL is an SVO language, having been SOV originally. Friedman (1976) claims that no basic ordering preference exists in ASL. Liddell (1977) supports Fischer in viewing ASL as

The Role of Visual Space in Sign Language Development

165

following SVO order ... Yau (1977a) suggests that sign languages are SOV and offers an extra-linguistic constraint in explanation. Anderson (1978) says that VO varies with OV in ASL and that order is governed by an old-to-new principle.” At present, I still maintain my position stated in 1977a and 1979, and insist on differences between conventional and natural sign orders. In the former order, the order may be subject to various local syntactic interpretations, but in the latter, its ultimate justification lies in its cognitive constraints. 32 In Yau (1979), I have also presented evidence in child languages, showing that French or Chinese children adopt a descriptive strategy producing utterances similar to the one here given by Armstrong et al. (1995). 33 Nakanishi (1994) is cited in Masone et al. (2004). 34 I use NN’ R (SOV) and N’NR (OSV) to represent the orders observed in CSL dialects such as sign dialects in Beijing, Shanghai, Nanjing and Hong Kong. For more details, see Yau 1977a, and the following section in this paper. 35 Back to the case of ASL for a comparable sequence. Fischer (1975) writes: “there are two ways one could translate a sentence like “The girl kicked the boy” into ASL. The most straightforward way of doing so would be GIRL KICK BOY. ... If, however, the sentence occurs in a more extended discourse, especially if one will wish to refer to the boy and/or the girl again, the most natural way to sign the meaning of (this sentence) is BOY (HERE) GIRL (HERE) SHE-KICK-HIM. (right hand) (left hand) left “kick” right, from direction of location of girl to location of boy. Moreover, she says that SOV is possible in sequences with reversible NP’s, where “the subject and object are reversible, i.e., if they could be reversed and one would still have a semantically plausible utterance.” (Fischer 1975) 36 A further scrutiny of my filmed gestural data reveals that the N agent not only marks the departing point of R but is also incorporated with its action at the initial stage. Likewise, N’ patient does not only mark the terminal point of R but is visually present through out the performance of R. Hence a more faithful schematic representation of a basic declarative gestural sequence should be NnRn’N’ (spatial) and NN’nRn’ or N’Nn’Rn (temporal), instead of NRN’ (spatial) and NN’R or N’NR (temporal). The n and n’ indicate respectively the renewed reference made to the starting point and the ending point of R. In the case of OL, n and n’ are the referential constituents with N and N’ as their respective referents. The projection of NN’R or N’NR into NN’nRn’ or N’Nn’Rn, all on the temporal register, is of immediate significance in the study of word order in child language. See Yau 1979. 37 As a sign of gratitude, I published a reprint version of Valade’s book in 1979 in Editions Langages Croisés, Paris, with a new preface by William C. Stokoe. 38 Following again this hypothesis of mine, an OL of the type NRN’ would reflect some features characteristic of the spatial register while an OL of the type NN’R or N’NR would reflect some of those of the temporal. It is beyond the scope of this paper to present the related arguments. The reader is requested to consult my papers of 1977a and 1979.

166

39

S. Yau

In Yau (1978) I report an experiment which confirms that people without training will represent a visual event such as “Someone reaches for the telephone” by first drawing a telephone, then a hand near to it. I interpret the appearance of these two elements in drawing as equivalent to N’NR (OSV) in which the hand standing for “someone” is incorporated to the R (V) “to reach”. In another article (1983), I argue that the temporal order in the composition of Archaic Chinese pictograms is also NN’R (SOV), or N’NR (OSV). 40 In many other SLs, CSL for example, orders like HUNGRY VERY are still strictly respected. 41 Many ASL researchers confirm this influence. In Battison (1978: 91, 96) we read “... the influence of the English language on ASL is pervasive.” and “ASL borrows linguistic forms and patterns from English ... .” (Battison 1978: 96) 42 Normally, “right”, “left”, “front” and “back” are determined with regard to the signer’s body, but sometimes they refer to an external entity. In case the entity in question does not have an inherent orientation, the position of the various directional positions in the signing space depends on the signer’s choice of a reference strategy or perspective, which in turn determined by his or her cultural environment. Thus the signer might adopt the so-called “ego-aligned” strategy, or on the contrary, the “ego-opposed” strategy, equivalent respectively to Lee’s (1996) “alignment strategy” and “facing strategy” in the context of child speech. See also Engberg-Pedersen, 1993: 199 (which also quotes Pierre Oléron (1978, 124); Yau 1988; 2004. As for the time lines, see Frishberg and Gough 1973; Kyle and Woll 1985; Engberg-Pedersen 1993: 80–90 where she also quotes Brito 1983, Traugott 1978, Benveniste 1974, and many others. 43 Other researchers may use the term “locations” for the context dependent locatives, and I do not know whether they would include the context dependent locatives under the same term. For example, Loncke et al. (1986) write: “Yau (1977) and Liddell (1980) stressed that location has a tendency to come first in sentences.” (My italics) 44 It is notewhorthy that there is no attempt whatsoever to explain this initial position of the locatives by means of topicalisation as in the case of shifting the N’ patient in front of the N agent and the verb. 45 Apart from the locatives, there are probably some other cases where visual constraints might still resist modification in sign sequencing. For example, Coulter (1977) points out that “There are a number of two clause constructions in ASL, in which the order of the two clauses is fixed. In casual conversations no conjunctions are used (unless emphasis is placed on the conjunction, e.g., for contrastive purposes). Sequences of clauses describing events which happen in succession are ordered in conversation in the same sequence as the events themselves.” 46 I have left out the details in subscription in the presentation of the sequence. See Baker et al. 1980: 364. 47 Liddell (1978) notes: “Though many signers will use WHO as a relative pronoun, the deaf researchers here at the Salk Institute regard this as an intrusion of English.” 48 Later in the paper Liddell adds: “In ASL the type of relative clause described here is really ambiguous. ... The context is usally sufficient to let the addressee know which noun phrase is

The Role of Visual Space in Sign Language Development

167

functioning as the head of the relative clause. Further help is provided by semantic plausibility (i.e., in the following example DOG is more likely subject of BARK than CAT). ... process of spatial indexing can be used in the relative clause.” However, in Liddell (1995) where he cites this example again, only “The dog which chased the cat recently came home.” is given, without mentioning that the sequence is ambiguous. Therefore we do not know whether the former ambiguity has since disappeared. 49 I feel that Liddell’s precaution does not constitute a guarantee as he might expect. Having the examples several days before hand, the deaf interpreters might have a lot of time for improvising means to meet the request. On the other hand, if the deaf signers do possess this structure before the assignement, there is no need to take this precaution measure. 50 Here is the example given by Baker et al. (1980) The grammatical signals are composed of a brow raise, cheek and upper lip raise, and a backward tilt of the head. There is no pause between the relative clause and the rest of the sentence: ____________________________rel.cl. THAT-ONE MAN WITH DOG MY GOOD-FRIEND That guy with the dog is a real good friend of mine. (Baker et al. 1980) 51 Coulter also supports his argument by citing the case of nodding, saying that “this combination of the ‘nod’ plus THAT is reasonably translated into English as ‘that’s the one.’ “JOHN DOCTOR means simply ‘John’s doctor’ but JOHN DOCTORnod means ‘John is a doctor. ‘ME THINK FAT LADYnod means ‘I think the fat lady is the one.’ or ‘I think the fat lady did it.’” “Finally, some relative clauses do have this head nod. In these cases the nod marks a question, and the signer waits for an answer before continuing. This happens, for example, when the signer isn’t sure that the addressee can identify the referent from the information which is provided in the relative clause, and wants confirmation of this from the addressee before continuing with the matrix clause. In other words the relative clause with the lean is also a question, although the sentence as a whole is not.” 52 This criticism of mine also applies to the case of subordination structures in FSL presented by Jouison in section 3 on nonmanual features in SL. Since Jouison’s paper, I do not know whether other researchers on FSL have done any follow-up studies on the same topic, and in particuar, whether the subordinate structure signaled by nonmanual features has developed further. 53 I would like to thank Professors Benci Woll, Elisabeth Engberg-Pedersen and Scott K. Liddell for their prompt replies. They all feel that they do not have any papers on comparatives in SL to recommend. Professor Woll, in her reply, provides me with information about comparative in BSL which I reproduce partly below: “In response to your specific questions I would sign those examples as follows: ‘5 is greater than 3’: THREE (non-dominant hand) FIVE (dominant hand) FIVE MORE ‘3 is smaller than 5’: FIVE (non-dominant hand) THREE (dominant hand) THREE LESS” (Bencie Woll, September 2005, personal communication). 54 See the case of the deaf signer Carpenter of Attawapiskat in Yau (1992).

168

55

S. Yau

The form of the sign bǐ “than” is: fist, palm downward, index and thumb extended, with index pointing from Entity 1 to Entity 2. See Yau 1977: 13. In this sequence, it is observed that the subject “I” is merged wih the subsequent WRITE, i.e., at the terminal position of WRITE the finger tips are turned slightly towards the signer’s chest. 56 More information will be provided in another paper ‘Comparatives in CSL’ in preparation.

The Role of Visual Space in Sign Language Development

169

References Anderson, Lloyd B. (1978). Historical change and stability of American sign language. Working paper, Gallaudet College, Linguistics Research Laboratory, Washington, D.C. Armstrong, David, W. Stokoe & S. E. Wilcox. (1995). Gesture and the Nature of Language. Cambridge University Press. Armstrong, David F. (2002). Signing Metaphorically. Sign Language Studies, vol. 2(4), 441-451. Gallaudet University, USA. Baker, Charlotte & Dennis Cokely. (1980). American Sign Language, a Teacher’s Resource Text on Grammar and Culture. Silver Spring, Maryland. USA: T. J. Publishers Inc. Barrière, Michèle et al. (1975). Comparing comparatives. Colloques Internationaux du C.N.R.S., n°206. – Problèmes Actuels en Psycholinguistique. Paris. Battison, Robbin. (1978). Lexical Borrowing in American Sign Language, pp. 240, Linstok Press, Silver Spring, Maryland, USA. Bautista, Ma. Lourdes S. (1983) Word order in Tagalog Child Language. Philippine Journal of Linguistics, 14(1), 25–44. Benveniste, Emile. (1974). Le langage et l’expérience humaine. In E. Benventiste (Ed.), Problèmes de linguistique générale II (pp. 67–78). Paris: Gallimard. Boyes-Braem, Penny. (1986). Two Aspects of Psycholinguistic Research: Iconicity and Temporal Structure. In B. Tervoort (Ed.), Signs of Life, Proceedings of the Second European Congress on Sign Language Research (pp. 65–74). The Institute of General Linguistics of the University of Amsterdam. Brito, Lucinda Ferreira. (1983). A Comparative Study of Signs for Time and Space in Sao Paulo and Urubu-Kaapor Sign Languages. In W. Stokoe & V. Volterra (Eds.), SLR ’83 (pp. 262–268). USA Istituto di psicologia CNR Italy: Linstok Press. Chang Jung-hsing, Shiou-fen Su & James H-Y. Tai (2005). Classifier predicates reanalysed, with special reference to Taiwan Sign Language. In Language and Linguistics, pp. 247–278. Academia Sinica, April 2005, Taiwan. Coulter, Geoffrey R. (1977). American Sign Language Relative Clauses. Working paper, pp. 23, The Salk Institute for Biological Studies, La Jolla, CA. Emmorey, Karen. (1991). The Confluence of Space and Language in Signed Languages. In Paul et al. (Eds.), Bloom (pp. 171–209). Emmorey, Karen. (1996). The Confluence of Space and Language in Signed Languages. In Bloom, Paul et al. (Eds.), Language and Space (1991, pp. 171–209). Cambridge MA: MIT Press. Engberg-Pedersen, Elisabeth. (1993). Space in Danish Sign Language – the Semantics and Morphosyntax of the use of Space in a Visual Language. Hamburg: Signum-Verlag. Fauconnier, Gilles. (1985). Mental Spaces – Aspects of Meaning Construction in Natural Language. Cambridge Mass M.I.T. Press. Fischer, Susan. (1975). Influences of Word Order Change in American Sign Language. In C. N. Li (Ed.), Word Order and Word Order Change (pp. 1–25). Austin London: University of Texas Press. Friedman, Lynn A. (1976). The Manifestation of Subject, Object and Topic in the American Sign Language. In C. N. Li (Ed.), Subject and Topic (pp. 127–148). New York: Academic Press. Frishberg, Nancy. (1973). Two Processes of Reduplication in ASL. Foundations of Language, 9, 469–480. Frishberg, Nancy. (1975). Arbitrariness and Iconicity: Historical Change in American Sign Language. Language 51, 696–719.

170

S. Yau

Frishberg, Nancy. (1979). Historical Changes in Sign Language. In Børge Forkjaer-Jensen (Ed.), The Sciences of Deaf Signing. Københavns Universitet. Frishberg, Nancy & Bonnie Gough. (1973). Time on Our Hands, read at the 3rd Annual May California Linguistic Meeting, Stanford, California, pp. 28. Givon, Talmy. (1985). Iconicity, Isomorphism and Non-arbitrary Coding in Syntax. In John Haiman (Ed.), Iconicity in Syntax (pp. 187– 219). Amsterdam Philadelphia: John Benjamins Publishing Co. Greenberg, Joseph H. (1963). Some Universals of Grammar with Particular Reference to the Order of Meaningful Elements. In J. H. Greenberg (Ed.), Universals of Language (pp. 58–90). Cambridge Massachusetts: M.I.T. Press. Grosjean, François. (1981). Sign and Word Recognition: A First Comparison. Sign Language Studies 32, 195–220. Linstok Press. Haiman, John, ed. (1985). Iconicity in Syntax, pp. 403. Amsterdam Philadelphia: John Benjamins Publishing Company. Hewes, Gordon W. (1974). Language in Early Hominids. In Roger. W. Wescott (Ed.), Language Origins (pp. 1–33). USA: Linstok Press. Humbolt, Wilhelm von. (1836). Linguistic Variability and Intellectual Development (translated by G. C. Buck & F. A. Raven, 1971), (pp. 296). Philadelphia: University of Pennsylvania Press. Humbolt, Wilhelm von., (1836). On Language: the diversity of human language-structure and it influence on the mental development of mankind (Translated by Peter Heath with an introduction by Hans Àarsleff. 1988). (pp. 296). Cambridge University Press. Isenhath, J. O. (1990). The Linguistics of American Sign Language. USA Jefferson, N.C.: MacFarland and Company. Janzen, Terry, Barbara Odea, & Barbara Shaffer. (2001). The Construal of Events: Passives in American Sign Language. Sign Language Studies, 1(3), 281–310. USA: Gallaudet University. Jouison, Paul. (1986). The Role of the Body in the Organization of Signed Expressive Discourse. In B. Tervoort (Ed.), Signs of Life, Proceedings of the Second European Congress on Sign Language Research (pp. 95–110). The Institute of General Linguistics of the University of Amsterdam. Keep, J. (1869). The language of signs. American Annals of the Deaf and Dumb, 14(2), 89–95. Keep, J. (1871). The sign language. American Annals of the Deaf and Dumb, 16(4), 221–234. Kegl, Judy. (1990). Predicate argument structure and verb-class organization in the ASL lexicon. In Ceil Lucas (Ed.), Sign language research. Theoretical issues. (Proceedings of the International Conference, Theoretical Issues in Sign Language Research, II, May 18–21, 1988 at Gallaudet University) (pp. 146–175). Washington: Gallaudet University Press. Kegl, Judy, Annie Senghas, & Marie Coppola. (1999). Creation through contact: sign language emergence and sign language change in Nicaragua. In DeGraff, Michel (Ed.), Language creation and language change: creolization, diachrony, and development (pp. 179–237). Cambridge Mass.: MIT Press. Klima, E. & Bellugi U. (1979). The Signs of Language (pp. 417). Cambridge Massachusetts: Harvard University Press. Kyle, J. G. & B. Woll (Ed.). (1985). Sign Language: the study of deaf people and their language. Cambridge University Press. Lane, Harlan et al. (1976). Preliminaries to a Distinctive Feature Analysis of Handshapes in American Sign Language. Cognitive Psychology 8, 263–289.

The Role of Visual Space in Sign Language Development

171

Lee, Thomas Hun-Tak. (1996). Theoretical issues in language development and Chinese child language. In C.-T. James Huang & Y.-H Audrey Li (Eds.), New Horizons in Chinese Linguistics (pp. 293–356). Dordrecht Boston London: Kluwer Academic Publishers. Lehmann, Winfred P. (1972). Proto-Germanic Syntax. In Frans van Coetsem & H. L. Kufner (Eds.), Toward a Grammar of Proto-Germanic (pp. 240–269). Max Niemeyer Verlag, Tübingen. Liddell, Scott K. (1977). An investigation into the syntactic structure of American Sign Language. Unpublished doctoral dissertation, University of California, San Diego. Liddell, Scott K. (1978). Nonmanual signals and relative clauses in American Sign Language. In Proceedings of the First National Symposium on Sign Language Research and Teaching. Silver Srping, MD: National Association of the Deaf, 193–228. Or in Understanding Language through Sign Language Research, ed. P. Siple, 59–90, Academic Press, NY. Liddell, Scott K. (1980). American Sign Language Syntax, Mouton, The Hague, pp. 194. Liddell, Scott K. (1984). THINK and BELIEVE: Sequentiality in American Sign Language signs. Language 60(2), 372–399. Liddell, Scott K. (1995). Real, Surrogate, and Token Spaces: Grammatical Consequences in ASL. In Emmorey, Karen, Judy S. Reilly (Eds.), (pp. 19–41). Liddell, Scott K. (2004). Grammar, Gesture, and Meaning in American Sign Language. Cambridge University Press. Linden, Eugene. (1974). Apes, Men and Language. (pp. 304). N.Y.: Penguin Books. Loncke, Filip et al. (1986). Sign Order in Deaf Children and Adolescents. In B. Tervoort (Ed.), Signs of Life, Proceedings of the Second European Congress on Sign Language Research (pp. 182–193). The Institute of General Linguistics of the University of Amsterdam. Long, Joseph Schuyler. (1918). The sign language: a manual of signs (1st printing 1918). Washington: Gallaudet University (1962). Mandel, Mark. (1977). Iconic Devices in American Sign Language. In Lynn A. Friedman (Ed.), On the Other Hand, (pp. 57–107). New York: Academic Press. Masone, Maria Ignacia, & Monica Curiel. (2004). Sign Order in Argentine Sign Language. Sign Language Studies, 5(1), 63–93. USA: Gallaudet University. McIntire, Marina. (1980). Locatives in American Sign Language. Ph.D. dissertation, University of California, Los Angeles. McIntire, Marina. (1982). Constituent Order and Location in ASL. Sign Language Studies 37, 345–386. USA: Linstok Press. Moody, Bill. (1983). Introduction à l’histoire et à la grammaire de la langue des signes. International visual theatre – Centre socio-culturel des sourds (pp. 187). France: Vincennes. Oléron, Pierre (1978). Le langage gestuel des sourds: Syntax et communication. Centre National de la Recherche Scientifique, Paris. Park, Tscheng-Zin. (1970). The Acquisition of German Syntax. Working paper (pp. 51). University of Bern. Poizner, Howard et al. (1981). Perception of American Sign Language in Dynamic Point-light Displays. Journal of Experimental Psychology: Human Perception and Performance, 7(2), 430–440. Poulin, Christine, & Christopher Miller. (1995). On narrative discourse and point of view in Quebec Sign Language. In Emmorey, Karen, Judy S. Reilly (Ed.), (pp. 117–13). Ritzzolatti, Giacomo & Michael A. Arbib. (1998). Language within our grasp. Trends in Neuroscience, 21(5).

172

S. Yau

Saeed, J. & L. Leeson. (1999). Detransitivization in Irish Sign Language. Paper presented at the Third Intersign Workshop, Siena, Italy. Sanders, Gerald. (1970). Constraints on Constituent Ordering. In A. L. Vanek (Ed.), Papers in Linguistics (pp. 460–502). Springfield Illinois: Linguistic Research Inc. Sandler, Wendy, Irit Meir, Carol Padden, & Mark Aronoff. (2005). The emergence of grammar: Systematic structure in a new language. Jeremy A. Sabloff (Ed.). University of Pennsylvania Museum of Archaeology and Anthropology, Philadelphia, PA, and approved January 3, 2005. Senghas, Ann. (2005). Language Emergence: Clues from a New Bedouin Sign Language. Current Biology, 15(12). Siple, Patricia. (1978). Visual Constraints for Sign Language Communication. Sign Language Studies, 19, 95–110. USA MD: Linstok Press. Slobin, Dan I. (1966). The Acquisition of Russian as a Native Language. In Frank Smith & George A. Miller (Eds.), The Genesis of Language, a Psycholinguistic Approach (pp. 129–148). Cambridge Mass: M.I.T. Press. Stokoe, William. (2000). Modal, Signs and Universal Rules. Sign Language Studies, 1(1), 10–16. USA: Gallaudet University. Stokoe, William, Dorothy Casterline, & Carl G. Croneberg. (1965). A Dictionary of American Sign Language on Linguistic Principles (pp. 346). Washington D.C.: Gallaudet College Press. Supalla, Samuel. (1981). Changes in Repetition and Movement in ASL. Working paper (pp. 8). Supalla, Ted. (1980). Morphology of Verbs of Motion and Location in American Sign Language. Proceedings of the Second National Symposium on Sign Language and Teaching ‘pp. 27–45). Maryland: National Association of the Deaf. Supalla, Ted. (1982). Structure and Acquisition of Verbs of Motion and Location in American Sign Language. Doctoral dissertation, University of California, San Diego. Talmy, Leonard. (2003). The Representatin of Spatial Structure in Spoken and Signed Language. In Emmorey, Karen (Ed.), Perspectives in Classifer Construction in Sign Languages (pp. 169–195). New Jersey: Edition L.E.A. Tang, Gladys. (2003). Berbs of Motion and Location in Hong Kong Sign Language: Conflation and Lexicalization. In Emmorey, Karen (Ed.), Perspectives in Classifer Construction in Sign Languages (pp. 143–165). New Jersey: Edition L.E.A. Thompson, Henry. (1977). The Lack of Subordination in American Sign Language. In L. A. Friedman (Ed.), On the Other Hand (pp. 181–195). N. Y.: Academic Press. Townsend, David J. & Thomas G. Bever. (1977). Main and Subordinate Clauses: a Study in Figure and Ground, 24. Indiana Univ. Linguistics Club, Bloomington (manuscript). Traugott, Elizabeth Closs. (1978). On the expression of spatio-temporal relations in language. In Joseph Greenberg et al. (Eds.), Universals of Human Language Vol. III (pp. 369–400). Stanford: Stanford University Press. Tylor, Edward Burnett, (1964) [1865]. Early History of Mankind. The University of Chicago Press. Tylor, Edward Burnett. (1958) [reprint]. The Origins of Culture (pp. 416). New York: Harper and Brothers Publishers. Valade, Y.-L. Rémi, (1854) [reprint 1979]. Etudes sur la lexicologie et la grammaire du langage naturel des signes. Librairie philosophique de Ladrange, Paris. reprint Langages Croisés, CRLAO, Paris. Vogt-Svendsen, Marit. (1981). Mouth Position and Mouth Movement in Norwegian Sign Language. Sign Language Studies 33, 363–376. USA: Linstok Press.

The Role of Visual Space in Sign Language Development

173

Vogt-Svendsen, Marit. (1983). Lip Movements in Norwegian Sign Language. In J. G. Kyle & B. Woll (Eds.), Language in Sign:an International Perspective on Sign Language (pp. 85–96). London Canberra: Croom Helm. Vogt-Svendsen, Marit. (1990). Eye Gaze in Norwegian Sign Language Interrogatives. In SLR 87, Papers from the Fourth International Symposium on Sign Language Research, Lappeenranta Finland, July 15–19 1987, 153–162, Signum Press, Hamburg. Volterra, Virginia et al. (1984). Italian Sign Language: the Order of Elements in the Declarative Sentence. In F. Loncke et al. (Eds.), Recent Research on European Sign Languages (pp. 19–48). Lisse Netherlands: Swets & Zeitlinger. Wilbur, Ronnie B. (1979). American Sign Language and Sign Systems. Baltimore MD: University Park Press. Wilbur, Ronnie B. (1987). American Sign Language: Linguistic and Applied dimensions. Boston: College-Hill Press. Wilcox, S. E. (1996). Hands and Bodies, Minds and Souls: Or How a Sign Linguist Learned to Stop Worrying and Love Gesture. Paper presented at the Workshop on Intergrating Language and Gesture, University of Delaware, Newwark. Wittgenstein, L. (1953). Philosophical Investigation (Translated by G.E.M. Anscombe). N.Y.: MacMillan. Woll, Bencie (1981). Question Structure in British Sign Language. In B. Woll et al. (Eds.), Perspectives on BSL and Deafness (pp. 136–149). London: Croom Helm. Woodward, James. (1976). Signs of Change: Historical Variation in American Sign Language. Sign Language Studies 10, 81–94. USA: Linstok Press. Woodward, James & De Santis S. (1977). Negative Incorporation in French and American Sign Language. Language in Society 6(3), 379–388. Yau, Shun-chiu. (1977a) [1982]. Constraints on Basic Sign Order and Word Order Universals. In K. M. Ritchie (Ed.), Nonverbal Communication Today: Current Research (pp. 140–153). The Hague: Mouton Publishers. Yau, Shun-chiu. (1977b). The Chinese Signs - Lexicon of the Standard Sign Language for the Deaf in China (pp. 130). Paris: Editions Langages Croisés. Yau, Shun-chiu. (1978). Element Ordering in Gestural Languages and in Archaic Chinese Ideograms. Cahiers de linguistique Asie Orientale, n°3, 51–65. Paris: Centre de recherches linguistiques sur l’Asie Orientale. Yau, Shun-chiu. (1979). Natural Word Order in Child Language. International Journal of Psycholinguistics, 6–2(14), 21–43, The Hague: Mouton Publishers. Yau, Shun-chiu. (1983). Temporal Order in the Composition of Archaic Chinese Ideograms. Journal of Chinese Linguistics, XI(2), 187–213. University of California, Berkeley. Yau, Shun-chiu. (1985a). Sociological and Cognitive Factors in the Creation of a Sign Language by an Isolated Deaf Person within a Hearing Community. In W. Stokoe & V. Volterra (Eds.), Proceedings of the III International Symposium on Sign Language Research (pp. 299–306). USA Istituto di Psicologia CNR Italy: Linstok Press. Yau, Shun-chiu. (1985b). The Two Strategies in Sign Creation. In B. T. Tervoort (Ed.), Proceedings of the Second European Congress on Sign Language Research in Amsterdam. The Dutch Foundation for the Deaf and Hearing Impaired Child, The Institute of General Linguistics of the University of Amsterdam, and The Dutch Council of the Deaf. Yau, Shun-chiu. (1987). Chinese Sign Language. The Gallaudet Encyclopedia of Deaf People 3, 65–67. New York: McGraw-Hill.

174

S. Yau

Yau, Shun-chiu. (1988). A Rashomonian Analysis: the Orientation of Time in Languages. Computational Analyses of Asian and African Languages, 30, 135–147. Tokyo: Institute of Languages and Cultures of Asia and Africa. Yau, Shun-chiu. (1989). Moulded Gestures and Guided Syntax: Scenario of a Linguistic Breakthrough. In Jan Wind et al. (Eds.), Studies in Language Origins 1 (pp. 33–42). Amsterdam Philadelphia: John Benjamins Publishing Co. Yau, Shun-chiu. (1990). Lexical Branching in Sign Language. In Susan D. Fischer & Patricia A. Siple (Eds.), Theoretical Issues in Sign Language Research, Vol. 1: Linguistics (pp. 261–278). Chicago University Press. Yau, Shun-chiu. (1992). Création gestuelle et débuts du langage (pp. 501). Paris: Editions Langages Croisés, CRLAO - EHESS. Yau, Shun-chiu. (1993). Sign Languages in Asia. In The Encyclopedia of Language and Linguistics. Oxford UK: Pergamon Press. Yau, Shun-chiu. (2004). Semiotics out of the past. In Kenichi Takashima & Jiang Shaoyu (Eds.), Meaning and Form: Essays in Pre-Modern Chinese Grammar (pp. 211–239). München: LINCOM EUROPE. YAU, Shun-chiu, & He Jingxian (1987). Zhongguo gedi shouyü diming (Place names in CSL). In Shou yü long (Signs & Deafness) (pp. 3–4 and 38–138). China: Guangzhou School for the Deaf, Guangzhou. Zeshan, Ulrike. (2003). Indo-Paistani Sign Language Grammar: A Typological Outline. Sign Language Studies, 3(2), 157–212. USA: Gallaudet University. Zhongguo shouyu [The Chinese Sign Language] (2003). Beijing: Edited by the Association of the Chinese Deaf et al., Huaxia Publication.

ASYMMETRY IN THE EXPRESSION OF SPACE IN CHINESE ––––THE CHINESE LANGUAGE MEETS TYPOLOGY Dan Xu INALCO/CRLAO, France Keywords: typology, space, asymmetry, motion verbs, cognitive Abstract: Chinese changed typologically from a verb-framed language to a satellite-framed language. However, motion verbs are problematic in this dichotomy because some of them keep their verbal features in serial verb constructions when the Figure is an agent. As in other human languages, the syntax and semantics of space expressions in Chinese present numerous cases of asymmetry. In addition, the motion verb ‘to come’ in classical Chinese was perceived in a different way from other languages. Interpreting time as a vertical movement as in contemporary Chinese also seems to be uncommon. These issues are treated from a cognitive point of view in this paper.

0. Introduction It has been taken for granted that representations of space are rather different from one language to another. As Talmy (2000) says, language structures space. Each language needs a set of terms to describe space in a very schematic manner. These can be static terms such as point, line, dimension and so forth, or dynamic terms such as ‘to go, to come, to remove, to put’. A mass can move by itself (“The baby crawls towards his mother”), or be moved by an outside force (“The baby is put down”). These two kinds of terms allow people to recognize the location or movement of an object in space. Accordingly, the space identified by the speaker or hearer is always relative and non-stable. Because of the biological configuration of human beings, our perceptual capacity is limited and our language is conditioned. It is clear that the front of the body is more sensitive to events than the back, since the perceptual apparatus consists of the eyes, nose, mouth, and ears, which favor the front of the human body, not the back (Clark, 1973). These biological characteristics make the spatial terms asymmetrical. ‘Front’ is prior on ‘back’ and is positive; ‘up’ is positive and prior on ‘down’ because of the canonical vertical posture of humans with respect to level ground (see Clark, 1973). We will see that the motion verbs ‘come’ and ‘go’, locative terms such as ‘up’ and ‘down’, ‘before’ and ‘after’ in Chinese reflect the same asymmetrical properties. The forward direction is favored since it is also the direction of perception; also, humans walk forward, not backward. These factors imply the asymmetrical use of spatial terms describing space and time. This phenomenon is common in Chinese, as in other languages. In languages, human activity verbs constitute basic verbs. Among these verbs, motion verbs are frequently used in all languages, and ‘come’ and ‘go’ represent the most fundamental motion verbs. In Mandarin Chinese, the motion verbs ‘come’ and

175 D. Xu (ed.), Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives, 175–198. © Springer Science +Business Media B.V. 2008

176

D. Xu

‘go’ can be used as main verbs or satellites to indicate a motion path. (see Talmy 2000, Slobin 2004, among others)1. These two verbs imply opposite directions with respect to the speaker, who is the only reference in space for these verbs. We will see in this study that they are often asymmetrical at syntactic and semantic levels. In contemporary Chinese2, the pairs qián ‘before, front’ and hòu ‘after, back’3, zhè ‘this’ and nà ‘that’ 4, shàng ‘above’ and xià ‘under’, etc. show clear asymmetry in their syntactic behaviors. It will be shown that the fundamental space terms ‘before’ and ‘after’, ‘up’ and ‘down’ display syntactic behaviors that are not symmetrical, and have meanings that are not always opposite. 1. Asymmetry in Motion Verbs Asymmetry is a frequent phenomenon found at different levels (phonetic, lexical and syntactic) in languages. In Chinese, it is often possible to find pairs of asymmetrical structures expressing opposite concepts: one structure without a counterpart or two symmetrical structures that do not express opposite meanings. These structures are especially attested in expressions related to space. This is true for the motion verbs lái ‘come’ and qù ‘go’. Before examining their asymmetrical behaviors, their status must be clarified with respect to typological classification established by Talmy. 1.1. Are ‘go’ and ‘come’ main verbs or satellites in Chinese? Contemporary Chinese is classified as a serial verb construction language. In fact, the Chinese language was and remains a serial verb construction language. This assumption has some implications, such as: (a) the main sources of grammaticalization are provided by the serial verb construction, as has been seen in other serial verb construction languages; (b) in contemporary Chinese, a grammaticalized element coming from a verb often has a double status, a main verb or a grammatical word. As has been mentioned in the Introduction, the Path of a motion event can be expressed by a main verb or a satellite. (see Talmy, 2000, see also Slobin, 2004). Languages encoding Path with a verb are verb-framed languages (V-language) while languages encoding Path with a satellite are satellite-framed languages (S-languages). It is legitimate to ask if Chinese should be categorized as a V-language or a Slanguage. Some discussions have been seen in the recent literature (see Lamarre in this volume). Actually, contemporary Chinese (Mandarin) behaves most like an Slanguage, while Old Chinese presents characteristics of V-languages (Xu Dan, 2006). However, contemporary Chinese motion verbs seem to be difficult to fit into one of these two patterns. Let us examine some examples. Lái and qù can be used as main verbs (examples 1 and 2) or as satellites (see example 4). The problem is how to classify the motion verbs in (3).

Asymmetry in the Expression of Space in Chinese

(1)

(2)

(3)

(4)

177

他去北京。 tā qù Běijīng he go Beijing ‘He goes to Beijing.’ 你來 兒！ nǐ lái zhèr you come here ‘Come here!’ 他向車站跑來/去。 tā xiàng chēzhàn pǎo lái/qù he toward station run come/go ‘He is running toward the station.’ (with lái, the speaker is at the station; with qù, the speaker is not at the station) 別 了把錢帶來/去 bié wàng le bǎ qián dài lai/qu NEG forget LE BA money bring come (here)/go (there) ‘Don’t forget to bring the money.’

As the beginning of this section states, grammatical words with origins as verbs often have two different behaviors, sometimes acting as full verbs and sometimes as grammatical words. In English, a typical S-language, motion verbs and satellites cannot behave in this way. In (3), lái and qù do not function as other satellites. Comparing (3) and (4), it is easy to see that in (3) the Figure (moving object) is the agent and in (4), the Figure is an object. The tone is full in (3) while it is lost in (4). It is evident that in (3) the verbal property is preserved. In this paper, the motion verbs lái and qù seen in examples (1) and (2) (type a), (3) and (4) (type b) will be analyzed. In type (a), lái and qù are main verbs. In type (b), two cases have to be distinguished: lái and qù keep more verbal properties when the agent is the Figure (see example 3); lái and qù lose verbal properties when the patient is the Figure moved by an outside force (see example 4). Besides these two types, some lexicalized expressions will be also observed. As has been mentioned, when the agent moves by itself, lái and qù keep more verbal properties in serial verb constructions such as 跑來 pǎo lái, run-come, ‘come (over) here’，走去 zǒu qù, walk-go, ‘walk away’, 飛來 fēi lái，fly-come, ‘fly towards’,游去 yóu qù, swim-go, ‘swim away’. In sentences where these combinations occur, the agent is the Figure and the motion verbs do not completely lose their verbal features. When the patient is the Figure following the word order “V (main verb)+ lai/qu”, the agent is a causer and the patient is a causee. Combinations such as 拿來 nálai, take-come, ‘bring (towards)’，帶去 dàiqu, bring-go, ‘take away’，買來 mǎilai, buy-come, ‘buy (and bring)’, 寄去 jìqu, send-go, ‘send’, actually express a

D. Xu

178

causative meaning. In this type of usage, lái and qù have lost their verbal properties and behave as path satellites. They have become directional suffixes and function as path morphemes. When lái and qù form lexicalized expressions, the original motion verbs completely lose their verbal properties at the syntactic level, as well as their motional meaning at semantic level. In some cases they are lexicalized as disyllabic verbs and cannot be segmented as two independent elements. In other words, lái and qù cannot be analyzed either as verbs or as satellites5. Considering the above sentences, it is clear that only (4) closely resembles Slanguages. Actually, most resultative verb compounds in contemporary Chinese correspond more or less to S-languages. The syntactic distribution is clear: most verbs can appear in V1 position (or verb root), and a few “verbs” can appear in V2 position (as satellites). Some verbs can occur in V1 or in V2 position (see Xu Dan, 2006, chapter 4). In Chinese resultative verb compounds, verbs occurring in V1 position often indicate a starting point in a process while verbs appearing in V2 position mark an endpoint. The motion verbs lái and qù, however, even when used in V2 position, do not necessarily express an endpoint. When they designate an endpoint, they are similar to other resultative verbs; when they do not focus on an endpoint, they can express a direction in space or aspect (for instance inchoative). Verbs that can be used as main verbs and satellites (such as lái and qù) are rare in contemporary Chinese. Even though they can occupy two syntactic positions, they present different properties at the semantic level. When they occur in V2 position, they are similar to adjectives, i.e. they indicate an end point of a process, unlike the motion verbs lái and qù which can indicate a bounded or unbounded event when used as satellites. 1.2. Asymmetry of motion verbs as main verbs In this section, we will observe the asymmetry of motion verbs from a cognitive point of view. How do Chinese people perceive space? In other words, do Chinese speakers interpret ‘come’ and ‘go’ in the same manner as speakers of other languages? The examples that we will examine are mainly diachronic. In existential sentences in contemporary Chinese, lái can be used while qù cannot6. (5)

前面來/*去了一個人 qiánmian lái/*qù le yí gè rén ahead come/*go LE a CL person ‘Someone is coming ahead.’

Example (5) suggests that the scale of lái’s use is wider than qù and Chinese favors spatial terms which are in sight of the speaker7. In (5), the locative noun qiánmian ‘ahead’ excludes the use of the verb ‘go’. Actually, the sentence is grammatical if qiánmian is replaced by a place word:

Asymmetry in the Expression of Space in Chinese

(6)

179

家裏來/去了一個人 jiā lǐ lái/qù le yí gè rén home inside come/go LE a CL person a. ‘Someone came into the house.’ [with ‘come’] b. ‘Someone from the family went (as a representative).’ [with ‘go’]

In (5), the movement ‘come’ is accomplished in front of the speaker, while in (6) the action does not necessarily take place in the speaker’s view. In both of them however, the verb ‘come’ can only take the speaker as a reference, i.e. the action is done towards him. It has to be noted that in (6b) the place word is the source, not a goal. In general, a place word marks a goal for the motion verbs ‘come’ and ‘go’, but occasionally it can indicate a source for the verb ‘go’ (6b). Since Middle Chinese, when qù ‘go’ changed its meaning from ‘leave’ to ‘go’ (see Xu Dan, 2005), the place word expresses a goal, not a source for ‘go’. This means that if (6b) is grammatical, it is due to a trace of its original use. With further examples in classical Chinese, we will see that the problem is not simple. The motion verb ‘come’ could express movement in the speaker’s view or behind the speaker. In other words, the action ‘come’ does not implicitly require that the speaker is facing the movement.8 We can ask if the movement ‘come’ in Chinese expresses a movement towards the speaker. Let us examine some examples from Old Chinese which illustrate the above comments: (7)

(8)

送往迎來 中庸 sòng wǎng yíng lái escort go welcome come ‘see off those who depart and welcome those who arrive’ (Zhōngyōng) 往者不可諫，來者猶可追。 論語 18/5 wǎng zhě bù kě jiàn, lái zhě yóu kě zhuī go ZHE NEG can rebuke, come ZHE however can chase ‘No use rebuking what has already past; but you can still give chase to what is yet to come!’ (Confucius, 18/5, R.T. Ames and H. Rosemont)

In examples (7) and (8), lái ‘come’ and wǎng ‘go’ are antonyms. The direction of the motion verb lái ‘come’ with respect to the speaker is interesting. With the verb 迎 yíng ‘welcome’ in (7), people understand that lái describes a movement towards the speaker. However, the verb 追 zhuī ‘chase, pursue’ in (8) suggests something passing the speaker from behind, which the speaker then chases (with the speaker facing the back of the pursued). The following sentences show that lái can indicate a motion from behind the speaker; in other words, the notion ‘come’ in classical Chinese could refer to a movement from behind, instead of a face-to-face movement as seen in most languages.

D. Xu

180

(9)

前不見古人，後不見來者 陳子昂 661–702 qián bù jiàn gǔrén，hòu bù jiàn lái zhě before NEG see ancients, behind NEG see come ZHE ‘In front of me no ancient is in sight, behind I see none coming.’

(10)

囘瞻後來者 岑參 714–770 huí zhān hòu lái zhě return-head see after come ZHE ‘Look back to see those who are coming’.

In examples (9) and (10), lái ‘come’ clearly indicates a situation in which the one coming is facing the back of the speaker. In (9), hòu ‘behind’ and in (10) huí zhān ‘look back and see’ reflect this observation. This conception of the movement ‘come’ no longer exists in contemporary Chinese, but it can be found in proverbs and other expressions which keep evolutionary traces. These contemporary expressions actually possess temporal value. As other motion verbs, ‘come’ was originally relative to space, but it can be into a temporal expression through metaphor. With the verb lái ‘come’, time comes from front to back through the speaker (as in other languages), or from back to front through the speaker (not common in other languages). This means that in these metaphoric expressions with lái, time’s face moves towards the face of the speaker, or towards the back of the speaker. In fact, the last expression (hòulái, ‘behind-come: afterward’) comes from a metaphor in which another person is moving towards the back of the speaker. When the expression is lexicalized, the human moving is replaced by time moving. The expression 後來 hòulái, ‘behind-come: afterward’ illustrates this phenomenon. In contemporary Chinese, hòulái is a lexicalized expression playing the role of conjunction or adverb. From examples (9) and (10), we can see that hòulái comes from a syntactic phrase. In (9) and (10), hòu combines with lái expressing a spatial concept ‘come from behind’, while in modern Chinese, the combination has been reanalyzed as a temporal metaphor meaning ‘afterwards’. Other examples retained in contemporary Chinese: (11)

(12)

後來居上 hòu lái jū shàng, behind come be up ‘Newcomers come to the fore.’ 革命自有後來人。 gémìng zì yǒu hòulái rén revolution self have behind come people ‘The revolution certainly has successors.’

Evidently, though hòulái still refers to a motion from behind the speaker, contemporary speakers take it as a whole word and forget its origin. These sentences

Asymmetry in the Expression of Space in Chinese

181

show that Chinese speakers do not perceive the motion ‘come’ in the same way: Classical Chinese clearly indicates that ‘come’ can be a movement from behind the speaker, while in modern Chinese this schema of coming in space has disappeared except in some fossilized expressions. This contrast is so strong that 方經民 Fāng Jīngmín (2004: 14 preprint) declares the spatial notion of ‘coming’ in Classical Chinese to be “not understandable”. Perhaps ancient Chinese speakers perceived the spatial motion ‘come’ differently from contemporary speakers. This has caused asymmetry in the motion verbs ‘come’ and ‘go’ and left traces in lexicalized items (see 1.4.). 1.3. Asymmetry of motion verbs in the construction “V+qǐlái/*qǐqù” Once lái and qù had become antonyms after the meaning change of qù from ‘to leave’ to ‘to go’, in most cases they had almost the same syntactic position expressing opposite meanings. In this section, the construction “V (main verb) + qǐlái/*qǐqù” will be analyzed. Compare the following contemporary Chinese examples in Table 1.1. Table 1.1 Verbs + lái/qù

來 lái come (toward the speaker) 去 qù go (away from the speaker)

上 shàng up 上來 shànglái go up 上去 shàngqù go up

下 xià down 下來 xiàlázi go down 下去 xiàqù go down

進 jìn enter 進來 jìnlái come in 進去 jìnqù come in

出 chū go out 出來 chūlái go out 出去 chūqù go out

起 qǐ get up 起來 qǐlái get up *起去 qǐqù

It is evident that lái and qù are symmetrical in most combinations except in 起 qǐ (*起去 qǐqù). However, originally the combination 起去 qǐ qù indicated two successive actions ‘to get up and go away’. In Modern Chinese9, 起去 qǐqù has been reinterpreted as a verb compound when used after another verb. But in today’s Mandarin, *qǐqù is no longer grammatical, causing the asymmetry between qǐlái and *qǐqù. The following factors cause the asymmetry between qǐlái and *qǐqù. Syntactically, the degree of the grammaticalization of qǐlái is higher than qǐqù (see 王燦龍 Wáng Cànlóng: 2004). Semantically and cognitively, the direction of lái and qù is no longer distinctive after the verb qǐ ‘to get up’. In other cases, lái and qù are neutralized when the motion is out of the sight of a speaker. Syntactically, qǐlái and qǐqù do not have the same status; qǐlái can be a verb compound when indicating a direction in space, or it can be reanalyzed as a dissyllabic directional suffix when indicating an aspect of process (inchoative or durative), while qǐqù remains a verb compound in most cases and its spatial meaning is always salient (see Wáng Cànlóng: 2004: 35). In other words, qǐlái is more

182

D. Xu

grammaticalized as a disyllabic suffix, while qǐqù mainly remains a verb phrase or verb compound and occasionally behaves as a dissyllabic suffix when the verb has the [+abstract] feature. Apparently, qǐqù died before its grammaticalization process was finished. In other combinations such as shànglai/shàngqu ‘go up’, xiàlai/xiàqu ‘go down’, jìnlai/jìnqu ‘enter’ and chūlai/chūqu ‘go out’, lái and qù are syntactically symmetrical. They are satellites in these verb compounds when the Figure is a patient or form disyllabic suffixes when used after another verb. The meaning of the verb qǐ ‘to get up’ causes the asymmetry between qǐlái and qǐqù. In other words, the semantic properties of qǐ did not favor the grammaticalization of the motion verb qù in their combination. In most expressions in Table 1.1, lái and qù are opposite and symmetrical, taking the speaker as a reference. They regularly express the meaning ‘to move towards’ or ‘to move away from’ the speaker, while in qǐlái and *qǐqù, a unique direction ‘to move upwards’ is indicated. When the verb combining with qǐlái or qǐqù designates an upward motion such as fēi ‘to fly’, the direction of the moving item has to be above the head of a speaker. When the verb combining with qǐlái or qǐqù indicates a movement parallel to the ground, at least two directions are possible with respect to the speaker, ‘towards’ or ‘away from’. The first case will be called ‘vertical movement’ while the second case will be called ‘horizontal movement’. It is clear that in vertical movement, lái and qù, which describe horizontal motion, are converted into vertical movement. As has been mentioned, in combinations such as fēi qǐlái ‘fly-up-come’ and fēi qǐqù ‘fly-up-go’, the meaning of ‘come’ and ‘go’ (lái and qù) is neutralized since the source becomes the same, i.e. from the speaker or the ground, while the goal is not precise. Only when the verb fēi ‘to fly’ expresses a level movement are lái and qù symmetrical: 飛過來 fēi guòlái ‘to fly toward the speaker’, 飛過去 fēi guòqù ‘to fly away from the speaker’. These two combinations form a symmetrical pair of expressions to indicate opposite directions, taking the speaker as the reference. The concept ‘high above the speaker’ is expressed by the verb 飛 fēi ‘to fly’ and the notion ‘toward’ or ‘away’ by the combination 過來/去 guòlái/qù, pass-come/go. As has been seen, *起去 qǐqù is out of use in today’s Mandarin, though it is found in 18th–19th century literature10 and especially in the novels of Lao She (1899– 1966) who was famous for his vernacular and representative style of dialect in Beijing. This fact suggests that forty years ago, V+qǐlái and V+qǐqù were symmetrical in Mandarin, but new generations no longer recognize the expression qǐqù. In principle, lái and qù should be symmetrical in those combinations; the asymmetric behavior of *qǐqù is thus out of the ordinary. Take some sentences drawn from Lao She’s works as examples. It is easy to see that in combinations of “V+qǐqù”, the possibilities for the main verb are quite restricted, with verbs like 藏 cáng ‘to hide’, 收 shōu ‘to take in, to put away’, 卷 juǎn ‘to roll up’, 飛 fēi ‘to fly’. It is interesting to note that these verbs precede qǐlái as well as qǐqù. Compare some pairs of examples in Lao She:

Asymmetry in the Expression of Space in Chinese

(13)

183

a. 你进去主把小缸儿藏起来 龍鬚溝 nǐ jìnqù bǎ xiǎo gāng er cáng qǐlái you enter, BA small bowl hide suffixes ‘Go in and hide the small (fish) bowl.’ (Lóngxūgōu) b.東陽把頭藏起去 四世同堂 Dōng Yáng bǎ tóu cáng qǐqù Dong Yang BA head hide get-up go ‘Dong Yang has hidden his head (in the blanket).’ (Sìshì tóngtáng)

(14)

a.他把後悔收起來 四世同堂 tā bǎ hòuhuǐ shōu qǐlái he BA regret draw-in suffixes ‘He has hidden his regret.’ (Sìshì tóngtáng) b.把目前傷心的事都趕緊收起去 四世同堂 bǎ mùqián shāngxīn de shì dōu gǎnjǐn shōu qǐqù BA at-the-moment sad MOD thing all hastily draw-in suffixes ‘Do not show the suffering to him (Quickly, hide away what’s grieving you.)’ (Sìshì tóngtáng)

(15)

a.你的飛起來，我的也馬上飛起去 小動物們續 nǐ de fēi qǐlái, wǒ de yě mǎshàng fēi qǐqù you MOD fly suffixes, my MOD also at-once fly get-up go ‘Let your pigeons take off, mine too at once.’ (Xiǎo dòngwùmen xù) b.他恨不能一伸胳膊就飛起去 四世同堂 tā hèn bù néng yì shēn gēbo jiù fēi qǐqù he hate NEG can one extend arm then fly get-up go ‘He yearns to fly away with the flap of a wing.’ (Sìshì tóngtáng)

In the above examples, it is obvious that qǐlái and qǐqù were symmetric in the period of Lao She, i.e. some forty years ago. The selection of verbs which could combine with qǐqù was rather limited. Apparently, those that could co-occur with qǐqù were also compatible with qǐlái. Actually, occurrences of V+qǐlái are much more numerous than V+ qǐqù in Lao She’s works (37:1 according to Wáng, 2004: 32), but the statistics are reversed when the main verbs are cáng ‘to hide’ or shōu ‘to arrange’. The verbs cáng and shōu preceding qǐqù in Lao She’s works share some common features: they express the disappearance or hiding of something. Table 1.2

D. Xu

184

shows that the combination Vqǐqù is more appropriate to describe this kind of situation, and its occurrences are more significant. In fact, the directional suffixes lái ‘toward the speaker’ and qù ‘away from the speaker’ would lose their semantic contrast, since the hidden item is out of the sight of the speaker anyway and the direction of motion seems to be less important. In contemporary Mandarin, only cáng qǐlái is used even when the hidden item/person is far from the speaker: (16)

他犯了法，跑到國外藏起來了。 Tā fàn le fǎ, pǎo dào guó wài cáng qǐlái le he commit LE law, run at country out hide suffixes LE ‘He has committed a crime, he has run and hidden in a foreign country.’

In the above sentence, the combination qǐqù, which would be more appropriate since it expresses a direction away from the speaker, is no longer possible in today’s Mandarin. The following table is based on the statistics of Wáng (2004): Table 1.2. V+qǐlái and V+ qǐqù in Lao She’s works V qǐlái/qù

cáng qǐlái ‘to hide’

cáng qǐqù ‘to hide’

occurrences

15

26

shōu qǐlái ‘to put away’ 3

shōu qǐqù ‘to put away’ 12

fēi qǐlái ‘to fly’

fēi qǐqù ‘to fly’

6

4

The verbs which can combine with qǐqù are limited. The fact that the combination “V+qǐqù” was seen less than “V+qǐlái ” in the period before its extinction shows that perhaps “V+qǐqù” was losing territory. At least the limited choice of the main verb did not favor its development. Since qǐlái was able to take the function of qǐqù, it may be supposed that lái and qù had merged because they no longer indicated direction in the cases “V+qǐlái/qù ” 11 . This reinterpretation had to be realized by new generations who did not make a distinction between lái and qù in these two structures. This means that the merger of lái and qù indicates the aspect of an action rather than the direction of an action. The asymmetry between lái and qù only after the verb qǐ ‘to get up’ suggests that they are symmetrical when they describe horizontal movement in view, and in vertical movement, they have to be opposite each other with respect to the speaker. The first condition is determined by natural human activity: human beings move level with the ground in normal cases; when a movement is not in sight, the direction is less sensitive to perception. The second condition explains that the speaker is the most important reference in all movements. When only the upward direction and away from the speaker are possible, the speaker reference is merged with the ground reference and the difference between lái and qù is neutralized. From these two conditions, one conclusion is drawn: a motion outside the speaker’s view or away from him does not necessarily require a direction.

Asymmetry in the Expression of Space in Chinese

185

1.4. Asymmetry of motion verbs in lexicalized expressions In this section, we will study two types of lexicalized words: temporal expressions which are metaphorized space expressions, and fossilized words in which motion verbs combine with perception verbs. We will examine lái ‘come’ and qù ‘go’ from a cognitive point of view, try to understand how motion verbs indicate time, and examine how motion verbs lost their original meaning when combined with perception verbs. After the meaning change of qù from ‘to leave’ to ‘to go’, it formed a syntactically symmetrical pair with lái. I will show examples of this in lexicalized expressions coming from syntactic phrases. Scholars have noticed (Shěn Jiāxuān, 1999) that lái and qù are often asymmetrical and that qù only refers to the past while lái is compatible with the future as well as the past12. Compare some examples in the following table: Table 1.3. Expressions with lái and qù in Mandarin lái past 向來 xiàng lái until now, have always 近來 jìnlái recently 從來 cónglái ever; have always; all along (in the negative) 後來 hòulái afterward

lái future 未來 wèilái time to come

qù past 過去 guòqù in the past

qù future –

將來 jiānglái in the future 來年 láinián next year

去世 qùshì pass away 去年 qùnián last year

–

來日方長 lái rì fāng cháng There will be ample time.

一來二去 yī lái èr qù in the course of contacts

–

–

In table 1.3, these expressions belong to different word categories. Some are adverbs such as jìnlái, jiānglái, some are nouns such as qùshì13, qùnián. As table 1.3 shows, lái is compatible with the past as well as the future while qù refers only to the past. More precisely, these temporal expressions imply different perception of motion verbs. In expressions such as 來日 láirì，come-day, future, 來年 láinián，come-year, next year, 來世 láishì, come-world, next life, lái ‘come’ indicates time moving towards the front of the speaker, while in expression such as 後來 hòulái，aftercome, afterward, lái indicates time moving towards the back of the speaker. In contrast, qù only shows time moving in expressions such as 過去 guòqù，pass-go, in the past, 去年 qùnián, go-year, last year. It is evident that in the case of lái, the most frequent case is for time to run from the front to the back of the speaker. The inverse case is seen in lexicalized

D. Xu

186

expressions coming from the verb phrase. As for the case of qù, the ‘front’/ ‘back’ notions are not pertinent since the direction of qù can be seen as radiation from the speaker. Now let us observe the second case in which lái and qù combine with perception verbs. In this case, they have completely lost their original meaning. Actually, they originated from disyllabic particles (kàn qǐlai and kàn shàngqu) and today present an elliptical form: kànlái and kànqù. This suggests that the degree of grammaticalization of these motion verbs is higher than other expressions, since the reduction is attested in the lexical form. The general tendency is that “morphophonological reduction accompanies grammaticalization” (Svorou, 2002: 132). These examples in Chinese confirm this trend. The mentioned expressions evolved from a phrase to a compound in which ‘come’ and ‘go’ were part of a disyllabic suffix. More precisely, lái and qù, the remaining morphemes, are more grammaticalized than qǐlai and shàngqu when combined with a perception verb. Compare some examples: (17)

(18)

看來/去沒問題 kàn lái/qù méi wèntí, look come/go NEG problem ‘It looks like there’s no problem.’ 聼來/去，他不大滿意。 tīng lái/qù，tā bù dà mǎnyì hear come/go, he NEG much satisfy a. ‘He doesn’t sound very satisfied.’ b. ‘He doesn’t look very satisfied./It seems that he is not very satisfied.’

In the above examples, it seems that the sentences with lái and qù are similar, and the meaning of ‘come’ and ‘go’ is merged. Actually, they are quite different in most cases. The particle lái can be used after perception verbs such as ‘look’, ‘hear’, ‘speak’, and ‘think’, while the list for qù is shorter and limited to ‘look’ and ‘hear’. Here are more examples: (19)

(20)

遠遠看去/*來， 色很美。 yuǎnyuǎn kàn qù/*lái，jǐngsè hěn měi far far look go/*come, landscape very beautiful ‘From far away, the view is splendid.’ 看來/*去，會有不少麻煩。 kàn lái /*qù，huì yǒu bùshǎo máfan look come/*go, can have NEG few troublesome ‘It seems that we will have a lot of trouble.’

Asymmetry in the Expression of Space in Chinese

(21)

(22)

187

說來/*去，多虧他幫 。 shuō lái/*qù，duōkuī tā bāngmáng speak come/*go, luckily he help ‘Actually, this is thanks to his help.’ 想來/*去， 是他的主意。 xiǎng lái/*qù，zhè shì tā de zhúyi think come/*go, this be him MOD idea ‘It may be assumed that it is his idea.’

In (19–22), it is evident that lái and qù express different meanings, and their capacity for combining with perception verbs is also different. With qù, the meaning is concrete, while with lái, it is abstract. Expressions with lái often require a derived meaning, whereas qù (such as kàn qù ‘look-go, look likely’ and tīng qù ‘hear-go, with these words, hearing this’) often gives a direct meaning. In (17), the verbs ‘to look’ combining with méi wèntí NEG problem, ‘no problem’ implies two possibilities, i.e. a concrete meaning when qù is used and an abstract meaning when lái is used. With qù, the sentence ‘It looks like there’s no problem’ in (17) refers to a concrete problem such as a flaw or a defect, while with lái, it expresses a trouble, an inconvenience. Kàn qù means ‘it looks like’, kàn lái expresses ‘it seems that’. The same analysis can be applied to (18): with qù, the sentence ‘He sounds unsatisfied’ can express a concrete meaning ‘by the sound of his words’, ‘it sounds’ (see 18a), or an abstract signification ‘according to his words’, ‘it seems like’ (18b). That is why the perception verb ‘to hear’ can be interpreted as ‘to look’. The perceptive meaning is changed from ‘to hear’ to ‘to look’. 2. Asymmetry in Locative Terms I will analyze a set of fundamental spatial terms, ‘before’ vs. ‘behind’ and ‘on/over’ vs. ‘down/below’, from a cognitive point of view. This will illuminate how Chinese structures space and exploits spatial metaphors. First, their (a)symmetrical behaviors will be presented; then their specific manner of expressing time will be analyzed. The phenomena of “moving time” vs. “moving ego” (see Clack, 1973) and “horizontal time” vs. “vertical time” will be examined. 2.1. Moving time and moving ego In Chinese, ‘before’ and ‘after’ are more or less symmetrical in spatial expressions. Chinese sign language confirms this phenomenon. However this pair of words is asymmetrical in temporal expressions. We will base our work on Clark’s (1973) framework. Clark (1973) noticed that some English expressions use moving time and others use moving ego. According to Clark, “Time can be viewed as a highway consisting of a succession of discrete events. We humans are seen in one of two ways with respect to this highway: either (1) we are moving along it, with future time ahead of us and the past behind us; or (2) the highway is moving past us from front to back. These two metaphors might be called the moving ego and moving time metaphors,

D. Xu

188

respectively” (p50), “for moving ego, the equations are front = ‘future’ and back = ‘past’, whereas for moving time, the equations are front = ‘past’ and back = ‘future’ (p. 51)”. In this section, we will observe how Mandarin views time and which metaphor it selects. We will examine a pair of examples, qián ‘ahead, before’ and hòu ‘behind, after’. When these expressions are used with a moving ego, space is salient—they are used to mean ‘ahead’ and ‘behind’, and usually suggest a concrete notion. When moving time is implied, time is prominent—the two terms are used to mean ‘before’ and ‘after’, and usually refer to an abstract notion. Actually, the meanings of ‘ahead’ and ‘behind’ are determined by the front and back of the speaker. In other words, spatial terms expressing ‘ahead’ and ‘behind’ are not absolute. The Waxiang dialect (Yunji Wu in this volume), spoken in Hunan, China, uses “mountain” and “river” as references to direction or location. Different patterns exist in other languages (Levinson, 2003). In Mandarin, the speaker is a reference in space even if we say “we will meet by the front of the door of the school”. The door of the school is an explicit reference; the speaker and the hearer, however, are implicit references, because they are supposed to be facing the door of the school. In this way, they give the door a front and a back. If we are inside the school, the same sentence may cause confusion. In Mandarin, we can choose the word 大門口 dàménkǒu ‘entrance, doorway’ (by the main gate) or 大門前 dàmén qián ‘at the front of the door’ (outside/in front of the main gate) to disambiguate the location. Now some examples are provided to illustrate moving time and moving ego in Mandarin. –Examples of moving ego: (23) (24) (25)

前程 qiánchéng ahead distance, ‘prospect, future’ 前 qiánjǐng ahead view, ‘perspective, future’ 前途 qiántú ahead route, ‘prospect, future’. –Examples for moving time:

(26) (27) (28)

前天 qiántiān before day, ‘day before yesterday’ 前年 qiánnián before year, ‘year before last’ 前人 qiánrén before man, ‘older generation’

(29) (30) (31)

後天 hòutiān after day, ‘day after tomorrow’ 後年 hòunián after year, ‘year after next’ 後人 hòurén after man, ‘later generation’.

Qián and hòu are opposite when they express space 14 , but two distinct orientations are possible when they indicate time. This suggests that Mandarin Chinese structures space statically, but time is treated dynamically. In the above examples, it is clear that qián ‘ahead, before’ and hòu ‘back, behind’ are asymmetrical15. For qián, time can be viewed with moving ego or with

Asymmetry in the Expression of Space in Chinese

189

moving time, whereas for hòu, only moving time is attested. In (23–25), moving ego is implied. Future is ahead; a reference in front of the speaker is assumed. In (26–31), moving time is implied. The past is the “front” of time, which is moving, passing us from front to back. As has been mentioned, qián uses two approaches to indicate time, time moving towards the speaker or the speaker moving forward. The time expressions in examples (23), (24) and (25) are metaphors from spatial expressions. The “distance”, the “view” and the “route” are all ahead the speaker or in his sight. The front of the speaker is facing these things in space or in his imaginary space. These spatial notions changed into temporal notions indicating the future. Thus expressions with qián include ones with moving ego as well as ones with moving time. In examples (29) –(31), “behind” is the future for moving time. The past is excluded in most expressions with hòu ‘back, behind’. In a few expressions such as hòulái (see section 1.4.), hòu can indicate the future in a past context. In the mentioned expressions with hòu, time is moving in one direction and towards the speaker. This is most likely because human’s back cannot normally perceive things. In other words, human beings do not walk backwards, and moving ego is not favored in these expressions. The above analyses are also attested in the standard sign language for Chinese deaf signers16. In this sign language, ‘ahead, before’ refers to something in front of the speaker in space. One has to point ahead with one finger to express ‘ahead, before’. To express ‘perspective, future’ or ‘prospect, future’, the language uses the same gesture (pointing forwards) first and adds a second gesture to express ‘view’ or ‘route’. In the sign language, as in spoken Mandarin, the signer expresses ‘the day before yesterday’ with two fingers at the level of the temples facing behind his own body. This means that in this sign language, ‘ahead, before’ has two opposite gestures, one facing ahead when space is expressed, the other facing behind when time is described. In spoken Mandarin, we have seen the same situation. The sole difference is that Mandarin expresses these notions orally while the sign language expresses it gesturally. In fact, this sign language has visualized spoken Mandarin’s conceptualization of these ideas. In sign language, the spatial notion ‘behind’ is expressed by the signer with one finger at the level of the shoulders pointing behind his own body. As for the temporal notion ‘after’, one has to wave the palm forwards. To express the notion of ‘before’ another time, one waves the palm at shoulder level towards the back. It is clear that the spatial notion ‘behind’ and the temporal notion ‘before’ use a gesture in the same direction (towards one’s back) while ‘behind’ and ‘after’ take two directional gestures: in a backwards direction when the spatial notion ‘behind’ is expressed, and in a forwards direction when the temporal notion ‘after’ is signed. Again, these gestures match very well with the conceptualizations of the corresponding expressions in Mandarin. 2.2. Horizontal time and vertical time In most languages, time is described as a river. This means that time can be viewed by human beings as something moving horizontally. For this reason, horizontal space expressions such as ‘ahead, behind, before, after’ are frequently

D. Xu

190

borrowed as time expressions. In the above section, we have seen that qián ‘ahead, before’ and hòu ‘back, behind’ are words typically used to express past or future, since a human takes himself as a spatial reference. In fact, not only horizontal time expressions are attested in Chinese; vertical time expressions are also seen: Table 2.1 Vertical time expressions 上 shàng up

day 上午 shàngwǔ forenoon/morning

week 上周 shàngzhōu last week

month 上個 shàng ge yuè last month

下 xià down

下午 xiàwǔ afternoon

下周 xiàzhōu next week

下個 xià ge yuè next month

year 上半年 shàng bànnián first half of the year 下半年 xià bànnián second half of the year

It is evident that in the above examples, Mandarin uses vertical notions to describe time. In these examples, shàng ‘above’ and xià ‘below’ indicate an order of events. The terms for ‘morning’ and ‘afternoon’ refer to the movement of the sun, and time moves in a circle. When the sun is rising, ‘above’ is employed expressing ‘fore-noon/morning’; when the sun is setting, ‘below’ is used expressing ‘afternoon’ 17 . This means that in some time expressions, Chinese takes the sun as a reference. It also explains why Chinese has vertical terms to express time. Vertical terms are also extended to other time expressions. In other examples, shàng ‘above’ corresponds to the past and xià ‘below’, to the future. If we understand the solar metaphor, we are not surprised that ‘above’ indicates ‘anterior’ and ‘below’ ‘posterior’. Actually, ‘above’ and ‘below’ allude to an order, i.e. earlier and later. When vertical terms are converted into horizontal terms, the asymmetry can be seen. Here are some examples: (32)

a. 眼下 yǎnxià, eye below, ‘right now’ b. 眼前 yǎnqián, eye ahead, ‘before [one’s] eyes, at the moment’

(33)

a. 下一代 xià yīdài, down one generation, ‘next generation’ b. 後一代 hòu yīdài after one generation, ‘next generation’.

In (32), ‘down’ and ‘before’ are spatial expressions. These expressions are metaphorized into time expressions meaning ‘right now’. In (33), ‘down’ and ‘after’ are synonyms. They express the future. As we saw in the last section, in most cases hòu indicates the future while xià indicates ‘later’ in an order. It is interesting that the vertical term ‘below’ can correspond to the horizontal terms ‘before’ as well as to ‘after’.

Asymmetry in the Expression of Space in Chinese

191

The solar metaphor ‘over’ and ‘down’ is extended in horizontal space/time expressions. For example, trains run level with the ground. It is a typical horizontal movement. However, instead of using the horizontal terms ‘before’ and ‘after’, Chinese uses a pair of vertical terms to describe stations: (34)

a 上一站 shàng yī zhàn, above one station, ‘the previous station’. b 下一站 xià yī zhàn, below one station, ‘the next station’.

In the above examples, the stations are presented as vertical items. The train (or the bus) seems to move like water, which runs from a higher place to a lower place. This time, horizontal space/time is expressed by a vertical term. It has been shown that vertical time expressions are symmetrical because the ego moving is not the canonical movement of humans in space. Human beings move level with the ground but not upwards or downwards. However, when these vertical times are changed into horizontal times, asymmetry is exposed, as the horizontal time terms are profoundly asymmetrical. Actually, other languages also use vertical terms to describe time. For example in English we can say ‘over the summer’ (see Traugott, 1978) and ‘from that time onwards’. In these words, ‘over’ and ‘on-’ clearly use spatial metaphors as time expressions. What is interesting is the equivalent of ‘onwards’ in Chinese: if we examine the word by word translation, we will see that the word ‘onwards’ in English corresponds to ‘down’ or ‘after’ in Chinese. Observe the following example: (35)

from now on 從今以後 cóng jīn yǐhòu from today after

(36)

from the Han onwards 自漢以降 zì Hàn yǐ jiàng, since Han from down

In the above examples, it seems that English speakers and Chinese speakers do not see temporal movement in the same way. In English, time passes in an upwards direction (for example “Time is up”18), while in Chinese, time runs downwards. The word ‘update’ in English refers to the present, which is later than the past. Cook (1996: 457) points out that in English idiomatic expressions such as “who’s up?”, ‘up’ means ‘next’ in time or space. Again, ‘up’ must be translated as ‘down’ in Chinese: (37)

Who’s up? 下一個是誰？xià yí ge shì shuí down one CL be who

Apparently, vertical time is not perceived in the same manner in these two languages. According to Cook, the directional particle a’e ‘up’ in Hawaiian indicates ‘next in sequence’. Here is an example cited by Cook from Hawkins:

D. Xu

192

(38)

ke kanaka a’e, the person up, ‘the next person’

The use of a’e in (38) behaves like the English particle ‘up’ in (37). It is to be noted that in Hawaiian, a’e ‘up’ and iho ‘down’ are not always symmetrical since both of them can mark the future or the recent past when combined with nei ‘last’ (a’e nei ‘recent past’, iho nei ‘recent past’. See Cook, 1996: 458). It is clear that vertical time is seen not only in Chinese but also in other languages, though it is not very common. 3. Conclusion Contemporary Chinese is a satellite-framed language while Old Chinese was a verb-framed language. This typological change must be considered in studies. Motion verbs do not fit very well into this dichotomy, because some of them keep their verbal features in serial verb constructions when the Figure is an agent. Motion verbs are grammaticalized as satellites in a verbal compound when the Figure is a patient. Asymmetry is one of the characteristics of human languages. The Chinese language also presents numerous cases of asymmetry. Due to the configuration of the human body, the front-back reference of the speaker plays a key role in space and time expressions. Asymmetry in language reflects the asymmetric perceptions of the speaker in space. The motion verbs lái and qù, the space terms ‘before’ vs. ‘after’ are asymmetrical at the syntactic level as well as the semantic level. These verbs and locative terms express a horizontal movement which is always relative and cannot be stable in space. The vertical time expressions in Chinese are symmetrical and look rather uncommon among other languages. But this symmetry is immediately destroyed when the vertical time expressions are converted into horizontal time expressions. The latter are the most commonly attested in other languages. More investigation should certainly be done to understand how languages grasp the movement of time. The following tables show how Chinese perceives space and how time is viewed in this language: Table 3.1 Asymmetry for ‘come’ and ‘go’ in lexicalized expressions in Chinese19 SPATIAL

TEMPORAL

lái ‘come’ a. towards the face of the speaker b. towards the back of the speaker a. past b. future

qù ‘go’ in any direction away from the speaker

past

Asymmetry in the Expression of Space in Chinese

193

Table 3.2 Asymmetry of spatial terms in Chinese20

SPATIAL TEMPORAL

qián ‘front, ahead’ anterior forwards temporal anteriority (moving ego) temporal posteriority (moving time)

hòu ‘back, after’ posterior backwards temporal posteriority (moving time)

shàng ‘top, up’ superior

xià ‘bottom, down’ inferior

earlier, former

later, latter

In these two tables, the asymmetric properties of the spatial descriptions (motion verbs and space terms) are evident. What is interesting is to compare them with other languages from a typological point of view. Italicized words indicate aspects specific to the Chinese language, and the other words present properties seen in other languages. Based on twenty six genetically unrelated languages, the samples found by Svorou (2002) do not show the interpretations mentioned in the above tables21. It is not common for the notion of ‘front’ to imply temporal posteriority, or for ‘top’ and ‘bottom’ to designate an order of time. Apparently, the Chinese language presents an interesting sample for typological study. Abbreviations ASP BA CL LE LOC MOD NEG PREP VR SUO ZHE

aspect marker direct object marker classifier aspect marker, final particle location modification marker negation preposition verb+resultative verb nominalizing particle of patient nominalizing particle of agent, topic marker

Notes 1

These motion verbs have played an important role in the rise of resultative verb compounds in Chinese, since the verb+resultative pattern had a subsequent impact on Chinese syntax. 2 See 沈家煊 Shěn Jiāxuān (1999), 方經民 Fāng Jīngmín (2004, preprint), 齊滬揚 Qí Hùyáng (1998). 3 See Dan Xu (1996), section 3 of Chapter 3, see also Fuxiang Wu in this book. 4 See 徐丹 Xú Dān (1988).

194

5

D. Xu

Here are some examples: 説來 shuōlái, speak-come, ‘it is a long story to say that’，想來 xiǎnglái, think-come, ‘it may be assumed that’, 看來 kànlái, look-come, ‘it seems like’and 聼 來 tīnglái, ‘it sounds like’. Evidently, lái and qù have lost their original meaning and no motion is implied. See section 1.4. 6 Cf 劉 華 Liú Yuèhuá et al. (1998). The same thing happens when lái/qù is used after another verb in a serial verb construction with an existential meaning: 樹上飛來/*去一只鳥 shù shàng fēi lái/*qù yì zhī niǎo, tree up fly come/*go a CL bird. ‘A bird is flying to a tree.’ 7 This causes the asymmetry between lái ‘come’ and qù ‘go’. The same phenomenon happens for other terms such as 前 qián ‘ahead before’ and 後 hòu ‘back, behind’, 这 zhè ‘this’ and 那 nà ‘that’. Among these three pairs of terms, the first counterparts, i.e. ‘come’, ‘ahead’ and ‘this’ are preferential. We can also say that these terms are ‘positive’ and their antonyms are ‘negative’. 8 It is to be noted that in OC the lexical choice was different, the motion verb ‘go’ was wǎng not qù in Old Chinese. For example: 孟子去齊,宿於晝。(孟子 2B/11) Mèngzǐ qù Qí, sù yú Zhòu. Mencius-leave-Qi, lodge for the night-at-Zhou. Mencius left Ch’i and on his way put up at Chou (Mencius, 2B/11, Lau). Here, the motion verb qù preceding an NP [+locative] expresses ‘to leave a place’. In the Tang Dynasty (618–907), the verb wǎng ‘to go’ had been grammaticalized into a preposition introducing a location (see 馬貝加 Mǎ Bèijiā, 2002: 83) and the motion verb qù had progressively gained a new meaning ‘to go’. 9 The language used in the 紅 樓 夢 Hónglóumèng (18th century) and 兒 女 英 雄 傳 Ernǚyīngxióngzhuàn (19th century) represent modern Mandarin accurately. 10 For example in the 水 滸 傳 Shuǐhǔzhuàn, 紅 樓 夢 Hónglóumèng and 兒 女 英 雄 傳 Ernǚyīngxióng zhuàn. 11 王 燦 龍 Wáng Cànlóng (2004) has studied the problem of qǐqù indicating that its grammaticalization is not completed. I agree with this approach but not with his analysis on cognitive salience of lái and qù to explain the disappearance of qǐqù (p. 34), because this cannot clarify why in other parallel cases, V lái and V qù are almost symmetrical except qǐlái and qǐqù. 12 李崇 Lǐ Chóngxīng (1990) indicates that qù expresses the past as well as the future in diachronic texts. 13 As many dissyllabic words in Chinese, qùshì can be used as a verb or a noun. 14 The perception of qián and hòu are not symmetrical in space. 15 They are symmetric in some space expressions. For example: 前面 qiánmian, ahead, in front of, 後面 hòumian，behind, at the back of, 前臺 qiántái stage, 後臺 hòutái, back stage, etc. 16 All examples in sign language are taken from Zhōngguó shǒuyǔ 中國手語, a method for deaf mute Chinese people published in 2003. 17 中 zhōng ‘middle’ is used for ‘noon’, i.e. 中午 zhōngwǔ. 18 Thanks to Craig Baker for his constructive comments. 19 The elliptical forms with perception verbs are not included here. 20 Cf the table established by Svorou (2002: 138) with typological approach. Table 3.2 follows her model with Chinese data. Thanks to Fuxiang Wu who has given me this source. 21 With Svorou’s terms, “spatial grams” are concerned. A spatial gram is a “grammatical element that incorporates the expression of the region designator and the relator”, Svorou, 2002: 139.

Asymmetry in the Expression of Space in Chinese

195

References Ameka, Felix K. (1995). The linguistic construction of space in Ewe. Cognitive Linguistics. 62/3, 139–181. Ames, R. T. & H. Rosemont, Jr. (1998). The Analects of Confucius. New York: Ballantine Books. Bellavia, Elena. (1996). The German über. In Martin Pütz & René Dirven (Eds.), The Construal of Space in Language and Thought (pp. 73–107). Berlin New York: Mouton de Gruyter. Bilmes, Leela. (1995). The Grammaticalization of Thai ‘Come’ and ‘Go’. In Proceedings of the Twenty–first Annual Meeting of the Berkeley Linguistics Society (pp. 33–46) Berkeley: Berkeley Linguistics Society. Bybee, Joan, Perkins, Revere & Pagliuca, William. (1994). The Evolution of Grammar. Chicago London: University of Chicago Press. Cáo Guǎngshùn 廣 . (1995). Jìndài hànyǔ zhùcí 近 代 漢 語 助 詞 . Běijīng: Yǔwén chūbǎnshè. 北京: 語文出版社. Carlson, Robert. (1991). Grammaticalization of Postpositions and Word Order in Senufo Languages. In Elizabeth Closs Traugott & Bernd Heine (Eds.), Approaches to Grammaticalization Vol. II (pp. 201–223). Amsterdam Philadelphia: John Benjamins Publishing Company. Clark, Herbert H. (1973). Space, Time, Semantics, and the Child. In Timothy E. Moore (Ed.), Cognitive Development and the Acquisition of Language (pp. 27–63). New York London: Academic Press. Cook, Kenneth W. (1996). The temporal use of Hawaiian directional particles. In Martin Pütz & René Dirven (Eds.), The Construal of Space in Language and Thought (pp. 455– 466). Berlin New York: Mouton de Gruyter. Emmorey, Karen. (1996). The Confluence of Space and Language in Signed Languages. In Paul Bloom et al. (Eds.), Language and Space (pp. 171–209). Cambridge, Mass.: MIT Press. Ernst, Thomas. (1988). Chinese Postpositions? –Again. Journal of Chinese Linguistics, 16(2), 219–244. Fāng Jīngmín 方經民. (2004). Lùn hànyǔ kōnjiān qūyù fànchóu jíqí biǎodá fāngshì de yǔfǎhuà 論漢語空間區域範疇及其表達方式的語法化(Preprint). (Conference at Institute of Linguistics, CASS, March 2004). Heine, Bernd, Ulrike Claudi & Friederike Hünnemeyer. (1991). From Cognition to Grammar– Evidence from African Languages. In Elizabeth Closs Traugott & Bernd Heine (Eds.), Approaches to grammaticalization Vol. 1 (pp. 149–187). Amsterdam: John Benjamins Publishing Company. Hopper, P. J. & Sandra A. Thompson. (1980). Transitivity in grammar and discourse. Language, 56, 251–299. Jiāng Lánshēng 江 藍 生 . (1984). Gàishù cí lái de lìshǐ kǎochá 概 數 詞 來 的 歷 史 考 察 . Zhōngguó yǔwén 中國語文, 2, 145–153. Jiǎng Shàoyú 蔣紹愚. (1999). Hànyǔ dòngjié shì chǎnshēng de shídài 漢語動結式產生的時 代. Guóxué yánjiū 國學研究, 6, 327–347. Běijīng Dàxué chūbǎnshè 北京大學出版 社. Kang, Jian, (2001). Perfective Aspect Particles or Telic Aktionsart Markers? – Studies of the Directional Verbs Compounds. Journal of Chinese Linguistics, 29(2), 281–339.

196

D. Xu

Lamarre, Christine. (2005). Tǎolùn yí ge fēi diǎnxíng de shùqūshì：’zǒuqù lèi zǔhé 討論一個 非典型的述趨式：’走去’類組合. In yǔfǎhuà yǔ yǔfǎ yánjiū èr 語法化與語法 研究 二 (pp. 53–68). Běijīng：Shāngwù yìnshū guǎn. 北京：商務印書館. Levinson, Stephen C. (2003). Space in Language and Cognition: Exploration in Cognitive Diversity. Cambridge: Cambridge University Press. Lǐ chóng xīng 李崇 . (1990). Zǔ táng jí zhōng de zhùcí qù 祖堂集中的助詞去. In Jiǎng Shàoyú 蔣紹愚，Jiāng Lánshēng 江藍生 (Eds.), Jìndài Hànyǔ yánjiū èr 近代漢 語研究 二 (pp. 214–221). 北京：商務印書館， Běijīng：Shāngwù yìnshū guǎn. Lǐ Míng 李明. Qūxiàng dòngcí láiqù de yòngfǎ jíqí yǔfǎhuà 趨向動詞來去的用法及其語法 化. Yǔyánxué lùncóng 語言學論 , 29, 291–313. Běijīng: Shāngwù yìnshūguǎn 北京: 商務印書館. Lǐ Rúlóng 李如龍. (1996). Quánzhōu fāngyán de tǐ 泉州方言的體. In Zhāng Shuāngqìng 張 雙 慶 (Ed), Dòngcí de tǐ 動 詞 的 體 (pp. 195–224). Xiānggǎng: Xiānggǎng Zhōngwén Dàxué Zhōngguó wénhuà yánjiūsuǒ Wúduōtài Zhōngguó yǔwén yánjiū zhōngxīn 香港: 香港中文大學中國文化研究所吳多泰中國語文研究中心. Liú Dānqīng 劉丹青. (2001). Fāngsuǒ tíyuán de ruògān lèixíng xué cānxiàng 方所題元的若 干類型學參項. Zhōngguó yǔwén yánjiū 中國語文研究, 1, 11–23. Liú Dānqīng 劉丹青. (2003). Yǔxù lèixíng xué yǔ jiècí lǐlùn 語序類型學與 詞理論. Běijīng: Shāngwù yìnshūguǎn 北京: 商務印書館. Liú yuèhuá 劉 華 (Ed.). (1998). Qūxiàng bǔyǔ tōng shì 趨向補語通 . Běijīng: Běijīng yǔyán wénhuà dàxué chūbǎnshè 北京: 北京語言文化大學出版社. Mǎ Bèijiā 馬貝加. (2002). Jìndài hànyǔ jiècí 近代漢語 詞. Běijīng: Zhōnghuá shūjú 北京: 中華書局. Munnich, Edward et al. (2001). Spatial Language and spatial representation: a cross-linguistic comparison. Cognition, 81, 171–207. Peyraube, Alain. (2003). On the History of Place Words and Locatizers in Chinese: a Cognitive Approach. In Yen-Hui Audrey LI & Andrew Sinpson (Eds.), Functional Structure, Form and Interpretation (pp. 180–198). London: Routledge-Curzon. Qí Hùyáng 齊滬揚. (1998). Xiàndài hànyǔ kōngjiān wèntí yánjiū 現代漢語空間問題研究. Shànghǎi: xuélín chūbǎnshè 上海：學林出版社. Senft, Gunter. (1992). Everything we always thought we knew about space-but did not bother to question. Working Paper. No 10. Nijmegen: Cognitive Anthropology Research Group, Max Planck Institute for Psycholinguistics. Shěn Jiāxuān 沈家煊. (1999). Bù duìchèn hé biāojì lùn 不對稱和標記論. Nánchāng ： Jiāngxī jiàoyù chūbǎnshè. 南昌：江西教育出版社. Slobin, Dan I. (2001). Language and thought online: cognitive consequences of linguistic relativity (draft). Slobin, Dan I. (2004). The many ways to search for a frog: Linguistic typology and the expression of motion events. In Strömqvist & Verhoeven (Eds.), Relating events in narrative: Vol. 2. Typological and contextual perspectives (pp. 219–257). London: Lawrence Erlbaum. Svorou, Soteria. (1993). The Grammar of Space. Amsterdam Philadelphia: John Benjamins Publishing Company. Svorou, Soteria. (2002). Semantic constraints in the grammaticalization of locative constructions. In Ilse Wischer & Gabriele Diewald (Eds.), New Reflections on Grammaticalization (pp. 121–142). Amsterdam/Philadelphia: John Benjamins Publishing Company.

Asymmetry in the Expression of Space in Chinese

197

Talmy, L. (2000). How Language Structures Space. Toward a Cognitive Semantics, Vol. 1, 177–254. Cambridge, Mass.: MIT Press. Talmy, L. A Typology of Event Intergration. Toward a Cognitive Semantics, Vol. 2, 213–288. Cambridge, Mass.: MIT Press. Traugott, E. C. (1978). On the Expression of Spatio-Temporal Relations in Language. In Greenberg et al. (Eds.), Universals of Human Language (pp. 369–400). Stanford: Stanford University Press. Wáng Cànlóng 王燦龍. (2004). Qǐqù de yǔfǎhuà wèi wánchéng jíqí rènzhī dòngyīn 起去的語 法化未完成及其認知動因. Shìjiè hànyǔ jiàoxué 世界漢語教學, 27–37. Wáng Guóshuān 王國栓. (2003). Qù cóng lí yì dào wǎng yì de biànhuà shìxì 去從離義到往 義的變化試析. Yǔyánxué lùncóng 語言學論 , 27, 324–328. Běijīng: Shāngwù yìnshūguǎn 北京: 商務印書館. Wáng Jǐnhuì 王錦慧. (2004). Wǎng lái qù lìshí yǎnbiàn zōng lùn 往來去歷時演變綜論. Taipei: Lǐrén shūjú 臺北：里仁書局. Wú Fúxiáng 吳福祥. (1996). Dūnhuáng biànwén yǔfǎ yánjiū 敦煌變文語法研究. Chángshā: Yuèlù chūbǎnshè 長沙: 岳麓出版社. Wǔ Yúnjī 伍云姬. (1996). Chángshā fāngyán dòngtài zhùcí qùlái hé gādā de duìlì yǔ hùbǔ 長 沙方言動態助詞去來和咖噠的對立與 補. In Zhāng Shuāngqìng 張雙慶 (Ed.), Dòngcí de tǐ 動詞的體 (pp. 114–124). Xiānggǎng: Xiānggǎng Zhōngwén Dàxué Zhōngguó wénhuà yánjiūsuǒ Wúduōtài Zhōngguó yǔwén yánjiū zhōngxīn 香港: 香 港中文大學中國文化研究所吳多泰中國語文研究中心. Xíng Fúyì 福義. (2002). Qǐqù de pǔ fāng gǔ jiǎnshì 起去的 方古檢視. 方言 Fāngyán, 97–107. Xú Dān 徐丹. (1988). Qiǎn tán zhè/nà de bù duìchèng xìng. Zhōngguó Yǔwén 中國語文, 2, 128–130. Xú Dān 徐丹. (1994). Guānyú hànyǔ lǐ dòngcí+X+dìdiǎncí de jùxíng 關於漢語裏動詞+X+ 地點詞的 型. Zhōngguó Yǔwén 中國語文, 3, 180–185. Xu, Dan (1996). Initiation à la syntaxe chinoise. Paris: l’Asiathèque. Xú Dān 徐丹. (2000). Dòngbǔ jiégòu zhōng de shàng zì yǔ xià zì 動補結構中的上字與下字. Yǔfǎ yánjiū hé tànsuǒ 語法研究和探索. Běijīng: Shāngwù yìnshūguǎn 北京: 商務 印書館. 10, 112–120. Xú Dān 徐 丹 . (2001). Cóng dòngbǔ jiégòu de xíngchéng kàn yǔyì duì jùfǎ jiégòu de yǐngxiǎng 從動補結構的形成看語義對 法結構的影響. Yǔwén yánjiū 語文研究, 2, 5–12. Xú Dān 徐丹. (2005). Qūxiàng dòngcí lái/qù yǔ yǔfǎhuà 趨向動詞來去與語法化. Guóxué yánjiū 國學研究, 14, 315–330. Běijīng Dàxué chūbǎnshè 北京大學出版社. Xú Dān 徐丹. (2006). Mǒuxiē chūtǔ jiǎnbó wénxiàn lǐ suǒ jiàn de qūxiàng dòngcí lái hé qù de yòngfǎ 某些出土文簡帛文獻裏所見的趨向動詞來和去的用法. 簡帛語言文字研 究 Jiǎnbó yǔyán wénzì yánjiū, 2, 2–14. 成都：巴蜀書社 Chéngdū：Bā-Shǔ shūshè. Xu, Dan. 2006. Typological change in Chinese syntax. Oxford University Press.

198

D. Xu

Yáng Kèdìng 楊克定. (1992). Cóng Shìshuō xīnyǔ Sōu shén jì děng shū kàn Wèi Jìn shíqī dòngcí lái qù yǔyì biǎodá hé yǔfǎ gōngnéng 從世說新語搜 記等書看魏晉時期 動詞來去語義表達和語法功能. In Chéng Xiāngqīng 程湘清 (Ed.), Wèi Jìn Nán Běi cháo hànyǔ yánjiū 魏晉南北朝漢語研究 (pp. 240–275). Jǐnán: Shāndōng jiàoyù chūbǎnshè 濟南：山東教育出版社. Zhū Měifāng 朱美芳. (2003). L’emploi des verbes Lai et Khi en minnan taiwanais. Mémoire de DEA, INALCO, France (unpublished). Zhù Mǐnchè 敏徹 & Shàng Chūnshēng 尚春生. (1984). Dūnhuáng biànwén zhōng de jǐ ge xíngwéi dòngcí 敦煌變文中的幾個行爲動詞. Yǔwén yánjiū 語文研究, 1, 32–34. Zhōngguó shǒuyǔ 中國手語. (2003). Zhōngguó lóng rén xiéhuì biān 中國聾人協會編. Huáxià chūbǎnshè 華夏出版社. Zhōng Zhàohuá 鈡兆華. (1988). Dòngcí qǐqù hé tā de xiāoshī 動詞起去和它的消失. Zhōngguó Yǔwén 中國語文, 5, 380–385.

TWO CONDITIONS AND GRAMMATICALIZATION OF THE CHINESE LOCATIVE* Chaofen Sun Asian Languages Stanford University Keywords: Chinese locative construction, grammaticalization, semantic change, spatial terms, prepositional phrases Abstract: This paper argues that there is neither postposition nor circumposition in Standard Modern Chinese that is typologically similar to a Maya language Yukatec where there are only a few spatial adpositions but many spatial nominals playing a more important role. A selectional condition and a multi-syllabic constraint are proposed to account for the distributions of the Chinese locative construction that includes a semantically underspecified preposition zài, a spatial NP, and an NP enclitic. These two conditions define the Chinese locative construction by placing the definite locative, specific locative and general locative on a non-discrete continuum due to various semantic or pragmatic, as well as historical, reasons. Historical evidence will be given to demonstrate this grammaticalization process, or the formation of the two proposed conditions, in the history of Chinese as a renewal process.

1. Introduction This paper focuses on the structural status of the Chinese locative construction (in Standard Modern Chinese) and proposes two conditions: a selectional condition and a multi-syllabic constraint; showing step by step how they played out in the grammaticalization of the Chinese locative construction as a renewal process. Furthermore, these two conditions can also define the prototypical Chinese locative construction on a continuum that includes three non-discrete categories: definite locative, specific locative, and general locative. In the available literature, the syntactic status of the locative particles (or localizers (see Chappell and Peyraube in this volume)) in present-day Chinese such as shàng 上 ‘up,’ xià 下 ‘down,’ lĭ 里 ‘in,’ etc. has long been controversial. On the one hand, many (Peyraube 1980, Li and Thompson 1981, Ernst 1988, Wu 2005) treated the locative particles as postpositions. Li (1990) nevertheless treated them as nouns. F. Liu (1998), on the other hand, argued that they should be properly treated as NP clitics. Recently, D. Liu (2003 and in this volume) proposed still another view arguing that prepositional phrases like zài N shàng/xià/etc. given in (1) constitute circumpositions similar to those in Amharic observed by Greenberg (1995). (1)

a. 她在山上看日出 tā zài shān shàng kàn rìchū 3rd at hill up see sun-out ‘She watched the sunrise on the hill.’

199 D. Xu (ed.), Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives, 199–227. © Springer Science +Business Media B.V. 2008

200

C. Sun

b. 她在山 种地 tā zài shān xià zhòngdì 3rd at hill down grow-ground ‘She worked in the field at the bottom of the hill.’

In this paper, certain distributional properties and typological features of the Chinese locative particles will be examined in reference to certain crosslinguistic observations to show that they are neither postpositions nor circumposition such as those in Amharic. The data in (2) show that the nominal shān is a noun in (2b) and need not co-occur with any spatial particle like shàng if not occurring in a locative construction. However, after the locative preposition zài, it is unacceptable to occur without a spatial particle like shàng or xià in (1). (2)

a. *她在山看日出/种地 tā zài shān kàn rìchū/ zhòngdì 3rd at hill see sun-exit/grow-ground ‘She watched sunrise on the hill/farmed the land on the hill.’ b. 香港有山又有水 xianggang you shan you you shui Hong Kong have hill again have water ‘There are mountains and water in Hong Kong.’

Historical evidence will be given to demonstrate that a simpler and fuller account may follow from F. Liu’s clitic hypothesis (1998). The monosyllabic locative particles are all NP enclitics as they typically occur after a non-spatial nominal stem. The NP enclitics were originally full-fledged nouns that were grammaticalized into enclitics when the two conditions on the locative construction headed by the preposition zài were formed around the 14th century CE: a selectional restriction for either a definite NP, or a spatial NP (with syntactic significance, see below), and a multi-syllabic constraint on the head of the NP in a locative construction. Further-more, only one of the two conditions is applicable on the polar ends of of a proposed non-discrete scale of the Chinese locative construction due to various semantic and pragmatic, as well as historical, reasons. A Chinese spatial nominal can take either an argument position, shìjiè shàng ‘in the world’ in (3c), or an adjunct position together with a preposition such as zài shān shàng ‘on the hill’ in (1a) or zài shān xià (1b). J. Huang (1987) observes that there can be a selectional restriction between a spatial nominal and the main predicate such as the data in (3) show. Whereas the predicate shì 'to be’ in (3a) selects a non-spatial nominal shìjiè ‘the world,’ the existential predicate méiyŏu in (3c) selects a spatial nominal shìjiè shàng ‘in the world’. (3) a. 世界是你们的1 shì -jiè shì nĭ-mén de world-border be 2nd-men GEN ‘The world is yours.’

Two Conditions and Grammaticalization of the Chinese L ocative

201

b. *世界上是你们的 shì -jiè shàng shì ni-mén de world-border up be 2nd-men GEN c. 世界上没有无缘无故的爱 shì -jiè shàng méiyŏu wú-yuán-wú -gù de ài world-border up Neg-have no-reason-no-origin GEN love ‘There is no such thing as love without reason or cause in the world.’ d. *世界没有无缘无故的爱 shì -jiè méiyŏu wú-yuán-wú -gù de ài world-border Neg-have no-reason-no-origin GEN love. Although in many cases the necessary occurrence of a spatial nominal is primarily determined by the main verb, the goal of this paper is, nevertheless, to uncover the structural nature that motivated the historical developments of the Chinese locative construction in its common adjunct position. The relationship between a predicate and a spatial nominal in argument position, though relevant, falls out of the focus of this paper. Furthermore, how different spatial nominals interact with various geometric, or topological, meanings (Fang 2004) such as shān shàng ‘on the hill’ and shān xià ‘under the hill’ in (1) is also beyond the scope of the current paper. This paper focuses on uncovering the regularity, and irregularities, between a generic preposition zài and its ensuing NPs that together make up the Chinese locative construction. For the sake of clarity and ease of illustration, the two conditions proposed in this paper are solely based on the data relating to the generic locative preposition zài and its following NPs. Even though the two conditions are applicable to other Chinese spatial prepositional phrases headed by spatial prepositions such as cóng ‘from,’ xiàng ‘toward,’ and dào ‘to/at,’ etc., I will not go into detailed discussion of them here in order to maintain the set focus of this paper. The historical account of the NP enclitics developing from nominals will be presented within the theoretical framework of grammaticalization as has been advocated by Hopper and Traugott (2003) and Himmelmann (2004). The modern Chinese locative construction as defined by the phonological constraint and selectional restriction originated in analogy to the constraint that first came to exist for place names in Old Chinese. That is, spatial verbs, the host-class of the construction in Himmelmann’s terminology, collocating with a special type of nominal, which is called spatial nominals in this paper, expanded to include verbs of other kinds. Subsequently, zài at first commonly used as the main verb of a sentence to host spatial nominals began to show up more and more in a verbal series and ultimately became grammaticalized as a generic, underspecified preposition marking a locative phrase in adjunct position. Furthermore, Old Chinese spatial nominals like shàng ‘upside’ and xià ‘downside,’ etc. became grammaticalized into NP enclitics (perhaps as G2 in Brinton and Traugott’s (2005) framework) marking a derived specific spatial NP and other more abstract meanings. Finally, in some specific contexts, the selectional restriction became optional for the general locative (listing

202

C. Sun

and contrastive) and the multi-syllabic constraint became optional for the directionbearing locative, a type of definite locative (see section 3). This paper is divided into four sections. Section 2 deals with the typological features of the Chinese locative construction and proposes two conditions for the construction. Section 3 discusses various historical factors in the making of the presentday Chinese locative construction over time as a renewal process of grammaticalization and proposes a non-discrete semantic continuum to account for various pragmatically motivated uses of the present-day Chinese locative construction. Section 4 is the summary. 2. Typology In this section, I will first talk about the problems in treating the Chinese NP enclitics as postpositions or parts of circumpositions. Second, I will address the syntactic differences between what Greenberg observed about Amharic and the Chinese history, and, third, why these two are not comparable cases. I will then argue that the Chinese locative construction, instead, should be treated as similar to a corresponding construction in a Mayan language like Yukatec that also has many important spatial nominals with only a few spatial prepositions (Levinson et al. 2003). 2.1. Problems with a postpositional treatment The biggest problem for the postpositional, and circumposition, hypothesis is that the Chinese locative particles are not always allowed in a locative construction (4a). This is then unexplainable by a circumposition hypothesis which by definition treats zài …shàng/xià/etc. as two discontinuous constituents of a single unit. If the hypothesis were correct, the two constituents should have always co-occurred. The data in (4), instead, imply that the preposition zài selects a special type of noun rather than forming a circumposition with the spatial particles. In light of the data given in (1) and (2), it follows that the place names and NPs marked by the spatial enclitics such as shàng/xià/nèi etc. are in complementary distribution, two of the same kind occurring after the preposition zài. That is, locative particles like shàng/xià/nèi occur in the construction of an NP that makes a non-place name noun into a category similar to it. It then follows to hypothesize that the NPs with a spatial enclitic are structurally equivalent to the place names in Chinese. (4) a. 他在美国工作 tā zài měiguó gōngzuò 3rd at USA work ‘He worked in the USA’. b. *他在美国内工作 tā zài měiguó nèi gōngzuò 3rd at USA in work. The data in (5) are also troublesome for both circumposition and postpositional hypotheses. The sentence in (5a) allows a phonetic pause between the preposition

Two Conditions and Grammaticalization of the Chinese Locative

203

zài and the following unit guó nèi, and the one in (5b) does not. This once again shows that guó and nèi form a phonological unit, whereas zài and nèi do not. Therefore, this is another piece of evidence in favor of treating the elements after zài as one syntactic unit, most probably a type of NP. (5)

a. 他在__国内工作 tā zài __ guó nèi gōngzuò 3rd at state in work ‘He works in (his) home country.’ b. 他在国* __ 内工作 tā zài guó __ nèi gōngzuò 3rd at state in work.

__ indicates a pause in (5)

Furthermore, according to Zwicky (1985), clitics are typically phonologically dependent and form a phonological unit with an independent word. The pause pattern revealed by the data in (5) then suggests nèi behaves more like a clitic forming a phonological unit, as a dependent, with the independent word guó ‘state.’ As the spatial particles always occur after a nominal stem, they are, thus, treated as NP enclitics in this paper. The data in (6) confirm that guó nèi ‘(home) country’ and měiguó ‘the USA’ can participate in a syntactic operation as two structurally equivalent NPs coordinated by a conjunctive marker hé, functioning as two NP objects of the preposition zài. Moreover, prosodically, the morpheme guó ‘state’ and the enclitic nèi ‘in’ form a phonological unit just like the compound měiguó ‘USA’. (For a more detailed discussion, see F. Liu 1998.) (6) a. 他在国内和美国都工作过 tā zài guó nèi hé měiguó dōu gōngzuò-guò 3rd at state in and USA all work-ASP ‘He has the experience to work in (his home) country and in the USA.’ b. PP[在 NP[N[国]enclitic[内]] == PP[在 NP[美国]] zài guó nèizài měiguó at state-inat USA ‘in home country’ ‘in USA’. c. pp[在 phonological unit[[国内]] == pp[在 phonological unit [美国]]. Recent proponents for both circumposition (D. Liu 2003 and in this volume) and postposition treatments (Wu 2005) cited Greenberg’s (1995) theory that the emergence of circumposition had followed the word-order changes from Noun Adjective to Adjective Noun and from Noun Genitive to Genitive Noun in Amharic. At first glance there does appear a strong resemblance between the Amharic (7a) and Chinese (7b) locative constructions, both of which begin and end with a different spatial term. However, also according to Greenberg in the same paper, the Amharic prepositions such as bä- in (7a) are both older than the postposition wəst and bound to the following noun. On the other hand wəst ‘interior’ is not only

204

C. Sun

newer but also free. However, in Chinese the so-called postpositions and the lexeme zài are equally old (see section 3). The data given in (1) through (6) also demonstrate that the so-called postpositional elements are in construction of a noun phrase with another noun as an enclitic, a morpheme that is morphologically more bound than an adposition but freer than a bound morpheme. Furthermore, as a pause can take place between a preposition and a noun but not between a noun and an enclitic. It is then obvious that the Chinese and Amharic PPs have a different internal structure and are not alike at all. (7)

a. [[bä-bet] NP wəst]PP(Amharic) at house interior ‘inside (the) house’ b. [在 [[房] N [里]enclitic] NP]PP (Chinese) zài fáng lĭ at room in ‘inside (the) room’

Greenberg (1995: 155) further hypothesized that in Amharic ‘the first change is from VSO to SOV, followed by changes NA Æ AN, NG Æ GN, and finally preposition to postposition.’ That is, the postpositional wəst in Amharic presumably moved into the position following the head noun bet ‘house’ after a genitive marker of the language had moved out of it to the pre-head position. However, no Chinese word-order change of this kind can be found over the last 3,500 years during which the post nominal spatial terms emerged as a grammatical element. The data in (8) show that genitive markers, zhī (8a) for Old Chinese (700 BCE-200 CE) and de (8c) for Modern Chinese (after 19th century), throughout Chinese history occur before the head noun of an NP, thus showing no sign of any word-order change comparable to the hypothetical series of changes in Amharic. Without any comparable word-order change, no matter how similar the Amharic locative construction is, it is simply not a parallel case that would provide any immediate explanation of the Chinese data. Proponents of the Chinese postpositional treatment will need to explain independently how a grammatical position for the postposition came into being and how the process was internally motivated. (8)

a.

NP[

X 之 N](Old Chinese) 葬之[郐城之下](左传：僖公 33) zàng zhī kuài chéng zhī xià bury 3rd Name city GEN. downside ‘Bury him under the Kuai city wall.’ b. NP[ X enclitic](Modern Chinese) 在[西安城下] zài xī’ān chéng xià at Name city under ‘under the city wall of Xi’an’

Two Conditions and Grammaticalization of the Chinese Locative

c.

205

NP[

X 的 N](Modern Chinese) 把他埋在[西安城墙的下面] bă tā mái zài xī’ān chéngqiáng de xiàmiàn TOP 3rd bury at Name city-wall GEN. downside ‘Bury him under the city-wall of Xi’an.’

The xiàmian ‘bottom’ in (8c) is a Modern Chinese spatial nominal functioning as the head of the NP. In other words, unlike a language such as English where spatial references are closely associated with prepositions, similar spatial references in Old Chinese are signaled by nouns that take up the head position of an Old Chinese NP. In Modern Chinese, these Old Chinese spatial nominals are grammaticalized into enclitics retaining the grammatical status as a functional head of the NPs, such as the xià in (8b), even though it now forms a phonological unit with the preceded nominal stem chéng ‘city’. The Chinese enclitics such as shàng ‘on/up’ and xià ‘under’ in (1), nèi ‘inside’ in (5), lĭ ‘inside’ in (7b) are actually all functional heads of the NPs in which they occur. The NP enclitics are gram-maticalized from Old Chinese relevant spatial nominals such as the one in (8a), where the Old Chinese noun xià ‘downside’ marked by a relative marker zhī is grammaticalized into a Modern Chinese enclitic ‘under’. (The grammaticalization of spatial nominals will be further addressed in section 3) 2.2. The selectional restriction of the preposition zài In a study exploring ‘natural concepts’ in spatial topological domains crosslinguistically, Levinson et al. (2003) observed from a set of nine languages that there are languages that can have more than 100 spatial adpositions and only three to five unimportant spatial nominals such as the Cariban language Tiriyó. They also note that there are languages that may have relatively few spatial adpositions but many important spatial nominals. One of these is a Mayan language, Yukatec, that has only two to four (one generic) spatial adpositions and many important spatial nominals. Chinese is probably a language somewhat similar to Yukatec as it has no more than ten spatial prepositions, cóng ‘from,’ dào ‘to,’ lĭ ‘apart from,’ cháo ‘toward,’ xiàng ‘toward,’ zì ‘from,’ zài ‘at,’ wăng ‘toward,’ etc., and many important spatial nominals. In other words, there are no Chinese spatial preposition comparable to the English in, on, above, over, below, behind, across, etc. that embody relatively richer spatial meanings including some natural concepts in geometric or topological domains. As a matter of fact, in present-day Chinese similar topological, or geometric, meanings are typically signaled by the preposition zài and a spatial NP jointly in forming a complex locative construction. For example in (7b), the spatial PP in a room is expressed through a generic preposition zài together with a spatial NP fáng lĭ ‘room in,’ in which the NP enclitic lĭ expresses a topological meaning represented by one English preposition in. It is in this sense that the Chinese locative construction is a more complex, two-tier system. The data noted above in (4-7) also suggest that the underspecified, or generic, spatial preposition zài tends to co-occur with a spatial NP: either a place name or a non-spatial nominal marked by an NP enclitic. I will therefore expand the term to

206

C. Sun

include place names in Chinese on the ground of their ability to occur after the locative marker zài. Moreover, a selectional restriction for the Chinese locative construction is proposed in (9) that can be thought of as a kind of default condition on a prototypical Chinese locative construction. (9)

The selectional restriction of the preposition zài: In non-contrastive or nonlisting contexts, the underspecified preposition zài 在 normally selects either a. a spatial NP or b. an explicitly marked, definite and inanimate NP.

In the following, section 2.2.1 portrays the formation of the Chinese spatial NP, section 2.2.2 deals with the explicitly marked definite, inanimate NP, and the exceptional uses of the locative construction in contrastive and listing contexts will be dealt with together with the multi-syllabic constraint in sections 2.3 and 3. 2.2.1. Spatial NPs Chinese spatial NPs are composed of two types: place names and derived spatial nominals. In accordance with the selectional restriction in (9a), the nominal after the preposition zài can be either a place name (10a) or a derived spatial NP (10b). (10) a. 在北京 ài z běijīng at P.N. in ‘ Beijing’ b. 在城里 ài z chéng lĭ at city in in ‘ the city’ c. *在城, ài z chéng at city

在上海 zài shànghăi at P.N ‘in Shanghai’ 在河上 zài hé shàng at river up ‘on the river’ *在河 zài hé at river

在中国 zài zhōngguó at P.N ‘in China’ 在公园里 zài gōngyuán lĭ zai park in ‘in the park’ *在公园 zài gōngyuán at park

在纽约 zài niŭyuē at P.N ‘in New York’ 在桌子上 zài zhuōzi shàng zai table up ‘on the table’ *在桌子 zài zhuōzi, at table

The data in (10c) show that non-spatial nominals are generally not allowed to follow the underspecified preposition zài in the Chinese locative construction. But their corresponding counterparts in (10b) are all good. With the addition of the NP enclitics such as shàng and lĭ, the non-spatial nominals in (10c) are changed into spatial nominals in (10b), which will be identified as derived spatial nominals (an open class excluding animate nominals) in the subsequent discussion. In other words, the NP enclitics with clear spatial meanings are recruited to signal specific locational meanings that are lacking in the underspecified preposition.

Two Conditions and Grammaticalization of the Chinese Locative

207

2.2.2. An explicitly marked definite and inanimate NP The data in this sub-section show why condition (9b) is necessary. First of all, the head of the NPs in a locative construction is typically inanimate.2 Secondly, note that the NP enclitic shàng in (11a) is optional, as the sentence in (11b) is still good without it. What makes (11b) acceptable is the demonstrative quantifier phrase, zhèběn, that explicitly marks the NP as definite. However, without this explicit definite marking, the sentence in (11c) is not acceptable, in spite of the fact that a bare nominal in Chinese can be definite, such as shū ‘the book’ in (11d). (11) a. 我在这本书上加了很多插图 wŏ zài zhè -běn shū shàng jiā -le hěn duō chā-tú 1st at DEM-CL book up add -PFV very much insert-picture ‘I added to this book many illustrations.’ b. 我在这本书加了很多插图 wŏ zài zhè -běn shū jiā -le hěn duō chā-tú at DEM-CL book add -PFV very much insert-picture ‘I added to this book many illustrations.’ c. *我在书加了很多插图 wŏ zài shū jiā -le hěn duō chā-tú 1st at book add -PFV very much insert-picture d. 书他买了 shū tā măi le book 3rd buy PFV ‘The book, he bought it.’ The data in (12) show that the same definite effect can be signaled by an ordinal construct dì yī ‘the first.’ (12) a. 我在第一页上打了个记号 wŏ zài dì yī yè shàng dă le ge jìhào 1st at Ord. one page up hit ASP CL mark ‘I made a mark on the first page.’ b. 我在第一页打了个记号 wŏ zài dì yī yè dă le ge jìhào 1st at Ord. one page hit ASP CL mark ‘I made a mark on the first page.’ The data in (13b) show that a relative clause, such as gāng măi lái de ‘the one that (he) has just bought’ or a pronominal wŏ that explicitly marks the NP as definite enables an NP to occur after zài without a relevant spatial enclitic lĭ. (13) a. 他在刚买来的房子里搭了个阳台 tā zài gāng măi lái de fángzi lĭ dā le ge yángtái 3rd at just buy come GEN house in add –PFV CL balcony ‘He added a balcony to that house that he just bought.’

208

C. Sun

b. 他在刚买来的房子搭了个阳台 tā zài gāng măi lái de fángzi dā le ge yángtái 3rd at just buy come GEN house add –PFV CL balcony ‘He added a balcony to that house that he just bought.’ c. 他在我的房子搭了个阳台 tā zài wŏ de fángzi dā le ge yángtái 3rd at 1st GEN house add –PFV CL balcony ‘He added a balcony to my house.’ Therefore, a non-spatial nominal after the underspecified zài is acceptable only if it is explicitly marked as definite by a demonstrative, an ordinal marker, or a relative clause. The primary function of a locative construction in natural discourse is perhaps to locate a certain participant of an event to either a known, or easily identifiable, place to facilitate ease in communication. It should not then be a surprise that the nominal after the preposition zài tends to be either shared information, NPs with additional spatial reference (derived spatial NPs), or explicitly marked, definite NPs. 2.3. The multi-syllabic constraint In addition to the selectional restriction in (9), there is also a multi-syllabic constraint (14) on the non-directional nominals after the generic preposition zài. The directional nominals are, nevertheless, exempted from this phonological constraint. (14)

A phonological constraint on the NPs after the preposition zài: A non-directional NP after the preposition zài must not be monosyllabic.

The clearest evidence for such a constraint is the fact that no Chinese place name is monosyllabic. An additional piece of evidence is that even monosyllabic foreign place names are all represented multi-syllabically in Chinese. For instance, the Chinese names for the unincorporated territory of the United States Guam and the German city Bonn that are both monosyllabic in the original are transliterated by the disyllabic guāndăo 关岛 ‘Guam’ and bō’ēn 波恩 ‘Bonn’ respectively. However, as the selectional restriction (9) practically makes most NPs after the preposition zài multi-syllabic, one may question if the phonological constraint in (14) is necessary. But it is truly necessary for two reasons: A directional NP can be monosyllabic, and a multi-syllabic, but not a monosyllabic, non-spatial NP can be used without an explicit definite marker or a spatial enclitic in listing and contrastive contexts where more than one location is involved. The lexeme 在 zài can function either as a preposition or as a verb meaning ‘to exist’ exemplified by (15), where there are three clauses with zài functioning as the predicates of the first two clauses. When it is used as such, a set of direction-bearing nominals can occur grammatically without being subject to a multi-syllabic constraint (see section 2.3) like other spatial nominals. They can be monosyllabic, such as dōng ‘east,’ xī ‘west,’ nán ‘south,’ běi ‘north,’ shàng ‘top,’ xià ‘bottom,’ lĭ

Two Conditions and Grammaticalization of the Chinese Locative

209

‘inside,’ wài ‘outside,’ qián ‘front,’ hòu ‘back,’ zhèr ‘here,’ nàr ‘there,’ jiā ‘home,’ etc., or multi-syllabic, such as dōng-bian ‘east side,’ shàng-mian ‘upside,’ qíán-tou ‘front side,’ etc. However, when zài occurs in the adjunct position, a multi-syllabic spatial nominal is generally preferred, as is evidenced by the contrast between the monosyllabic hòu in the ungrammatical (16a) and the multi-syllabic hòumian in (16b). (15)

他在前，我在后，什么都看不见 tā zài qián, wŏ zài hòu, shěnme dōu kàn bú jiàn 3rd at front, 1st at back, what all look not see ‘He is at the front, and I am in the back, (therefore) I cannot see anything.’

(16) a. *我在后什么都看不见 wŏ zài hòu shěnme dōu kàn bú jiàn 1st at back what all look not see b. 我在后面什么都看不见 wŏ zài hòumian shěnme dōu kàn bú jiàn 1st at back-side what all look not see ‘I cannot see anything from behind.’ Therefore, it appears that the collocation constraints on the locative construction in adjunct position are more robust than those in argument position. However, there are three monosyllabic nominals, zhèr ‘here,’ nàr ‘there,’ and jià ‘home,’ that can occur in the adjunct position as part of a locative construction such as those in (17). (17) a. 我今天在家看书 wŏ jīntiān zài jiā kàn shū 1st today at home read book ‘I am reading at home today.’ b. 我今天在这看书 wŏ jīntiān zài zhèr kàn shū 1st today at this read book ‘I am reading here today.’ c. 我今天在那看书 wŏ jīntiān zài nàr kàn shū 1st today at that read book ‘I am reading there today.’ Note that zhè ‘this’ and nà ‘that’ are demonstratives that can function as explicit definite markers for the NPs selected by restriction (9b). Furthermore, zhèr ‘here’ and nàr ‘there’ are deitic and inherently definite by nature with a strong sense of direction like those in (16a) that can take up the argument position when following the verbal zài. Still another reason is that these three items are among the

210

C. Sun

most commonly used vocabulary. They have been used as such for such a long time so commonly that they constitute a layering phenomenon that Hopper and Traugott (2003) refer to as the co-existence of older forms and meanings with newer forms and meanings. The two conditions came into being around the 14th century (see below). With respect to jiā ‘home,’ which is a term closely related to all human beings, it is entirely possible that all human beings are conceptually taken to have a home to return to after a day’s activities. As everyone will have a different home, except for those who form a common home, a home can be construed to be a definite and unique one for each. Jiā ‘home’ may then be a goal-orientated directional term evidenced by the English phrases, ‘to head home’ in which home can be substituted by other directional words such as north in ‘to head north,’ etc. Chinese directional spatial terms are not subject to the multi-syllabic constraint in adjunct position and satisfy the selectional restriction without any additional marking as they are universally definite. For instance, in (18), after the directional preposition wàng ‘toward,’ both nominals, běi ‘north’ and jiā ‘home,’ are monosyllabic. (18) a. 我往北走 wŏ wàng běi zŏu 1st toward north walk ‘I walked toward north.’ b. 我往家走 wŏ wàng jiā zŏu 1st toward home walk ‘I walked toward home.’

As a matter of fact, Chu (2004) noted that a non-spatial nominal hēibăn ‘blackboard,’ which is nevertheless multi-syllabic in (19a), occurs after the underspecified preposition zài grammatically without a spatial enclitic. Although sentence (19c) with a derived spatial nominal, zài hēibăn shàng ‘on the blackboard,’ is perfectly good, the sentence (19a) with a disyllabic non-spatial nominal hēibăn appears to be just as good. In light of the unacceptable sentence (19b) because of the monosyllabic non-spatial nominal shū ‘book’ in the locative construction, it appears that the multi-syllabic constraint is necessary. (19) a. 你在黑板写，我在书上写 nĭ zài hēibăn xiě, wŏ zài shū shàng xiě 2nd at blackboard write, I at book up write ‘You write on the blackboard, and I write on the book.’ b. *你在黑板写，我在书写 nĭ zài hēibăn xiě, wŏ zài shū xiě 2nd at blackboard write, I at book write

Two Conditions and Grammaticalization of the Chinese Locative

211

c. 你在黑板上写，我在书上写 nĭ zài hēibăn shàng xiě, wŏ zài shū shàng xiě 2nd at blackboard up write, I at book up write ‘You write on the blackboard, and I write on the book.’ A closer examination of the data resembling (19) reveals that this kind of usage without observing the selectional restriction is possible in contrastive contexts where more than one location is involved. The example in (20) exhibits that exceptions to selectional restriction (9) are also allowed in listing locations without identifying a specific one. (20)

我们常常在电梯，在超级市场，在很多地方都能看到这广告。 wŏmen chángchang zài diàntī, zài chāojí shìchăng, zài hěnduō dìfāng kàndào zhe guănggào 1st-PL often at elevator, at super market, at very much places see-reach this advertisement ‘We frequently see this ad in elevators, supermarkets, and many other places.’

It is necessary to emphasize that the examples in (19–20) do not necessarily falsify selectional restriction (9). Rather, the restriction matches up with the general function of a locative construction, i.e., to locate a certain participant of an event in reference to either a known or easily identifiable place to facilitate ease in communication. Furthermore, the phonological condition is clearly applicable to all non-directional nominals in the locative construction. All this simply suggests that additional discourse factors might be at work in these cases. 3. The Grammaticalization of the Complex Locative Construction In section 2, I have shown that, pragmatic and historical factors aside, there are two conditions, selectional restriction (9) and multi-syllabic constraint (14), controling various uses of the Modern Chinese locative construction. In this section I will discuss how they came into being as a renewal process within a grammaticalization framework. Wu (2005) proposes that the Chinese spatial particles were grammaticalized into postpositions because of a universal tendency in languages with a genitive noun word order. However, in section 2.1, I have shown that without a similar word-order change motivating the emergence of postpositions, no explanation of the Chinese data would follow from the Chinese genitive noun word order. It is, therefore, not clear how and from where Chinese developed a grammatical slot for the postposition. In this section, I am proposing a simpler theory to show how, step by step, the spatial nominals developed and how some spatial nominals are grammaticalized into NP enclitics that at first maintain a binominal complex structure and function as the head of the derived spatial nominals, as is schematized in (21), and subsequently lose the binominal structure in special uses in some discourse contexts.

212

(21)

C. Sun

於山上 (Old Chinese) [yú [[shān]N [shàng]N]NP]PP at hill top ‘on the hill’

Æ

在山上 [zài[[shān]N[shàng]enclitic]NP]PP (Modern Chinese)

The changes of a nominal into an enclitic described in section 2.1 reveal that the formation of the locative construction is definitely a grammaticalization process. Semantically, the spatial enclitics are not always the same as an ungrammaticalized nominal. The data in (22–24) demonstrate that the grammaticalized enclitic shàng always has a more abstract meaning than the corresponding lexical nouns. The pair in (22) exhibits that the spatial nominal shàngmian marked by a relative marker and the spatial nominal shàng can be synonymous in referencing physical spatial meaning. (22) a. 在报纸的上面有枝笔 zài bàozhĭ de shàngmian yŏu zhī bĭ at paper REL top exist CL pen ‘There is a pen on the newspaper.’ b. 在报纸上有枝笔 zài bàozhĭ shàng yŏu zhī bĭ at paper top you zhi bi ‘There is a pen on the newspaper.’ However, the enclitic shàng can signal a more abstract part-whole relationship, whereas the lexical noun shàngmian does not. (23) a. 在报纸上有条中国新闻 zài bàozhĭ shàng yŏu tiáo zhōngguó xīnwén at paper top exist CL China news ‘There is a piece of news about China in the newspaper.’ b. *在报纸的上面3有条中国新闻 zài bàozhĭ de shàngmian yŏu tiáo zhōngguó xīnwén at paper REL top exist CL China news. Similarly, the enclitic shàng can imply a more abstract meaning of ‘in the area of,’ whereas the lexical noun shàngmian does not. (24) a. 这个问题在理论上不难解决 zhè-ge wéntí zài lĭlùn shàng bū nán jiĕjué DEM-CL problem at theory up NEG difficult solve ‘This problem is not difficult to solve in theory.’ b. *这个问题在理论的上面不难解决 zhè-ge wèntí zài lĭlùn de shàngmian bū nán jiĕjué DEM-CL problem at theory top NEG difficult solve

Two Conditions and Grammaticalization of the Chinese Locative

213

In the following, I will discuss how grammaticalization of the locative construction played out in Chinese history. Section 3.1 deals with an analogical change that motivated the multi-syllabic constraint. Section 3.2 argues that the selectional constraint follows from the multi-syllabic constraint as the verbal zài ‘to exist’ becomes grammaticalized into a generic preposition heading a PP most frequently in the adjunct position. 3.1. Multi-syllabic constraint and the formation of spatial NPs In Old Chinese the locative construction at the time did not have any syllabic constraint (see Chappell and Peyraube in this volume). In (25) the monosyllabic xià was a full-fledged lexical noun as it can be marked by a genitive marker zhī in (25a) or followed another noun chéng ‘city’ directly to form a locative construction in (25b) without the Old Chinese preposition yú. Furthermore, the lexeme xià occurs as a noun denoting a topological sense corresponding to the English preposition under. That is, spatial meanings of this kind in Chinese traditionally are represented by nominals and are typologically different from that in English which uses preposition to signal such meanings. This is then the ultimate reason why spatial nominals are so important in present-day Chinese. (25) a. 葬之郐城之下(左传：僖公 33) zàng zhī kuài chéng zhī xià bury 3rd Name city GEN. downside ‘Bury him under the city wall of Kuai.’ b. 齐梁之兵连于城下(史记：张仪列传) qí liáng zhī bīng lián yú chéng xià Name Name GEN soldier connect at city downside ‘The troops of Qi and Liang engaged under the city wall.’ Cheng (1992) observed that in the history of Chinese a process of multisyllabification started roughly in Old Chinese (700 BCE) and gained momentum significantly about the time of the Qin dynasty (221–207 BCE). Pertaining to the history of spatial nominals, place names in Old Chinese already became predominantly multi-syllabic at the beginning of the Common Era. According to the historical maps by Tan (1982), only during the Shang dynasty (before 11th century BCE) were place names in Central China, zhōng yuán 中原 in Chinese, predominantly monosyllabic (approximately 4:1). But in the 700 years after the Shang dynasty, the number of monosyllabic place names decreased dramatically and became extremely rare by the time of the Western Han. As a matter of fact, in the map of the Eastern Han (25 CE-220) produced by Tan, not a single monosyllabic place name can be found in the area under investigation.

214

C. Sun

(26) The Old Chinese place names in the area of 340–360 north and 1110–1160 east: Time Shang dynasty (1100 BCE) Autumn and Spring (777–476 BCE) Qin dynasty (221–207 BCE) Western Han dynasty (206 BCE-24 CE)

Monosyllabic 40

Multi-syllabic 9

42

61

4

78

1

138

Right about this time, according to Li (1992) and Zhang (2001), occurrences of complex noun phrases like chéng xià in (25b) became more and more numerous. The table in (27) shows that the multi-syllabic nominals became more and more closely associated with the lexeme zài, which later changed into a generic preposition in Chinese. In the very Old Chinese text Zuŏzhuàn, the ratio between multi-syllabic and monosyllabic nominals after zài was about the same. But the frequency of the multi-syllabic NPs had increased to 21% higher than the frequency of monosyllabic ones in the Middle Chinese text Shìshuō xīnyŭ (3rd century CE). (27) The development of derived spatial nominals after zài: Texts Zuŏzhuàn 左传 隐、桓、庄、闵、僖、文 Shĭjì 史记 117-128 Shìshuō xīnyŭ 世说新语 德行、言语、政事、文学

Multi-syllabic NP Monosyllabic NP Without NP 29 (47%)

27 (44%)

6 (10%)

27 (44%)

17 (29%)

7 (12%)

34 (49%)

20 (28%)

17 (24%)

In consideration of the findings characterized in (26) and (27), it makes sense to hypothesize that the development of the derived spatial nominals with the multisyllabic constraint after the lexeme zài ‘to exist,’ is a process of change analogous to that of the place names which by this time had already become multi-syllabic. An additional piece of evidence for such a hypothesis is the common strategy for forming both place names and derived spatial nominals. That is, they both utilize a spatial term as the second morpheme, as exemplified by the data in (28), in which the spatial term xià ‘downside’ occurs in the same position of a binominal structure. The difference is that whereas the directional, topological, or geometric, meanings need not always be transparent for the place names, such meanings are almost always obvious for the derived spatial NPs. In Modern Chinese the formation of place names with a nominal morpheme and a spatial term is still a highly productive strategy (for more examples, see (33) below ) .

Two Conditions and Grammaticalization of the Chinese Locative

215

(28) The analogy between place names and derived spatial NPs in Shĭjìmorpheme+spatial term place names:

spatial NPs:

稷下 jì-xià 垓下 gāi-xià 戏下 xì-xià

NP[N

+ down] + down] NP[N + down]

place name place name place name

城下 chéng xià

NP[N

+ down]

阶下 jiē xià

NP[N

+ down]

桑下 sāng xià

NP[N

+ down]

‘the foot of the city wall’ ‘the bottom of the steps’ ‘the bottom of the tree’

NP[N

This analogical hypothesis is then superior to the postpositional hypothesis (Wu 2005) in that the enclitic position follows from the grammaticalization process and need not look for additional motivation for the postpositions. It is, therefore, a simpler solution. 3.2. The formation of the Modern Chinese locative construction The analogical process proposed in the previous section took place over a very long period of time. The data in (29) with both monosyllabic and multi-syllabic nominals after zài in adjunct position show that the locative construction was not subject to the multi-syllabic constraint in Middle Chinese. (29) a. 殷在妾房眠 世说新语：言语 (3rd century CE) yīn zài ciè fáng mián N at concubine room sleep ‘Yin slept in the concubine’s room.’ b. 在道所经 世说新语：言语 zài dào suŏ jīng at road PRON. Pass ‘What one passed on the road.’ The situation remains nearly the same in the Early Modern Chinese text of Zhūzĭ Yŭlèi (13th century CE), exemplified by the data in (30), where shān-dĭng shàng (30a) is a tri-syllabic, spatial nominal at the argument position, while lĭ (30b) and yī (30c) are monosyllabic in adjunct position. (30) a. 便能一跃在山顶上 朱子语类：卷一百六 biàn néng yī yuè zài shān-dĭng shàng then can one jump at hill top up ‘(One) can then jump all the way to the top of the hill.’

216

C. Sun

b. 终日在里默坐 朱子语类：卷一百一十三 zhōng-rì zài lĭ mò zuò all-day at inside quiet sit ‘Sit quietly inside all day long.’ c. 在邑恣行 朱子语类：卷一百一十二 zài yī sì xíng at city unscrupulous practice ‘(Someone) behaved unscrupulously in the city.’ The situation changes to the modern pattern after the Yuan dynasty (14th century). In the texts of Lǎoqĭdà Yànjiĕ and Piǎotōngshì Yànjiĕ, Chinese language textbooks compiled in Korea for Korean diplomats and merchants at the time, the nominals used after zài in the adjunct position revealed a pattern that is consistent with the pattern of the present-day Chinese. The data in (31) are all taken from the Korean textbooks. In (31a and b), zài-qián ‘before’ and zài-xiān ‘past’ that have gone out of fashion in present-day Chinese are frozen expressions used adverbially indicating time. Frozen expressions aside, the Chinese locative construction at the time patterns essentially like its present-day counterpart exemplified by the sentence in (31d). The two sentences in (31d and e) are two cases found in the Korean Chinese textbook in which the monosyllabic jiā ‘home’ follows the lexeme zài in the argument position, which is also consistent with the present-day Chinese usage. (31) a. 比在前十分好(老乞大谚解) bĭ zài-qián shífēn hǎo COMP. at-front complete good ‘(It) is better than before.’ b. 在先只是土搭的桥(老乞大谚解) zài-xiān zhĭshì tŭ dā de qiáo at-early only mud pile GEN. bridge ‘The bridge that was made by mud in the past …’ c. 在辽东城里住(老乞大谚解) zài liáodōng chéng lĭ zhù at Place N. city inside live ‘(I) lived inside the city of Liaodong.’ d. 我相公不在家(朴通事谚解) wŏ xiànggōng bú zài jiā 1st master NEG exist home ‘Our master is not home.’ e. 小人每日不在家(朴通事谚解) xiǎorén mĕirì bú zài jiā little man yesterday NEG exist home ‘I was not home yesterday.’ In the 17th century Ming dynasty novel Jĭngshì tōngyán by Feng Menglong, zài continues to have the dual functions in marking a locative construction in the

Two Conditions and Grammaticalization of the Chinese Locative

217

adjunct position or taking up the predicate position as a verb. As far as the grammatical status of the locative construction is concerned, except for a few idiomatic expressions, the two conditions appear to be well in place at this time (see 32a and b). The example in (32e) that has a monosyllabic non-spatial nominal, shēn ‘body,’ in the argument position is another case of set expression. The monosyllabic NP in adjunct position (32c and d) is a demonstrative pronoun cĭ, a variant of the modern zhèr consistent with the modern pattern in (17). Therefore, it appears that the multi-syllabic constraint is more entrenched in the adjunct locative construction than in the argument position. (32) a. 准在江边侍立供候 警世通言：俞伯牙率琴谢知音 zhŭn zài jiāng biān shìlì gōnghòu certain at river side stand wait ‘(I) certainly will stand waiting by the river.’ b. 东坡在湖州做官 警世通言：王安石三难苏学士 dōngpō zài húzhōu zuò guān Name at Place N. make official ‘Dongpo was holding an official position in Huzhou.’ c. 留得在此 警世通言：庄子休鼓盘成大道 liú dé zài cĭ leave can at DEM ‘(Something) can be left here.’ d. 何期在此相遇 警世通言：吕大郎还金完骨肉 hé qī zài cĭ xiāng yù wh expect at DEM mutual meet ‘Never expected to run into each other here.’ e. 君命在身 警世通言：俞伯牙率琴谢知音 jūn mìng zài shēn lord order exist body ‘(I) am under the order of the lord.’ In short, it is clear that, in spite of some Early Modern Chinese (14th to 17th centuries) idiomatic expressions that have gone out of fashion in present-day Chinese (after the 19th century), the two conditions proposed in section 2 were in position about the 14th century. Furthermore, the forms and meanings represented by older grammar appear to persist more strongly in the argument position than they do in the adjunct position. 3.3. The grammaticalization Grammaticallization typically does not merely seize a single word or morpheme but the whole construction formed by the syntagmatic relation of the elements in question (Lehmann: 1992). Himmelmann (2004) recently contends

218

C. Sun

that the correlation of three types of expansion (host-class, syntactic, semanticpragmatic) is essential to grammaticalization. Brinton and Traugott (2005: 99) observe that ‘Grammaticalization is the change whereby in certain linguistic contexts speakers use parts of a construction with a grammatical function. Over time the resulting grammatical item may become more grammatical by acquiring more grammatical functions and expanding its host-classes.’ Involving place name formation and the expansion of a phonological constraint resulting from such, the path from argument position to adjunct position, and a whole range of semantic-pragmatic factors, the grammaticalization of the Chinese locative construction as a renewal process developed almost exactly in accord with this line of thinking. Before the grammaticalization of the Chinese locative construction, Chinese place names went through a process of lexical change giving rise to a multi-syllabic constraint on spatial NPs. One common place-name forming strategy is to incorporate a spatial morpheme into the name for a place such as zhōng ‘middle’ in the name for the country 中国 zhōng-guó middle-country ‘China.’ To further illustrate the commonality of this strategy, more present-day Chinese place names are given in (33). (33) 北京 bĕi-jīng northcapital

上海 广东 广西 shàng-hǎi guăng-dōng guăng-xī

台北 tái-bĕi

台中 白下 tái-zhōng bái-xià

up-ocean

moundnorth

mound-mid whitedown

vast-east

vast-west

‘Beijing’ ‘Shanghai’ ‘Guangdong’ ‘Guangxi’ ‘Taipei’ ‘Taizhong’ ‘Baixia’ It appears that a spatial morpheme more commonly occurs at the end of a place name than at the beginning, with a [N N(spatial)]. This then constitutes the model for the spatial nominal to take shape as a grammatically significant category in the position after spatial prepositions. 3.3.1. Host-class expansion The oldest host-class for spatial nominals in Old Chinese is a set of verbs of placement (hereafter spatial verbs) that are subcategorized for an argument that can be filled by either a spatial NP (34a) or a prepositional phrase (34b) synonymously. Zhang (2001) observes that this set of verbs include verbs like 如 rú ‘to go,’ 入 rù ‘to enter,’ 抵 dĭ ‘to arrive,’ 在 zài ‘to exist,’ etc. (34) a. 至盟津 史记：本纪：帝纣 zhì méngjīn arrive P. N. ‘(The troops) arrived at Mengjin.’

Two Conditions and Grammaticalization of the Chinese Locative

219

b. 至于盟津 史记：本纪：武王 zhì yú méngjīn arrive at P.N. ‘(The troops) arrived at Mengjin.’ c. 先至城下(史记：绛侯周勃世家) xiān zhì chéng xià first arrive city downside ‘(They) first arrived below the city.’ The collocation constraint between a spatial nominal and a spatial verb emerged from the context in (34a and c) where a locative preposition like yú (34b) is optional. The set of verbs was expanded to include other kinds in the Han dynasty (after 206 BCE) such as hùi ‘to meet’ and pò ‘to defeat’ in (35) which are not spatial verbs, thus extending the collocation constraint and encouraging usage of spatial nominals like jĭng shàng ‘on the border’ and gàixià ‘place name.’ (35) a. 臣尝从大王与燕王会境上(史记：廉颇蔺相如列传) chén cháng cóng dà wáng yŭ yān wáng huì jĭng shàng subject once follow great king with Name king meet border on ‘I once followed your majesty to meet with King Yan on the border.’ b. 破之垓下 史记：黥布列传) pò zhī gāixià defeat him P.N. ‘They defeated him at Gaixia.’ Furthermore, the analogical process of the multi-syllabic constraint extended from place names to the derived spatial NPs, with the majority having either a binominal structure [[Y]NP zhī X]NP (36) (zhī as a genitive enclitic can also mark a relative clause) or two nominal stems [[Y]N[X]N] NP (35). (36)

葬之郐城之下(左传：僖公 33) zàng zhī kuài chéng zhī xià bury 3rd Name city GEN. downside ‘Bury him at the bottom of the Kuai city wall.’

Finally, as the Old Chinese locative preposition yú (34b) and the present-day counterpart zài co-existed for a very long time in history, the former became greatly outnumbered by the latter (approximately 1:4 as a locative marker) (Zhang 2001: 231) in vernacular Chinese just about the time of Yuan dynasty (14th century) when the two conditions came into in effect. It was only then that zài assumed the function of the primary locative marker. Recall that the Old Chinese spatial nominals first occurred after spatial verbs without the locative preposition yú. As the host class for spatial nominals expanded to include other verbs in Middle Chinese, the usage of yú declined and has gone out of use in present-day Chinese. Zài thus took over the

220

C. Sun

function of the preverbal, adjunct position. Therefore, this also qualifies a process of renewal (Hopper and Traugott 2003). 3.3.2. Syntactic expansion The spatial nominals of the data in (34 to 36) are all in argument position. Zhang (2001) noted a few early cases in which the lexeme zài already shows up in front of another verb in the Analects, resembling the modern locative construction in adjunct position. However, she argued that it was then perhaps still a verb, since such cases as those shown in (37) were so few. (37) a. 子在齐闻韶(论语：述而) zĭ zài qí wén sháo Name exist P.N. hear P.N. ‘Confucius hear (the music) of Shao in Qi.’ b. 子在川上曰 … 论语：子罕 zĭ zài chuān shàng yuè Name exist river up say ‘Confucius said by the river...’ Following Zhang’s insight, the zài NP V sequence in (37) are then a series of two VPs that is allowed by the grammar as some sort of coordinate construct. Zhang (2001: 72) observed that zài in such a sequence became a preposition in the Western Han (206 BCE to 24 CCE) and in the Buddhist texts of the Eastern Han (25 CE-220) zài was used as a preposition either postverbally (17 cases like (38a)) or preverbally (43 cases like (38b)). Therefore, as far as the grammaticalization of the locative construction is concerned, the development from argument position (35) to nonargument position in (38)4 can then be taken as a process of syntactic expansion according to Himmelmann’s framework. (38) a. 分散在民间 论衡 fēnsàn zài mín jiān spread at people among ‘(Something) spread among the people.’ b. 时在后院池中沐浴 修行本起经 shí zài hòu yuàn chí zhōng mùyù time at back yard pond middle bathe ‘A the time, (someone) was bathing in the pond of the backyard.’ Nevertheless, the two conditions proposed for the present-day Chinese locative construction did not form immediately after zài assumed the function of marking the locative case in late Old Chinese. For example, even in the 13th century, the multisyllabic constraint had not completely formed, since the locative construction still allows the monosyllabic lĭ in (39a). It was in the 14th century that the multi-syllabic constraint and the construction became organized and applicable to the locative

Two Conditions and Grammaticalization of the Chinese L ocative

221

construction as is evidenced by the disyllabic spatial nominal lĭmian in (39b) as lĭ is grammaticalized into an enclitic that cannot occur without a stem. (see section 3.2) (39) a. 终日在里默坐 朱子语类：卷一百一十三 zhōng-rì zài lĭ mò zuò all-day at inside quiet sit ‘Sit quietly inside all day long.’ b. 有几个秀才在里面 警世通言 yǒu jĭ-ge xiùcăi zài lĭmian have several-CL scholar at inside ‘There are several scholars inside.’ The change of the Old Chinese monosyllabic spatial nouns like 上 shàng,下 xià, 里 lĭ, 内 nèi, 外 wài, etc. into modern Chinese NP enclitics is the most obvious formal change indicative of grammaticalization. Schematically, I want to present the structural change of the locative construction in (40), in which a binominal structure of the objective NP is changed into an NP headed by an enclitic that forms a phonological unit with the immediately preceding element of the initial NP. (40) Old Chinese

Modern Chinese

Modern Chinese (pragmatically motivated) [yú [[Y]NP [X]N]NP]PPÆ [zài [[Y]NP Xenclitic]NP]PPÆ [zài [[X]NP]PP

In section 3.3.3, I will talk about the pragmatically motivated uses of some locative NPs in which the selectional restriction is optional, thus resulting in a single NP without the enclitic after the preposition zài. 3.3.3. Semantic-pragmatic expansion The selection of the lexeme zài ‘to exist’ to head a locative construction as a preposition signaling a spatially underspecified meaning is itself a semantic change of abstraction. The pragmatic expansion of the Chinese locative construction into a more inclusive metalinguistic term interacting with various pragmatic factors can be viewed in at least two significant ways. First, the locative construction with a spatial enclitic shàng expands its original meaning of referencing physical space to the more abstract meanings of part-whole relationship (23) and ‘in the sphere of ’ (24), both of which are not available in the locative construction without the more grammatical spatial enclitic. Second, in present-day Chinese, the correct use of the locative construction cannot be divorced from discourse/pragmatics, as in a contrastive context (19) or a listing context (20) where the application of the default selectional restriction in (9) is optional. The omission of a spatial enclitic in these discourse contexts, giving rise to still another structure of the Chinese locative construction in which the NP after zài is no longer binominal but mono-nominal with a functional head characterized by the pragmatically motivated schema in (40). All this suggests that the formation of the present-day Chinese locative construction

222

C. Sun

is definitely a case of grammaticalization marked by semantic/pragmatic factors as emphasized by Himmelmann (2004). The primary function of a locative construction is to locate a certain participant of an event in reference to either a known or easily identifiable place to facilitate ease in communication. The most fundamental idea of selectional restriction (9) is to capture the fact that a spatial nominal is typically used after the locative preposition zài. (9)

The selectional restriction of the preposition zài: In non-contrastive or nonlisting contexts, the underspecified preposition zài 在 normally selects either a. a spatial NP or b. an explicitly marked, definite and inanimate NP.

However, the proposed restriction only works with three major qualifications, as it is subject to three types of exceptions in contrastive contexts, in listing contexts, and with explicitly definite markers. That is, the grammaticality of the locative construction cannot be decided, to a large extent, without looking at a given discourse context in actual use. With an explicit definite marker, a non-spatial nominal can occur without a spatial enclitic (41). In (41c) the demonstrative pronoun is construed as both definite and directional, thus satisfying the selectional restriction and exempted, at the same time, from the multi-syllabic constraint that is applicable to the non-directional nominals only. (41) a. 我在第一页上打了个记号 wǒ zài dì yī yè shàng dă le ge jìhào 1st at Ord. one page up hit ASP CL mark ‘I made a mark on the first page.’ b. 我在第一页打了个记号 wǒ zài dì yī yè dă le ge jìhào 1st at Ord. one page hit ASP CL mark ‘I made a mark on the first page.’ c. 我在那儿打了个记号 wǒ zài nàr dă le ge jìhào 1st at DEM hit ASP CL mark ‘I made a mark there.’ In discourse contexts such as listing and contrasting when multiple locations are involved, non-spatial nominals can occur in the locative construction without a spatial enclitic, constituting a general locative construction. Levinson et al. (2003) witnessed that crosslinguistically some adpositions are more specific than the others. That is, some specific features of a subordinate term can be missing in a superordinate term. For example, the English inside has specific features that are missing in the more general in. In this light, I propose a contextually based continuum ranging from definite locative to specific locative to general locative in order to characterize the present-day Chinese locative construction in (42). When a universal direction-bearing marker like 东 dōng ‘east,’ 西 xī ‘west,’ 家 jiā ‘home,’ 这 zhèr ‘this,’ 那 nàr ‘that,’ etc. is involved, the nominal of the construction is

Two Conditions and Grammaticalization of the Chinese Locative

223

subject to the selectional restriction but not to the multi-syllabic restriction. The universal direction-bearing markers are used as such mirroring the tolerance of monosyllabic nominals in the locative constructions by older grammar as some kind of layering effect (Hopper and Traugott 2003). Although the two conditions control most of the nominals of the locative construction, only the multi-syllabic constraint has to be applied to the general locative construction, thus allowing multi-syllabic non-spatial nominals to occur in listing and contrasting contexts without spatial enclitics. However, the relaxation of the selectional restriction for the general locative reflects a newest structural change motivated by pragmatic factors in making contrast and listing. (42) definite locative specific locative general locative (listing and contrasting)

directional, with definite markers, place names, derived spatial NPs, non-spatial nominals

Selectional restriction Multi-syllabic constraint

Definite locative, specific locative, and general locative on the scale can be defined by the applicability of the two conditions and the coding devices such as directional, derived spatial nominals, etc. A prototypical locative construction in Chinese is, therefore, definite or specific and subject to both the selectional restriction and the multi-syllabic constraint. The spatial NPs with enclitics carry more specific spatial meanings, be it topological or geometric, than those without enclitics. Furthermore, there are three semantically non-discrete categories with no clear-cut boundaries separating them. For example, all of the monosyllabic terms for universal directionbearing markers have available multi-syllabic terms to express the same ideas such as 东方 dōngfāng or 东面 dōngmian for 东 dōng ‘east,’ or 这里 zhèli ‘here,’ 这面 zhèmian ‘this side,’ or 这边 zhèbian ‘this side’ for 这儿 zhèr ‘here.’ In terms of referential properties, universal directional terms, NPs with definite markers and place names are all definite in nature. Specific locative construction overlaps with most definite and general locatives as they are all subject to the multi-syllabic constraint. In general locative construction, a speaker always has an option to use either a derived spatial NP or simply a non-spatial nominal, as additional topological, or geometric, meanings are generally not a necessary ingredient in the discourse contexts of listing and contrasting. 4. Summary In summary, two conditions on the Chinese locative construction emerged over the last two millennia resulting from grammatcalization to renew a locative construction whereby a verb meaning ‘to exist’ changes into a preposition signaling

224

C. Sun

generic locative meaning, and a spatial nominal changes into an NP enclitic indicating a specific spatial meaning. This grammaticalization results in changes of the construction in different grammatical domains, phonology, morphology, syntax, semantics and pragmatics. These changes that emerged over the course of shaping the locative construction measure up very well with known crosslinguistic grammaticalization patterns (Hopper and Traugott 2003, Himmelmann 2004). Furthermore, it has been demonstrated that the two-tier locative construction is prototypically composed of a preposition and a spatial nominal such as a derived spatial NP or a place name. The two conditions as they occurred during the last millennium define the prototypical uses of the locative construction that exists in present-day Chinese. It is also demonstrated that the locative Chinese PP is neither postpositional nor circumpositional.

* I would like to acknowledge my gratitude to Stanford Center for East Asian Studies for two grants enabling me to invite Professor Changcai Zhao of CASS to work with me for three months (2004) and to present my initial ideas for the paper at a conference sponsored by EHESS, France and hosted by the University of Melbourne in 2005. I want to thank Professor Zhao for his kindness and willingness to work with me. Unfortunately, during his visit, mostly because of my wrong ideas and failure to think differently, we were unable to come up with a paper that would make sense. I want to thank Professor Traugott for her patience and comments on a very rough version of the paper before I went to Melbourne. When I presented my initial ideas in Melbourne, I benefited from the discussion with all the participants. Earlier versions of this paper were presented to an International Conference (May 30–31, 2006) in honor of Professor Li Fang-Kuei at Academia Sinica, Taiwan and NACCL-18 (June 2006) at the University of Western Washington. I want to thank all those at the two conferences who commented on my presentations, making me recognize my unclear thinking and revise my hypotheses. Finally, I want to thank the editor of this volume, Professor Dan Xu, for her patience and helpful comments. All the problems and errors in this paper are completely mine.

Notes 1

Note that it is also possible to have PP instead of a noun phrase at this position, as a locative PP is also a spatial expression. Furthermore, a demonstrative can also turn it into a grammatical sentence because of the special function definite NPs may have in the Chinese locative construction (see section 3). a. 在世界上没有无缘无故的爱 zài shì-jiè -shang méiyŏu wú-yuán-wú -gù de ài Prep world-border-up Neg no-reason-no-origin POS love ‘There is no such thing as love without reason or cause in the world.’

Two Conditions and Grammaticalization of the Chinese Locative

b.

225

这个世界没有无缘无故的爱 zhè-ge shì -jiè -shang méiyŏu wú-yuán-wú -gù de ài Dem-CL world-border-up Neg no-reason-no-origin POS love ‘There is no such thing as love without reason or cause in this world.’ 2 I was most grateful to Professor Audrey Li for pointing out this property to me. 3 Note that shangmian can have the more abstract meaning when the relative marker de is absent. In other words, the disyllabic shàngmian is ambiguous and can have either a physical spatial reference as a noun or an abstract reference like an enclitic. 在报纸上面有条中国新闻 zài bàozhĭ shàngmian yŏu tiáo zhōngguó xīnwén at paper REL top exist CL China news ‘There is a piece of news about China in the newspaper.’ 4 These two examples are taken from Zhang (2001: 54).

226

C. Sun

References Brinton, Laurel J. & Elizabeth Closs Traugott. (2005). Lexicalization and Language Change. Cambrige, United Kingdom: Cambridge University Press. Cheng, Xianqing. (1992). Lianghan hanyu yanjiu [A study of the Chinese during the two Han dynasties]. Jinan: Shandong Jiaoyu Chubanshe. Chu, Zexiang. (2004). Hanyu ‘zai+fangwei duanyu’li fangweici de yinxian jizhi [The mechanisms responsible for the deletion of locative particles in ‘zai locative phrases]. Zhongguo yuwen 299(2), 112–22. Croft, William. (2001). Radical construction grammar: Syntactic theory in typological perspective. Oxford: Oxford University Press. Ernst, Thomas. (1988). Chinese postpositions? – Again. Journal of Chinese Linguistics, 16(2), 219–44. Fang, Jingmin. (2002). Lun hanyu kongjian quyue fanchoude xingzhi he leixing [On the properties and typology of the Chinese topological domain]. Shijie hanyu jiaoxue, 61(3), 37–48. Fang, Jingmin. (2004). Xiandai hanyu fangwei chengfende fenhua he yufahua [The divisions of locative elements and grammaticalization in Modern Chinese]. Shijie hanyu jiaoxue, 68(2), 5–15. Greenberg, Joseph H. (1995). The diachronic typological approach to language. In Masayoshi Shibatani & Theodora Bynon (Eds.), Approaches to Language Typology (pp. 145– 66). Oxford: Clarendon Press. Herskovits, Annette. (1986). Language and spatial cognition. Cambridge: Cambridge University Press. Himmelmann, Nikolaus P. (2004). Lexicalization and grammaticalization: Opposite or orthogonal?. In Walter Bisang, Nikolaus P. Himmelmann, & Björn Wiemer (Eds.), What makes Grammaticalization? (pp. 21–42). Berlin and New York: Mouton de Gruyter. Hopper, Paul J. & Elizabeth Closs Traugott. (2003). Grammaticalization (second edition). Cambridge: Cambridge University Press. Huang, C.-T. James. (1987). Existential sentences in Chinese and indefiniteness. In Reuland & Meulen (Eds.), The representations of (in)definiteness (pp. 226–53). Cambridge: MIT Press. Levinson, Stephen, Sergio Meira, & the language and cognition group, Max Planck Institute for Psycholinguistics. (2003). ‘Natural concepts’ in the spatial topological domain—adpositional meanings in crosslinguistic perspective: An exercise in semantic typology. Language, 79(3), 485–516. Li, Audrey Y.-H. (1990). Order and Constituency in Mandarin Chinese. Dordrecht: Kluwer Academic Publishers. Li, Chongxing. (1992). Chusuoci fazhanlishi de chubu kaocha [An preliminary investigation of the historical development of the locative phrases]. In Hu zhu’an, Naisi Yang, Shaoyu Jiang (Eds.), Jindai Hanyu Yanjiu. Beijing: Commercial Press. Li, Charles & Sandra Thompson. (1981). Mandarin Chinese: A functional Reference grammar. Berkeley: University of California Press. Liu, Danqing. (2003). Yuxu leixingxue yu jieci lilun [Word-order typology and theories on prepositions]. Beijing: Commercial Press. Liu, Feng-Hsi. (1998). A clitic analysis of locative particles. Journal of Chinese Linguistics, 28(1), 48–70. Miller, George Armitage & Philip N. Johnson-Laird. (1976). Language and perception. Cambridge, Massachusetts: The Belknap Press of Harvard University Press.

Two Conditions and Grammaticalization of the Chinese L ocative

227

Sun, Chaofen & T. Givón. (1985). On the so-called SOV word order in Mandarin Chinese: A quantified text study and its implications. Language, 61, 329–51. Tan, Qixiang. (1982). zhongguo lishi dituji [Historical maps of China]. Shanghai: Ditu chubanshe. Wu, Fuxiang (2005). Hanyu yufahua yanbian de ji-ge leixingxue tezhen [Some typological features in the grammaticalization changes in Chinese]. Zhongguo Yuwen, 309, 483–94. Zhang, Cheng. (2001). Xiandai hanyu jieci cizu ‘zai L’ yu dongci bingyu de cixu guilü [The word-order regularity of the syntactic object and the prepositional phrase ‘at L’ in Modern Chinese]. Zwicky, Arnold. (1985). Clitics and particles. Language, 61, 283–305.

ORIGIN AND EVOLUTION OF THE LOCATIVE TERM HÒU ‘BACK’ IN CHINESE Fuxiang Wu Institute of Linguistics, CASS Beijing, China Keywords: the locative term hou, semantic change, metonymy, subjectification Abstract: This paper focuses on the source model and semantic changes of the locative term hou in Chinese, especially the semantic and morphosyntactic changes taking place when hou occupied the head position in genitive structures, based on historical literature. The chief conclusions are as follows: (i) The locative term hou in Chinese originated in the conception of motional process. Although the source model does not precisely correspond to the mainstream source models of BACK-REGION grams revealed by cross-linguistic comparison, it represents a typological feature in human languages rather than a language-specific characteristic. (ii) The pattern of semantic changes including a series of meanings and uses in the diachronic development of the locative term hou corresponds precisely to the universal tendencies of BACK-REGION grams revealed by cross-linguistic research. (iii) The arising of various meanings of hou is motivated, and metonymy serves as the most fundamental mechanism in the semantic change.

1. Introduction The source models and semantic change of locative terms are issues that have received widespread attention in recent studies of cognitive linguistics, historical semantics and grammaticalization (for example, Heine et al., 1991; Heine, 1997 on African languages; Brugman and Macaulay, 1986; Maclaury, 1989, De Leon,1991; Hollenbach, 1995 and Lillehaugen, 2003, 2004 on Mesoamerican languages ; Bowden, 1992 on Oceanic Languages; Matsumoto, 1999 on Japanese; particularly the crosslinguistic examinations of Svorou, 1986, 1993, 2003). Cross-linguistic studies suggest that there are notable universal commonalities in the source models and evolutionary paths of locative terms in human languages. Svorou (1986, 1993) observes that the fundamental source models of locative terms in human languages consist of body part terms, and then environmental landmarks as well as relational part terms. For instance in Estonian, Finnish and Ewe, the locative term ‘UP’ originated from body part terms such as head, face, back (animal), etc., but in Ewe, Basque and Guaymi they developed out of environmental landmarks like heaven/sky, summit and so on, whereas in Tigre, Abkhaz, Igbo and Vietnamese, they are rooted in relational part terms of objects such as top/upper part. Heine (Heine et al., 1991; Heine, 1997) holds that relational part terms, instead of being primitive concepts, are derived from body part terms or environmental landmarks in many languages. Therefore, in his theory, the fundamental source models of locative terms are body part terms and environmental landmarks; locative

229 D. Xu (ed.), Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives, 229–247. © Springer Science +Business Media B.V. 2008

230

F. Wu

terms may also come from such dynamic concepts as activities. The source models of locative terms proposed by Heine are illustrated in table1.1.

Table 1.1 Common source models of spatial locative terms (After Heine, 1997: 38) Source model

expressions of spatial orientation

Body parts Landmarks Dynamic concepts

Uses parts of the human body in its upright position as a model Uses environmental landmarks Uses activities

On the other hand, some linguists suggest that morphosyntactic and semantic changes in locative terms (those from human body parts in particular) usually proceed along predictable paths, which present cross-linguistic universal tendencies (Svorou, 1986, 2003; Heine et al., 1991; Heine, 1997). Svorou (1986) put forward two important hypotheses in view of this: (i) Evolution of linguistic material should be represented as a continuum where various points are intervals and constitute stages. (ii) Meaning is the driving force of any linguistic change: this requires there to be two parallel continua in the evolution of locative expressions, one being the evolution of meaning and the other the evolution of form. Svorou (1993) later revised the two continua as follows: (i) The evolutionary path of spatial grams from nouns (Svorou 1993: 90): animal body part human body part

relational part of object

location adjacent to object part

location in region of object part

environmental landmark

(ii) The continuum of morphosyntactic evolution of spatial grams (Svorou 1993: 101): Genitive construction Noun

adposition

affix

0

adverb This paper tentatively adopts the analytical framework and theoretical achievements in research of locative terms in general linguistics in order to discuss the origin and evolution of the locative term hòu ‘BACK’ in Chinese, especially the semantic

The Locative Term hòu ‘BACK’ in Chinese

231

and morphosyntactic changes taking place when hòu occupies the head position in genitive structures. Our major concerns are: (i) What is the relationship between the source model of the locative term hòu and the universal tendencies exposed by cross-linguistic research on the locative term “BACK”? (ii) What kinds of semantic and morphosyntactic changes has the locative term hòu undergone, and by which paths and mechanisms? 2. The Origin and Evolution of the Locative Term hòu hòu (後) was originally a verb referring to a motional situation, and its original meaning is believed to have been “to walk and follow (someone else) at a distance”. Instances of this meaning, however, appeared comparatively late in literature, for example: 1

子路從而後，遇丈人，以杖荷蓧。 論語·微子 zǐ lù cóng ér hòu, yù zhàng rén, yǐ zhàng hè diào. Name follow Conj. hòu, meet old man, with rod carry weeding tool ‘Zǐ lù walked behind (Confucius), and came across an old man who carried a weeding tool with a stick.’

The spatial concept “posterior in spatial position” arose from constructions of this kind, and hòu consequently changed into a prototypical locative noun, as in: 2

禦者執策（ 立于馬後。 儀禮·既夕禮 yù zhě zhí cè, lì yú mǎ hòu. rider hold whip, stand Prep. horse hòu ‘The rider is holding a whip standing behind the horse.’

Generally speaking, hòu is used to designate the spatial position of the Figure in comparison with the Ground, when acting as a locative term expressing spatial relationships. As in (2), hòu signifies the posterior order of the Figure rider compared with the Ground horse. In Pre-Qin literature, hòu, which was rarely used for spatial relations, can be found abundantly in constructions indicating temporal concepts; especially when (zhī) hòu follows temporal phrases referring to time intervals, it indicates that a situation occurs or is completed after the end of the time interval. For example: 3

八世之後，莫之與京。 左傳·莊公二十年 bā shì zhī hòu, mò zhī yú jīng. eight generation zhī hòu, none Particle Prep.-comparative achieve-greatness ‘After eight generations, no other family will achieve such greatness as his.’

During the Spring-Autumn and Warring States Periods (770–221 BC), the abovementioned structure “ NP+( zhī) hòu” was ordinarily adopted to express the concept “after the end of a time interval”. It is probable that ( zhī) hòu had been reanalyzed as

232

F. Wu

a postposition; in other words, the former genitive marker zhī and the locative noun hòu had fused into a two-syllable postposition by reanalysis. Temporal nouns were not the only means of expressing time intervals. Verbs and verb phrases referring to events and activities could also be used, since an event or activity always covers an interval of time. Consequently, (zhī) hòu could also follow a VP to designate temporal concepts in Pre-Qin times, as in (4): 4

凡我同盟之人，既盟之後，言歸於好。 左傳·僖公九年 fán wǒ tóng méng zhī rén, jì méng zhī hòu, yán guī yú hǎo. all our allies Particle people, perfect-marker alliance zhī hòu, words return good ‘All the states should be friendly to each other, since we have formed an alliance.’

In the Han (206 BC–220 AD), the element preceding ( zhī) hòu extended from a VP to a full clause. ( zhī) hòu, which indicated that the situation expressed by the main clause was realized or completed after the realization or completion of the situation expressed by the subordinate clause, evolved into a typical temporal clause connective (a subordinate clause connective). For example: 5

a. 陵死後，蒼為丞相。 史記·張丞相列傳 líng sǐ hòu, Cāng wéi chéng xiàng. name die hòu, name become the prime minister ‘After Líng’s death, Cāng became the prime minister.’ b. 堯崩之後，舜老，亦以傳於禹。 論衡·書虛篇 yáo bēng zhī hòu, Shùn lǎo, yì yǐ chuán yú Yǔ. name die zhī hòu, name old, also Prep.(the throne) pass Prep. name ‘After the death of Yáo, when Shùn became old, he also passed the throne to Yǔ.’

From the Wei and Jin (220–420 AD) to the Sui and Tang (581–907 AD), (zhī) hòu was even more freely used as a temporal clause connective, as in the following examples: 6

a. 何驃騎亡後，征褚公入。 世說新語·言語 hé piào qí wáng hòu, zhēng chǔ gōng rù. name general die hòu, call-up name enter ‘After General Hé was dead, Sir Chǔ was called up (by the King).’ b. 吳王死後，夫差太子為吳王。 敦煌變文·伍子胥變文 wú wáng sǐ hòu, Fū chāi tài zǐ wéi Wú wáng. wú king die hòu, name crown-prince become Wú King ‘After the death of the emperor in the kingdom of Wú, the crown prince Fūchāi came to the throne.’

The Locative Term hòu ‘BACK’ in Chinese

233

It is worthwhile to pay attention to the fact that the temporal clause connective hòu began to be widely used in hypothetical and subjunctive conditional clauses in the literature of the Tang. For example: 7

持世若教成道後，魔家眷屬定須摧。 敦煌變文·維摩詰經講經文 chí shì ruò jiào chéng dào hòu, mó jiā juàn shǔ dìng xū cuī. name if can achieve Tao hòu, devil family-dependents definitely must destroy ‘If (Bodhisattva) Chíshì has achieved Tao, the family of devils will be destroyed.’

The temporal reference of hòu was weakened in contexts such as in (7), whereas hòu in the following sentences can seemingly be construed as either a temporal clause connective or a hypothetical conditional connective: 8

a. 若能曉了驪珠後，只這驪珠在我身。 祖堂集·丹霞和尚 ruò néng xiǎo liǎo lí zhū hòu,zhǐ zhè lí zhū zài wǒ shēn. if can understand name bead hòu, only this name bead be-at my body “If/After I understood the bead Li, this bead only belongs to me.’ b. 師云： 為汝不薦祖。 僧曰： 薦後如何? 師云： 方知不是祖。 祖堂集·大光和尚 shī yún: “wéi rǔ bú jiàn zǔ.” Sēng yún: “jiàn hòu rú hé?” Shī yún: “fāng zhī bú shì zǔ.” master say: “for you NEG. esteem grandmaster.” monk say: “esteem hòu like what?” master say: “then know NEG. is grandmaster.” “The master said: ‘because you do not esteem the grandmaster.’ The monk asked: ‘What if I esteem him/What happens when I esteem him?’ The master said: ‘Then you will know he is not the grandmaster.’”

By the Song and Jin (960–1279 AD), hòu had become a comparatively typical connective for hypothetical conditional clauses, as in: 9

是人後疾忙快分說，是鬼後應速滅。 董解元西廂記·卷六 shì rén hòu jí máng kuài fēn shuō, shì guǐ hòu yīng sù miè. be human-being hòu fast talk, be ghost hòu must immediately destroy ‘If you are a human being, you shall tell me at once, if you are a ghost, you must be destroyed immediately by me.’

10

都知說得果無謬，若非今日隨喜後 （ 著丹青畫出來不信道有。 董解元 西廂記·卷一 dōu zhī shuō de guǒ wú miù, ruò fēi jīn rì suí xǐ hòu, zhuó dān qīng huà chū lái bú xìn dào yǒu.

234

F. Wu

all know say Particle. indeed NEG. false, if NEG. today visit (the temple) hòu, use color draw out NEG. believe say exist. ‘I have heard that it is indeed true, but if I did not visit (the temple) today, I will not fully believe it even if you draw it on a picture of it in color.’ Furthermore, the hypothetical meaning of the connective hòu was bleached in some contexts, and hòu gradually changed into a topic marker when following a VP without a subject, for example: 11

得後是自家采，不得後是自家命。 董解元西廂記·卷一 dé hòu shì zì jiā cǎi, bù dé hòu shì zì jiā mìng. get hòu be myself luck, NEG. get hòu be myself fate ‘It is lucky for me to get it, and it is for fate’s sake not to get.’

Finally, the topic marker hòu began to follow non-topic elements via generalization, and further changed into an intra-sentence pausing marker, as in (12): 12

把酒問春因底意，為誰來後為誰歸。 王周：問春 bǎ jiǔ wèn chūn yīn dǐ yì, wèi shuí lái hòu wèi shuí guī. hold wine ask spring for what thoughts, for who come hòu for who go-back “ propose a toast to Spring and ask about her thoughts: ‘For whom do you come and for whom do you leave?’”

The above examination traces the evolutionary path of hòu after becoming a temporal postposition. It successively acquired meanings and uses including “after a time interval”, “after the realization or completion of a situation”, “hypothetical conditional marker”, “topic marker” and “pausing marker”. We also observe a semantic change that took place in at least the Tang, in which the meaning “stative commencement” derived from the meaning “after a time interval” when hòu came after a stative verb. Such examples are quite common in the 全唐诗 Quán Táng Shī : 13 a. 騎來未省將鞭觸（病後長教覓藥醫。 全唐詩 qí lái wèi shěng jiāng biān chù, bìng hòu cháng jiào mì yào yī. ride come never with whip touch, fall-ill hòu often let find medicine heal. ‘I had never whipped the horse with whip when riding, but I always looked for medicine to heal it when the horse got ill.’ b. 醉後未能別，待醒方送君。 全唐詩 zuì hòu wèi néng bié, dài xǐng fāng sòng jūn. drunk hòu NEG. can bid-farewell, wait sober not-until see-off you “I was too drunk to bid you farewell, and I did not see you off until I was sober.”

The Locative Term hòu ‘BACK’ in Chinese

235

Apart from the semantic and syntactic changes mentioned above, in its evolution the locative term hòu underwent a kind of morphological change starting in the Six Dynasties Period. In the process, hòu, following the substantive terms, evolved from a postposition to an intra-word component. For example: 14 a. 餘兩小簏，著背後。 世說新語·雅量 yú liǎng xiǎo lù, zhuó bèi hòu. remain two small box-made-of-bamboo, put back hòu ‘there were two small bamboo boxes left, so he hid them behind his back.’ b. 千人莫引于前頭，萬騎罷隨於背後。 敦煌變文校注 qiān rén mò yǐn yú qián tóu, wàn jì bà suí yú bèi hòu. thousand people NEG. lead at front, ten-thousand riders NEG. follow at back hòu. ‘Do not be escorted by hundreds of guards ahead and thousands of riders behind.’ In (14a), bèi hòu (背後) is an adpositional construction, with hòu indicating the thematic role and regional dimension of bèi, whereas bèi hòu in (14b) has already lexicalized into a single locative noun. 3. The Mechanisms of Semantic Change of hòu I have given a brief survey of the origin and evolution of hòu and accordingly summarize the diachronic semantic changes and the consequent morphosyntactic changes as follows: (15a) The path of semantic changes of hòu: motional process

the commencement of a state

spatial position behind the object of reference after a time interval after the realization or completion of a situation hypothetical marker topic marker pausing marker

236

F. Wu

(15b) The path of morphosyntactic evolution of hòu: process verb→ →locative noun→ →postposition→ → temporal clause connective → intra-word morpheme conditional clause connective→ → topic particle → pausing particle What we must now answer is this: How did the various functions and meanings associated with hòu as in (15a) arise over time? In other words, what mechanisms led to the semantic changes of hòu? Metaphor and metonymy are frequently cited in the linguistics literature relating to mechanisms of semantic change. Traugott and König 1991, Traugott and Dasher Metaphor refers to the mapping of concepts from different cognitive domains; a comparatively concrete concept in the source domain is mapped onto a more abstract concept in the target domain. The mapping is motivated by analogy and iconic relationships between the concepts instead of being random, and is unidirectional. Metaphor involves two different cognitive domains, so it is generally believed to bring about abrupt changes operating on the paradigmatic level. Metonymy, on the other hand, is defined as a transition within one cognitive domain, from one concept to a contiguous concept, and is motivated by the contiguity and indexicality of the concept as well as by cognitive association. Because metonymy is a transition restricted to one cognitive domain, it is generally believed to cause gradual changes operating on the syntagmatic level. Traditionally, metaphor is regarded as the major process at work in semantic change, while metonymy is a minor process. However, recent studies of semantic change and grammaticalization suggest that actually, metonymy plays a more fundamental role compared to metaphor (Barcelona 2000; Brinton 1988; Brinton and Traugott 2005; Bybee et al. 1994; Traugott 1989, 2006; Traugott and König 1991; Traugott and Dasher 2002; Hopper and Traugott 2003). First of all, the fundamental mechanism of semantic change is naturally metonymy, not metaphor, since the process of semantic change consists of gradual change rather than abrupt change. Secondly, many instances of semantic change once attributed to metaphor are actually the result of metonymy (Traugott and König 1991, Traugott and Dasher 2002，Bybee et al. 1994，Hopper and Traugott 2003). Third, in essence metaphor is a comparison and generalization between sources and targets from different cognitive domains correlated with semantic change, and can be regarded as the result of metonymy on the synchronic level (Traugott and Dasher 2002, Hopper and Traugott 2003, Brinton and Traugott 2005, Traugott 2006). Metonymy may not necessarily lead to metaphor, but it is probable that nearly all instances of metaphor consist of a process of metonymy. The following discussion will focus on the fundamental role of metonymy in the semantic evolution of the locative term hòu in Chinese.

The Locative Term hòu ‘BACK’ in Chinese

237

We start from the development of meaning . When hòu expresses the motional process meaning “ to walk behind ”, it typically presents a scene in which the Figure and the Ground move in the same direction and the spatial position of the Figure is behind that of the Ground. The themes of motion (the movers, namely the Figure and the Ground), the motional process and the mover’s relative spatial positions in the motional process, are all in the same cognitive frame. Consequently, in particular contexts (e.g. (1)), the conceptual contiguity between the motional process and the relative spatial positions of the motional themes can bring about this pragmatic inference: if X behind Y and moves forward (e.g. walk behind in (1)), then the location of X would normally be behind Y (the spatial position posterior to the object of reference) in the motional process. After this pragmatic inference semanticized, hòu acquired the coded meaning “the spatial position posterior to the object of reference”. Therefore, the transfer from meaning to meaning was a metonymical process motivated by pragmatic inference. The evolution from meaning (“the spatial position behind the object of reference”) to meaning (“after a time interval”) seems to fall into the category of conceptual metaphor if judged by its result, because the two meanings share extremely similar imaginary schemas involving “posterior order”: In the first meaning, the Figure is posterior to the Ground in spatial position; while in the second, the Figure (situation) is posterior to the Ground (time unit) in temporal sequence. Nevertheless, metonymy is again involved in the change if the intermediate stage of “temporal posterior order” is considered. Svorou (1993) observes that “temporal posterior order” functions as an important stage in the evolution of the BACK-REGION grams, from which other temporal uses are derived directly or indirectly in many languages, such as the postposition pāchūin Bihari: (16) hamara-sà pāchū rāma œlāhaa (Jha 1958: 336) me-ABL after Rama come. PAST ‘Rama came after me.’ In Chinese, hòu is used to indicate temporal posterior order, but unlike pāchū in Bihari, hòu is an adverb instead of a postposition in this sense, for example: (17)

齊桓公合諸侯，衛人後至。 呂氏春秋·精諭 qí huán gōng hé zhū hóu, Wèi rén hòu zhì. proper name summon vassal-lords, proper name people hòu arrive ‘Huán, the lord of Qí, summoned all the other vassal lords to his presence, while the diplomatic mission from Wèi arrived late.’

238

F. Wu

What leads to the derivation of “temporal posterior order” from “spatial posterior order”? We hold that a metonymical process is at work. Spatial relationships between entities can be stative, as in “there is a tree in front of the gate”, reflecting the stative existence of the spatial relationship. They can also be dynamic, for instance, “X runs after Y to the cafeteria”. In this sentence, the spatial relation between X and Y is reflected by a dynamic situation, with their specific spatial positions continually changing in motion but their relative order unchanged. In such a scene where both the Figure and the Ground are moving in the same direction, there is a close relation between the relative spatial position of the two moving objects and their temporal order in reaching the destination, which may invite this reasoning: if X(=Figure) is behind Y(=Ground) in spatial position in a motional process, it is most probable that X will reach the destination later than Y. This reasoning is based on our life experiences and empirical knowledge of motional situations, and essentially belongs to pragmatic inferencing, which brought about the change from spatial order to temporal order in the BACK-REGION grams (including hòu in Chinese). On the other hand, metonymy can also be found in the development from “temporal posterior order ”(the intermediate stage) to “after a time interval” (meaning ). Both have something to do with temporal order relating to the Ground. In the former, the Figure is in temporal posterior order when reaching the destination, whereas in the latter, the realization or the completion of a situation is in posterior order. The two meanings differ in the qualities of the Figure and the Ground: two animate objects in the former and a situation and a time interval respectively in the latter. Thus it can be seen that the mechanism of metonymy was also brought into play in the change from “the spatial position posterior to the object of reference” to “after a time interval”, though the outcome may be regarded synchronically as metaphor. This mechanism also underlies the change from meaning (“after a time interval”) to meaning (“the commencement of a state”). As is mentioned above, the elements preceding hòu with meaning are often temporal phrases referring to a time interval (such as bāshì (八世) in (3)). Time intervals of this kind usually possess in themselves an inherent process and clear boundaries, and the locative term hòu designates that a situation begins or is completed after the end of a time interval. However, a time interval can also be expressed as a situation with an inherent temporal process and boundaries (such as méng (盟) in (4)), following which hòu expresses that a situation begins or is completed after the end of the time interval denoted by another situation. In some contexts, however, the construction “motion verb + hòu + VP” can be construed as a posterior temporal order of the VP in relation to either the end or the beginning of a particular activity. For instance, yǔ hòu (雨后) in (18b) can be construed as “after the situation ‘rain’ begins” or “after the situation ‘rain ’ comes to an end”; while in (18c), the better interpretation is “after the situation rain begins” in contrast to qíng tiān (clearing-out); but then the yǔ hòu in (18a) can only be construed as “after the situation ‘rain’ comes to an end” with reference to the context (“the rain last night” in the first clause).

The Locative Term hòu ‘BACK’ in Chinese

239

(18) a. 煙深草濕昨夜雨，雨後秋風渡漕河。 全唐詩 yān shēn cǎo shī zuó yè yǔ, yǔ hòu qiū fēng dù cáo hé. mist thick grass wet last-night rain, rain hòu autumn-wind ferry canal ‘After the rain last night, the mist turned thick, the leaves of grass became moist, and the autumn wind passed the canal for transporting tribute rice.’ b. 池荷雨後依香起，庭草春深綬帶長。 全唐詩 chí hé yǔ hòu yī xiāng qǐ, tíng cǎo chūn shēn shòu dài cháng. pond lotus rain hòu sweet-smell up, garden grass spring deep ribbon long ‘The lotus flowers in the pond blossom in/after the rain, and the grass all over the garden flourishes in late spring.’ c. 雨後煙景綠，晴天散餘霞。 全唐詩 yǔ hòu yān jǐng lù, qíng tiān sàn yú xiá. rain hòu misty scene green, clearing-up dispel rest rosy-clouds ‘The scene turns into a beautiful misty green picture in the rain, and the rest of the rosy clouds are dispelled (by the sun) after clearing up.’ When the elements preceding hòu extended to stative verbs, the reading “after a time interval” was blocked in the context “stative verb + hòu + VP”; by contrast, the meaning of “stative commencement” was further strengthened and finally became the coded meaning of hòu (see (13)). The process of metonymy again took place when hòu with meaning “after a time interval” evolved into hòu with meaning “after the realization or completion of a situation”, both of which indicate the temporal order between the Figure and the Ground. In the former, a situation begins or finishes after the end of a time interval, while in the latter, a situation begins or finishes after another situation finishes. The difference rests in the fact that the Ground in the former is a time interval and in the latter it is a situation. Conceptually, the situation and its durative time resemble the relationship between the part and the whole of an object, and the contiguity in concepts motivates the metonymy of the durative time by its relevant situation. Therefore the change from meaning to meaning is actually the conceptualization of a situation into a time interval, namely, a metonymy from the time interval of a situation to the situation itself. The development of meaning (“hypothetic marker”) from meaning of hòu also involves metonymy. hòu with meaning usually appears in eventive clauses expressing a realistic situation with bounded verbs as its predicates. In such contexts, there are clear temporal boundaries in the clause containing hòu, which typically indicates that a situation begins after another situation is realized or has completed (see (6)). In certain contexts, however, the temporal relationship between the situation expressed by the temporal clause containing hòu and the situation expressed by the main clause may invite this pragmatic inference: the situation expressed by the temporal clause containing hòu acts as not only the temporal frame of reference but also as the precondition or prerequisite of the situation expressed by the main clause.

240

F. Wu

So in (6a), “the death of the emperor in the kingdom of Wu” is the temporal frame of reference and, at the same time, the precondition or prerequisite of the situation “the crown prince Fuchai came to the throne” in the main clause. Pragmatic inferencing is a type of pragmatic implicature depending upon certain contexts, and can be cancelled. However, if the sentence is a non-eventive clause expressing an irrealistic situation, the temporal boundaries of the clause containing hòu become vague and the temporal relationship between the two situations in the sentence is weakened. Especially in complex sentences containing a hypothetical conditional clause, the logical relation (condition-result) between the two situations often dominates over the temporal relation between the two situations. In such contexts, meaning (“after the realization or completion of a situation”) of hòu begins to recede, and the inferential meaning of “hypothesis or conditional” from the contiguous relationship of utterances becomes predominant (see (7), (8)). Lastly, if the clause containing hòu is a hypothetical or subjunctive conditional clause without preposing hypothetic connectives having a stative verb as its predicate, hòu can no longer be construed as a temporal connective; on the contrary, it can be analyzed only as a connective of hypothetical conditional clauses (as in (9)). Moreover, the hypothetical clause connective hòu, developing one step forward, can be used at the end of a clause with a preposing hypothetic connective, combining into a circum-hypothetic connective (see (10)). To sum up, pragmatic inferencing, facilitated by the contiguous relation between utterances, led to the evolution of hòu from a temporal clause connective meaning “after the realization or completion of a situation” to a hypothetical conditional clause marker, therefore the fundamental mechanism at work was metonymy. Metonymy also took part in the change from hòu from a hypothetical conditional clause marker to a topic marker. Hypothetical conditional clause markers and topic markers are two closely related concepts. From a crosslinguistic perspective, Haiman (1978) proved the close relation between the conditional and the topic. Primarily, conditional clause markers simultaneously act as topic markers in many diverse languages; secondly, they are highly parallel in semantic and discourse functions: “A conditional clause is (perhaps only hypothetically) a part of the knowledge shared by the speaker and his listener. As such, it constitutes the framework which has been selected for the following discourse (Haiman 1978: 583)”; on the other hand, “The topic represents an entity whose existence is agreed upon by the speaker and his audience. As such, it constitutes the framework which has been selected for the following discourse. (Haiman 1978: 585)”. Haiman (1978) even asserted that conditional clauses are identical to sentence topics in view of their parallel formal markings, semantic properties and discourse functions. Nevertheless, we think that the two belong to their different respective linguistic categories despite their conceptual relationship. The most apparent difference seems to be as follows: in a sentence with a conditional clause, there is a logical relationship of derivation between the protasis and the apodosis (if VP1, then VP2; such as (9)); whereas in sentences containing a topic, VP1 as the topic is explained or interpreted by VP2 (VP1 is the object under explanation or interpretation, and VP2 is the content of explanation or interpretation; such as (11)). It can be assumed that the listener probably reanalyzes hòu as the topic

The Locative Term hòu ‘BACK’ in Chinese

241

marker by abduction in some contexts like (9) in which the logical relation of condition-result recedes. The change of hòu from topic marker to intra-sentence pause marker also involved metonymy. An intra-sentence pause marker, or pausing particle, refers to a particle denoting the phonetic pauses in a sentence, which sometimes designates the following items as the focal information, but the elements preposing it are not the topic. The topic marker is in fact a kind of pausing marker, which always comes after the topic. In this sense, the relationship between the topic marker and the intrasentence pausing marker is part to whole, and the change from the former to the latter is a metonymical process. We have briefly examined the semantic changes of hòu and the mechanisms of change. In the process of semantic change, the formation of each meaning and use is motivated; the fundamental mechanism is metonymy. 4. Cross-Linguistic Survey of the Source Model and Semantic Change of the Locative Term hòu in Chinese In the above chapter, we discussed the origin and evolution of the locative term hòu from the perspective of semantics and morphosyntax. We now consider this question: What is the relationship between the source model and semantic change of the locative term hòu and the universal tendencies revealed by crosslinguistic examination of the BACK-REGION grams? In other words, do the source model and semantic changes of hòu reflect the characteristic and typological features of the Chinese language, or do they represent to some extent the universal commonalities in general language evolution? We will enter into details in this chapter. 4.1 The crosslinguistic source model of the BACK-REGION grams According to the crosslinguistic research by Svorou (1993), the BACK-REGION grams in human languages mainly developed out of body part terms (of human or animals), environmental landmarks and relational part terms as well as some abstract nouns, among which body part terms (BACK in particular) act as the most basic source model, as is shown in table 4.1: Table 4.1 The source models of BACK-REGION grams (After Svorou 1993: 248–258) Spatial grams BACK

Source model Body part terms (of human or animals) Environmental landmarks Relational part terms

Back, Anus, Loins, Buttocks, Bones Trace/ Footprint Tip/End

Sample Bari, English, Finnish, Persian

Abkhaz, Navajo, Mid.Welsh Estonian, Burmese

242

F. Wu

Moreover, Heine et al. (Heine 1989, Heine et al. 1991) have made quantitative analyses of the sources of locative terms in 125 African languages (see table 4.2 for the statistics of locative term “BACK”). Table 4.2: Source domains for African Locatives (after Heine, 1989, Heine et al. 1991: 128) (sample 125 African languages) Target BACK Source body parts 103 landmarks

0

relational concepts

1

other sources

2

no etymology available

15

no data available Total

13 134

In this quantitative research, the majority (103) of the 134 BACK-REGION grams under examination originated in human or animal body part terms, which conforms precisely with the conclusion of Svorou (1993). However, neither Svorou (1993) nor Heine et al (1991) has made clear mention of verbal sources for the locative term “BACK”. Then is it possible that verbal sources exist in other languages? According to the quantitative analyses of locative terms “BACK” in 104 Oceanic languages made by Bowden (1992), the locative terms in some Oceanic languages corresponding to the locative term hòu “BACK” in Chinese have identifiable verbal sources (see table 4.3). Table 4.3: Source domains for Oceanic Locatives (after Bowden 1992: 35) ( for 104 Oceanic languages) Target BACK Source body parts landmarks

60

63%

29

30%

7

7%

landmark/body part locative verbal source others no etymology Total

26 122

The Locative Term hòu ‘BACK’ in Chinese

243

Bowden (1992) found 122 locative terms with the meaning “BACK”, among which 7 terms have verbal sources accounting for 7% of the 96 terms with available etymology. What is more interesting, the 7 terms developed out of motion verbs meaning “to follow” with no exception (Bowden 1992: 38), closely resembling the source of the locative term hòu “BACK” in conception. It is thus evident that the source model and conceptual basis of the locative term hòu “BACK” in Chinese reflect typological commonalities instead of characteristics of one language, though this is not common in other languages, since we have found parallel source models in the Oceanic languages. 4.2 The semantic map of BACK-REGION grams In her crosslinguistic research Svorou (1993) observes that BACK-REGION grams in human languages generally possess different kinds of non-spatial uses and temporal uses apart from spatial uses. All the distinct uses involving different cognitive domains are related to the BACK-REGION grams in different ways, causing the latter to present various patterns of polysemy in different languages. Svorou (1993) summarizes the uses of the BACK-REGION grams under examination in the following semantic map, based upon crosslinguistic comparison and reconstruction of diachronic changes.

POSTRIOR STATE TEMPORAL COMMENCE POSTERIOR ORDER

AFTER

POSTERIOR

BENEFACTIVE UNDER 

   

HETEROLATERAL THROUGH CIRCUMFERENTIAL BACK TO EVERY

COMITATIVE BASE INSTRUMENT

Fig. 4.1 BACK-REGION semantic map (Svorou 1993: 171)

244

F. Wu

Svorou (1993) suggests that the various uses cluster in particular BACK-REGION grams in a core-peripheral manner, with the use POSTERIOR as their core and the other uses as peripheral uses of BACK-REGION grams deriving from the core. Furthermore, peripheral uses divide into several groups (the three groups in figure1) on the basis of conceptual similarities and contiguity, and uses in the same group have closer relationships in function and meaning than uses in different groups. It is worthwhile to notice that not all the uses in figure 1 will necessarily be included in every language, but all the uses of BACK-REGION grams from languages under investigation are believed to be embodied by the semantic map. The semantic map model lays down restrictions on particular linguistic elements in human languages by building up the conceptual space of those elements in order to generalize the universal of semantic organization, whereas the diversity of a language consists of the division of conceptual space. Now we turn to the relationship between the evolutional path of the locative term hòu in Chinese and the semantic map above. By comparing (15a) with figure 1, we find that the locative term hòu has the core use“POSTERIOR” and acquired the upper group of peripheral uses without developing the other two groups in figure1. On the other hand, several uses of the locative term hòu such as hypothesis marker, topic marker and pausing marker are not included in the map. How do we explain such a phenomenon? It is necessary to clarify that Svorou (1993) came to her conclusions from crosslinguistic comparison of the current uses of BACK-REGION grams, without considering uses that might have emerged in evolution but do not remain in the present day languages. In other words, Svorou’s (1993) crosslinguistic comparison is mainly synchronic. By contrast, our analyses and conclusions on the uses of the locative term hòu are mainly diachronic, which means that the different uses of hòu did not necessarily come into being at one particular stage in history. Actually, the hypothesis marker, topic marker and pausing marker uses have disappeared since the Ming, at least in the mainstream Chinese language. Therefore those uses would not take part in a synchronic semantic map even if Svorou had taken the Chinese language into account in her research, considering that the uses no longer exist in the modern Chinese language. In this light, the semantic evolution of the locative term hòu in Chinese does not contradict the crosslinguistic generalizations made by Svorou. In the preceding sections, we have looked into the source model and semantic changes of the locative term hòu from a crosslinguistic perspective. Although the source model of hòu (motional process) does not correspond to the major source models of BACK-REGION grams (body part terms, environmental landmarks and relational part terms) exposed by crosslinguistic comparison, there are still parallel models in other languages; in other words, the source model of hòu should not be regarded as characteristic of the Chinese language. In addition, the semantic evolution of the locative term hòu conforms precisely to the universal tendencies revealed by crosslinguistic research.

The Locative Term hòu ‘BACK’ in Chinese

245

5. Conclusion The paper focuses on the source model and semantic change of the locative term hòu in Chinese, especially the semantic and morphosyntactic changes that took place when hòu occupied the head position in genitive structures, based on historical literature. The chief conclusions are as follows: (i) The locative term hòu in Chinese originated in the concept of a motional process. Although this source model does not precisely correspond to the mainstream source models of BACK-REGION grams (body part terms, environmental landmarks and relational part terms) revealed by crosslinguistic comparison, it represents a typological feature in human languages rather than a language-specific characteristic. (ii) The pattern of semantic change including a series of meanings and uses in the diachronic development of the locative term hòu corresponds precisely with the universal tendencies of BACK-REGION grams revealed by crosslinguistic research. (iii) The formation of various meanings of hòu is motivated, and metonymy serves as the most fundamental mechanism in the semantic change.

Abbreviations CONJ PREP NEG

conjunction preposition negation

246

F. Wu

References Barcelona, Antonio. (2000). On the plausibility of claiming a metonymic motivation for conceptual metaphor. In Antonio Barcelona (Eds.), Metaphor and Metonymy at the crossroads: a cognitive perspective (pp. 31–58). Berlin: Mouton de Gruyter. Bowden, John. (1992). Behind the preposition: the grammaticalization of locatives in Oceanic languages. Canberra: Australian National Universtity. Brinton, Laurel. (1988). The development of English aspectual systems: Aspectualizers and postverbal particles. Cambridge: Cambridge University Press. Brinton, Laurel & Elizabeth Closs Traugott. (2005). Lexicalization and language change. Cambridge: Cambridge University Press. Brugman, Claudy M. (1983). The use of body-part terms as locatives in Chalcatongo Mixtec. Survey of Californian and Other Indian Languages, 4, 235–90. Brugman,Claudy M. & Monica Macaulay. (1986). Interacting semantic systems: Mixtec expressions of location. Berkeley Linguistics Society, 12, 315–27. Bybee, Joan, Revere Perkins, & William Pagliuca. (1994). The evolution of grammar: Tense, aspect and modality in the languages of the world. Chicago: University of Chicago Press. De Leon, Lourdes. December (1991). Body parts and location in Tzotzil: ongoing grammaticalization. Paper presented at the Workshop on Space in Mesoamerican languages at the Max Planck Cognitive Anthropology Research Group of the Max Planck Institute of Psycolinguistics: Nijmegen. Haiman, John. (1978). Conditionals are topics. Language, 54, 564–589. Heine, Bernd. (1989). Adpositions in African languages. Linguistique Africaine, 2, 77–127. Heine, Bernd. (1997). Cognitive foundations of grammar. New York, Oxford: Oxford University Press. Heine, Bernd, Ulrike Claudi, & Friederike Hünnemeyer. (1991). Grammaticalization: A conceptual Framework. Chicago: University of Chicago Press. Hollenbach, Barbara E. (1995). Semantic and Syntactic Extensions of Body Part terms in Mixtecan: The Case of ‘Face’ and ‘Foot’. International Journal of American Linguistics, 61(2), 168–190. Hopper,Paul J. & Elizabeth Closs Traugott. (2003). Grammaticalization (Second edition). Cambridge: Cambridge University Press. Jha, Subhadra. (1958). The formation of the Maithili language. London: Luzac & Company. Lillehaugen, Brook Danielle. (2003a). The Categorial Status of Body Part Prepositions in ValleyZapotec. MA Thesis: University of California, Los Angeles. http://www.linguistics.ucla.edu/people/grads/lillehaugen/ Lillenhaugen, Brook Danielle (2003b). The Syntactic and Semantic Development of Body Part Prepositions in Valley Zapotec Languages. University of California, Los Angeles. http:// www. linguistics.ucla.edu/people/grads/lillehaugen/ Lillenhaugen, Brook Danielle. (2004). Modified Body Parts are not Prepositions, University of California, Los Angeles. http://www.linguistics.ucla.edu/people/grads/lillehaugen. MacLaury, Robert E. (1989). Zapotec body-part locatives: Prototypes and metaphoric extensions. International Journal of American Linguistics, 55(2), 119–154. Matsumoto, Yo. (1999). On the extension of body-part nouns to object-part nouns and spatial adpositions. In Barbara A.Fox, Dan Jurafsky & Laura A. Michaelis (Eds.), Cognition and Function in Language. Stanford, California: CSLI Publications. Svorou, Soteria. (1986). On the evolution paths of locative constructions. Berkeley Linguistics Society, 12, 515–527. Svorou, Soteria. (1993). The grammar of space. Amsterdam: Benjamins.

The Locative Term hòu ‘BACK’ in Chinese

247

Svorou, Soteria. (2003). Semantic constraints in the grammaticalization of locative constructions. In Wischer & Diewald (Eds.), New reflections on Grammaticalization. Amsterdam: Benjamins. Traugott, Elizabeth C. (2006). Historical Pragmatics. In Laurence R. Horn and Gregory Ward (Eds.), Handbook of Pragmatics (pp. 538–56). Blackwell Publishing. 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. & Richard Dasher. (2002). Regularity in semantic change. Cambridge University Press. Traugott, Elizabeth C. & Ekkehard König. (1991). The semantics-pragmatics of grammaticalization revisited. In eds. Traugott Elizabeth C. & Bernd Heine (Eds.), Approaches to grammaticalization (1, pp. 189 –218). Amsterdam: Benjamins.

THE INESSIVE STRUCTURE IN ARCHAIC AND MEDIEVAL CHINESE: AN EVOLUTIONARY STUDY OF INESSIVE DEMONSTRATIVE USES FROM ARCHAIC TO EARLY MODERN CHINESE Qingzhi Zhu

Wenjie Chen

Peking University

Nanking University∗

Keywords: GDP, IDP, evolution, Medieval Chinese, Chinese translations of Buddhist scripture Abstract: Location is an important category of space. In Chinese, from the times of Oracle-bone inscriptions to now, one of the means used to indicate space or location is the demonstrative pronoun. But before the medieval period there only was one kind of demonstrative pronoun which can be both used to demonstrate person and location. During the medieval period, an evolution occurred in the Chinese demonstrative pronoun system. In the pre-modern period Chinese had developed a system of inessive demonstrative pronoun, such as zhèlǐ 这里 ‘here’, nàlǐ 那里 ‘there’, zhèbiān 这边 ‘this side’, nàbiān 那边 ‘that side’, which are special demonstrative pronouns used only to indicate space and location. This paper focuses on the evolution and tries to answer the following questions: how did the evolution occur? What is the reason of the evolution?

1. Introduction Location is an important category of space. In the Chinese language, the demonstrative pronoun is an important means by which one indicates the notion of space, and the indication of space as the principal application of the demonstrative pronoun is present as early as the oracle-bone inscriptions1 of Archaic Chinese-the earliest recording extant writing samples (Zhāng Yùjīn 2001). Past studies reveal that the Chinese inessive demonstrative pronoun (IDP) had undergone the following evolution from the archaic to early modern period2: In the archaic (before Eastern Hàn) and medieval (Eastern Hàn through the early Táng, with the Wèi-Jìn and Northern and Southern Dynasties as its core) periods, there was no specific IDP. The IDPs and the personal pronouns shared the same form, which we refer to as the general demonstrative pronoun (GDP).3 All these pronouns are monosyllabic, such as “cǐ 此”, “zī ”, “shì 是”, “zhī 之”, “bǐ 彼”, etc., (Lǐ Chóngxīng, 1992; Zhāng Yùjīn, 2001, 2004). In the early modern times, Chinese developed its own system of specified IDPs: zhèlǐ 这里 ‘here’, nàlǐ 那里 ‘there’, zhèbiān 这边 ‘this side’, nàbiān 那边 ‘that side’, etc. are the most common ones, and all are disyllabic (Ōta Tatsuo, 1957; Shimura Ryōji, 1982)4. Such a diachronic evolution of the IDPs reveals two critical points: the first, from the state of general to the state of specific; the second, from the monosyllabic to the disyllabic. The former is dependent on the latter, for in a syllabic language like Chinese, the specified IDP cannot evolve through a monosyllabic form. It is also noteworthy that the former point in fact begets the evolution of the latter, for in both Archaic and Medieval Chinese, the monosyllabic demonstrative pronouns, for the 249 D. Xu (ed.), Space in Languages of China: Cross-linguistic, Synchronic and Diachronic Perspectives, 249–266. © Springer Science +Business Media B.V. 2008

250

Q. Zhu and W. Chen

most part, have the characteristic of generality. From the perspective of typology, this does not accord with the universal law of language. As far as Chinese is concerned, the realistic solution is to attach onto the monosyllabic GDP a marker of locality, turning the original monosyllabic word into a disyllabic word. Such a shift also corresponds to the disyllabic preference prevalent in the evolution of Chinese vocabulary. When did these evolutions occur? What are their intermediate stages? And why did they occur in this fashion? These are some interesting questions that have attracted much attention in research. In the past, few scholars have paid much attention to the matter of IDPs. Lǐ Chóngxīng (1992) is the only paper of some significance. In that paper, there is a special section entitled “The Development of IDPs”, which gives a rough and conclusive description, in three phases, of the development of IDPs. He said: 1. There is no specific IDP in Archaic Chinese. 2. No later than the Northern and Southern Dynasties period had there existed such expressions as cǐ dì 此地 ‘this place’, cǐ chù 此处 ‘this location’, cǐ zhōng 此中 ‘this inside’ and cǐ jiān 此间 ‘this space’. (1)

(2)

(3)

(4)

此地 有崇山峻岭，茂林修竹。 东晋·王羲之 兰亭集序 cǐ dì yǒu chóng shān jùn lǐng, mào lín xiū zhú. (Lántíngjí xù by Wáng Xīzhī in East Jìn) this place have high mountain high ridge luxuriant forest long bamboo In this place there are high mountains and lofty hills, dense forests and tall bamboos.

国记 此处 有百枚小塔。 法显 cǐ chù yǒu bǎi méi xiǎo tǎ. (Fóguó jì by Fǎxiǎn in East Jìn) this place have hundred quantifier small pagoda In this place there are one hundred small pagodas. (Giles 1877: 38) 卿 此中何所有？ 答曰： 此中 空洞无物。 (刘义庆 世说新语·排调 ‘qīng cǐ zhōng hé suǒ yǒu?’ Dá yuē: ‘Cǐ zhōng kōng dòng wú wù.’ ( Shì shuō xīn yǔ by Liú Yìqìng in Sòng of North and South Dynasties) you this inside what part. have answer say this inside empty cavern no thing “What have you got in there?” [Chou] replied, “In here is an empty cavern with nothing in it.” (Mather 2002: 441) 日既向暮，此间大可畏，君作何计？ 刘义庆 幽明录 rì jì xiàng mù, cǐ jiān dà kě wèi, jūn zuò hé jì? (Yōu míng lù by Liú Yìqìng in Sòng of North and South Dynasties) Sun already towards sunset this space greatly need fear you do what plan The sun has set already. This space is really terrifying and what are you going to do?

The Inessive Structure in Archaic and Medieval Chinese

251

This should be considered as the beginning of the separation of the first IDPs from the GDPs. 3. Demonstrative pronouns zhè 这 ‘this’ and nà 那 ‘that’ first appeared in the texts of the Táng Dynasty, and were never used to solely indicate location. So we can imagine that before the appearance of the disyllabic forms “zhèlǐ 这里” and “nàlǐ 那里”, the task of indicating location fell to the demonstrative pronouns of the “cǐ 此” and “bǐ 彼” series. “Zhèlǐ 这里” “nàlǐ 那里” were first seen in the texts of the late Táng. Lǐ Chóngxīng also presents the following explanation for the origin of these expressions: The reason of the appearance of “cǐ chù 此处” and “bǐ chù 彼处” is to overcome the syntactic shortcomings of “cǐ 此” and “bǐ 彼”, …because when “cǐ 此” and “bǐ 彼” appear as the subject or object of a sentence, they generally can not be used to indicate location. It is quite clear that the author is trying to address the questions we proposed above. However, perhaps due to space limitations, his discussion seems overly simple. He did not illustrate the process by which he reaches his conclusions, nor does he provide us with an explanation of the cited language data. We are also not presented with precise definitions for these “expressions”, e.g., “cǐ dì 此地”, “bǐ suǒ 彼所” etc.，which he considered as the beginning of the separation of the IDPs from the GDPs.5 As such, we feel there is room to return to the question of IDP evolution for another, more in depth observation. 2. Language Data The expressions, such as “cǐ dì 此地” “cǐ chù 此处” “cǐ zhōng 此中” and “cǐ jiān 此间”, put forward by Lǐ Chóngxīng (1992) could be structurally described as “demonstrative pronoun(monosyllabic) inessive noun (including phrase)”. After the combination of the demonstrative pronoun and inessive noun, they semantically confine each other: The demonstrative pronoun makes the inessive noun possess a sense of demonstration, while the inessive noun confines the demonstrative function of the pronoun within the category of location, and also removes the pronoun’s abstract substitution function. As such, we refer to this kind of “expression” as the “inessive demonstrative structure” (IDS), or more briefly as the “inessive structure” (IS). In such a structure, one of the functions of the inessive noun is to confine the applicable extension of the GDP within the category of location. These nouns have the similar function as such morphemes as “-lǐ 里” and “-biān 边” in specified IDPs. In this paper, we will refer to them as “markers of locality”. Thanks to large-scale digital databases,6 we may investigate this issue broadly, through a search of demonstrative pronouns in texts prior to the early Tang period, and using such data to demonstrate the application of the IS in Archaic and Medieval Chinese. We make three analyses of note:

252

Q. Zhu and W. Chen

I. The IS is widely present in the language reflected in the archaic texts. For example: (5)

嗟我妇子，曰 改 ，入此室处。 诗·豳风·七 jiē wǒ fù zǐ, yuē wéi gǎi suì, rù cǐ shì chǔ. (Qīyuè in Shī jīng) sigh I wife child say be change year enter this room dwell “Come, wife and children, The change of the year is at hand. Come and live in this house.” (Waley 1996:121) cǐ shì: this room.

(6)

思 多士，生 此王国。 诗·大 ·文王 sī huáng duō shì, shēng cǐ wáng guó. (Wénwáng in Shī jīng) part. great many bachelor born this king state Mighty were the many knights That brought this kingdom to its birth. (Waley 1996:227) cǐ wáng guó: this kingdom.

(7)

韩厥登，举爵曰： 臣之不敢爱死， 两君之在 此堂也。 左传·成公三年 Hán Juē dēng, jǔ jué yuē:‘Chén zhī bū gǎn ài sǐ, wèi liǎng jūn zhī zài cǐ táng yě. (The third year of Chénggōng in Zuǒ zhuàn) Hán Juē ascend hold wine-vessel say subject part. not dare stint die for two king part. at this hall part. Hán Juē ascended the hall, held the wine vessel in hand and said: “For the sake of the two kings drinking hard and restoring good relation, your humble servant does not dare to mean die.” cǐ táng: this hall.

(8)

登 此昆吾之 ，绵绵生之瓜。 左传·哀公十七年 dēng cǐ Kūnwǔ zhī xū, miánmián shēng zhī guā. (The 17th year of igōng in Zuǒ zhuàn) ascend this Kūnwǔ part. site continuously grow part. melon Ascending this Kūnwǔ Mountain you will find many melons growing up continuously. cǐ Kūnwǔ zhī xū: this Kūnwǔ Mountain.

(9)

爽鸠氏始居此地，季荝因之。 左传·昭公二十年 xī Shuǎngjiūshì shǐ jū cǐ dì, Jìzé yīn zhī. (The 20th year of Zhāogōng in Zuǒ zhuàn)

7

The Inessive Structure in Archaic and Medieval Chinese

253

past Shuǎngjiū clan begin reside this place Jìzé follow him In the past Shuǎngjiū resided in this place first, and Jìzé carried on as before. cǐ dì: this place. (10) 寡人有此国也，而曰田成有之，何也？ 韩非子·外储说 上 guǎ rén yǒu cǐ guó yě, ěr yuē Tián Chéng yǒu zhī, hé yě? (Wài chǔ shuō yòu shàng in Hánfēizǐ) sovereign possess this state part. but say Tián Chéng possess it why part. I possess this state, but you say Tián Chéng possess it. Why? cǐ guó: this state. (11) 惟时俗兮疾正，弗可久兮 此方。 西汉王褒 九怀·思忠 wéi shí sú xī jí zhèng, fó kě jiǔ xī cǐ fāng. (Sī zhōng in Jiǔ huái by Wáng Bāo in Western Hàn) think present custom part. hate standard not can long part. this area I think that at present, the common sentiment hates the honest man, so I can’t stay here longer. cǐ fāng: this area. The above is the “cǐ 此” series. (12) 陟 彼高冈，我马玄黄。 诗·周南·卷耳 zhì bǐ gāo gāng, wǒ mǎ xuán huáng. (Juǎn ěr in Shī jīng) ascend that high ridge I horse black yellow I am climbing that high ridge, My horses are sick and spent. (Waley 1996: 7) bǐ gāo gāng: that high-ridge. (13) 泛 彼柏舟，在 彼中河。 诗·鄘风·柏舟 fàn bǐ bǎi zhōu, zài bǐ zhōng hé. (Bǎi zhōu in Shī jīng) float that cypress boat at that middle river Unsteady is that cypress boat In the middle of the river. (Waley 1996: 15) bǐ bǎi zhōu: that cypress-boat. bǐ zhōng hé: middle of that river. (14) 于以采藻，于 彼行潦。

诗·召南·采苹

yú yǐ cǎi zǎ o, yú bǐ xíng liáo. (Cǎi píng in Shī jīng) at where pick algae at that flowing seeper

254

Q. Zhu and W. Chen

Where are algae to be found? Along the margin of a pond. (Wāng and Rén 1994: 59) bǐ xíng liáo: that flowing seeper. The above is the “bǐ 彼” series. (15) 它山 之石，可以 错。 诗·小 ·鹤鸣 tā shān zhī shí, kě yǐ wéi cuò. (Hè míng in Shī jīng) other hill part. stone may use as grinding tools There are other hills whose stones Are good for grinding tools. (Waley 1996: 184) tā shān: other hills. (16) 徙于他邑， 从而授之。 周礼·地官·司徒 xí yú tā yì, zé cóng ér shòu zhī. (Sītú in Zhōu lǐ) move to other city then follow and give it If one’s people move to other cities, he should follow them and hand over immigration certification to local officials. tā yì: other cities/countries. (17) 兄弟 在 他邦，加一等。 仪礼·士丧礼 xiōngdì jiē zài tā bāng, jiā yī děng. (Shì sānglǐ in Yí lǐ ) brother all in other country add one rank If one’s brothers stayed in another country (among them someone was dead), the brothers who were in mourning for his death should add one rank of mourning apparel. tā bāng: other countries. (18)

居而不居，越而之他处，主死国亡。 西汉刘安 淮南子·天文 dāng jū ér bù jū, yuè ér zhī tā chù, zhǔ sǐ guó wáng. (Tiān wén in Huáinánzǐ by Liú n in Western Hàn) should stay but not stay exceed and go other place host die nation subjugated When one ought to stay but rather leaves for another locale, then the host will die and the nation will be subjugated. tā chù: other places.

(19) 侯邑之在其国者，毕从之 他所。 西汉贾谊 新书·淮难 hóu yì zhī zài qí guó zhě, bì cóng zhī tā suǒ. (Huái nàn in Xīn shū by Jiǎ Yí in Western Hàn) feod city part. in his state part. all follow go other location

The Inessive Structure in Archaic and Medieval Chinese

255

The nobles whose marquisates are in his state, all move with him to other locations. tā suǒ: other locations. The above is the “tā 他” series.8 Aside from the above, there is also a single occurrence of “shì jiān 是间”. (20) 晋人御师必于崤，崤有二陵焉。其南陵，夏后皋之墓也 其北陵，文 王之所避风雨也。必死 是间。 左传·僖公三十三年 Jìn rén yù shī bì yú Xiáo, Xiáo yǒu èr lín yān. Qí nán lín, Xià Hòugāo zhī mù yě, qí běi lín, Wénwáng zhī suǒ bì fēng yǔ yě. Bì sǐ shì jiān. (The 33rd year of Xǐgōng in Zuǒ zhuàn) Jìn people resist troops must at Xiáo Xiáo have two mounds part. its south mound Xià Hòugāo part. tomb its north mound Wén King part. part. avoid wind rain part. must die this space If Jìn people fight against the troops they must be at Xiáo. There are two mounds there. The south mound was Xià Hòugāo’s tomb, and the north mound is King Wén’s shelter from wind and rain. You must die at this place (between the two mounds). shì jiān: this place (between two mounds). II. Some of the inessive nouns (phrases) of the IS have a very specific reference, such as “shì 室” “wáng guó 王国” “Kūnwǔ zhī xū 昆吾之虚” etc, and are present in monosyllabic, disyllabic and polysyllabic forms. Others do not refer to anything specific, but rather have a relatively higher degree of generality, such as “dì 地” “fāng 方” “jiān 间” etc., 9 and tend to be monosyllabic. An examination of the data reveals the importance of distinguishing between these two IS manifestations, for they are, in fact, the product of different phases of the evolution of the IS. For the purposes of distinguishing the two, we call the former the “specific IS” and the later the “abstract IS”. We can see from the above text that all the examples Lǐ Chóngxīng puts forward are examples of the abstract IS. In Archaic Chinese, the specific IS is already a common occurrence, while the abstract IS occurs relatively rarely. The above mentioned “cǐ dì 此地” “cǐ fāng 此方” and “cǐ jiān 此间” examples are the only usages that were found in the database. Moving forward from Archaic to Medieval Chinese, the abstract IS became very popular in its usage, and can be said to be one of the distinct characteristics of the language of the period. Therefore, although the abstract IS had already emerged in the archaic texts, we may still reasonably regard it as one of the characteristics of Medieval Chinese.

256

Q. Zhu and W. Chen

Out of the data from Archaic and Medieval Chinese, 31 occurrences of abstract IS were found (see appendix below). If we categorize them according to the demonstrative pronoun, among the 31 IS there are 13 proximate pronouns, 7 obviative pronouns and 13 general pronouns. There are 7 distinct demonstrative pronouns, which are cǐ 此 ‘this’, shì 是 ‘this’, bǐ 彼 ‘that’, tā 他 ‘other’, bié 别 ‘different/other’, yú 余 ‘remaining/other’ and yì 异 ‘different/other’, and 7 inessive nouns, which are dì 地 ‘place’, fāng 方 ‘direction’, chù 处 ‘location’, suǒ 所 ‘place’, zhōng 中 ‘inside’, jiān 间 ‘space’ and biān 边 ‘side’. The 7 inessive nouns are also different from each other in character. According to their degree of abstraction, they can be classified into two groups: one group includes the general abstract inessive nouns “dì 地” “fāng 方” “chù 处” and “suǒ 所”, while the other group is comprised of the nouns of locality “zhōng 中” “jiān 间” and “biān 边”. The latter group has a higher degree of abstraction than the former, and as such is referred to as the “general abstract IS”, the latter being the “specific abstract IS”. Table 1. The overall table of the abstract IS in Medieval Chinese General abstract IS

nouns pronouns cǐ此 proximate shì是 bǐ彼 obviative qí其 tā他 exclusive bié别 yú余 yì异

dì地

fāng方

chù处

＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋

＋ ＋ ＋ ＋ ＋ ＋ ＋ －

＋ ＋ ＋ ＋ ＋ ＋ ＋ ＋

Specific abstract IS suǒ所 zhōng中 jiān间 ＋ － ＋ ＋ ＋ ＋ ＋ －

＋ ＋ ＋ ＋ － － － －

＋ ＋ ＋ ＋ － － － －

biān边 ＋ ＋ ＋ － ＋ － － －

An analysis of the above table yields three conclusions which differ significantly with the data quoted by Lǐ Chóngxīng: The pattern. The pattern of IS is much richer than that of Lǐ’s “cǐ dì 此地” “cǐ chù 此处” “cǐ zhōng 此中” and “cǐ jiān 此间”. It includes not only the proximate type, but also the obviative and the exclusive type; not only the “cǐ 此” series, but also the “shì 是” series, the “bǐ 彼” series and the “tā 他” series.

The Inessive Structure in Archaic and Medieval Chinese

257

The time. The IS existed commonly in Archaic and Medieval Chinese-it did not appear, as Lǐ suggests, in the Northern and Southern Dynasties. In Archaic Chinese the main use of the IS is the specific IS, and in the later period of Archaic Chinese there are several examples of the abstract IS. It is after the Eastern Hàn period-not the Northern and Southern Dynasties-that the IS began to appear in large numbers. The position. Although the abstract IS could appear at the subjective and attributive position, they most commonly occur in the objective position, including the verbal and prepositional object. Since all the IS were located in the subjective position in Lǐ’s examples, one would draw the conclusion that the cause of the emergence of these “expressions” was “to overcome the syntactic shortcomings of ‘cǐ 此’ and ‘bǐ 彼’”. All the above suggest that Lǐ’s conclusions are in fact in contrast with the broad-scale linguistic realities of the period. III. Investigation of these examples shows that the majority of the abstract IS appeared first in the Chinese translations of Buddhist scriptures: 5 examples out of the 7 “cǐ 此” series, 5 examples out of the 6 “shì 是” series，6 examples out of the 7 “bǐ 彼” series, 3 examples out of the 7 “qí 其” series, 1 example out of the 5 “tā 他” series, 2 examples out of the 4 “bié 别” series, as well as all of the 4 examples of the “yú 余” series and 2 examples out of the 3 “yì 异” series. This suggests that the abstract IS seems to have a kind of special relationship with the Medieval Chinese translations of Buddhist scripture. Table 2. The overall table of the first application of the abstract IS (1–4) I cǐ-series

Texts & times BT10 NT cǐ dì EH11 PQ cǐ fāng TK WH cǐ chù TK NS cǐ suǒ WJ — cǐ zhōng EH WJ cǐ jiān EH TK cǐ biān NS SS total 5 2

II Texts & shì-series times BT NT shì dì WJ NS shì fāng T — shì chù EH NS shì suǒ — — shì zhōng EH — shì jiān EH PQ shì biān EJ — total 5 1

III bǐ-series

Texts & times BT NT bǐ dì S WH bǐ fāng EJ NS bǐ chù EJ T bǐ suǒ EJ T bǐ zhōng EH NS bǐ jiān EH T bǐ biān EJ — total 6 1

IV Texts & qí-series times BT NT qí dì ?12 ? qí fāng ? ? qí chù ? ? qí suǒ ? ? qí zhōng ? ? qí jiān ? ? qí biān ? ? total ? ?

258

Q. Zhu and W. Chen

Table 3. The overall table of the first application of the abstract IS (5–8) V Texts & VI Texts & VII Texts & VIII tā-series times bié-series times yú-series times yì-series BT NT BT NT BT NT tā dì WJ EH bié dì — T yú dì EJ NS yì dì tā fāng EH EH bié fāng — T yú fāng EJ T yì fāng tā chù EJ WH bié chù WJ T yú chù EH NW yì chù tā suǒ EJ WH bié suǒ S T yú suǒ WJ T yì suǒ tā zhōng — — bié zhōng — — yú zhōng — — yì zhōng tā jiān — — bié jiān — — yú jiān — — yì jiān tā biān EJ — bié biān — — yú biān — — yì biān total 2 4 total 2 0 total 4 0 total

Texts & times BT NT — T EH — NW — — — — — — — — — 2 0

Our further discussions will be based on the above discoveries. 3. Discussions We have mentioned above that Lǐ Chóngxīng (1992) has already made some discussions about the abstract usage of the IS. What might have aroused Lǐ’s attention is Lǚ Shūxiāng’book (1942–44 published in the first half of the 1940s, which could be considered as the foundational work on the study of Modern Chinese grammar, covering most of the phenomena characteristic of the language. In this book, the author mentioned a kind of “demonstrative locative word”. He grouped these words into two systems: “zhèr 这儿” and “nàr 那儿” belong to the vernacular system, while “cǐ dì 此地” “cǐ zhōng 此中” “cǐ jiān 此间” and “bǐ zhōng 彼中” belong to the classical style system. From an investigations of the examples, we can see that Lǚ Shūxiāng equated the characteristics of the abstract IS with that of the specified IDPs. He thought the only distinction between the two was stylistic variation, but obviously this view is not in accord with the linguistic reality as reflected in the medieval texts. 13 First of all, let’s take a look at the time factor, which is the most distinct point of difference between the two. The IDPs did not appear until after the Táng Dynasty, but are present in Modern Chinese. The abstract IS which emerged as early as archaic times, was widely applied in medieval times, but disappeared gradually in modern usage. As such, the abstract IS could be viewed as a distinguishing feature of Medieval Chinese. So we can say that there is no complementary distribution relationship between the two at a synchronic level.

The Inessive Structure in Archaic and Medieval Chinese

259

Secondly, let’s talk about the factor of stylistic variation. In the language of Medieval Chinese texts, the abstract IS occurs most frequently in the Chinese translations of Buddhist scriptures. It is generally believed that the stylistic variation present in these ancient translations is result of a combination of both vernacular and classical styles. Compared with the other stylistic variation of the same time, it most closely resembles the spoken language of the time. Taking into consideration the clear discrepancy in the frequency of the IS in Chinese translations of Buddhist scriptures and other Medieval Chinese texts, as well as its parallel relationship with monosyllabic GDPs, it is more reasonable to treat them as vernacular elements rather than classical stylistic elements. Finally let’s have a look at the factor of meaning. Whereas a sole GDP cannot be used to indicate a specific location, the IS, on the other hand, can be used to realize this function. Also noteworthy is, as opposed to undemonstrative nouns, the IS brings an attached demonstrative sense to nouns. Such duplicated meanings have both similarities and differences with later appearing IDPs such as “zhèlǐ 这里” and “nàlǐ 那里”. Their similarity lies in the simultaneous presence of both demonstrative and locative sememes. Their difference is that the abstract IS still conveys some concrete meaning-that is, they have not completely weakened in a semantic sense. The whole structure does not possess as highly abstractive a substitution function as the specified demonstrative pronoun does. It must be made clear that the abstract IS and the specified IDPs are two kinds of language elements with different characteristics. The former one is situated at the lower phase of the grammaticalization process, while the latter one belongs to a higher phase. It seems that Lǐ Chóngxīng does not accept Lǚ Shūxiāng’s opinion, but is not quite sure about the characteristic of this linguistic fact himself. For he refers to the IS with, instead of a clear definition, the rather ambiguous term “expression” on the one hand, and attributes such phrases as “cǐ dì 此地” and “cǐ zhōng 此中” into the so called “cǐ 此” “bǐ 彼” series of demonstrative pronouns on the other hand. This betrays an obvious contradiction. It must be admitted that Lǐ’s attempt to regard the abstract IS as “the beginning of the derivation of the IDPs from the GDPs” does have its merits. He shows a keen insight in associating such a language phenomenon, for the first time in history, with the emergence of the IDPs of early Modern Chinese. But the investigation of the examples does not support his opinion as a whole. As we have stated above, the abstract IS first appeared at the later period of Archaic Chinese, and became a prevailing phenomenon in Medieval Chinese. Yet this kind of structure did not appear suddenly, but rather on the basis of the specific IS. The main differences between these two kinds of IS are the degree of abstraction and the amount of the syllables of the inessive nouns. However, these differences are not essential, or we may say they have not reached the extent of qualitative change. Therefore, in order to have a rather precise comprehension of the

260

Q. Zhu and W. Chen

characteristics of the abstract IS, we should put them into the framework of the “demonstrative pronouns + inessive nouns” structure, and discuss them with the specific IS together. Our opinion is that, all IS with the “demonstrative pronouns + inessive nouns” structure as their framework, rather than the mere abstract IS, is the means of indicating location with a demonstrative sense that generated in the archaic times and evolved notably in medieval times. Neither the specific nor the abstract ISs can be considered “words”. They belong to a class of linguistic units larger than words. As soon as the IS came into being, they became a part of the grammaticalization /lexicalization process that developed a set of specified IDPs for Chinese. The whole process can be further divided into three phases: the specific IS, the general abstract IS and the specific abstract IS. For this reason, we may say they are the primitive forms of IDPs in Chinese that bridge the evolution of the inessive demonstrative means from Archaic to early Modern Chinese. It is not difficult to testify the above statement. We can get started with three aspects as follows: the structural similarities between the IS and the IDP; the chronological sequence of their emergence; and the historical origins of the three kinds of markers of locality. Firstly, the structural similarities between the IS and the disyllabic IDPs: From the structural point of view, the IS, whether specific (“cǐ shì 此室” and “cǐ wángguó 此王国”) or abstract (“cǐ chù 此处” “cǐ suǒ 此所” “bǐ zhōng 彼中” and “bǐ biān 彼边”,) present themselves no differently from the disyllabic specific IDPs of Modern Chinese, such as “zhèlǐ 这里” “nàlǐ 那里” “zhèbiān 这边” and “nàbiān 那边” etc. They all share the same form pattern “GDP + marker of locality”. Surely this is not accidental or meaningless. It shows there must exist between the two some kind of relationship. The disyllabic specific IDP obviously evolved from the IS, while the “abstract IS” must have come from the “specific IS”. Of course, since the modern demonstrative pronouns “zhè 这” and “nà 那” do not share a common root with the archaic “cǐ 此” and “bǐ 彼”, such a relationship is reflected more in the form pattern and the markers of locality used. Secondly, the chronological sequence of emergence of the IS and the disyllabic IDPs: The investigation of the examples above clearly show that the IS had existed in the texts of the Spring and Autumn Period at the latest, and all of them expressed specific locations. At the later period of archaic times, there appear several cases of IS referring to abstract locations, such as “cǐ chù 此处” in text of the Warring States Period, “tā chù 他处” and “tā suǒ 他所” in the Western Hàn Period. Such a shift can be explained as the result of the abstraction of the inessive nouns of the IS-a natural evolution of the IS. As we move forward into the medieval period, the abstract IS emerges in vast numbers, and degree of generality continuously increases with time. Beside the general inessive nouns, nouns of locality, which are of a higher degree of generality, were also applied in the IS. As this tendency continued, the IDP emerged eventually in the modern times.

The Inessive Structure in Archaic and Medieval Chinese

261

We have such a chronological chain presented before us:

Specific IS Æ abstract IS Æ disyllabic IDP Archaic periodÆ Æ medieval period Æ modern period

In such an evolutionary process, the “specific IS” is at the starting point, the “disyllabic IDP” is at the terminal point, the “abstract IS” stands in between. Such a sequence is completely in line with the law of cognition: from particulars to generalities. The language of the medieval times first employed the specific, nongeneralized combination of “demonstrative pronoun + specified inessive noun”, and then later evolved the combination of “demonstrative pronoun + abstracted inessive noun”, which possess certain degree of generality and is more freely applicable. For the main difference among the three was their degree of generality. Following the direction of the arrow, the farther we progress rightward, the higher is the degree of generality. From a pragmatic point of view, this was also reflected in the frequency of the two combinations. In the archaic texts, there were only a limited number of cases of the latter, far less than the former; while in the medieval texts, although there still existed many cases of the former, the latter became prevalent. Thirdly, the relationship among the three kinds of markers of locality and the process of grammaticalization: In the “specific IS”, all the markers of locality are general inessive nouns with a lower degree of generality. In the IDPs, the markers of locality such as “-lǐ 里” and “-biān 边” etc. surely came from the nouns of locality “lǐ 里 or lǐtou 里头 ” and “biān 边 or biānshang 边上 ” ， but by now their lexical meanings have weakened thoroughly and have turned into highly abstractive grammatical elements that indicate the category of location. The markers of locality of the “abstract IS” such as “-dì 地” “-fāng 方” “-chù 处” “-suǒ 所” and “-zhōng 中” “-jiān 间” “-biān 边” etc., semantically and in terms of degree of grammaticalization, are situated in the middle of the above two. Compared with the markers of locality in IDPs, they still possess relatively obvious lexical meanings; while compared with the markers of locality of the “specific IS”, their lexical meanings have generalized considerably. From the specific inessive nouns to the abstract inessive nouns, and then from the nouns of locality to finally pure grammatical markers indicating location, all these changes reflect the grammaticalization process of such word-formation elements as “-lǐ 里” and “-biān 边” etc., in the demonstrative pronouns of Modern Chinese. This is a typical process of morphologization. As a rule, it underwent the three evolutionary phases as item (notional words), clitic morphemes and affix in turn. 14 We can make an observation of the “abstract IS” from another aspect. In Modern and Contemporary Chinese, even after the emergences of such specific IDPs as “zhèlǐ 这里” and “nàlǐ 那里” etc., the “specific IS”, as a category, still exists, but use of the “abstract IS” has decreased remarkably. Except for a few cases

262

Q. Zhu and W. Chen

in pre-medieval literary language (“qí zhōng 其中”, “cǐ jiān 此间” etc.), the demonstrative pronouns “zhè 这” and “nà 那”, when combined with the “abstract inessive markers” (the only exception being “-biān 边”,) such as “-dì 地” “-fāng 方” “ -chù 处” “ -suǒ 所” “-zhōng 中” and “-jiān 间” etc., would either be considered as untenable (as in the combination of “ *zhè jiān 这间” “ *nà jiān 那间” “*zhè suǒ 这所” and “ *nà suǒ 那所”), or have to narrow their extension greatly, no longer possessing the same degree of generality as in the medieval period. For example, “zhè dì 这地” would refer to “a plot of field” in a particular situation. As for the reason of such a result, it may well be attributed to the emergence of such demonstrative pronouns as “zhèbiān 这边” and “nàbiān 那边” etc., which take the place of the “abstract LS” that was situated between the most specific and the most general extremes. This proves again that before the emergence of the specified demonstrative pronouns, it is more or less the “abstract IS” that played the same role. If our knowledge of the development of the Chinese demonstrative pronouns was previously restricted to an acknowledgement of the evolution from general to specific, from one syllable to two syllables, we now have a much clearer picture of the process of that evolution. No later than the middle period of archaic times, GDPs started to get combined with inessive nouns, and thus emerged a new means of indicating location—the IS. The demonstrative pronouns and the inessive nouns restrict each other semantically after they are combined. The demonstrative pronoun attached a sense of demonstration, while the inessive noun confines the demonstrative function of the demonstrative pronoun within the category of location, thereby losing the abstracttive substitution function of a pronoun. At this stage of development, there is no special limitation on the characteristics of the inessive nouns and the amount of the syllable. At around the end of the archaic period, the abstract IS emerged. There also existed some limitations on the characteristics of the inessive nouns and the amount of syllables: the meaning of the word has a rather broad extension, as well as a relatively higher degree of generality and abstraction. All such nouns should possess only one syllable. This shift also facilitated the recovery of the substitution function of the demonstrative pronouns in the IS. In medieval times, the abstract IS were applied more frequently. The nouns of locality begin to appear in the IS,15 the inessive meaning of the IS getting further and further abstractive. Correspondingly, the substitution function of the demonstrative pronouns recovered a step further. The above evolution continued through the modern period, eventually leading to the generation of the specified IDPs. The structural framework of IDP is same as IS, but in the IDP, the meaning of locality is weakened totally, and the substitution function of pronouns is recovered thoroughly.

The Inessive Structure in Archaic and Medieval Chinese

263

Abbreviations BT EH EJ GDP IDP IDS IS NS NT NW PQ S SS T TK WH WJ

Buddhist translation texts East Hàn Dynasty (25AD–220AD) East Jìn Dynasty (317AD–420AD) general demonstrative pronoun inessive demonstrative pronoun inessive demonstrative structure inessive structure, the brief form of IDS North and South Dynasties (420AD–589AD) native texts North Wèi Dynasty (386AD–534AD) Pre-Qín period (before 221BCE) Suí Dynasty (589AD–618AD) South Sòng Dynasty (1127AD–1279AD) Táng Dynasty (618AD–907AD) Three Kingdoms period (220AD–280AD) West Hàn Dynasty (206BCE–24AD) West Jìn Dynasty (281AD–316AD)

Notes ∗

English translation assistance for this paper was provided by Mr. Lǐ Déchāo 李德超, Professor Yáng Yǒnglín 杨永林, Dr. Shuài Zhìsōng 帅志嵩, Professor Zhāng Hóngmíng 张洪明 and Mr. Eric Zine, and is greatly appreciated. 1 There are only two demonstrative pronouns present in oracle-bone inscriptions: “zī ” and “zhī 之”. They can be used to refer to either general matters or locations. For example (Zhāng Yùjīn, 2001): 庚午卜，內贞：王勿作邑才 在 茲，帝若 诺 ？ 甲骨文合集 14201 In the day of Gēngwǔ, there was a practice of augury. A man called Nèi asked: The King do not build a city at this place. Does the God allow him to do so? (Jiǎgǔwén héjí 14201) “ 亥卜，內贞：王侑石在鹿北东，作邑於之？” 甲骨文合集 13505 In the day of Yǐhài, there was a practice of augury. A man called Nèi asked: Does the King offer a sacrifice to the God in the northeastern Lù and build the city in that place? (Jiǎgǔwén héjí 13505) 2 From the late Táng through the Sòng and Yuán Dynasties. 3 The only exception is “yuán ”. It is particularly employed as a reference to location, but disregarding distance. As it most likely is rooted in regional dialect, we will not discuss it in this paper. 4 The words “zhèr 这儿” and “nàr 那儿” in the spoken Beijing dialect, according to Lǚ Shūxiāng (1942–44/1990:197), were originated from “zhèlǐ 这里” and “nàlǐ 那里”. 5 At first, the author referred to the words “cǐ chù 此处” and “bǐ chù 彼处” as “expressions”, and considered them as the mark of the beginning of the separation of the IDPs from the GDPs, though he mentions that it is difficult to determine their characteristics. Though he does later seem to attribute the cǐ此 and bǐ 彼 series into the class of demonstrative pronouns.

264

6

Q. Zhu and W. Chen

Guóxué Bǎodiǎn 国学宝典 (The Treasured Chinese Classical Works) (The special version of the 50th anniversary of the History Department of the Héběi University, 2003.5.), edited and issued by Yǐn Xiǎolín, including all of the archaic and the overwhelming majority of the medieval texts, and CBETA Chinese Electronic Tripitaka Collection 电子 典集成, version 2006, edited and issued by Chinese Buddhist Electronic Text Association, which includes all known Medieval Chinese Buddhist texts. 7 The other examples are: 吾先君成侯，受诏襄王，以 此地也。 战国策·魏策四 My deceased monarch Marquis Chén, received the order from King Xiāng to defend the territory of this place. (Wèi cè, in Zhànguó cè) 齐、鲁譬诸疾，疥癣也， 能涉江、淮而与我争此地哉？ 国语·吴语 Qí and Lǔ, which were compared to all diseases, were scabies, and couldn’t they wade through the Yangtze River and Huái River to scramble for this place? (Wú yǔ in Guó yǔ) 晋 公过九原而叹曰：“嗟呼！此地之蕴吾良臣多矣，若使死者起也，吾将谁与归乎？” 西汉刘向 新序·杂事 When Duke Jìnpíng went through Jiǔyuán, he acclaimed: ‘Alas! Many officials are buried under the earth. If someone could make up the dead, who can I go home with?’ (Zá shì in Xīn xù by Liú Xiàng in Western Hàn) 此地民有贤于不齐者五人，不齐事之， 教不齐所以治之术。 西汉刘向 说苑·政 理 There are five people wiser than Būqí in this area. Būqí served them, and they all taught Būqí skills of governance. (Zhèng lǐ in Shuō yuàn by Liú Xiàng) 8 In the Pre-Qín Period, the word “qí 其” is commonly employed as the third personal pronoun, as well as the obviative demonstrative pronoun. But when it is combined with the nouns of locality, it is often hard to tell whether it is a personal pronoun or a demonstrative pronoun. For example: 九四， 其墉，弗克攻， 。 周易·上经 Nine in the fourth place means: He climbs up on that/his city wall; he cannot attack. Good fortune. (Shàng jīng in Yì jīng) 久非其位，安得禽也。 周易·下经 A man, who persists in stalking game in that/his place where there is none may wait forever without finding any. (Xià jīng in Yì jīng) 其室 迩，其人甚远。 诗·郑风·东门之 His/that house is close by, and that person is too far away. (Dōngmén zhī shàn in Shī jīng) Cf. Although his house is so close by, his figure seldom meets my eye. (Wāng and Rén 1994: 355) 9 “qí其” has an obviously stronger capability than other pronouns in combining with the relatively generalized nouns of locality. But it is still hard to tell exactly whether it is a demonstrative pronoun or not. 10 BT: Buddhist translation texts; NT: native texts. 11 EH: East Hàn Dynasty (25AD–220AD); TK: Three Kingdoms period (220AD–280AD); WJ: West Jìn Dynasty (281AD–316AD); NS: North and South Dynasties (420AD–589AD); PQ: Pre-Qín period (before 221BCE); WH: West Hàn Dynasty (206BCE–24AD); SS: South Sòng Dynasty (1127AD–1279AD); T: Táng Dynasty (618AD–907AD); EJ: East Jìn Dynasty (317AD–420AD); S: Suí Dynasty (589AD–618AD); NW: North Wèi Dynasty (386AD– 534AD). 12 See endnotes 8 and 9.

The Inessive Structure in Archaic and Medieval Chinese

13

265

Professor Liú Dānqīng 刘丹青 once told me that in the Wú dialect, there is an IDP “cidì 此地”. It is a rather literary way of speaking. 14 Hopper and Traugott, 1993: 132. 15 In the archaic texts, there is only one case of nouns of locality, ie., “shì jiān 是间” in Zuǒ zhuàn 左传. Compared with the one in Dào xíng bō rě jīng 道行般若经 (translated by Lokasema, Eastern Hàn): “诸天人来到是间听受般若波罗蜜，作礼承事。” ‘The devas and human beings came to this place, listened for and received the Prajñāpāramitā, worshiped and served Buddha.’ The word jiān 间 in the phrase “shì jiān 是间” of Zuǒ zhuàn seems apparently more specific in semantic meaning, whereas the jiān 间 in shíjiān 时间 ‘time’ is an absolutely weakened example.

266

Q. Zhu and W. Chen

References Chén Wénjié 陈文杰. (1999). Cóng zǎoqī hànyì fódiǎn kàn zhōnggǔ biǎo fāng suǒ de zhǐshì dàicí 从早期汉译 典看中古表方所的指示代词 [A survey of instructive pronoun for location and direction in Chinese Buddhist translations of Medieval times]. Gǔ Hànyǔ Yánjiū 古汉语研究 4. Chángshā. Giles, Herbert Allen. (1877). Record of the buddhistic kingdoms (translated from Chinese). Shanghai: Kelly & Walsh. Lǐ Chóngxīng 李崇兴. (1992). Chùsuǒ cí fāzhǎn lìshǐ de chūbù kǎochá 处所词发展历史的初 考察 [Initiative Inspections on the Development of Nouns of Locality]. In Hú Zhú’ān (Ed.), Jìndài Hànyǔ Yánjiū近代汉语研究. Beijing: Shāngwù Yìnshūguǎn. Lǚ Shūxiāng 吕叔湘. (1942–44/1990). Zhōngguó wénfǎ yàolüè 中国文法要略 [A Summary of Chinese Grammar]. In Lǚ Shūxiāng Wénjí吕叔湘文集 I. Beijing: Shāngwù Yìnshūguǎn. Hopper, P., & Traugott, E. C. (1993). Grammaticalization. Cambridge: Cambridge University Press. Mather, Richard B. (2002). A New Account of Tales of the World (translated with introduction and notes by. Ann Arbor). Center for Chinese Studies, University of Michigan. Ōta Tatsuo 田辰 . (1958/2003). 中国語歷史文法[A Historical grammar of Modern Chinese]. 江南書院. (Chinese translation by Jiǎng & Xu). Beijing: Beijīng Daxué Chūbanshe. Shimura Ryōji 志村良治. (1982/1995). 中国中世語法史研究[A Study on Grammar of Medieval Chinese]. 三冬社. (Chinese translation by Jiāng & Bái). Beijing: Zhōnghuá Shūjú. Waley, Arthur. (1996). The Book of Songs. New York: Grove Press. Wāng Róngpéi 汪榕培主 & Rén Xiùhuà 任秀桦. (1994). Shijīng Yīng Yì 诗经英译 [The Book of Poetry]. Shěnyáng: Liáoníng Jiàoyù Chūbǎnshè. Zhāng Yùjīn 张玉金. (2001). Jiǎgǔwén Yǔfǎ Xué 甲骨文语法学 [Study on Grammar of the Oracle-Bone Inscriptions]. Beijng: Xuélin Chūbanshe. Zhāng Yùjīn 张玉金. (2004). Xīzhōu Hànyǔ Yǔfǎ Yánjiū 西周汉语语法研究 [Study on Grammar of the West Zhōu]. Beijing: Shāngwù Yìnshūguǎn.

Subject Index

close vs. further participant 126 clusters 3, 6 cognitive 8, 144, 178, 185, 187 cognitive domains 236, 243 combination 251, 259, 261, 262 conceptual space 244 conceptualization 239 conditional clause connective 236 contact 134, 138 contact with the action 122 continuum 202, 222 contrastive 202, 206, 211, 221-223 current location 132, 137 cyclic movement 34

abstract inessive markers 262 abstract inessive noun 261 abstract IS 255-262 abstract location 260 accusative 119, 139 addressee 129, 131, 132, 138 adjunct 200, 201, 209, 213, 215-217 adpositions 4-6, 22, 40, 41, 59-61, 204, 230 adverb 230, 237 affected 134, 140 affix 230, 261 analogy 201, 215 andative 71, 73 around 137 asymmetry 8, 176, 178, 181, 191, 192 autonomous motion 75, 83, 87

dative 134, 137, 139 dative constructions 123, 125, 128, 132 dative of place 138 dative-like nouns 138 definite locative 199, 202, 208, 209, 223 definite markers 222 degree of generality 255, 260 deictic directionals 71 deictic path 74 demonstrative and locative sememes 259 demonstrative locative word 258 demonstrative pronouns 249, 251, 256, 259, 260, 262 dependent-marking 41, 52, 61 derived spatial nominals 214 derived spatial NPs 206, 208, 219, 223, 224 diachronic evolution 249 direct object 120, 123 direct vs. indirect 139 direct vs. indirect object 123 directional 208, 210, 211, 214, 222, 223 directional complements 72 distinct demonstrative pronouns 256 disyllabic 249, 250, 255

back 121, 140 BACK-REGION 237, 238, 241, 243, 244 Basic Meaning model 33 benefactive 138 binominal 214, 219, 221 body part terms 229, 244, 245 bounding 93 case affix 41 case markers 41, 62 cases 39, 40 cause of motion 77 caused motion 75, 86 central participant 128 central vs. peripheral participant 126, 131 Chinese translations of Buddhist scriptures 257, 259 choice of postposition 134 circumpositions 46, 47, 55, 199, 200, 202, 203 clitic morphemes 261 clitics 199, 200 267

268

Subject Index

disyllabic IDP 260 disyllabic localizers 16, 25, 27, 31 disyllabic specific IDP 260 double object 123, 127, 131 durative marker zhe 136 enclitics 6, 201, 205, 206, 211, 221, 223 environmental landmarks 229, 230, 242, 245 evolution 249, 251, 260, 262 evolution of Chinese vocabulary 250 expansion 218 experiencer 129 Figure 5, 6, 155, 177, 231, 237, 238 frames of referenes (FoR) 99, 115 front 121, 140 functional head 205, 221 further 132 GDP 251, 259, 262 general abstract inessive noun 256 general abstract IS 256, 260 general demonstrative pronouns 249 general location 123 general locative 199, 201 general point of reference 134 general pronoun 256 generic preposition 201, 205, 208, 213, 214 genetic 3 genitive 203, 204, 211, 213, 219 genitive construction 230 goal-oriented 72 goal 73, 78, 134, 138, 179 grammaticalization 4 ,6, 8, 25, 28, 31,32, 43, 46, 54, 63, 64,158, 181, 199, 201, 205, 211, 213, 217, 220, 221, 229, 259-261 Ground 70, 77, 155, 231, 237, 238, 239 head-marking 41, 52, 60, 62 horizontal time 189, 191, 192 host-class 201, 218 hypothetical conditional connective 233 hypothetical marker 235 iconicity 144 inchoative 178 indirect beneficiary 130

indirect object 123, 124, 128, 138 inessive demonstrative pronoun (IDP) 249, 251, 259, 261 inessive demonstrative structure (IDS) 251 inessive meaning 262 inessive nouns 251, 255, 260 inessive structure (IS) 251, 255, 256, 262 inside 103 intended recipient 126 intra-sentence pausing marker 234, 241 intra-word component 235 intra-word morpheme 236 left 111 lexicalization 260 lexicalization of two postpositions 136 lexicalized word 185 listing 201, 206, 208, 211, 221-223 literary language 262 localizers 16, 17, 21, 24, 25, 27, 28, 31, 32 location 121, 126, 128, 131, 132, 136, 250, 251, 254 location marker 135 locative cases 137 locative construction 199-201, 204-206, 208-211, 213, 215-218, 220, 222, 223 locative marker 119 locative noun 231, 235, 236 locative particles 199, 200, 202 locative prepositions 17, 22, 136 locative term 229-231, 235, 237, 241, 243-245 locative words 99 main clause 232, 240 manner of motion 77 marker of locality 250, 260 metaphor 236-238 metonymy 236-239, 241, 245 modern Chinese grammar 258 monosyllabic demonstrative pronoun 249 monosyllabic GDP 250, 259 monosyllabic localizers 16, 23, 25, 27, 33 morphological cases 41, 42, 47, 61, 65 morphologization 261 morphology 3, 6 motion verbs 5, 8, 185, 192 motional situations 231, 238 moving ego 187, 191 moving time 187, 188

Subject Index multi-syllabic constraint 199, 200, 202, 206, 208, 210, 213-215, 217-220, 222, 223 near/around 136 nondeictic path 77 non-discrete categories 199, 223 noun 230, 232, 241 object markers 131, 134, 137 object of preposition 131 object-locative affinity 137 obviative pronoun 256 on 137 on vs. around 121, 122, 135, 139 oracle-bone inscriptions 249 oral languages 143 outside 106, 108 part-whole relation 212, 221 path satellites 72 paths 137 pathways 139 pausing marker 234, 235, 241, 244 personal pronouns 249 place names 16, 24, 202, 205, 206, 208, 213, 214, 218, 219, 223, 224 place words 16 polysyllabic 255 postpositions 6,7, 40, 41, 43, 46-51, 55, 59, 61-63, 120, 139, 199, 200, 202204, 211, 215, 223, 232, 234, 236 pragmatic 200, 202, 211, 224 pragmatic inference 237, 240 precise position 25, 28 prepositions 40, 41, 43, 44, 46, 51-56, 59-63 Principle of Economy 61, 63, 64 process verb 236 pronominal objects 131 pronoun 20 prosodic 52, 63, 64 prototype models 28, 33 prototypical 199, 206, 223, 224 proximate pronoun 256 recipient 123, 126, 127, 131 recipient vs. source 126, 127, 131, 137 relational part terms 229, 242, 245

269

relator principle 39, 51, 52, 61, 63, 64 renewal 199, 202, 211, 218 resultative 6 resultative verb compounds 178 right 111, 112 satellite-framed 42, 70 selectional restriction 200, 201, 205, 206, 208, 211, 221-223 semantic change 32 semantic map 243, 244 semantic-pragmatic 218 semantic-pragmatic expansion 221 serial verb constructions 4, 8, 176, 177, 192 side 109 sign languages 5, 6, 143, 189 single object 128, 131 Sinitic languages 15 S-languages 5, 70, 176, 178 source 123, 126 source model 229, 230, 241, 244, 245 source-oriented 72 space 1, 7 spatial concept 231 spatial grams 230, 242 spatial nominals 201, 205, 208, 211, 213-215, 217-223 spatial NPs 42, 53-58, 60-63, 200, 201, 206, 208, 213, 218, 222 spatial preposition 201, 218 spatial references 205, 208 specific abstract IS 260 specific IDP 249, 251, 261 specific inessive noun 261 specific IS 255, 256, 259-261 specific location 259, 260 specific locative 199, 222 specified IDP 260, 262 stylistic variation 258 subjectification 229 subjectivity 81, 90 subordinate clause 232 subordinate clause connective 232 surface 121 surface contact 122 SVO 55, 59, 61 synchronic 258 syntactic expansion 220

270

Subject Index

temporal clause connective 232, 236, 240 theme 129, 130 tone systems 3 topic marker 234, 235, 241, 244 typology 6, 250 ultimate recipient 125 underspecified preposition 201, 206, 210, 222 vague position 25, 28 venitive 71 verb 231, 234, 239, 243

verb-final 49, 59, 61, 63 verb-framed 42, 69, 176 vernacular 259 vertical time 189, 191, 192 vicinity 128 visual space 144 V-languages 5, 70 word order 41, 44, 47, 58, 59, 61, 64 word order harmony 64 word order typology 55, 59, 61, 64 written style 82

Language Index

Finnish 229, 241 French 89, 91, 143

Abkhaz 229, 241 Al-Sayyid Bedouin Sign Language 151 Altaic 3 American Sign Language (ASL) 145, 146, 147 Amharic 40, 199, 200, 202-204 Archaic Chinese 15, 17, 43, 47, 249, 250, 252, 255, 256, 259

Gan 2, 65 German 40, 71, 74 Guaymi 229 Hakka 2, 31-33, 65 Hani 138 Hong Kong Sign Language (HKSL) 155 Huaxian dialect 29 Hui 2, 65 Hungarian 70, 75

Bai 3, 4 Bari 241 Basque 229 Beijing Mandarin 42, 45, 48-52, 55, 58, 60, 62, 63 Bihari 237 Breton 2 British Sign Language (BSL) 144, 152 Burmese 241

IE 138 Igbo 229 Indo-European 2 Italian Sign Language (ISL) 155

Cantonese 7, 43, 45, 54, 55, 56, 58, 59, 61, 63, 78 Changsha dialect 62 Chinese 229, 230, 237, 238, 241, 244, 245, 249, 262 Chinese dialects 39, 42, 47, 52, 59, 61-63 Chinese Sign Language (CSL) 1, 145 Classic Chinese 62 Contemporary Chinese 261

Jakaltek 71 Japanese 41, 85, 229 Jianchuan dialect 119, 136 Jianghuai Mandarin 62 Jiangjin dialect 29 Jin 1 Jinhua dialect 60 Jinuo 138 Jizhou 4

Dai 138 Dali dialect 119, 136 Danish Sign Language (DSL) 147

Karen, 3 Korean 74, 85, 216

early modern Chinese 259 English 6, 35, 78, 91, 187, 205, 210, 213, 222, 241 Estonian 229, 241 Ewe 40, 229

Lahu 138, 139 Latin 138 LEME dialect 119, 136 Lolo-Burmese 119

271

272

Language Index

Manchu 3 Mandarin 1, 43, 44, 45, 50, 52, 53, 58, 60, 61, 132 Medieval Chinese 15, 23, 25, 29, 32, 249, 251, 255, 256, 258 Miao-Yao 3 Middle Chinese 179 Middle Welsh 241 Min 2 Modern Chinese 28, 260, 261 Mongolian 3 Nanchang dialects 52 Nanjing dialects 52 Navajo 241 Naxi 138 Ningbo dialects 59, 61, 63 Northern Chinese 79

Shaoxing dialect 45, 51, 59, 61, 63 Sino-Tibetan 2, 3, 8, 119, 138 Spanish 143, 152 Standard Mandarin 42, 45, 48, 58, 60, 65, 78 Suzhou dialects 45, 50, 51 Tai 3 Taiwan Sign Language (TSL) 152 Taixing dialect Thai 3 Tibetan 138, 139 Tibeto-Burman 3, 40, 119, 138 Tigre 229 Tiriyó 205 Uyghur 40 Vietnamese 3, 229

Old English 138 Pashto 40, 138 Persian 241 Pinghua 2 Polish 71 Russian 41, 71 Shanghai dialect 30 Shanghainese 42, 43, 47, 48, 50, 53, 56, 60

Waxiang 4, 99, 102, 105, 108, 109, 111, 114, 115, 188 Wu 2, 42, 44, 48, 50-56, 58-61, 63 Xiang 2, 65 Yue 2 Yukatec 202, 205 Zhuoka 138

Author Index

Armstrong, D. 5, 146, 151, 158 Asifa, Majid 99

Fan, Jiyan 53 Fang, Jingmin 181, 201 Fauconnier, G. 144 Feng, Shengli 10, 52, 63 Fischer, S. 154 Frishberg, N. 145, 149 Fu, Jingqi 3

Baker, C. 157 Barcelona, A. 236 Baxter, W. H. 2 Bellugi, U. 148, 149 Berthele, R. 5, 70 Bo, Wen 3 Bowden, J. 229, 243 Boyes-Braem, P. 146, 150 Brinton, L.J. 201, 218, 236 Brugman, C.M. 229 Bybee, J. 236

Gough, B. 145 Greenberg, J.H. 34, 41, 152, 199, 202, 203 Grosjean, F. 150 Hagège, C. 17 Haudricourt 3 He, Leshi 18 Heine, B. 9, 30, 32, 41, 229, 230, 242 Hewes, G. 143 Hill, D. 79 Himmelmann, N. 9, 201, 217, 220, 222, 224 Hollenbach, B.E. 229 Hong, Bo 109 Hopper, P.J. 201, 210, 220, 223, 224, 236 Huang, Borong 30, 35 Huang, J. 200 Humbolt, W. von 161 Hünnemeyer, 30

Cartier, H. 17 Chang, Jung-hsing 152 Chao, Y.R. 16, 17, 65, 93 Chappell, H. 5, 30 - 32, 35, 199, 213 Chen, Wenjie 7 Cheng, Xianqing 213 Chu, Zexiang 17,18, 25, 34, 48, 210 Clark, H.H. 175 Claudi, U. 30, 32 Cook, K.W. 191 Coulter, G.R. 157 Craig, C. 71

Isenhath, J.O. 158 Dasher 236 De Leon, L. 229 DeLancey, S. 3, 138 Dik , S.C. 51, 61, 63 Dryer, M.S. 3, 47, 62

Janzen, T. 158 Jha, S. 237 Jiang, Lansheng 109 Jin, Peng 138

Emmorey, K. 155 Engberg-Pedersen, E. 147 Ernst, T. 199 Fan Xiao 50

Karlgren 2 Kegl, J. 158, 161 Klein, W. 28 Klima, E. 148, 149

273

274 Koënig, 236 Kyle, J.G. 144, 152 Lamarre, C. 5, 7, 31, 42, 52, 59, 176 Lao, She 182 Leeson, L. 158 Legge, J. 34 Lehmann, W.P. 144 Levinson, S.C. 188, 202, 205, 222 Li Charles 199 Li, Chongxing 17, 19, 21, 22, 25, 214, 249, 251, 255, 256, 258, 259 Li, Fanggui 2 Li, Jingxi 17 Liddell, S.K. 146 Lillehaugen, B.D. 229 Linden, E. 144 Litip Tohti 40 Liu, Danqing 3, 5, 7, 44, 46, 49, 52, 58, 59, 62, 65, 78, 199, 203 Liu, Feng-Hsi 199, 200, 203 Liu, Shiru 17 Loncke, F. 151 Lü, Shuxiang 17, 22, 25, 109, 258 Ma, Jianzhong 17 Macauley, M. 229 MacLaury, R.E. 229 Masone, M. I. 152 Mather 250 Matisoff, J. 138 Matsumoto, Y. 86, 229 McCullough, K.E. 99 McGregor, W. 30 McIntire, M. 156 Mei, Zulin 136 Moody, B. 159 Nakanishi, 151 Nichols, J. 41, 62 Norman, Jerry 3 Nüse, R. 28 Ogawa, K. 109 Paris , M.-C. 92 Parker, 144 Pasicki, 119, 134

Author Index Peyraube 4, 5, 16, 17, 18, 28, 33, 34, 43, 47, 76, 99, 199, 213 Poizner, H. 150 Poulin, C. 147 Pulleyblank 2 Ren, Xiuhua 253 Ren, Ying 65 Rijkhoff 51, 61 Rygaloff, A. 17, 92 Saeed, J. 158 Sagart, L. 2 Shen, Jiaxuan 185 Shi, Youwei 44 Shibatani, M. 86 Shimura, R. 249 Skalmowski, W. 139 Slobin, D.I. 5, 74, 144, 176 Song, Jae-Jung 42 Stokoe, W. 149 Sun, Chaofen 4, 8, 17, 34, 35, 43, 47 Svorou, S. 9, 139, 186, 229, 230, 237, 242, 243, 244 Talmy, L. 5, 71, 165, 175 Tang, G. 155 Tang, Zhenzhu 30 Thompson, H. 157 Thompson, S.A. 35, 199 Traugott, E.C. 191, 201, 210, 218, 220, 223, 224, 236 Tesnière 5 Tylor, E. 150 Valade, Y.-L. R. 151 Wälchli, A. 5 Waley 252, 253 Wang, Canlong 181 Wang, Ying 27, 28 Wen, Lian 17 Wilbur, R.B. 151, 158 Wilcox, S.E. 146 Wilkins, D. 79 Wittgenstein, L. 143 Woll, B. 144, 152 Wu, Fuxiang 4, 9, 199, 203, 211 Wu, Yunji 4, 8, 62, 109, 188

Author Index Xu, Baohua 30 Xu, Dan 3, 8, 62, 65, 76, 176, 179 Xu, Lin 3,119, 136 Yang, Bojun 18 Yau, Shun-chiu 4, 8, 144, 148 Zhang, Cheng 47, 219 Zhang, Wanqi 18 Zhang, Yujin 249 Zhang, Zhende 24

Zhao, Cheng 18 Zhao, Jie 10 Zhao, Yansun 119, 136 Zhou, Ren 6 Zhu, Dexi 17, 34 Zhu, Qingshi 4, 9

275